You are on page 1of 96

CASE STUDIES

FOR

ELECTROMAGNETIC TRANSIENTS

Third Edition

Hermann W. Dommel

August 2006
TABLE OF CONTENTS

1. LINEAR AC STEADY-STATE CASES

1.1 COUPLING BETWEEN POWER LINES

1.2 COUPLING FROM A POWER LINE TO A FENCE

2. STEADY-STATE THEVENIN EQUIVALENT CIRCUITS

2.1 INTRODUCTION

2.2 CALCULATION OF THEVENIN EQUIVALENT CIRCUIT

2.3 THEVENIN EQUIVALENT CIRCUIT FOR ADJACENT TRANSMISSION


LINES

2.4 THEVENIN EQUIVALENT CIRCUIT FOR 1100 KV TEST LINE OF


BONNEVILLE POWER ADMINISTRATION

3. AC STEADY STATE WITH HARMONIC DISTORTION

3.1 INTRODUCTION

3.2 HARMONICS FROM TRANSFORMER SATURATION

3.3 HARMONICS FROM SINGLE-PHASE DIODE BRIDGE RECTIFIERS

4. TRANSFORMER INRUSH CURRENTS

4.1 INTRODUCTION

4.2 COMPARISON BETWEEN FIELD TEST AND SIMULATION

ii
5. FERRORESONANCE

5.1 INTRODUCTION

5.2 FERRORESONANCE ON 1100 KV TEST LINE

5.3 FERRORESONANCE BETWEEN ENERGIZED AND DE-ENERGIZED


TRANSMISSION LINES

6. SUBSYNCHRONOUS RESONANCE

6.1 INTRODUCTION

6.2 FIRST BENCHMARK MODEL FOR SUBSYNCHRONOUS RESONANCE

6.3 SECOND BENCHMARK MODEL FOR SUBSYNCHRONOUS


RESONANCE

7. SATURATION IN CURRENT TRANSFORMERS

7.1 INTRODUCTION

7.2 COMPARISON WITH FIELD TEST

8. SWITCHING SURGES ON OVERHEAD TRANSMISSION LINES

8.1 INTRODUCTION

8.2 REDUCTION OF SWITCHING SURGE OVERVOLTAGES

8.3 COMPARISON OF SIMULATION RESULTS BY CIGRÉ

iii
The following additions are planned:

♦ Section 8: More cases on switching surges (until finished, see 2nd edition of Case Studies).

♦ Section 9: Fault overvoltages (until finished, see 2nd edition of Case Studies).

♦ Section 10: Transient recovery voltage (until finished, see 2nd edition of Case Studies).

Possible additions from the collection of Microtran data files:

♦ Lightning surges.

♦ Power electronics cases (my own and from [11]).

♦ Start-up of induction motors and synchronous motors.

♦ Back-to-back capacitor switching (case from Bartholomew).

If you have material that you would like to contribute to this document, please contact Hermann
Dommel at : hermannd@ece.ubc.ca

iv
FOREWORD TO THE THIRD EDITION

This third edition is a major revision, with contributions from Dr. Kwok-Wai Louie and
Mr. Jiang Ping. The earlier second edition was compiled with the help of Dr. Andrew Yan, Raiza
J. Ortiz de Marcano, and Adrubal Boscan Miliani. I wish to express my thanks to all of them.

The author also wishes to thank the following persons for their assistance in producing
this booklet: Mr. P. Williams, Dr. B. Garrett, the late Mr. R. M. Hasibar, Mr. A. C. Legate and
Mr. C. W. Taylor, for providing some of the test cases. Mr. Stewart, Dr. Luis Marti, Mr. B. C.
Chiu provided simulation results and figures for some of the test cases.

I also appreciate the help that I have been receiving from Professor Washington Neves of
the Federal University of Campina Grande, Paraiba, Brazil, and from his graduate students,
during my visits in Campina Grande in 2002 and 2004.

Much has changed since the first edition was published in May 1983. Most of all,
mainframe computers have given way to personal computers, which has increased the number of
potential users of EMTP-type programs tremendously. There are also many more versions of
EMTP-type programs available now. For those seeking information about the simulation of
electric power system transients, many references are available now, including a number of
books, as shown at the beginning in the Reference Section.

The cases described in this booklet were analyzed with MicroTran® , the EMTP version
from The University of British Columbia, Vancouver, Canada. Other EMTP versions may work
for many of the cases as well. The MicroTran input files that are needed for running the cases are
included with this documentation. They are listed at the beginning of each section.

Hermann W. Dommel, August 2006

v
1. LINEAR AC STEADY-STATE CASES

The Electromagnetic Transients Program MicroTran (MT), and most other EMTP
versions, have an option for the solution of linear ac steady-state cases, either at one specified
frequency, or over a frequency range from fmin to fmax (frequency scan option), or at certain
specified frequencies (harmonics option). Steady-state solutions were first implemented to obtain
initial conditions for the simulation of transients, but they have become very useful by
themselves. In MT, use Tstart < 0 on sinusoidal sources to request steady state, and tmax = 0 if
there is no transient simulation to follow the steady-state solution.

Suggestion No. 1:

It is often wise to ask for a steady-state solution by itself first, before embarking on a
simulation of transients, because it is much easier to check whether the steady-state results look
reasonable for the particular case. Finding errors created by wrong data or by wrong network
connections is much easier with steady-state results than with waveforms of transient results.

Suggestion No. 2:

Transmission lines are represented as π-circuits with admittance values in steady-state


solutions, even if the input data is in the form of distributed parameters. If a transmission line is

very short, then the series admittance matrix of its π-circuit, e.g., [Yseries ] =
1
l
[
Z per unit length ]
−1
for the

nominal π-circuit, becomes very large compared with its shunt admittance matrix, e.g.,

[Yshunt ] = l
2
[ ]
jω C per unit length . This shunt admittance matrix could get “swamped out” by the large

values of series admittances. Capacitive coupling effects between phases and capacitance-to-
ground effects would then get lost. In the case discussed next in Section 1.1, the 4th π-circuit
comes close to this problem, with Y11 shunt = 7.2 ⋅ 10 −6 S , compared to Y11 series ≈ 1 S . On a 32-bit

single-precision computer, the results differed by 10 % from the more accurate double precision
results. If only capacitive effects are of interest, it is best to leave the series impedances out, with
1-1
a method discussed in the MT Reference Manual in the section on π-circuits. That method was
used by the author in a study about the safety issues of storing large metal drainage pipes under a
500 kV line, where the pipes were 20 feet long.

1.1 COUPLING BETWEEN POWER LINES

Input files: LINECOUP.DAT, LINECOUP.PUN, EMTPCOUP.DAT, COUPSCAN.DAT

1.1.1 INTRODUCTION

Electrostatic and magnetic coupling effects from energized overhead transmission lines to
nearby objects or to other adjacent power or communication lines are of great importance for
safety issues. In this case study, the electrostatic coupling (and to a lesser extent the magnetic
coupling) among three three-phase ac transmission lines are investigated. The results show that
the induced voltages due to electrostatic effects from the energized lines on a nearby de-
energized line can be significant.

To study such coupling phenomena can be especially important in cases where a new
transmission line is being built adjacent to existing energized lines. According to references [1]
and [2], problems associated with the construction of a new transmission line near and parallel
with energized lines involve possible electrical shock hazards when laying the foundations,
building the towers and stringing the conductors. Shock currents can be either steady-state or
transient in nature [1]. If the coupling effect is predominantly electrostatic, the simulation of
transient shock currents can sometimes be done by obtaining a Thevenin equivalent circuit (VThev
behind C), and representing the nearby object as a resistive or resistive-inductive circuit.

Other objects such as vehicles, telephone lines, fences, and buildings, can also be the
source of electrical shock when they are near energized power transmission lines [1]. For
instance, vehicles near energized high and extra-high voltage transmission lines can cause two
possible safety problems: electrical shock and ignition of fuel vapors [1]. Also, voltages will be
induced in telephone lines running near and in parallel with an existing energized transmission
line.

1-2
1.1.2 POWER SYSTEM UNDER CONSIDERATION

The case described here was studied in co-operation with B.C. Hydro & Power Authority
when a new 550 kV transmission line was being built adjacent to two existing lines. One of the
existing lines was a 550 kV line, and the other was a 360 kV line. Mr. P. Williams measured the
induced voltages and grounding currents on the line under construction as part of a safety
training program, and the computer simulations were made by B. Garrett.

1.1.2.1 Power System Configuration

Figure 1.1 shows the tower configuration of the three lines on the same right of way, and
Figure 1.2 gives the transposition scheme of the lines over the length of exposure to the line
under construction.
L3 L2 L1
550 kV under 550 kV in 360 kV in
construction operation operation

135’ 126.5’

Figure 1.1 Tower configuration of adjacent transmission lines

1.1.2.2 Network Data

The major input data consists of the representation of the three coupled transmission
lines, as discussed in Section 1.1.2.3. In addition, voltage sources connected to the energized
lines are needed for the simulation of the electrostatic effects. For the simulation of the magnetic
effects data is needed for the loading of the transmission lines.

1-3
(1) Voltage sources: For line L1, Vsending end = 360 kV (RMS, line-to-line), f = 60 Hz; for line L2,
Vsending end = 550 kV (RMS, line-to-line), f = 60 Hz. Line L3 was under construction, and no
voltage source was therefore connected to it.

(2) Loading: The loading on line L1 is represented by a current source of 0.430 kA (RMS) at the
receiving end, and on line L2 by a current source of 0.860 kA (RMS) at the receiving end.

59 miles phase no.

3 1 13 20 22
L1
360 kV
C 9
B 8

A 7
L2
550 kV
A 6
B 5

C 4
L3
550 kV
A 3
B 2

C 1
6 5 4 3 2 1

Figure 1.2 Transposition scheme of adjacent transmission lines

1.1.2.3 Line Models

(1) Nominal π-circuit

Since each transposition section of the three coupled lines is short, a “nominal π-circuit”
is the proper line model for each section. For lines of moderate “electrical” length (typically less
than 100 km at 60 Hz, which is definitely true here), nominal π-circuits are accurate enough. At
this time, they are probably the best models to use for steady-state solutions at power frequency.
The series impedance matrix of the nominal π-circuit is obtained by simply multiplying the series
impedance matrix per unit length with the length, and the shunt capacitance matrix at both ends

1-4
is obtained by multiplying the shunt capacitance matrix per unit length with half the length
(Section 4.2.1.1 of [3]).

(2) Exact equivalent π-circuit

If the frequency for the steady-state solution were much higher than 60 Hz, then the lines
might have to be represented accurately with exact equivalent π-circuits, but not all EMTP
versions have exact equivalent π-circuit models. For the exact equivalent π-circuit, the series
sinh(γl )
impedance of a single-conductor line would be Z no min al , where Z no min al is the series
γl
impedance of the nominal π-circuit. For the M-phase case, one has to transform the equations to
modal parameters first, because the hyperbolic sine function cannot be evaluated for matrices,
and then transform the equations back to phase quantities (Section 4.1.5 in [3]).

(3) Distributed-parameter line models

For steady-state solutions, distributed-parameter line models are converted internally into
π-circuits, because travelling-wave solution methods do not work in the frequency domain.
Normally, the user does not have to be concerned about this conversion, but must be aware of the
accuracy problems associated with distributed-parameter models. The major accuracy problem in
this case study is related to the untransposed nature of the three circuits. For strongly unbalanced
lines (lines where the self impedances of the phases and the mutual impedances between the
phases are not equal or close to equal among themselves), distributed parameter models should
not be used for steady-state solutions. To show this, let us take the first 22-mile untransposed
section (from right to left in Figure 1.2), and model it as:

a. Nominal π-circuit.

b. Constant parameter line model (distributed L' and C'; R lumped in 3 places;
values calculated at 60 Hz; complex and frequency dependent transformation
matrix approximated as a real matrix).

c. Lossless high-frequency approximation of distributed-parameter line model, with


distributed L' and C' and with a transformation matrix that is real and constant
(Section 4.1.5.2 in [3]).

1-5
Results for capacitive coupling, with de-energized line open at both ends, and voltages
calculated at the receiving end:

Line model phase A phase B phase C

Π-circuit 28.77 kV 14.61 kV 8.34 kV

CP line 43.16 kV 2.70 kV 19.36 kV

lossless line 28.75 kV 14.59 kV 8.32 kV

The lossless high-frequency approximation has no errors in the transformation matrix, and the
capacitances remain therefore unchanged in the transformation between phase and mode
quantities. Its results are therefore very close to the nominal π-circuit results. The constant
parameter line model works with an approximate transformation matrix at 60 Hz, after
discarding the imaginary part, which is too inaccurate in this case.

Results for inductive coupling, with de-energized line grounded at both ends through
resistances of 0.1 Ω. The inductive coupling comes from a load current of 430 A on the 360 kV
line, and of 860 A on the 550 kV line. Currents through the grounding resistances at the
receiving end:

Line model phase A phase B phase C

Π-circuit 70.61 A 22.48 A 8.92 A

CP line 101.48 A 18.66 A 25.96 A

lossless line 64.49 A 19.46 A 8.99 A

lossy line*) 71.14 A 22.64 A 9.01 A

_______________________________________

*)with [R] + j[ωL


internal] lumped at receiving end

1-6
As can be seen, the errors from ignoring the resistances in the lossless high-frequency
approximation are less severe than the errors from discarding the imaginary part of the
transformation matrix. This is more true at 60 Hz, and less true at high frequencies. Lumping [R]
+ j[ωLinternal] at the receiving end makes the lossless line more accurate. F. Castellanos, under
the direction of J. R. Marti, developed a frequency-dependent line model for transient studies, by
lumping [R(ω)] + j[ωLinternal(ω)] in a number of places along the lossless line, and taking the
frequency dependence of these lumped impedances into account [5]. [R(ω)]+j[ωLinternal(ω)]
represents the resistances and internal inductances of the conductors and of earth return. Based
on this approach, Ting-Chung Yu made further improvements that work for underground cables
as well [6].

(4) Computation of the nominal π-circuits

To create the nominal π-circuits, a support program “MTLINE” is used, that takes the
tower configuration (Table 1.1) and the conductor characteristics (Table 1.2) as input. At the
time of writing, no graphical user interface for MTLINE had been released yet. The input file is
LINECOUP.DAT, as listed as Table 1.3. The full meaning of the numbers can only be
understood by consulting the MTLINE Reference Manual. To represent the three lines in
parallel, a cascade connection of five 9-phase nominal π-circuits is used, one for each
untransposed section. The [Z] and jω[C] matrices of these nominal π-circuits are obtained from
the per unit length matrices by multiplying them with the respective section length. The pair of
node names of each phase can be seen in Table 1.3. The first number in the node name (1…6)
indicates the node location from the right to the left, the next two characters (L1, L2, L3)
designate the line, and the last letter indicates the phase (A, B, C). The type of transposition is
achieved by simply using the appropriate node names: the first π-circuit has phase sequence
C,B,A on line L1, whereas the second π-circuit has this changed to B, A, C, and the 3rd π-circuit
to A, C, B etc. MTLINE produces an output file LINECOUP.PUN that contains the five 9-phase
π-circuits for direct pasting into the MT input file “EMTPCOUP.DAT”. The file extension
“.PUN” goes back to the days when this output file consisted of punched cards.

1-7
Table 1.1 Tower configuration

Circuit Height at tower Height at mid-span Separation between


(feet) (feet) phases (feet)

L1 73 55 35

L2 76 33 39

L3 77 33 35

Table 1.2 Conductor characteristics

Circuit L1 L2 L3

Number of conductors in 2 4 4
bundle

Bundle spacing (inches)


α = 0o α = 45o α = 45o
18 18
18

Conductor outside diameter 1.108 0.95 1.0


(inches)

Resistance at 60 Hz (Ω/mile) 0.1288 0.160 0.151

Geometric mean radius (inches) 0.4476 0.3876 0.4260

Earth resistivity (Ω-m) 100

1-8
1.1.2.4 Computation of Coupling Effects

The input data for the computation of the coupling effects with the program “MT” is
shown in Table 1.4, for the case where the currents through grounding resistances of 5 Ω were
computed. This input file was created with a DOS input processor MTDATA. Windows-based
graphical user interfaces for setting up the case are available as well.

Table 1.3 Input data file for MTLINE

Input file for mtLine to create nominal pi-circuits for file "EMTPCOUP.DAT".
BRITISH
1 .151 2 .426 1. -170. 77. 33. 4 18. 45.
2 .151 2 .426 1. -135. 77. 33. 4 18. 45.
3 .151 2 .426 1. -100. 77. 33. 4 18. 45.
4 .16 2 .3876 .95 -39. 76. 33. 4 18. 45.
5 .16 2 .3876 .95 0. 76. 33. 4 18. 45.
6 .16 2 .3876 .95 39. 76. 33. 4 18. 45.
7 .1288 2 .4476 1.108 91.5 73. 55. 2 18. 0.
8 .1288 2 .4476 1.108 126.5 73. 55. 2 18. 0.
9 .1288 2 .4476 1.108 161.5 73. 55. 2 18. 0.

100. 60. 1 1
C The next line defines the 22-mile section (negative sign means nominal π).
100. 60. -22. 4
C The next 9 lines define the 9 pairs of node names for the 22-mile section.
1L3C 2L3C
1L3B 2L3B
1L3A 2L3A
1L2C 2L2C
1L2B 2L2B
1L2A 2L2A
1L1C 2L1C
1L1B 2L1B
1L1A 2L1A
100. 60. -20. 4
2L3C 3L3C
2L3B 3L3B
2L3A 3L3A
2L2C 3L2C
2L2B 3L2B
2L2A 3L2A
2L1B 3L1B
2L1A 3L1A
2L1C 3L1C
.
.
.
100. 60. -3. 4
5L3C 6L3C
5L3B 6L3B
5L3A 6L3A
5L2C 6L2C
5L2B 6L2B
5L2A 6L2A
5L1A 6L1A
5L1B 6L1B
5L1C 6L1C

1-9
To get the electrostatic coupling effects, voltage sources of 550 kV RMS line-to-line (449
kV peak phase-to-ground) and 360 kV RMS line-to-line (294 kV peak phase-to-ground) are
connected to lines L1 and L2 at node 1. The line loading, with values of current thought to be
typical, is represented by current sources of 0.860 kA RMS (1.216 kA peak) and 0.430 kA RMS
(0.608 kA peak) at the far end of lines L2 and L1 at node 6.

Table 1.4 Input data file for MT

* File "EMTPCOUP.DAT".
C THIS IS A STEADY-STATE SOLUTION AND THERE ARE NO PLOTS!!!!!!
C TEST CASE FOR STEADY-STATE COUPLING EFFECTS AMONG ADJACENT POWER LINES.
C FROM H.W. DOMMEL ET AL, CASE STUDIES FOR ELECTROMAGNETIC TRANSIENTS, P.1,1991
ELECTROSTATIC COUPLING BETWEEN LINES 60. 60.
0.
1L3A 5. 1
1L3B 5. 1
1L3C 5. 1
(The line parameters are the same as in file LINECOUP.PUN produced by MTLINE. Either its
filename must be listed here with a certain format, or the data must be pasted in here)
$ = = End of level 1: Linear and nonlinear elements = = = = = = = = = = = =
$ = = = End of level 2: Switches and piecewise linear elements = = = = = = = =
14 1L2A 449. 60. 0. -1.
14 1L2B 449. 60. -120. -1.
14 1L2C 449. 60. 120. -1.
14 1L1A 294. 60. 0. -1.
14 1L1B 294. 60. -120. -1.
14 1L1C 294. 60. 120. -1.
14 6L2A-1 1.216 60. 0. -1.
14 6L2B-1 1.216 60. -120. -1.
14 6L2C-1 1.216 60. 120. -1.
14 6L1A-1 0.608 60. 0. -1.
14 6L1B-1 0.608 60. -120. -1.
14 6L1C-1 0.608 60. 120. -1.
$ = = = End of level 3: Sources = = = = = = = = = = = = = = = = = = = = = = =
1L3A 1L3B 1L3C
$ = = = End of level 4: User-defined voltage output = = = = = = = = = = = = =
$ = = = Level 5: End of data case = = = = = = = = = = = = = = = = = = = = = =

In the case study, the following two cases are simulated:

(1). Induced voltages on line L3 if both ends are open.

(2). Grounding currents if the right side of line L3 is grounded through a grounding
resistance of 5 Ω.

1 - 10
Simulation Results

The simulation results are summarized in Table 1.5. It compares calculated and measured
values. The values were calculated with voltages of 360.0 and 550.0 kV, respectively. When the
measurements became available, with voltages of 372.0 and 535.0 kV, respectively, the values
were recalculated again (last column in Table 1.6). Notice that the recalculated values do not
come closer to the field measurements. One source of errors may have been a phase difference
between the voltages on the 360.0 and 550.0 kV lines. This was not measured and assumed to be
zero in the calculation. The coupling in this case is mostly electrostatic. The results are shown in
Table 1.6, where the values are compared with the case where the line loading currents of lines
L1 and L2 are set to zero.

Table 1.5 Comparison between simulation results and measurements (with 360 and 550 kV)

Calculated results Measured results


A 28.200 kV 30.0 kV
Induced voltages B 14.196 kV 15.0 kV
on open line L3 C 8.053 kV 10.0 kV

Grounding currents A 10.784 A 11.0 A


for grounding B 3.256 A 5.0 A
through 5Ω in node 1 C 1.536 A 1.0 A

Table 1.6 Influence of line loading on results (with 372 and 535 kV)

Line loading of 430.0 A on


L1 and 860.0A on L2 Line loading currents zero

A 27.46 kV 26.96 kV
Induced voltages B 13.83 kV 13.41 kV
on open line L3 C 7.850 kV 7.475 kV

1 - 11
1.2 COUPLING FROM A POWER LINE TO A FENCE

Input files: FENCELIN.DAT, FENCELIN.PUN, FENCEMT1.DAT, …FENCEMT3.DAT

1.2.1 INTRODUCTION

Reference [3] discusses a case on page 4-26 to 4-28 where a power line runs in parallel
with a 2 km long fence, as shown in Figure 1.3. Reference [3] asks the reader to calculate the
capacitive coupling into the open and ungrounded fence from the capacitance matrix, and then to
calculate the circulating current in the fence, when grounded at both ends, from the series
impedance matrix. Both effects can be found with the same nominal π-circuit, as explained next.

Figure 1.3 Fence in parallel with three-phase power line

1.2.2 SOLUTION WITH NOMINAL Π-CIRCUIT

First we find the four-phase nominal π-circuit from the tower configuration and
conductor characteristics of Fig. 1.3, with FENCELIN.DAT as the input file to MTLINE, and
FENCELIN.PUN as the output file for the π-circuit, to be copied into the MT input file
FENCEMT1.DAT. Then we find the capacitive coupling effects with voltage sources of V = 69
kV (RMS, line-to-line) applied to the power line, with the fence open-ended at both ends. By
simply short-circuiting both ends of the fence through a very small resistance (to obtain the
current), we obtain the circulating current in the fence for the inductive coupling effects if a short
circuit current of 10 kA (RMS) is drawn from phase A of the line. The answers agree with the
1 - 12
values in [3]. For capacitive coupling, we obtain a voltage of 793.2 V on the fence (in [3] it is
793.2*345/69 = 3966 V because of an unrealistic operating voltage of 345 kV there). The
circulating current in the fence is 1.526 kA, if grounded at both ends.

Input files FENCEMT2.DAT and FENCEMT3.DAT can be used to simulate the transient
shock currents if the ungrounded fence is touched, with the person touching the fence
represented as a resistance of 1000 Ω. Note the high current of 0.7 A at the beginning, which
settles down to the 4.17 mA (RMS) of the steady-state solution. This steady-state value can also
be obtained with a steady-state solution by itself, with Tclose = -1.0 for the switch specification.
Note that the π-circuit model may not be good enough for the high frequencies involved in the
pulse. The input file FENCEMT3.DAT has a simplified fence-coupling model, in which only the
capacitance matrix is used.

Figure 1.4 Shock current when fence is touched through 1000 Ω

1 - 13
2. STEADY-STATE THEVENIN EQUIVALENT CIRCUITS

2.1 INTRODUCTION

For the analysis of large electric power systems, it is often necessary to simplify the
network with an equivalent circuit of reduced size. The simplest equivalent circuit is the
Thevenin equivalent circuit. It is particularly suitable for steady-state studies, because in that
case the Thevenin equivalent circuit is an exact representation of the larger network, provided
the system is linear. For the simulation of transients, equivalent circuits are more difficult to
obtain, because its impedances becomes frequency-dependent (see, for example, reference [7]).
In this section, Thevenin equivalent circuits are found for two cases.

Thevenin’s Theorem is also known as the Theorem of Helmholtz, who proposed it in


Germany in 1853 before Thévenin proposed it in France in 1883 [8]. It states that the current
through an impedance Zl connected between any two nodes in a network is given by V/(Z + Zl),
where V is the voltage across these terminals when Zl is removed and Z is the impedance of the
network seen from the two terminals. If the behaviour of the large network is only of interest as
it affects an “external” network connected to these terminal nodes, then any network having two
terminals can be replaced by a voltage source in series with an impedance, between these
terminals. The value of the voltage source is the voltage between the terminals when the external
circuit is disconnected (“open-circuit voltage”). The value of the impedance is the internal
impedance of the large network between the terminals, when all its internal sources are replaced
by their internal impedances (“short-circuit impedance”). As will be seen later, we will extend
this concept to N pairs of terminals between the external and internal networks.

If many steady-state solutions are to be found for a large network in which only a few
parameters are changed and the rest of the network remains unchanged, then it may be best to
reduce the unchanged part of the network first to a Thevenin equivalent circuit. Parameter
variations can then be studied much faster with this equivalent circuit if the number of steady-
state solutions with parameter variations is large. For an external network with N branches, it
takes N+1 solutions with the large external network to find the Thevenin equivalent circuit.

2-1
Therefore, one must be looking for more than N+1 steady-state solutions before savings can be
realized. Reduction to Thevenin equivalent circuits may be less important nowadays with fast
computers and large memories.

Thevenin equivalent circuits can also be useful in some transient studies. Since they are
only exact at a particular steady-state frequency at which they are calculated, some judgement is
required to assess the validity of the circuits for transients. In particular, one has to ask whether
the circuit is inductive or capacitive, and how valid the equivalent circuit would be over the
frequency range of interest.

In the derivation of the Thevenin equivalent circuits here, the following assumptions are
used:

1. All voltage and current sources have the same frequency.

2. The network is linear.

3. The N external branches which are to be modeled in detail, or whose parameters are to be
varied, have no inductive coupling to the internal network which is to be reduced to a
Thevenin equivalent circuit. However, inductive or capacitive coupling within these N
branches is permitted.

2.2 CALCULATION OF THEVENIN EQUIVALENT CIRCUIT

To obtain the Thevenin equivalent circuit for a power system, seen from N pairs
of terminals a, b, c, …N, the following procedure can be followed:

1. Remove the N branches a, b, c, …N from the original network, and model the unchanged
(external) part in detail with all its branches and voltage and current sources (Figure 2.1).
Find the open-circuit voltages across the node pairs 1a-2a, 1b-2b, … of the removed branches
with one steady-state solution. If the branches go from node to ground, then we simply ask
for node voltage output. If the branches go from node to node, then we have to ask for branch
voltage output. In some EMTP versions it may not be possible to ask for node-to-node

2-2
voltages directly; in that case, insert resistance branches across the node pairs with very high
resistance values, e.g. 10 20 Ω. This steady-state solution produces

1N
1c
1b
1a
UNCHANGED
PART OF THE
NETWORK TO
BE REDUCED TO a b c ...... N
A THEVENIN
EQUIVALENT
CIRCUIT
2a
2b
2c
2N

Figure 2.1 Thevenin equivalent circuit seen from branches a, b, …, N

 V1a − 2 a 
V 
[VThev ] =  1b − 2b  (2.1)
 M 
 
V1N − 2 N 

2. Find the impedance matrix

 Z aa Z ab ⋅ ⋅ ⋅ Z aN 
Z Z bb ⋅ ⋅ ⋅ Z bN 
[ZThev ] =  ba . (2.2)
 ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ 
 
 Z Nz Z Nb ⋅ ⋅ ⋅ Z NN 

of the Thevenin equivalent circuit column by column with N steady-state solutions as


follows:

(a) Short-circuit all voltage sources (the easiest way to do this is to set the amplitudes to
zero; don’t just remove the voltage sources because that would create an open circuit),

2-3
and cancel all current sources (simply set their amplitudes to zero), in the unchanged part
of the network.

(b) Connect a current source of +1.0 A RMS (or +1.4142136 peak on source input) to
terminal 1i, and of –1.0 A RMS (or –1.4142136 peak on source input) to terminal 2i,
and find the branch voltages Va(i), Vb(i), . . . VN(i). If either node 1i or node 2i is
grounded, then the current source into that ground node can be omitted. Make N such
steady-state solutions for

i = a, b, c,⋅ ⋅ ⋅N (2.3)

Then the i-th column of [ZThev] is simply the RMS-voltage solution as follows:

 Z ai  Va (i ) 
 Z   (i ) 
 bi  = Vb  , i = a, b, c,⋅ ⋅ ⋅N (2.4)
 M   M 
   (i ) 
 Z Ni  VN 

After knowing the open-circuit voltage [VThev] and the impedance matrix [ZThev], the
unchanged part of the network is described by the N-branch Thevenin equivalent circuit of
Figure 2.2, with

[V ] = [VThev ] − [Z Thev ][I ] (2.5)

where [V] and [I] are the voltages across and the currents through the N branches. If the
equations for these branches are

[V ] = [Z BRANCH ][I ] (2.6)

then

[I ] = [ [Z Thev ] + [Z BRANCH ] ]−1 [VThev ] (2.7)

2-4
N
.
[ZThev] .
.
a

[VThev]

Figure 2.2 N-branch Thevenin equivalent circuit

2.3 THEVENIN EQUIVALENT CIRCUIT FOR ADJACENT

TRANSMISSION LINES

Input file: THEV_BCH.DAT

Let us take the case of the three adjacent transmission lines described in Section 1.1, with
Figures 1.1 and 1.2, and reduce the cascade connection of the five 9-phase π-circuits (Fig. 2.3) to
a Thevenin equivalent circuit as seen from nodes 1L3. To find the Thevenin equivalent circuit,
we use the same data as described in Table 1.4, except that the three 5 Ω resistances from nodes
1L3A, 1L3B and 1L3C to ground are omitted.

As described in the preceding Section 2.2, we first find a steady-state solution after that
modification, to obtain the open-circuit voltages. We then set the voltage and current magnitudes
to zero, add currrent sources of 1.0 A RMS into one phase of nodes 1L3, one at a time, and
repeat the solution. The voltages in 1L3 will then give us the impedances.

2-5
360 kV Phase No

A 9
B 8 L1
C 7
550 kV
A 6
B 5 L2
C 4
6 5 4 3 2 1
A 3
B 2 L3
C 1

Figure 2.3 The complete system before reduction


(line L1 transposition not shown here; see Fig. 1.2)

2.3.1 SINGLE-PHASE THEVENIN EQUIVALENT CIRCUIT

The Thevenin equivalent circuit for phase A at L3 (node 1L3A) is given in Figure 2.4.
The open-circuit voltage between 1L3A and ground is simply the node voltage at 1L3A, which is
then the source voltage of the Thevenin equivalent circuit,

VThev = 28.20∠ + 128.9o kV (RMS)

With the amplitude of all voltage and curent sources set to zero, a current source of 1 A
(RMS) is then injected into node 1L3A, which produces the impedance of the Thevenin
equivalent circuit looking into node 1L3A as

ZThev = 2.43∠ − 89.9o kΩ at 60 Hz.

ZThev

VThev I 5Ω

Figure 2.4 Single phase Thevenin equivalent circuit

2-6
Since we know that the effect is basically capacitive coupling, we can assume that this is
a capacitance not only at 60 Hz, but also a capacitance up to some higher frequency. The
Thevenin equivalent circuit can therefore not only be used for steady-state studies, but with some
precautions for transient studies with frequencies above 60 Hz as well. To see how good the
capacitance representation is, a frequency scan was performed, with the same data that was used
for obtaining the impedance. Instead of injecting a current of 1 A (RMS), which gave us the
impedance before, we now connect a voltage source of 1 V (RMS) to node 1L3A through a very
small resistance, or through a switch. The current in that small resistance or switch will then give
us the admittance, since I = YV , and since V = 1( RMS ) . The respective input file is
“COUPSCAN.DAT”, which includes instructions on how to change from impedance to
admittance calculations, or vice versa. Figures 2.5 and 2.6 show the magnitude and phase angle
of that admittance up to 500 Hz. As can be seen, the admittance rises linearly with frequency up
to approximately 300 Hz, and the phase angle remains close to –90 0 to approximately 300 Hz.
The capacitive Thevenin equivalent circuit of Fig. 2.4 could therefore be used for transient
studies with frequencies up to 300 Hz.

Fig. 2.5 Magnitude of admittance seen from node 1L3A


2-7
Fig. 2.6 Phase angle of admittance seen from node 1L3A

2.3.2 THREE-PHASE THEVENIN EQUIVALENT CIRCUIT

The three-phase Thevenin equivalent circuit looking into nodes 1L3A, 1L3B and 1L3C is
shown in Figure 2.7. The procedure for finding its parameters is as follows:

1). Find the open-circuit voltages of phases A, B and C, as described in Section 2.2.

2). Find the 3×3 impedance matrix [Z THEV ] by injecting a current first into node 1L3A, then

into 1L3B and finally into 1L3C, as described in Section 2.2.

The results are summarized in Table 2.1. The validity of this Thevenin equivalent circuit
can easily be checked. First add the 5 Ω resistances to ground at 1L3A, 1L3B and 1L3C of the
equivalent circuit. Then compute the grounding currents through the 3 resistances. The results
are identical with the results obtained from the complete network, as shown in Table 2.2. By
using high accuracy input, no round-off errors are visible. The input file is “THEV_BCH.DAT”.

2-8
A
[ZThev] B L3
C

5Ω 0 0
 0 5Ω 0 
 
 0 0 5Ω
[VThev]

Figure 2.7 Three-phase Thevenin equivalent circuit

Table 2.1 Parameters of three-phase Thevenin equivalent circuit

Open-circuit voltages Equivalent impedance


(kV) (Ω)

1L3A: 1.9209403 1.2967234 1.3483427


28.19986∠128.9360° -j 2429.6192 -j 500.98862 -j 240.21622

1L3B: 1.2967234 2.1363337 1.4692897


14.19608∠128.3667° -j 500.98862 -j 2457.8891 -j 516.14881

1L3C: 1.3483427 1.4692897 2.2554161


8.053402∠126.7944° -j 240.21622 -j 516.14881 -j 2466.141

Table 2.2 Grounding currents

Calculated branch currents


(A)

Branch Thevenin equivalent circuit Complete circuit

1L3A – Ground 10.784∠-141.13° 10.784∠-141.13°

1L3B – Ground 3.2559∠-141.86 ° 3.2559∠-141.86°

1L3C – Ground 1.5361∠-145.69° 1.5361∠-145.69°

2-9
2.4 THEVENIN EQUIVALENT CIRCUIT FOR 1100 KV TEST LINE OF

BONNEVILLE POWER ADMINISTRATION

Input files: FINDTHEV.DAT

This case describes the modeling of a 1100 kV test line (1.3 miles long), that Bonneville
Power Administration operated in the 1970’s [9]. The data was supplied by Mr. R. M. Hasibar.
Ferroresonance occurred in phase A when that phase was switched off on the low-voltage side of
the step-up transformer (Fig. 2.8). Phase C was not yet connected to a transformer at that time.

SW A

Vs

Vs
C

Figure 2.8 1100 kV test line of Bonneville Power


Administration

For the simulation of the ferroresonance phenomenon, it is easiest to use the full system
shown in Fig. 2.8. Since we are interested in network reduction to Thevenin equivalent circuits
in this section, we can also reduce the part within the dotted lines to a voltage source behind a
capacitance, and then use this reduced network together with the details of phase A in the
simulation.

2 - 10
2.4.1 NETWORK DATA

The network data for the test line of Fig. 2.8, for the 230 kV supply side, and for the step-
up transformer are as follows:

(1) Source:

The sequence reactances of the source, referred to the 230 kV side, are Xpos = 7.04 Ω and
Xzero = 7.88 Ω at 60 Hz. The resistances can be neglected. The voltage is assumed to be such that
it produces 1100 kV (RMS, line-to-line) on the high side of the transformer.

2) Transmission Line:

The transmission line is represented with constant parameters at 60 Hz, and its data is
given by the series impedance matrix [Z’] and shunt capacitance matrix [C’] per mile:

0.0951750 + j 0.920867 Symmetric 



[Z' ] = 0.0872880 + j 0.453280 0.0915829 + j 0.925271  Ω/mile;

0.0890750 + j 0.450562 0.0872880 + j 0.453280 0.0951750 + j 0.920867 

 20.6022 Symmetric 

[C' ] = − 3.55119 20.2788  nF/mile;

 − 3.11128 − 3.55119 20.6022 
line length = 1.3 miles.

(3) Transformer:

The transformers are single-phase autotransformer units, each rated 50 MVA,


635.1/139.4 kV. The short-circuit input impedance is 18.4%, and the load losses are 139966 W.

Table 2.3 Data for magnetizing impedance

Voltage (%) Exciting current (%) Excitation loss (W)

90 0.56 63129
100 1.50 89794
110 4.01 134394

2 - 11
2.4.2 PARAMETERS OF THE THEVENIN EQUIVALENT CIRCUIT

Actual values are used, rather than per unit values, and all values are referred to the 1100
kV side of the transformer:

(1) Source:

The sequence reactances are converted to a three-phase [X]-matrix in phase quantities.


The diagonal element Xs and the off-diagonal element Xm are

Xs =
1
(2 X pos + X zero ) and X m = 1 (X zero − X pos ) ,
3 3

or Xs = 7.32 Ω, Xm = 0.28 Ω seen on the 230 kV side. The values supplied for the 230 kV side
are then multiplied with the square of the transformer turns ratio of (635.1/139.4)2 , to convert
them to the 1100 kV side, which produces Xs = 151.94 Ω, Xm = 5.81 Ω seen on the 1100 kV
side. For deriving the Thevenin equivalent circuit, the three-phase [X]-matrix is used, but for
working with the Thevenin equivalent circuit together with the details of phase A later on, the
mutual coupling reactance Xm to phase B will be ignored because it is so small. The source
reactance for phase A then becomes a single-phase reactance Xs.

(2) Transmission Line:

A nominal π-circuit is used here, and the matrices [Z’] and [C’] per mile are simply
multiplied with the length of the line.

(3) Transformer:

For conversion from per unit to actual values, the base power is Sbase = 50 MVA (single
phase), and the base voltage is Vbase = 635.1 kV (RMS phase to ground). This gives a base
impedance of Zbase = 635.12/50 Ω. The single-phase transformer in phase B is modeled as a T-
circuit, as shown in Figure 2.9. For the ferroresonance study later on, a T-circuit is also used for
phase A, but the magnetizing inductance will then become nonlinear. The short-circuit
impedance for each transformer is calculated as
2 - 12
Pload loss 2
R= ⋅ Z base , Z = Z pu ⋅ Z base and X = Z − R 2 .Then R=22.58 Ω , X=1483.82 Ω. One
S base
half of R+jX is placed on the low and high side. The magnetizing impedance consists of Rc for
iron core losses in parallel with a linear magnetizing reactance Xm for the exciting current, with

2
Vbase Z base
Rc = = 4.49 MΩ and X m = = 0.538 MΩ from 1 p.u. excitation data.
Pexcitation loss I exciting current ( pu )

LOW MID HIGH

Rc Lm

Figure 2.9 T-circuit for single-phase transformer, with parameters referred to high side

We first find the open-circuit voltage at HIGH-A (sending end of transmission line in
phase A in input file FINDTHEV.DAT). Then we short the voltage sources and add a current
source of 1 A (RMS) to node HIGH-A. This gives us the Thevenin equivalent circuit of Figure
2.10, with the values of Table 2.4.

The magnetizing impedance has very little influence on the results of the Thevenin
equivalent circuit. Treating it as linear, as is required for network reduction, is therefore
acceptable, as shown in Table 2.5 for magnetizing impedances at 90%, 100% and 110%
excitation. For the ferroresonance study later on, the magnetizing inductance of phase A must of
course be modelled in more detail with a nonlinear flux-current characteristic.

SW ZThev

VThev

Figure 2.10 Details of phase A with Thevenin equivalent


circuit for coupling to phase B

2 - 13
Table 2.4. Parameters of Thevenin equivalent circuit (magnetizing impedance of transformer:
Rc = 4.49 MΩ, Xm = 0.538 MΩ)

Open-circuit voltage (kV) Equivalent impedance (kΩ)

130.71∠-120° 101.27∠-90°

Table 2.5. Comparison of the results with different magnetizing impedances

Calculated parameters of Thevenin


equivalent circuit

Magnetizing impedance Open-circuit voltage Equivalent impedance


of the transformer (MΩ) (kV) (kΩ)

Rc = 4.49, Xm = 0.538 130.71∠-120° 101.27∠-90°


(100% rated source voltage)
Rc = 6.39, Xm = 1.44 130.85∠-120° 101.27∠-90°
(90% rated source voltage)
Rc = 3.00, Xm = 0.201 130.34∠-120° 101.27∠-90°
(110% rated source voltage)

2 - 14
3. AC STEADY STATE WITH HARMONIC DISTORTION

3.1 INTRODUCTION

Harmonics caused by various devices are becoming more serious in electric power
systems. To analyze their effects, and to find ways of reducing them, is one of the topics of
power quality studies. Some sources of harmonics are rotating ac machines, fluorescent lighting,
glow discharge lighting, static Var compensators, overexcited transformers, adjustable speed
drives, light dimmers, controlled rectifiers and electric heating controllers. In rectifiers and
inverters, the magnitude of harmonics is reasonably well known, and these harmonics can
therefore be represented as given current or voltage sources in harmonic load flow programs that
are specifically designed for harmonics studies. In contrast, harmonics generated by transformer
saturation depend critically on the peak magnitude and waveform of the voltage at the
transformer terminals, which in turn are influenced by the harmonic currents and the frequency-
dependent network impedances.

Ac steady-state solutions with harmonic distortions can often be obtained with EMTP
programs by starting the simulation from approximate linear ac steady-state conditions at power
frequency. In the approximate linear ac steady-state solution, nonlinear elements are represented
as linear (e.g., with their unsaturated values). This steady-state solution is followed by a transient
simulation over a time span to tmax that is long enough to let the phenomena settle down to the
distorted ac steady state. Unsaturated values L1 of “switched inductances” (or “piecewise linear
inductances”) are automatically included in the ac steady-state solution by MicroTran. For
nonlinear elements defined point by point, the user must specify the unsaturated value to be
included in the steady-state solution. By starting the simulation from an approximate linear
steady state, the only disturbance will be the deviation between the linear and nonlinear
characteristics. This is reasonably simple for cases with transformer saturation. For power
electronics cases, the linearized starting point is not as easy to define.

The transients caused by the deviation from the approximate linear ac steady state will
often settle down to the steady-state conditions with harmonics fairly quickly, especially in cases

3-1
with high damping. The waveforms will contain the harmonics with reasonable accuracy up to a
certain order, depending on the time step size ∆t. Fourier analysis programs, such as the support
routine HARMONIC in Microtran, can then be used to obtain the harmonic content of the
waveforms.

Some newer versions of the EMTP can perform steady-state initializations that already
include the major harmonics, as described in [12]. This feature has been included in the
DCG/EPRI EMTP versions EMTP96 and EMTPRV. MicroTran includes the second harmonics
in the rotor windings of synchronous machines in its steady-state initialization (Section 8.4.2 of
[3]), which are produced by (fundamental frequency) negative sequence currents in the stator.

Two case studies are presented here, one from a Brazilian electric utility company where
transformer saturation causes the harmonics, and another case where power electronics devices
cause the harmonics.

3.2 HARMONICS FROM TRANSFORMER SATURATION

Input files: HARMON_1.DAT, HARMON_2.DAT

The following case comes from the Brazilian utility company CEMIG, where field test
data for saturation effects was available. In this particular case, the harmonics were primarily
positive sequence, with a dominant 7th harmonic. A single-phase, positive sequence network
representation might therefore have been good enough for the simulation. Since we normally do
not know what types and orders of harmonics to expect, it is best to use three-phase models,
because they will show negative and zero sequence harmonics as well. A three-phase model is
therefore used here.

3.2.1 POWER SYSTEM CONFIGURATION

Figure 3.1 shows the network configuration that was used for this test [10]. The power
plant at Jaguara, with three generators and two three-winding step-up transformers, was
connected to a single 345 kV transmission line to Taquaril (open-circuited receiving end). The
3-2
line had a shunt reactor connected to it at the sending end. The source of the harmonics was
transformer saturation.

THREE-WINDING
TRANSFORMER

S
P
T SENDING RECEIVING
END END

S
P
T

Figure 3.1 Network configuration for harmonics test in Brazilian utility


company CEMIG

3.2.2 NETWORK DATA

Input data is needed for the generators, the transformers, the shunt reactors and the
transmission line.

(1) Generators:

Each of the three generators is rated S = 112 MVA (three-phase), Xd’’ = 0.1385 p.u. based
on 100 MVA, f = 60 Hz. One transformer had only one generator connected to it, because at the
time of testing, its second generator had not yet been installed. The generators will be
represented with E’’ behind Ld’’, where Lpos = Lneg = Ld’’. Lzero is immaterial because of the delta
connections of the transformers on the generator side (no zero sequence currents can flow into
the generator in this case). The accuracy of the simulation results might be improved with a more
detailed generator model, but such a model was not yet available when these tests and
simulations were first done. It is reasonable to assume, however, that synchronous machine
inductances are fairly close to Ld” at harmonic frequencies.

3-3
(2) Transformers:

Each transformer is Ygrounded /delta/delta connected, and rated 120 MVA (three-phase).
The short-circuit reactances (primary P, secondary S, and tertiary T as in Figure 3.1), based on
100 MVA (three-phase) and the transformer voltage ratings (345 kV, phase-to-phase on the
primary side) are as follows:

XPS = 0.117 p.u.; XPT = 0.115 p.u.; XST = 0.241 p.u.

(3) Shunt reactor:

The shunt reactor is rated Q = 91 Mvar (three-phase), at a voltage rating of 440 kV


(phase-to-phase). It is a three-phase unit, with Xzero = 0.35 Xpos.

(4) Transmission line:

For the transmission line, a distributed-parameter model with constant parameters will be
used. The values were calculated at 500 Hz, with a Line Constants Program such as MTLINE,
because it was known that the dominant harmonic is the 7th. The reactances X were then
converted from 500 Hz to 60 Hz, because the rest of the input data was specified at 60 Hz. For
60 Hz input, the data is

Rpos = 0.057 Ω/km, Rzero = 1.320 Ω/km;

Xpos = 0.375 Ω/km, Xzero = 0.885 Ω/km;

Cpos = 0.0118 µF/km, Czero = 0.008 µF/km;

length = 398 km.

3.2.3 INPUT DATA

In this study, we represent the power plant as a Thevenin equivalent circuit seen from the
345 kV side that includes the generators and the three-winding transformers, as shown in Figure
3.2. This Thevenin equivalent circuit is obtained as follows:

3-4
(1) Find the values of the star circuit for the three-winding transformers with p.u. values
based on 100 MVA. Resistances can be ignored. This star circuit is well known from power flow
and short-circuit studies. The p.u. reactances of the star circuit are

1
XP = ( X PS + X PT − X ST ), X S = 1 ( X PS + X ST − X PT ), X T = 1 ( X PT + X ST − X PS ) ,
2 2 2

which gives us XP = -0.0045 p.u., XS = 0.1215 p.u., and XT = 0.1195 p.u. We could also have

used the preprocessor MTDATA, or a graphical user interface, to input short-circuit reactances

of each three-winding transformer directly, and add the generator reactances directly.

Positive sequence

S
P
T Negative sequence

S
P Zero sequence
T

Figure 3.2 Thevenin equivalent circuit for power plant

(2) By adding Xgenerator = 0.1385 p.u. between side S and T of the first transformer and
ground, and by looking into the circuit from the primary side, we get the positive sequence
reactance of the first transformer according to Figure 3.3 as

1
Xpos-1 = − 0.0045 + = 0.124998 p.u..
1 1
+
0.1385 + 0.1215 0.1385 + 0.1195

For the second transformer with one generator, we get


3-5
Xpos-2 = − 0.0045 + 0.1215 + 0.1385 = 0.2555 p.u.

Figure 3.3 Reactances for finding Thevenin equivalent circuit

Paralleling the two reactances finally gives us Xpos = 0.0839347 p.u. For the zero sequence
reactance, we must remember that the delta sides of the transformers appear as short circuits, or
1
for each transformer X = −0.0045 + = 0.0557458 p.u., or for the two
1 1
+
0.1215 0.1195
transformers in parallel, Xzero = 0.0278729 p.u.

(3) Transform the results from step 2 into the physical quantities seen from the 345 kV
side. This gives us Xpos = 99.903277 Ω and Xzero = 33.175719 Ω.

(4) Convert the sequence reactances Xpos, Xneg, Xzero to a three-phase [X]-matrix with self
and mutual reactances. Assume that Xneg = Xpos. Rounded to 4 digits, this gives us Xs = 77.66 Ω
and Xm = -22.24 Ω.

3-6
(5) Find Xpos and Xzero (Xneg = Xpos) and the three-phase [X]-matrix for the shunt
2
Vrating 440 2
reactor: X pos = = = 2127.4725 Ω and X zero = 744.61538 Ω , or rounded to 5 digits,
S rating 91

Xs = 1666.5 Ω and Xm = -461.0 Ω.

(6) Since saturation of the transformers plays the important role in producing the
harmonics, their nonlinear magnetizing inductances must be modeled. In this study, we add the
magnetizing inductance to the 345 kV side of each transformer, phase to ground, and for input
into the EMTP, we parallel the two into one nonlinear inductance. In the original study of [10], a
two-slope piecewise linear representation was used, as shown in Fig. 3.4, with the following
values per transformer (divide the R- and ωL-values by 2.0 for the two transformers in parallel).
This representation is used in the input file HARMON_1.DAT.

For iron core losses: RFe = 740 kΩ;

“knee point” λs = 775 Vs;

ωL1 = 348000 Ω, ωL2 = 806 Ω at 60 Hz.

L2
λs

RFe
L1

Figure 3.4 Magnetizing branch per phase for one transformer

It would be more accurate to connect the magnetizing branch across the terminal of that
winding which is closest to the iron core (which would probably be one of the delta-connected
windings), but then one could no longer use the simple Thevenin equivalent circuit chosen here.

3-7
A more accurate saturation characteristic with 4 data points is used in input file
HARMON_2.DAT, with the λ-i curve specified point by point according to Figure 3.5, with data
given in Table 3.1, both for the two transformers in parallel.

Table 3.1 Flux-current relationship

i (A) λ (Vs)

0.00 0.00
1.56 718.00
4.66 760.00
15.60 800.00
62.40 855.00

Figure 3.5 Flux current relationship

(7) The transmission line can no longer be modelled as a nominal π-circuit, because that
model would not be accurate for harmonic frequencies. A distributed-parameter model with
constant parameters is used instead. In that model, the resistances are lumped in the middle and
at both ends, and the inductances and capacitances are distributed in the two line sections.

3-8
Rtotal Z surge
Starting with MicroTran Version 2.06, must be less than , to assure reasonable
4 10
results. That is why the line is split into 3 sections.

(8) The step size ∆t must be chosen such that the highest frequency of interest is still
represented reasonably well. If we want to obtain harmonics up to the 15th order, we should have
approximately 10 points per cycle at its frequency of 900 Hz. This gives us a step size of ∆t =
T/10, where the period T for the 15th harmonic is 1/900 s. We will use the resulting ∆t = 111.111
µs, rather than rounding it down to 100 µs, to make the Fourier analysis of the waveform more
accurate. With that step size, the period for the fundamental frequency will then be an integer
multiple of ∆t.

3.2.4 SIMULATION RESULTS

Figure 3.6 from the original study of [10] compares the simulation results with the field
tests, both for the sending end and the receiving end. The simulation results with the two-slope
piecewise linear inductance (HARMON_1.DAT) and with the 4-point nonlinear inductance of
Table 3.1 (HARMON_2.DAT) are compared in Figure 3.7 for the sending end in “JAGA”, and
in Figure 3.8 for the receiving end in “TAQA”. Table 3.2 shows the results of a harmonic
analysis with the support routine HARMONIC up to the 15th harmonic. As can be seen, the 7th
harmonic is the largest harmonic.

To see whether the simulation follows the specified flux-current characteristic, a very
large inductance of 1020 H was inserted in parallel with the 4-point nonlinear inductance (input
file HARMON_2.DAT). The value is so large that it will obviously not influence the results.
Since its flux linkage λ is λ = L ⋅ i , we can use the current through it and multiply it with the
factor 1020 to obtain the flux linkage, which can then be plotted from the transients plot file as a
function of the current. This is shown in Figure 3.9, in which time is a parameter (not included in
the plot). As can be seen, the characteristic is followed correctly in the simulation.

3-9
Figure 3.6 Comparison between simulation and field test from [10]

3 - 10
Figure 3.7 Comparison between 2-slope and 4-point nonlinear inductance; sending end

Figure 3.8 Comparison between 2-slope and 4-point nonlinear inductance; receiving end

3 - 11
Table 3.2 Harmonic content after t = 0.08 s in nodes JAQA and TAQA
Period (rounded to nearest integer of Deltat)= 1.6666650E-02 seconds.
Analysis starts at 7.9999922E-02 seconds, fundamental frequency is
6.0000058E+01 Hz.

FOURIER COEFFICIENTS
HARMONIC MAGNITUDE IN JAQ MAGNITUDE IN TAQ
0 2.854686E+01 7.538403E+00
1 3.268292E+05 3.752933E+05
2 1.881242E+03 3.732090E+03
3 1.913835E+03 2.839941E+03
4 1.867009E+02 3.168774E+02
5 3.538089E+03 4.206045E+03
6 3.591766E+02 4.460591E+02
7 2.714493E+04 3.005954E+04
8 1.837664E+02 1.415761E+02
9 6.409219E+02 7.168201E+02
10 1.479132E+02 1.094878E+02
11 3.462770E+02 4.107837E+02
12 3.531106E+02 2.862839E+02
13 1.415308E+03 1.554581E+03
14 3.491369E+01 5.682588E+01
15 5.223129E+02 4.458299E+02
TOTAL HARMONIC DISTORTION TO 15TH HARMONIC = 8.4335 % in JAQA,
8.1997 % in TAQA.

Figure 3.9 Nonlinear characteristic that is followed in transient simulation

3 - 12
3.3 HARMONICS FROM SINGLE-PHASE DIODE BRIDGE RECTIFIERS

Input files: EX3A.DAT, EX3.DAT

This case comes from N. Mohan’s collection of computer exercises for power electronics
education [11]. It consists of a three-phase voltage source (wye grounded) with internal
inductances of 1 mH in each phase (Fig. 3.10). Between each phase and the neutral conductor, a
single-phase diode bridge rectifier is connected that has a load on the dc side of R = 20 Ω in
parallel with C = 1000 µF. The aim of the study is to find the current in the neutral conductor.
Please look at the input files for more hints about snubber circuits, numerical oscillations, etc.

Figure 3.10 Three single-phase bridge rectifiers on a three-phase supply

3.3.1 ZERO INITIAL CONDITIONS

By starting the simulation from zero initial conditions ( parameter Tstart = 0 on the source
specifications), the harmonics will develop as time proceeds. Figure 3.11 shows the current in
phase A from 0 to 0.25 s. As can be seen, it takes some time until steady state with harmonics is
reached. Figure 3.12 shows that steady state for all three phases between 0.20 s and 0.25 s.

3 - 13
Figure 3.11 Current in phase A, starting from zero initial conditions

Figure 3.12 Currents in phases A, B and C after reaching steady state

In Figure 3.13 we can see the high 3rd harmonic current in the neutral that is caused by
the diode bridge rectifier loads in each of the three phases. Its magnitude is as high as the
currents in the phase conductors, an effect that is frequently observed in commercial buildings.

3 - 14
Figure 3.13 Current in neutral conductor after reaching steady state

3.3.2 USER-SUPPLIED INITIAL CONDITIONS

Most EMTP versions allow the user to supply initial conditions. In this case, they were
found from the preceding case of Section 3.3.1 (input file EX3A.DAT), by requesting voltage
and current output at tmax = 0.25 s (with "1" in column 26 of the time specification line). These
values are then read in as initial conditions in input file EX3.DAT. In addition, two diodes of
phase C are already conducting at t = 0, which requires the user to set Tclose < 0 for these two
diodes in the input file. Also, Tstart on the source specifications must be negative, to start from an
ac steady-state solution. Figure 3.14, created by running input file EX3.DAT, shows correct
steady-state behaviour with harmonics from the very beginning.

3 - 15
Figure 3.14 Current in phase A with correct initial conditions

3 - 16
4. TRANSFORMER INRUSH CURRENTS

4.1 INTRODUCTION

A single-phase two-winding transformer (windings from phase to ground) can be


represented reasonably well as a T-circuit, if the impedances are referred to one side. The shunt
branch between its midpoint and ground is the nonlinear magnetizing inductance, sometimes
with a nonlinear resistance in parallel to represent iron core losses. The series impedances on
each side of the midpoint are one half of its short-circuit impedance.

When such a transformer is energized from one side, without a load connected to the
other side, then the impedance is essentially the nonlinear magnetizing impedance, assuming that
the source impedance and short-circuit impedance are small compared to the magnetizing
impedance. That assumption is correct in the unsaturated region. In the saturated region it may
no longer be quite correct, depending on the slope of the flux-current curve in the saturated
region.

Now assume that we energize the transformer through a circuit breaker at an instant at
which the source voltage is just going through zero. With

v = Vmax sin(ωt ) for t ≥ 0 , (4.1)

appearing across the nonlinear inductance, the flux being the integral over the voltage will be

λ = Λ max [1 − cos(ωt )] for t ≥ 0 , (4.2)

with the correct integration constant to make λ (0 ) = 0 at t = 0 . Equation (4.2) shows that the
flux will reach 2.0 p.u. of its normal value one half cycle after circuit breaker closing. Modern
transformers have very low air core inductances, which is essentially the slope dλ/di in the
saturated region of the magnetizing inductance.

Figure 4.1 shows a typical saturation curve obtained from measurements [13]. Typical
values for air-core inductances are 2Lshort (Lshort = short-circuit inductance) for two-winding
4-1
transformers with separate windings [16]. CIGRE guidelines [17] mention 1 to 1.5 times Lshort
for transformers with separate windings, and 3 to 4 times Lshort for autotransformers (= 4 to 5
times Lshort seen from the high voltage side if the nonlinear inductance is connected across the
low-voltage winding). If we try to find the current from this flux-current curve for λ = 2.0 p.u., it
will obviously have a very high value. This “inrush current” can be as high as 10-times rated
current. Due to damping, it will eventually decay to the steady-state exciting current, that is
typically less than 1 % (0.01 p.u.) of the rated current on large transformers.

Figure 4.1 Measured saturation curve of large transformer [13]

The relay protection of transformers must be able to distinguish the high inrush currents
from short-circuit currents, e.g., by analyzing the harmonic content. Inrush currents contain a
strong second and third harmonic.

If the circuit breaker were to close when the voltage is just at its peak, or
v = Vmax cos(ωt ) for t ≥ 0 , then the flux would be λ = Λ max sin(ωt ) for t ≥ 0 with a peak value

of only 1.0 p.u. (no 1.0 p.u. offset as before). In this case the inrush current would be the low
steady-state exciting current. Older circuit breakers with slow contacts had a tendency to
prestrike close to maximum voltage, and therefore produced lower inrush currents. Modern

4-2
circuit breakers close at high speed, and closing at zero voltage is as probable as closing at peak
voltage.

Reality is further complicated by the fact that transformers usually have a residual flux
left in the iron core, which may either increase the inrush current or decrease it. The inrush
current also depends on the tap position of the load tap changer, and by positioning it
conveniently, the inrush currents can be reduced [20]. Furthermore, if other transformers are
already in operation close to the one being energized, there is “sympathetic interaction” between
them that influences the inrush currents [21].

By monitoring the flux in the transformer, and by controlling the closing of the circuit
breaker contacts, it becomes possible to close at just the right moment to reduce the inrush
current to very small values similar to the steady-state exciting current ([22], [44], [45], [46]).
Figure 4.2 shows a field test on a 230/138kV transformer (wye grounded / wye grounded) with
uncontrolled closing of the circuit breaker contacts, with noticeable high inrush currents of -1111
A, and Figure 4.3 shows controlled closing with practically no inrush current (4.54 A in phase C)
[22]. The high spikes in the beginning are not true current values and should be ignored.

Figure 4.2. Inrush currents with normal (uncontrolled) closing of circuit breaker contacts (values at crossbar: Ia =
145.95 A, Ib = 209.70 A, Ic = -1111.00 A; Va = -93.26 kV, Vb = 70.14 kV, Vc = 6.36 kV)
4-3
There may also be high-frequency overvoltages when three-phase transformer banks are
energized if the closing times of the three phases are more than 5 ms apart [15].

Figure 4.3. Elimination of inrush currents with controlled closing of circuit breaker contacts(values at crossbar: Ia =
2.77 A, Ib = -0.47 A, Ic = 4.54 A; Va = 115.14 kV, Vb = -50.56 kV, Vc = -62.40 kV)

4.2 COMPARISON BETWEEN FIELD TEST AND SIMULATION

Input file: INRUSH1.DAT

A comparison between a filed test in Sweden and simulation results is fully documented
in [14]. Since that reference is available on the Microtran website www.microtran.com, it will
not be reproduced here, except for a brief description of the network configuration and one
comparison between field test and simulation results. Figure 4.4 shows the network configuration
during the test. An unloaded transformer at the end of a 220 kV transmission line was energized
through a circuit breaker. In addition to the unloaded secondary windings the transformer also
had delta-connected tertiary windings without any load. Figure 4.5 shows one comparison
between the field test and the simulation, for the current in phase A. The little asterisks in that
figure are the peak values recorded in the field test.
4-4
SOURC LINE BREAK PRIM SEC

Source Transmission line Circuit breaker

TERT

Figure 4.4 Network configuration during inrush current field test

Current in phase A on primary side (* = measured).


1.40

1.20

1.00
Primary side current, Phase A (kA)

0.80

0.60

0.40

0.20

0.00
10 20 30 40 50 60 70 80 90 100 110 120 130
Time (ms)
-0.20

BREAKA to PRIM-A var 1

Figure 4.5 Comparison between measured and simulated inrush current

4-5
5. FERRORESONANCE

5.1 INTRODUCTION

Ferroresonance is a nonlinear resonance phenomenon that can occur in circuits with a


capacitance and a nonlinear inductance. The latter is usually associated with saturation effects of
the iron cores in power transformers and instrument potential transformers of the wound type.
The capacitance may be the coupling capacitance between an energized and de-energized
transmission line, the zero sequence shunt capacitance of ungrounded systems, or the grading
capacitors of open circuit breakers [18], [19]. For guidelines on computer simulations see [29].

Ferroresonance occurs normally after a switching event. Due to the nonlinearity of the
magnetizing inductance of transformers, there can be two or more distinct stable operating states
([23], [24]). In one of them, the transformer is operating in the linear region of the magnetization
curve. Which one of the possible states will be reached after the switching event depends very
much on the initial conditions. EMTP programs can be used for the simulation of ferroresonance
phenomena, although the simulation will not give insight into all possible steady-state
conditions. In that sense, the EMTP simulation is somewhat similar to transient stability
simulations, which also do not give global answers about the overall stability of the system.

Two cases will be studied here. The first case comes from an occurrence on an 1100 kV
test line that was already used in Section 2.4 for the derivation of Thevenin equivalent circuits. A
detailed analysis of this case can be found in [25]. The second case comes from an unexpected
ferroresonance occurrence between an energized and de-energized transmission line.

5.2 FERRORESONANCE ON 1100 KV TEST LINE

Input files: FERROBPA.DAT, FERRO_TH.DAT, FERROLIN.DAT

Since the 1100 kV transmission line and the associated network data have already been
described in Section 2.4, only the additional data needed for the ferroresonance simulation will
be discussed in detail here. For a better understanding, Figure 2.8 is reproduced here as Fig. 5.1.

5-1
SW A

Vs

Vs
C

Figure 5.1 1100 kV test line of Bonneville Power


Administration

For the ferroresonance simulation, a second transformer must be added to phase A, and a
circuit breaker must be added that opens phase A. The transformer data for phase A is the same
as for phase B, except that its linear magnetizing inductance must be replaced by a nonlinear
inductance.

Table 2.3 of Section 2.4.1 shows the exciting current at 90 %, 100 % and 110 % of the
rated voltage. These 3 points give us a crude representation of the nonlinear characteristic, that
nonetheless produces reasonably good results. The voltages and currents in Table 2.3 are RMS
values that must be converted to a flux-current curve with peak values. As the flux linkage
becomes higher, the harmonic content of the current will increase. A simple multiplication with
√2 to obtain the peak value of the current, and multiplication with √2VRMS/ω to obtain the flux
linkage from the voltage, is therefore not good enough. Most EMTP versions have a support
routine (“SATURATE” in Microtran, or “CONVERT” in some other programs) that performs
this conversion. Reference [3] on page 6-54 describes the conversion method and uses the data of
Table 2.3 as an example. The output of Microtran’s SATURATE support routine is shown in
Table 5.1.

5-2
Table 5.1 Ouput file of support routine SATURATE

1100 kV test line

Vbase= 6.35100E+02 kV, Sbase= 5.00000E+01 MVA,


frequency= 6.00000E+01 Hz, Xshort= 0.00000E+00 P.U.

RMS CURRENT IN P.U. RMS VOLTAGE IN P.U.


1 0.000000E+00 0.000000E+00
2 5.600000E-03 9.000000E-01
3 1.500000E-02 1.000000E+00
4 4.010000E-02 1.100000E+00

The following table shows the flux-current curve, extended


10.0 times beyond the last calculated current point using
the slope between the last two points.If you want to change
the factor 10.0, type it in columns 41 ... 50 in the line
of the input file with Vbase, Sbase, f, Xshort.

The table starts with ".BEGIN" and ends with ".END", so that
you can plot the curve with MTPLOT,to check whether it looks
reasonable.

PEAK CURRENT IN A PEAK FLUX IN VS


.BEGIN
1 0.000000E+00 0.000000E+00
2 6.234921E-01 2.144216E+03
3 2.723762E+00 2.382462E+03
4 7.248722E+00 2.620708E+03
5 7.248722E+01 6.055617E+03
.END

RMS CURVE IN P.U. RE-COMPUTED FROM FLUX CURVE


2 5.600000E-03 9.000000E-01
3 1.500000E-02 1.000000E+00
4 4.010000E-02 1.100000E+00

The following lines are in a form ready to be copied into


the MicroTran input file for the nonlinear characteristic:
(omitted to shorten the table)

Strictly speaking, the values for the exciting current should have been reduced slightly
because they not only contain the current through the nonlinear inductance but through the
parallel resistance Rc for iron core losses as well. A more accurate conversion includes the
excitation loss in the input data as well, and produces a nonlinear inductance in parallel with a
nonlinear resistance [25].

Figure 5.2 shows the measurements after ferroresonance occurred when the circuit
breaker in phase A was opened on the 230 kV side. The top oscillogram shows the voltage of
phase B, while the middle and bottom oscillograms show the voltage of phase A. The middle

5-3
oscillogram was recorded from a bushing tap, while the bottom one was recorded with a voltage
divider.

Instead of the voltage in phase A taking on the value of 130.7 kV that one would expect
from the capacitive coupling to the energized phase B (see Table 2.4), the voltage in phase A
essentially maintains its magnitude, with the waveshape becoming more rectangular than
sinusoidal. That rectangular shape is typical of many ferroresonance curves.

Figure 5.3 shows the simulation results with the three-phase line modelled as a π-circuit
(input file FERROBPA.DAT). It does not match the field tests completely, but given the crude
model for saturation etc., it is a reasonable duplication of the ferroresonance phenomenon. When
the case is simulated with the simpler Thevenin equivalent circuit of Section 2.4 (input file
FERRO_TH.DAT), the answers are practically identical with those of Figure 5.3 from the more
detailed line model.

Figure 5.2 Measurements of the ferroresonance occurrence

5-4
Figure 5.3 Simulation results of the ferroresonance occurrence

Figure 5.4 shows the results if the transformer could be switched off from the high side.
In that case, the nonlinear inductance would no longer be in the circuit after circuit breaker
opening, and therefore there would be no ferroresonance. Instead, phase A would have a trapped
charge voltage, with the 60 Hz voltage from the coupling effect from phase B superimposed.
Since no circuit breakers are available for 1100 kV, that type of switching is not possible.

Figure 5.5 shows the results of a simulation in which it is assumed that the circuit breaker
of phase A is already open in the initial steady-state solution (input file FERROLIN.DAT). In
that case the solution is linear with a linear magnetizing reactance of 1.2965 MΩ (initial slope for
steady-state solution in input file and in Table 5.1) , with a voltage of 141.74 kV RMS in phase
A, which shows that there can be more than one answer if the initial conditions are different. The
voltage is slightly higher than the open-circuit voltage of 130.71 kV RMS of the Thevenin
equivalent circuit because of the series connection of the transformer impedances and the
coupling capacitance to phase B.

5-5
Figure 5.4 Circuit breaker opening on the high side of the transformer

Figure 5.5 Circuit breaker is already open in steady state

5-6
5.3 FERRORESONANCE BETWEEN ENERGIZED AND DE-
ENERGIZED TRANSMISSION LINES

Input files: FERRO2B.DAT, FERRO2A.DAT

This case describes a ferroresonance occurrence in the 500 kV transmission system of


Bonneville Power Administration [26]. Two 500 kV transmission lines ran parallel over a
distance of approximately 30 km, as shown in Fig. 5.6. One line, with a 500/230 kV transformer
at the remote end, was switched off there first on the 230 kV side. There was no circuit breaker
on the 500 kV side, because of lower costs for 230 kV circuit breakers. When the 500 kV circuit
breaker was opened next at the sending end, ferroresonance occurred between the magnetizing
inductances of the transformer still connected to the line at the remote end, and the coupling
capacitance to the parallel energized line.

The two parallel transmission lines were modelled as a six-phase π-circuit for the
simulation. The transformer was modelled with its winding resistance, some stray capacitance, a
resistance for iron core losses, and a nonlinear inductance with a piecewise linear characteristic
with 2 slopes. A reactance of 377·1020 Ω, or L = 1.0·1020 H, was connected in parallel with the
nonlinear inductance, to obtain the flux linkage for plotting the flux-current curve. Its value is so
high that it does not influence the results, and if its current is multiplied by 1.0·1020, we obtain
the flux linkage, since λ = L ⋅ i .

Figure 5.6 Two 500 kV transmission lines in parallel

Figures 5.7 and 5.8 show the simulation results. The 2-slope nonlinear inductances were
solved with the compensation method [3, Section 12.1.2], and the CDA scheme (critical damping

5-7
adjustment scheme) was used as well [27], [28]. Figure 5.7 shows the beginning of the
ferroresonance phenomenon, shortly after the circuit breaker contacts open at 55.65 ms (phase
B), 58.5 ms (phase A), and 62.05 ms (phase C). Figure 5.8 shows the sustained ferroresonance in
phase C after more than 1 s. It is interesting to note that the frequency is not 60 Hz, but 20 Hz
(3rd subharmonic). In Fig. 5.9, the flux is plotted as a function of the current for phase C. As we
can see, the simulation follows the specified 2-slope characteristic very well.

Figure 5.7 Beginning of ferroresonance phenomenon

This ferroresonance case seems to be very sensitive to very small changes in data, or to
small changes in the solution techniques. When the author first worked on this case with
piecewise linear inductances ("switched inductances") on a mainframe computer, he managed to
obtain the results shown here. When he worked on the case again in 1995, sustained
ferroresonance could no longer be obtained with MicroTran Version mt2.06, but it could be
obtained with the earlier Version mt2.04. In version mt2.04, the piecewise linear inductance
("switched inductance") had a slight offset when it switched from slope L1 to slope L2, whereas
version mt2.06 and newer versions no longer have that offset (supposedly an improvement).

5-8
Figure 5.8 Sustained ferroresonance in phase C

Figure 5.9 Flux-current curve obtained from transient simulation

5-9
The only way to get sustained ferroresonance with piecewise linear inductances
(“switched inductances”) from mt2.06 and newer versions was to run the case with mt2.04, ask
for a voltage and current output file ("punch file") at tmax, and use these values as initial
conditions in mt2.06 and newer versions. That is the input file "FERRO2A.DAT".

In the input file "FERRO2B.DAT used here, the nonlinear inductances are modelled as
three true nonlinear inductances and solved with the compensation method. The results show the
sustained ferroresonance explained in [26]. The simulation starts from ac steady-state conditions,
with the fluxes remaining in the linear region of the nonlinear inductance. Ferroresonance is
initiated by opening the switches at zero crossing of the current at the sending end. Note that
CDA is needed in this case to get the sustained ferroresonance results. Without CDA ("1" in
column 68 of the input line with ∆t, tmax, etc.), ferroresonance in phase C is not sustained, but
dies out as in phases A and B.

5 - 10
6. SUBSYNCHRONOUS RESONANCE

6.1 INTRODUCTION

Subsynchronous resonance is a resonance phenomenon at frequencies below the


synchronous power frequency. It causes torsional vibrations on the rotating mass and shaft
system of turbine-generators. They are excited by frequencies below power frequency in the
electric power system. Steam turbines of thermal power plants are vulnerable to it, because the
rotating masses and shafts of the turbine stages (typically high, intermediate and low pressure
stages), together with the masses and shafts of the generator and exciter machines, have
resonance frequencies below power frequency. Subsynchronous resonance does not seem to
occur with hydro turbines, because the masses of the turbine and generator are close together and
act more as a single mass.

The subsynchronous frequencies on the electric power system side are usually initiated
through switching events on transmission lines with series capacitors, where resonances exist
below power frequency. Subsynchronous resonance has also been observed in power plants close
to HVDC converter stations, where the firing control of the thyristors can possibly create
subsynchronous frequencies. If the frequency on the electric power system side is felectrical =
fsynchronous – ∆f, then subsynchronous resonance can occur if the mass-shaft system has a
resonance at fmechanical = ∆f. The sum of the resonance frequencies on the mechanical and
electrical side add up to fsynchronous. This assumes that the synchronous generator has 2 poles (1
pole pair); if not, the equations must be converted to those of an equivalent 2-pole machine.

The power industry became concerned about subsynchronous resonance phenomena in


the 1970’s, when shaft damage occurred in a steam power plant. This occurrence provided the
incentive for adding synchronous machine models to the EMTP [34], [35], because stability
programs for studying relative oscillations among turbine-generators were not able to simulate
phenomena at frequencies other than power frequency. Two benchmark models were developed
by IEEE to verify computer simulation results [30], [31].

6-1
6.2 FIRST BENCHMARK MODEL FOR SUBSYNCHRONOUS
RESONANCE

Input files: SSR1A.DAT, SSR1B.DAT, SSR1C.DAT

Since the documentation for this case is available on the Microtran website
www.microtran.com [32], the case will not be discussed in detail here, except for a brief
description of the network configuration with one simulation result. Figure 6.1 shows the
network configuration of IEEE’s first benchmark model.

Xo = 0.06
X1 = 0.06
A CAP B
INFI
Xo = 0.14 Xc = 0.371
X1 = 0.14 Ro = 0.50 Xo = 1.56
R1 = 0.02 X1 = 0.50

SWT
GEN EXC
Xo = 0.04
X1 = 0.04
HP IP LPA LPB

Figure 6.1 Network configuration of first IEEE benchmark model for subsynchrnous resonance

The case consists of a turbine-generator connected through a step-up transformer to a 500


kV transmission line with series capacitor compensation. The rotating masses consist of the high
pressure turbine stage, the intermediate pressure turbine stage, two low pressure turbine stages,
the synchronous generator, and the dc exciter machine. The electrical side of the generator is
modelled with 3 stator windings, the field winding, a winding on the quadrature axis to represent
eddy current effects, and one damper winding each on the direct and quadrature axis. For the
simulation, the data from the manufacturer, such as Xd, Xd’, Xd”, Tdo’, Tdo” etc., was converted to
self and mutual impedances with the highly accurate procedure described in Appendix VI of [3].
When armature resistances are ignored, this procedure reproduces the original data of the
manufacturer. The power system at the receiving end is represented as a Thevenin equivalent

6-2
circuit (infinite bus behind reactance). Subsynchronous torsional vibrations build up after a
three-phase fault of 4.5 cycle duration. The fault is simulated as a three-phase switch “SWT” at
node “B”.

Figure 6.2 shows the torque that develops after the fault on the shaft between the
synchronous generator and the exciter machine. As can be seen, torsional vibrations with a
frequency of approximately 40 Hz build up in this case.

2.00

1.00
Shaft torque GEN-EXC (x 1,000,000)

0.00
10 20 30 40 50 60 70 80 90 100 110 120 130 140 150 160 170 180 190 200
Time (ms)

-1.00

-2.00

MACH 1 -> SHTR 5

Figure 6.2 Torque on the shaft between the synchronous generator and the exciter machine

6-3
6.3 SECOND BENCHMARK MODEL FOR SUBSYNCHRONOUS
RESONANCE

Input file: SSR2.DAT

This case is also available on the Microtran website www.microtran.com [33], and only a
brief description of the network configuration with one simulation result will be given here.
Figure 6.3 shows the network configuration of IEEE’s second benchmark model. The case
chosen from the two systems in [31] consists of a single turbine-generator with high and low
pressure turbine stages and a dc exciter machine.

The electrical side of the generator is again modelled with 3 stator windings, the field
winding, a winding on the quadrature axis to represent eddy current effects, and one damper
winding each on the direct and quadrature axis. The most accurate data conversion to self and
mutual impedances fails on the quadrature axis in this case. An approximate conversion
recommended in [31] is then used for the failing axis. Since [31] uses this approximate conversion
for both axes, this is done here too, to make comparisons with other results easier.

The turbine-generator is connected through a step-up transformer to two parallel


transmission lines, of which one is series compensated. The power system at the receiving end is
represented as a Thevenin equivalent circuit (infinite bus behind impedance). A three-phase
fault is applied on the high side of the generator step-up transformer and the fault clearing time is
varied from 1 cycle to 4 cycles. The resultant torques vary greatly, and depend very much on the
fault clearing time.

Figure 6.4 shows the shaft torque between the synchronous generator and the low
pressure steam turbine, for the case with 17 ms clearing time.

6-4
4 3
2 1

GEN EXC
HP LP
22 kV/500 kV
R = 0.0002
X = 0.02

Bus 2
R1 = 0.0067
SWT R1 = 0.0074
Ro = 0.0186 fault
Ro = 0.022

X1 = 0.0739
Xo = 0.210 X1 = 0.080
Xo = 0.240

Bus C

1/ωC = 0.55 X1-line

Bus 1

Ro = R1 = 0.0014
Xo = X1 = 0.030

Bus 0

Infinite bus

Figure 6.3 Network configuration of second IEEE benchmark model for subsynchrnous resonance

6-5
8.00

6.00

4.00
Shaft torque (x 1,000,000)

2.00

0.00
100 200 300 400 500
Time (ms)
-2.00

-4.00

-6.00

-8.00

MACH 1 -> SHTR 2

Figure 6.4 Torque on the shaft between the synchronous generator and low pressure steam turbine

6-6
7. SATURATION IN CURRENT TRANSFORMERS

7.1 INTRODUCTION

Current transformers are designed to supply precise fractions of the measured current in
the primary winding to instruments, meters and relays connected to the secondary winding. A
current transformer is an iron-core device with a primary winding that consists of a few turns of
large current-carrying capacity. This winding must be sufficiently insulated from the core and
from the secondary winding, so that it can safely withstand the full power-circuit voltage. The
secondary winding consists of many turns of relatively small wire. The turns ratio is normally
designed to supply 5 A to the secondary circuit when the rated full-load current flows in the
primary circuit [36].

The current in the primary winding is determined, for all practical purposes, by the power
circuit itself in which the current transformer is installed. The current transformer impedance
seen from the primary side must therefore be extremely small. Ideally, the secondary winding
should be short-circuited, corresponding to zero “burden”. In actual practice, however, there is a
secondary load impedance, but it can be kept very low, consisting of only one or more
instrument coils. If the secondary load impedance becomes large enough to be significant, then
the burden is said to be too high. A current transformer must never be operated with the
secondary winding open-circuited; otherwise, high voltages would develop that would damage it.
As a standard practice, the secondary winding of a current transformer is short-circuited before
any of the devices are disconnected.

If the primary current becomes too large, the current transformer may go into saturation
and give incorrect readings on the secondary side. This can adversely affect the behaviour of
protective relays connected to the secondary side. It is therefore important to be aware of these
errors that can occur. It may be possible to correct for these errors with data processing
algorithms that model the saturation behaviour [38].

If precautions are taken to model the saturation curves properly, EMTP simulations come
reasonably close to laboratory or field tests [39].
7-1
7.2 COMPARISON WITH FIELD TEST

Data file: CT.DAT

This comparison came from C. W. Taylor of Bonneville Power Administration. The fault
current of a single-phase-to-ground fault was recorded through a current transformer, during a
test on the John Day Transmission Line [37]. The fault current can be simulated with reasonable
accuracy in this case with the R-L circuit shown in Fig. 7.1.

Rp Lp

SW

ip
Vsource

Figure 7.1 Simple circuit for simulating a single-phase-to-ground fault

7.2.1 FAULT CURRENT

For the R-L circuit of Fig. 7.1, a closed-form solution can be found for the primary fault
current ip(t), which can then be used as a current source going into the current transformer. The
voltage source in the primary circuit is

v source ( t ) = Vmax sin(ωt + α ) , (7.1)

525 2
with Vmax = kV , α = 45°, and f = 60 Hz.
3

The primary fault current ip(t) from the exact solution is:

Vmax  − 
t
i p (t ) =  sin(ωt + α − ψ ) − sin(α − ψ ) e  ,
T
(7.2)
R p2 + (ωL p )
2
 

7-2
 ωL p  L
with Rp = 0.8433 Ω and Lp = 0.0433 H, ψ = tan −1   = 87.04 o , and T = p = 0.0513 .
 R  Rp
 p 

The current transformer had a turns ratio of 320. With that ratio, we get the two
components of the current source, referred to the secondary side, as

1
i p ( t )sin usoidal = 81.953 cos(ωt − 132.04°) in A, (7.3)
N

t
1 −
i p ( t )exp onential = 54.88e 0.0513 in A. (7.4)
N

The sinusoidal component is specified in the EMTP input file in the usual way. For the
exponentially declining component, the double exponential surge function (source type 15) of
the form

(
f ( t ) = Fmax eαt − e βt ) (7.5)

can be used, with α = -1/T and β being such a large negative number that eβt is practically zero
for the first and all following time steps. The Microtran Reference Manual recommends β = -
500/∆t, which would be –107 for the chosen time step of 50 µs.

Instead of calculating the fault current from Eq. (7.2) and using it as a current source, one
could also include the primary circuit with a voltage source Vsource behind an Rp-Lp branch to
ground, and let the EMTP calculate the fault current. To convert the primary side to the
secondary side, an ideal transformer with a turns ratio 1:320 would have to be inserted as well.
This model is included in the input file CT.DAT for comparison purposes. The fault current in
the output and plot files appears as “ground TO ground”, because both terminals of the
secondary side were grounded (some EMTP versions may not allow branches that go from
ground to ground, but MicroTran allows it for branches that are coupled to other branches). One
could also convert the three values Vsource, Rp and Lp from the primary to the secondary side,
without adding an ideal transformer.

7-3
7.2.2 EQUIVALENT CIRCUIT FOR CURRENT TRANSFORMER

The equivalent circuit for the current transformer is shown in Fig. 7.2 All quantities are
referred to the secondary side, and the secondary leakage impedance is added to the burden
impedance on the secondary side, for a total resistance of R2 = 2.74 Ω and a total inductance of
L2 = 0.66 mH. The primary leakage impedance can be left off because it is extremely small
compared to the impedances of the power circuit. Even if we assume it is as high as the
secondary impedance Z2, which it is not, the total impedance would still only be 5.5 Ω referred
to the secondary side, or 54 µΩ on the primary side. The relationship between flux and current
on the secondary side came from the manufacturer, in the form of magnetomotive force (MMF)
in A⋅turn as a function of flux in Vs/turn, as shown in Table 7.1. With a turns ratio of 1600:5 (N
= 320), the flux must be multiplied by 320 to obtain the flux linkage in Vs, and the
magnetomotive force MMF must be divided by 320 to obtain currents in A, as included in Table
7.1. The last pair of values was obtained by extending the slope by another 100 A. A resistance

SW

i2

R2
ip
Rc Lm
N
ic im L2

iexc

Figure 7.2 Equivalent circuit for current transformer

Rc of 160 Ω, in parallel with the nonlinear inductance, approximates the iron core and hysteresis
losses.

7.2.3 SIMULATION RESULTS

Figure 7.3 shows the simulation results obtained by C. W. Taylor, together with the field
test results [37]. Using the input file CT.DAT, the same results are obtained, as can be seen in
Figure 7.4. Figure 7.5 shows the flux linkage as a function of the exciting current iexc, with minor
7-4
hysteresis loops. The flux linkage was obtained from the current through an inductance of 1020
H, after multiplying it with 10-20. While a simple model with a constant iron core resistance Rc
cannot produce the remanent flux correctly after the current declines to zero, it does give a crude
approximation for the losses in the iron core. A nonlinear resistance could improve the results
[3,Section 6.6.3].

Table 7.1 Relationship between current and flux

MMF(A⋅turn) i (A) φ(Vs/turn) λ (Vs)

0.0 0.0 0.0 0.0


500 1.5625 0.005 1.6
1200 3.75 0.0056 1.792
3000 9.375 0.0059 1.888
6000 18.75 0.00625 2.000
18000 56.25 0.007 2.24
28000 87.5 0.0071 2.272
187.5 2.3744

Fig. 7.3 Simulation and field test results of C. W. Taylor [37]

7-5
Fig. 7.4 Simulation results from input file CT.DAT

Fig. 7.5 Flux linkage as a function of current

7-6
8. SWITCHING SURGES ON OVERHEAD TRANSMISSION

LINES

8.1 CLOSING OPERATIONS

When an overhead transmission line is energized by closing the circuit breakers at both
ends, there is always a short time during which one circuit breaker has just closed and the other
one is still open. In such a situation, a travelling wave propagates along the line from the sending
end to the open-circuited receiving end. When it gets reflected at the open-circuited receiving
end, the voltage doubles, as shown in Fig. 8.1 for a case where a dc voltage of 1.0 V is applied to
the sending end of a lossless line. Applying a dc voltage is similar to applying an ac voltage
when its instantaneous value is just at its crest value. The frequency of the oscillation is

1
f open end = , (8.1)

with τ being the travel time, because of repeated reflections with a time period of 4τ. In reality, the
square wave shown here for a lossless line would become more and more rounded as time goes on
because of frequency-dependent losses.

Figure 8.1 Voltages at the receiving end of a lossless line after


connecting a dc voltage source of 1.0 V to the sending end

8-1
For a line terminated with a low impedance (less than the surge impedance of the line),
the oscilllation frequency would become

1
f low−Z = (8.2)

(dotted line in Fig. 8.1).

In reality, the overvoltages can be higher than 2.0 p.u. because of the following factors:

1. The sending end does not have a simple voltage source without internal impedance, but is
usually connected to a substation with other transmission lines, transformers etc.

2. The transmission line may still have “trapped charge” on it from a preceding opening
operation (see Section 8.2). Trapped charge means that the shunt capacitances of the line are
still charged up to some voltage level.

3. The three poles of the circuit breaker for phases A, B, C do not close simultaneously. In that
case waves start propagating at different times, and a slower zero sequence travelling wave
propagates as well.

Transmission line insulators must be able to withstand the switching surge overvoltages.
Otherwise, flashovers across insulators can occur during energizations at an unacceptable rate.
Occasional flashovers are tolerable because the overhead line insulation is “self-restoring”, that
is, the arc normally extinguishes without permanent damage to the insulator after the line is de-
energized temporarily and then re-energized again. The situation is different in underground
cables, where breakdown of the solid insulation would create permanent damage. Whether a
flashover will occur does not only depend on the magnitude of the overvoltage, but on its wave
shape and on the volt-time characteristic of the insulator as well. Flashover occurs, with a certain
probability, at the instant where the wave crosses the volt-time characteristic vBREAKDOWN of the
insulation, as shown in Fig. 8.2. For testing purposes, switching surge wave shapes such as
250/2500 µs for circuit breaker tests have been agreed upon [40], but in a real power system, the
wave shapes are rarely of a standard form. EMTP studies therefore usually ignore the wave
shape and look at the maximum magnitudes only.

8-2
Since the magnitudes depend on the instants when the circuit breaker poles close with
respect to the voltage source sine curve, and on the “pole spread” (times when the three poles
close with respect to each other), hundreds of simulations are usually performed with different
closing times to arrive at a statistical distribution of the overvoltages.

An example for a cumulative frequency distribution obtained from computer simulations


and from transient network analyzer studies is shown in Fig. 8.3, for the case discussed later in
Section 8.4. The graph in Fig. 8.3 has to be read as follows: at 10 % probability, the overvoltage
is approximately 2.6 p.u. (average among the curves). This means that in 10 % of the switching
operations, a 2.6 p.u. overvoltage may be reached or exceeded, or 90 % of the switching
operations will have overvoltages of 2.6 p.u. or below. If only one overvoltage value is
mentioned, it usually refers to the 2 % value. From a practical standpoint, a transmission line
designed for 2.6 p.u. overvoltage would require insulators that are approximately 2.6-times
longer than needed for the rated steady-state operating voltage. The towers would have to be
accordingly larger as well.

Fig. 8.2 Volt-time characteristic of insulation

Many EMTP versions have special switch types for running a large number of cases in
which the closing times are automatically varied. The output will then include the statistical
distribution of the overvoltages as well. The closing times are either varied statistically, or
systematically (see Section 10.2 of [3]).

8-3
Fig. 8.3 Cumulative frequency distribution of receiving end voltages from 100 digital computer and transient
network analyzer simulations [43]

8.2 RECLOSING OPERATIONS

Input file: SHUNT_L.DAT

The highest overvoltages can occur during reclosing operations, that is, if a transmission
line is first de-energized (switched off), usually because of a fault in one phase, and within a
short time (typically less than 1 s) re-energized again. When the line is reclosed, the line
capacitances of the unfaulted phases are still charged up, unless equipment is connected to the
line that could drain the charge off. If it so happens that the voltage in the feeding network
behind the circuit breaker has the opposite polarity of the trapped charge at the closing instant, a
voltage change of up to 2.0 p.u. enters the sending end of the line. This is illustrated in Fig. 8.4
8-4
for the lossless line of Fig. 8.1, with a trapped charge of –1.0 V. The 2 V step change doubles up
at the receiving end to 4 V, and after addition of the –1.0 V trapped charge, creates a 3.0 pu.
overvoltage.

Fig. 8.4 Voltage at the receiving end of a lossless line with trapped charge of –1.0 V

In the MicroTran version of the EMTP, the easiest way to specify trapped charge in the
input file is to read them in as user-supplied initial conditions. Both the voltages on the line, and
the currents going into the line (which are zero in this case) must be specified. For more details,
consult the MicroTran Reference Manual. Other EMTP versions may be using other approaches,
such as having voltage sources of very low frequency connected to the line during the steady-
state solution.

In the example of Fig. 8.4 with trapped charge, it is fairly easy to predict the voltage
across the circuit breaker contacts 10’s of milliseconds ahead if controlled closing is to be used,
where the electrical closing is to take place close to zero voltage across the circuit breaker
contacts. On the source side, the voltage is more or less sinusoidal at power frequency, and on
the line side it is a slowly decaying dc voltage.

If the line is shunt compensated with shunt reactors, this prediction becomes more
complicated, because of oscillations with a beat frequency on the line side, after the circuit
breaker opens [47]. Fig. 8.5 shows this beat phenomenon after circuit breaker opening for the

8-5
same case that was used in Fig. 8.4, except with 80 % shunt compensation and a 60 Hz voltage
source in this case. From surge impedance Z = 250 Ω and travel time τ = 1 ms of the 300 km
long transmission line, it follows that the shunt capacitance per unit length, multiplied by length,
becomes C = C' ⋅l = τ / Z = 4 ⋅ 10 −6 F . Half of this capacitance is compensated at the sending end,
and the other half at the receiving end. For 80 % shunt compensation, we get the shunt reactor
1
reactance value for each end as ωL = 2.0 ⋅ = 1657.86 Ω. .
0 .8 ω C

Fig. 8.5 Beat phenomenon with 80 % shunt compensation

8.3 REDUCTION OF SWITCHING SURGE OVERVOLTAGES

The incentive for reducing switching surge overvoltages came with the advance to higher
operating voltages, such as 500 kV. At much lower voltages, the costs of a transmission line may
not be all that different if the insulators are fairly long to withstand high (uncontrolled) switching
surges. At higher voltage levels such as 500 kV, there is a cost incentive to reduce the insulator
length and the tower size if measures can be taken to keep the switching surges lower.

There are a number of ways to reduce switching surge overvoltages:

1. Closing (pre-insertion) resistors: if a resistance R approximately equal to the surge


impedance Zsurge is inserted in series with the line, then the forward wave will see the two
resistances R and Zsurge in series before reflections return. If the pre-insertion resistance R
8-6
and the surge impedance Zsurge are equal, then only 50 % of the voltage will appear at the
sending end when the wave enters the line. As soon as reflections come back, the situation
becomes more complicated. Also, the resistors cannot remain in series with the line for a
long time because of unacceptably high losses. Closing resistors are typically shorted out
after 8 ms or so. A possible drawback of closing resistors is the more complicated contact
structure of the circuit breaker, with one auxiliary contact and its closing resistor and the
main contact. This may reduce the overall reliability of the circuit breaker.

2. Controlled closing: with modern circuit breakers, it is also possible to use “controlled
closing”, whereby the contacts close at an instant when the voltages on both sides of the
contact are more or less identical [15; 48-51]. It requires circuit breakers with closing times
that are predictable. The disadvantage of this method is that if the circuit breaker re-closes
into a fault, controlled closing produces maximum offset in the fault current.

3. Surge arresters: with zinc oxide surge arresters, it has become possible to use surge arresters
at both ends of the line. Keep in mind, however, that the highest overvoltage may appear
some distance from either the sending or receiving end, rather than at the terminals [41].

4. Draining off trapped charge: to reduce reclosing overvoltages when closing into trapped
charge, inductive potential transformers can be used to drain off the charge, possibly with
resistance grounding for additional damping. Properly rated grounding switches can also be
used in cases where auto-reclosing (automatic reclosing after a predetermined time) is not
used.

Fig. 8.6 shows the effectiveness of various techniques for reducing the switching surge
overvoltages, for a case where the circuit breaker recloses into a shunt compensated line ([51]
taken from [47]). Explanations for this figure:

• Staggered Closing: Controlled closing at instants that are best for each phase;

• One-Step PIR 400 Ω: One-step closing (pre-insertion) resistor. There are also circuit
breakers with two-step closing resistors (e.g., first resistor with 1000 Ω, second
resistor with 400 Ω, both inserted for 10’s of ms), but these circuit breakers are
probably more maintenance prone.
8-7
• CS + Three 1.5 pu MOSAs: Controlled closing, with three metal oxide surge arresters
rated at 1.5 pu, at both ends and in the middle of the line.

• Two MOSAs + Staggered C: Two metal oxide surge arresters at both ends of the line,
and staggered (controlled) closing.

Fig. 8.6 Statistical overvoltage distribution for reclosing into a shunt-compensated transmission line ([47] and [51])

8.4 COMPARISON OF SIMULATION RESULTS BY CIGRÉ

Input files: CIGRE.DAT, CIGRE1P.DAT, CIGRE10P.DAT, CIG_TRAP.DAT

CIGRÉ Working Group 13.05 defined a benchmark model for switching surge studies,
that was originally used to compare results from transient network analyzers [42]. Comparisons
with computer simulations were done later as well.

8.4.1 POWER SYSTEM CONFIGURATION

The test system is shown in Figure 8.7. The network data provided by CIGRÉ is as
follows:

1. Source: The source voltage is 1.0 p.u. (peak), f = 50 Hz.

8-8
2. Feeding network impedance: The sum of the source impedance and transformer impedance,
referred to the breaker side, is: Rpos = Rzero = 6.75 Ω; Xpos = Xzero = 127 Ω at 50 Hz.

3. Transmission line parameters: Z’pos = 0.04 + j0.318 Ω/km at 50 Hz, C’pos = 11.86 nF/km;
Z’zero = 0.26 + j 1.015 Ω/km at 50 Hz, C’zero = 7.66 nF/km;
length = 202.8 km.

4. Circuit breaker: the closing times are given with respect to the instant when the voltage in
phase A goes through zero from positive to negative; TCLOSE-A = 3.05 ms, TCLOSE-B = 8.05 ms,
TCLOSE-C = 5.55 ms.

5. Step size: a step size of ∆t = 50 µs is recommended.

Source Breaker Transmission


line

Figure 8.7 CIGRE benchmark model for switching surges

8.4.2 LINE MODELS

For the line, the distributed parameter line model with constant parameters is used,
because only constant parameters were provided. Since transient network analyzers use a
cascade connection of three-phase nominal π-circuits, that model was used as well for
comparison purposes. It is assumed that the line is balanced (perfectly transposed), again because
that was the only data available. On transient network analyzers, balanced lines are modelled
with 4 uncoupled branches. One branch each is used for phases A, B, C with positive sequence
parameters. The fourth common return branch has parameters that are equal to the mutual
impedance Zm and mutual capacitance Cm. Phase voltages must be measured with respect to this
return branch. This uncoupled approach comes out of transforming the equations for the series
voltage drop

8-9
 ∆V A   Z s Zm Z m  I A 
 ∆V  =  Z Zs Z m   I B  (8.3)
 B  m
∆VC   Z m Zm Z s   I C 

into the uncoupled equations

 ∆V A  (Z s − Z m )   I A   Z m (I A + I B + I C )
 ∆V  =  (Z s − Z m )   I  +  Z (I + I + I ) , (8.4)
 B   B  m A B C 

∆VC   (Z s − Z m )  I C  Z m (I A + I B + I C )
   

where the second term expresses the voltage drop through the common return branch with
impedance Zm and zero sequence current Izero = IA + IB + IC. In the first term, the impedance in
the diagonal elements is the positive sequence impedance Zpos = Zs – Zm. For limitations of the
four-branch approach, see Section 4.2.2 of [3].

In order to use a π-circuit model in Microtran, the positive and zero sequence parameters
must be converted to self and mutual parameters with

Zs =
1
(Z zero + 2Z pos ); C s = 1 (C zero + 2C pos ), (8.5)
3 3

Zm =
1
(Z zero − Z pos ); C m = 1 (C zero − C pos ) (8.6)
3 3

Some EMTP versions, as well as the preprocessor MTDATA of Microtran, allow input in
the form of sequence parameters as well. The transient network analyzer approach with four
uncoupled branches could be used as well, but 3*3 matrices in phase quantities are more
straightforward. The author of this document verified that the results from both approaches are
identical in EMTP simulations.

8.4.3 SIMULATION RESULTS

Figure 8.8 shows the transient network analyzer results for phase B, with the EMTP
simulation results superimposed. The results with a distributed parameter line from input file
CIGRE.DAT in Fig. 8.9, and a cascade connection of 10 π-circuits from input file

8 - 10
CIGRE10P.DAT are almost identical. If only one π-circuit is used (CIGRE1P.DAT), the results
become less accurate, as shown in Fig. 8.10, but not as bad as one might expect. The reason is
the filtering effect of the transformer inductance. If the line were not fed from an inductive
source, as in this case, results with one π-circuit would probably be unacceptable.

Fig. 8.8 Overvoltage at receiving end in phase B; computer results (dashed line) superimposed on family of
curves from transient network analyzer results; time count starts when wave arrives at receiving end

As mentioned in Section 8.2, the overvoltages can become higher if there are trapped
charges left on the transmission line after a preceding opening operation. The severity of the
overvoltages with trapped charges depends on the polarity of the trapped charges and the
switching instants of the breaker poles. In this CIGRE example, the peak overvoltages without
trapped charge are between 2.1 and 2.3 pu. For trapped charges of 0.9, 0.8, -0.8 p.u. on phases A,
B, C, the overvoltages would be reduced to between 1.4 and 1.5 p.u. If the polarities of the
trapped charges are reversed to –0.9, -0.8, 0.8 p.u., the overvoltages would increase to between
3.1 and 3.5 p.u. These values are summarized below, and the wave shape for trapped charges of
–0.9, -0.8, 0.8 p.u. is shown in Fig. 8.11.

Trapped charges (p.u.) Overvoltages (p.u.)


A B C A B C

0.0 0.0 0.0 2.068 2.166 2.287


0.9 0.8 -0.8 1.368 1.538 1.342
-0.9 -0.8 0.8 3.086 3.172 3.469
8 - 11
Fig. 8.9 Overvoltage at receiving end in phase B; distributed parameter line

Fig. 8.10 Comparison between distributed parameter line and one π-circuit (phase B)

8 - 12
Fig. 8.11 Overvoltage at receiving end in phase B with trapped charges; distributed parameter line

The following additions are planned:

♦ Section 8: More cases on switching surges (until finished, see 2nd edition of Case Studies).

♦ Section 9: Fault overvoltages (until finished, see 2nd edition of Case Studies).

♦ Section 10: Transient recovery voltage (until finished, see 2nd edition of Case Studies).

Possible additions from the collection of Microtran data files:

♦ Lightning surges.

♦ Power electronics cases (my own and from [11]).

♦ Start-up of induction motors and synchronous motors.

♦ Back-to-back capacitor switching (case from Bartholomew).

8 - 13
REFERENCES

GENERAL REFERENCES

In addition to the specific references quoted below, advice for Electromagnetic Transient Studies
can also be found in the following publications:

IEEE Publications:

J. A. Martinez-Velasco, editor, Computer Analysis of Electric Power System Transients. IEEE


Press, Piscataway, NJ, U.S.A., 1997. Collection of papers on 619 pages.

IEEE PES Special Publication, Modeling and Analysis of System Transients. IEEE Catalog No.
99TP133-0, IEEE Operations Center, Piscataway, NJ, U.S.A., 1998. Put together by a Working
Group chaired by A.J.F. Keri:

i. Modeling and Analysis of System Transients Using Digital Programs -


Introduction (A.J.F. Keri, A.M. Gole)

1. Digital Computation of Electromagnetic Transients in Power Systems: Current


Status (J.A. Martinez-Velasco)

2. Modeling Guidelines for Power Electronics in Electric Power Engineering


Applications (K.K. Sen and L. Tang, H. W. Dommel, K.G. Fehrle, A.M. Gole,
E.W. Gunther, I. Hassan, R. Iravani, A.J.F. Keri, R. Lasseter, J.R. Marti, J.A.
Martinez, M.F. McGranaghan, O.B. Nayak, C. Nwankpa, P.F. Ribeiro)

3. Modeling Guidelines for Low Frequency Transients (R. Iravani, A.K.S.


Chandhury, I.D. Hassan, J.A. Martinez, A.S. Morched, B.A. Mork, M. Parniani,
D. Shirmohammadi, R.A. Walling)

4. Modeling Guidelines for Switching Transients ( D.W. Durbak and A.M Gole,
E.H. Camm, M. Marz, R.C. Degeneff, R.P. O'Leary, R. Natarajan, J.A. Martinez-
Velasco, Kai-Chung Lee, A. Morched, R. Shanahan, E.R. Pratico, G.C. Thomann,
B. Shperling, A.J.F. Keri, D.A. Woodword, L. Rugeles, V. Rashkes, A. Sarshar)

5. Modeling Guidelines for Fast Front Transients (A.F. Imece, D.W. Durbak, H.
Elahi, S. Kolluri, A. Lux, D. Mader, T.E. McDermott, A. Morched, A.M. Moussa,
R. Natarajan, L. Rugeles, E. Tarasiewicz)

6. Modeling Guidelines for Very Fast Transients in Gas Insulated Substations (J.A.
Martinez and D. Povh, P. Chowdhuri, R. Iravani, A.J.F. Keri)

R-1
7. Modeling and Analysis of Transient Performance of Protection SystemsUsing
Digital Programs (A.K.S. Chaudhary and R.E. Wilson, M.T. Glinkowski, M.
Kezunovic, L. Kojovic, J.A. Martinez)

8. Bibliography on Modeling of System Transients Using Digital Programs (J.A.


Martinez-Velasco and T. E. Grebe)

IEEE Power Engineering Society, Tutorial on Electro- magnetic Transient Program Applications
to Power System Protection. A. Tziouvaras, Course Coordinator. IEEE Catalog No. 01TP150.

IEEE PES Task Force on Data for Modeling System Transients, “Parameter Determination for
Modeling System Transients – Part I: Overhead Lines; Part II: Insulated Cables; Part III:
Transformers; Part IV: Rotating Machines; Part V: Surge Arresters; Part VI: Circuit Breakers;
Part VII: Semiconductors”, IEEE Trans. on Power Systems, Vol. 20, pp. 2038-2094, July 2005.

Books (compiled with help from Dr. Luis Naredo):

H. H. Skilling, Transient Electric Currents. McGraw-Hill Book Company, Inc., 1937.

H. A. Peterson, Transients in Power Systems. Dover Publications, Inc., New York, 1966 (ISBN
0-486-61685-1).

R. Rüdenberg, Electrical Shock Waves in Power Systems. Harvard University Press, 1968.

J. P. Pickford, N. Mullineux, and J. R. Reed. Computation of Power-System Transients. IEE


Monograph Series 18, Peter Peregrinus Ltd., London, UK, 1976.

W. D. Humpage, Z-Transform Electromagnetic Transient Analysis in High Voltage Networks.


IEE Power Engineering Series 3, Peter Peregrinus Ltd., London, UK, 1982 (ISBN 0-906048-79-
6).

A. Greenwood, Electrical Transients in Power Systems, 2nd edition. John Wiley & Sons, 1992.

P. Chowdhuri, Electromagnetic Transients in Power Systems. Research Studies Press LTD, 1996
John Wiley and Sons, Inc. (ISBN 0 86380 180 3).

L. van der Sluis, Transients in Power Systems. John Wiley and Sons, LTD, 2001 (ISBN 0 471
48639 6).

N. Watson and J. Arrillaga, Power Systems Electromagnetic Transients Simulation. The


Institution of Electrical Engineers, United Kingdom, 2003.

L.C. Zanetta Jr., Transitórios Eletromagnéticos em Sistemas de Potencia (in Portuguese). Editora
da Universidade de Sao Paulo, Sao Paulo - SP, Brazil, 2003.

R-2
Antonio E. A. Araújo and Washington L. A. Neves, Transitórios Eletromagnéticos em Sistemas
de Energia (in Portuguese). Editora de Universidade Federal de Minas Gerais, Brazil, 2005.

H. W. Dommel, EMTP Theory Book, 2nd edition. Microtran Power System Analysis Corp.,
Vancouver, Canada, 1992, latest update April 1996. There seems to be an ATP edition as well.

SPECIFIC REFERENCES

[1] IEEE Working Group, “Electrostatic effects of overhead transmission lines”, IEEE Trans.
Power App. Syst., Vol. PAS-91, pp. 422-433, March/April 1972.

[2] IEEE Working Group, “Electromagnetic effects of overhead transmission lines. Practical
problems, safeguards, and methods of calculation”, IEEE Trans. Power App. Syst., Vol.
PAS-93, pp. 892-904, May/June 1974.

[3] H.W. Dommel, EMTP Theory Book, Second Edition, Microtran Power System Corporation,
Vancouver, BC, Canada, May 1992.

[4] Microtran Power System Corporation, Microtran Reference Manual, Vancouver, BC,
Canada, 1992.

[5] F. Castellanos and J. Martí, “Full frequency-dependent phase domain transmission line
model”, IEEE Trans. Power Syst., Vol. 12, pp. 1331-1339, Aug. 1997.

[6] Ting-Chung Yu and José R. Martí, “A robust phase-coordinates frequency dependent


underground cable model (zCable) for the EMTP”, IEEE Trans. Power Delivery, Vol. 18,
pp. 189-194, Jan. 2003.

[7] M.R. Iravani, W.C. Boaventura, A. Semlyen, and A. Lopes, “Robust sparse network
equivalent for large systems: Part I – Methodology; Part II – Performance evaluation”, to
appear in IEEE Trans. on Power Systems, February 2004.

[8] J.E. Brittain, “Thevenin’s theorem”, IEEE Spectrum, p. 42, March 1990.

R-3
[9] S.A. Annestrand and G.A. Parks, “Bonneville Power Administration’s 1100/1200 kV
transmission line project”, IEEE Trans. Power App. Syst., Vol. 96, pp. 357-366,
March/April 1977.

[10] C.A.F. Cunha and H.W. Dommel, “Computer Simulation of Field Tests on the 345 kV
Jaguara-Taquaril line”, (in Portuguese), Paper BH/GSP/12, Presented at “II Seminario
Nacional de Producao e Transmissao de Energia Eletrica” in Belo Horizonte, Brazil, 1973
(Portuguese paper or English translation by D.I. Cameron).

[11] N. Mohan, Computer Exercises for Power Electronics Education. Department of Electrical
Engineering, University of Minnesota, Minneapolis, MN, U.S.A., 1990.

[12] H.W. Dommel, A. Yan, Shi Wei, "Harmonics from transformer saturation", IEEE Trans.
Power Delivery, Vol. 1, pp. 209-215, April 1986.

[13] E.P. Dick and W. Watson, “ Transformer models for transient studies based on field
measurements”, IEEE Trans. Power App. Syst., Vol. PAS-100, pp. 409-419, Jan. 1981.

[14] Microtran Factsheet No. 3, “Transformer Inrush Current Test; Case 1”, April 2003 (available
on website www.microtran.com; click on “Tech Spot”).

[15] CIGRE Task Force13.00.1, “Controlled switching – a state-of-the-art survey”, Part I, Electra
No. 163, pp. 65-97, Dec. 1995. Part II, ELECTRA No. 164, pp. 39-61, Febr. 1996.

[16] D. Povh and W. Schulz, “Analysis of overvoltages caused by transformer magnetizing


inrush current”, IEEE Trans. Power App. Syst., Vol. PAS-97, pp. 1355-1365, July/Aug.
1978.

[17] CIGRE Working Group 33.02, Guidelines for representation of network elements when
calculating transients. Technical Brochure CE/SC GT/WG 02, 1990.

[18] A. Greenwood, Electrical Transients in Power Systems, Second Edition. John Wiley &
Sons, New York, N.Y., 1991.

R-4
[19] N. Germay, S. Mastero, and J. Vroman, "Review of ferroresonance phenomena in high-
voltage power systems, and presentation of a voltage transformer model for predetermining
them", CIGRE Report No. 33-18, Paris, 1974.

[20] H.S. Bronzeado and J.C. de Oliveira, “The influence of tap position on the magnitude of
transformer inrush current”, Proc. 3rd International Conference on Power Systems
Transients, Budapest, Hungary, June 20-24, 1999, 5 pages.

[21] H.S. Bronzeado, P.B. Brogan, and R. Yacamini, “Harmonic analysis of transient currents
during sympathetic interaction”, IEEE Trans. on Power Systems, Vol. 11, pp. 2051-2056,
Nov. 1996.

[22] H.S. Bronzeado, “CHESF R&D Project mitigates transformer inrush currents”, (in
Portuguese), Internal Report issued by CHESF/DOEL, Recife, PE, Brazil, December 2003.

[23] J.R. Marti and A.C. Soudack, “Ferroresonance in power systems: fundamental solutions",
IEE Proceedings-C, Vol. 138, pp. 321-329, July 1991.

[24] K. Al-Anbarri, R. Ramanujam, R. Saravanaselvan, and K. Kuppusamy, “Effect of iron core


loss nonlinearity on chaotic ferroresonance in power transformers”, Electric Power Systems
Research, Vol. 65, pp. 1-12, April 2003.

[25] W.L.A. Neves and H.W. Dommel, "On modeling iron core nonlinearities", IEEE Trans.
Power Systems, Vol. 8, pp. 417-425, May 1993.

[26] E.J. Dolan, D.A. Gillies, and E.W. Kimbark," Ferroresonance in a transformer switched
with an EHV line", IEEE Trans. Power App. Syst., Vol. 91, pp. 1273-1280, May/June
1972.

[27] J.R. Marti and J. Lin, "Suppression of numerical oscillations in the EMTP", IEEE Trans.
Power Systems, vol.4, pp. 739-749, May 1989.

[28] J. Lin and J.R. Marti, "Implementation of the CDA procedure in the EMTP", IEEE
Trans.Power Systems, vol. 5, pp. 394-402, May 1990.

R-5
[29] IEEE Task Force (B. Mork, lead author), “Modeling and analysis guidelines for slow
transients – Part III: The study of ferroresonance”, IEEE Trans. Power Delivery, Vol. 15,
pp. 255-265, Jan. 2000.

[30] IEEE Task Force, "First benchmark model for computer simulation of subsynchronous
resonance", IEEE Trans. Power App. Syst., vol. PAS-96, pp. 1565-1572, Sept./Oct. 1977.

[31] IEEE Task Force, "Second benchmark model For computer simulation of subsynchronous
resonance", IEEE Trans. Power App. Syst., Vol. PAS-104, pp. 1057-1066, May 1985.

[32] Microtran Factsheet No. 1, “Subsynchronous Resonance - Test Case 1”, April 2003 (available
on website www.microtran.com; click on “Tech Spot”).

[33] Microtran Factsheet No. 2, “Subsynchronous Resonance - Test Case 2”, April 2003 (available
on website www.microtran.com; click on “Tech Spot”).

[34] V. Brandwajn, Synchronous generator models for the analysis of electromagnetic transients.
Ph.D. thesis, University of British Columbia, Vancouver, Canada, 1977.

[35] G. Gross and M. Hall, “Synchronous machine and torsional dynamics simulation in the
computation of electromagnetic transients”, IEEE Trans. Power App. Syst., Vol. PAS-97, pp.
1074-1086, July/Aug. 1978.

[36] Howard M. Berlin and Frank C. Getz, Jr. Principles of Electronic Instrumentation And
Measurement, Merrill Publishing Company, Toronto, 1988.

[37] C. W. Taylor, Simulation of current transformers. Internal Memorandum, Bonneville Power


Administration, Portland, OR, U.S.A. Note: In this report, the values for Rp and Lp are three
times larger because they were used to find the symmetrical component currents Ipos = Ineg =
Izero first, whereas here we calculate the three times larger phase current Ia = 3Ipos directly.
Also, the polarity of the field test oscillograms is reversed compared with the simulation
results.

[38] Yong.-Cheol Kang, Seung-Hun Ok, and Sang-Hee Kang, “A CT saturation detection
algorithm”, IEEE Trans. Power Delivery, Vol. 19, pp. 78-85, Jan. 2004.

R-6
[39] M. Kezunović, LO. J. Kojović, A. Abur, C. W. Fromen, D. R. Sevcik, F. Phillips,
“Experimental evaluation of EMTP-based current transformer models for protective relay
transient studies”, IEEE Trans. Power Delivery, Vol. 9, pp. 405-413, Jan. 1994.

[40] IEEE Standard Test Code for High-Voltage Air Switches. IEEE C37.34e. The Institute of
Electrical & Electronics Engineers, New York, N. Y., 1979.

[41] L. Marti, and H. W. Dommel, “Calculation of voltage profiles along transmission lines”,
IEEE Trans. Power Delivery, Vol. 12, pp. 993-998, April 1997.

[42] CIGRÉ Working Group 13.05, “The calculation of switching surges. Part I. A comparison
of transient network analyzer results”, ELECTRA No. 19, pp. 67-78, Nov. 1971.

[43] CIGRÉ Working Group 13.05, “The calculation of switching surges. Part II. Network
representation for energization and re-energization studies on lines fed by an inductive
source. ELECTRA No. 32, pp. 17-42, 1974.

[44] J. H. Brunke, and K. J. Fröhlich, “Elimination of transformer inrush currents by controlled


switching, Part I”, IEEE Trans. Power Delivery, Vol. 16, pp. 276-280, April 2002.

[45] J. H. Brunke, and K. J. Fröhlich, “Elimination of transformer inrush currents by controlled


switching, Part II”, IEEE Trans. Power Delivery, Vol. 16, pp. 281-285, April 2002.

[46] E. Portales and A. Mercier, on behalf of CIGRÉ Working Group A3.07, “Controlled
switching of unloaded power transformers”, ELECTRA No. 212, pp. 39-47, Feb. 2004.

[47] K. Froehlich, C. Hoelzl, M. Stanek, A.C. Carvalho, W. Hofbauer, P. Hoegg, B.L. Avent,
D.F. Peelo, J.H. Sawada, “Controlled closing on shunt reactor compensated transmission
lines - Part I: Closing control device development - Part II: Application of closing control
device for high-speed autoreclosing on BC Hydro 500 kV transmission line”, IEEE Trans.
Power Delivery, Vol. 12, No. 2, pp. 734-746, April 1997.

[48] CIGRE Working Group 13.07, “Controlled switching of HVAC circuit breakers – Guide for
application – Part 1”, ELECTRA, No. 183, pp. 43-73, April 1999.

R-7
[49] CIGRE Working Group 13.07, “Controlled switching of HVAC circuit breakers – Guide for
application – Part 2”, ELECTRA, No. 185, pp. 37-57, Aug. 1999.

[50] CIGRE Working Group 13.07, “Controlled switching of HVAC circuit breakers – Planning,
specification and testing of controlled switching systems”, ELECTRA No. 197, pp. 23-33,
Aug. 2001.

[51] CIGRE Working Group 13.07, “Controlled switching of HVAC circuit breakers –Benefits
and economic aspects”, ELECTRA No. 217, pp. 37-47, Dec. 2004.

R-8

You might also like