2

C H A P T E R
EMTP: Modeling and
Simulation Capabilities

2.1 INTRODUCTION

We begin this chapter with a list of various digital computer programs currently in use for
computation of electromagnetic transients in power systems [1, 2]. We will then give a brief
history of EMTP and the versions of it currently existing. This is followed by the power system
component-modeling capabilities that are provided by a popular version of EMTP, Microtran.
The chapter concludes with the studies that can be carried out using EMTP and certain useful
special capabilities.
The term EMTP is used in this book for various versions of EMTP including Microtran.
Where it specifically refers to Microtran should be clear from the context. Also, at some places,
the words Microtran and mt are used interchangeably. mt refers to the transients simulation
module of the whole Microtran package.

2.2  DIGITAL COMPUTER PROGRAMS AVAILABLE

FOR COMPUTATION OF TRANSIENTS [1]

We give below the origin/attempts/existing computer programs for computation of transients in

electrical power systems:
• Technical University Munich 1963 (first published May 1964)
• Bonneville Power Administration (BPA) EMTP. It is a non-commercial version and was
distributed free of cost because of Freedom of Information Act in the USA.
• University of British Columbia (UBC) Microtran®. It is a commercial version owned by
UBC and is being distributed by Microtran Power System Analysis Corporation, Vancouver,
Canada
• NETOMAC Siemens; commercial.
• Morgat (Electricite de France); commercial.
• PSIM (H. Jin); commercial. Primarily for power electronics studies.
• SABER; for power electronics studies, commercial.
• SPICE, PSPICE; commercial. Occasionally used for power electronics.
8  Computational Electromagnetic Transients

about the contributions from the UBC Research Group, most of which found their way into
other versions such as BPA EMTP, DCG/EPRI EMTP (and later EMTP-RV) and ATP. Following
are the partial list of contributions:
• TACS of Laurent Dube, Transient Analysis of Control Systems [5]
• Multiphase untransposed transmission line with constant parameters of K. C. Lee [6]
• Frequency-dependent transmission line of J. R. Marti [7]
• Three-phase transformer models of H. W. Dommel and I. I. Dommel [8]
• Synchronous machine model of V. Brandwajn, Type 59 [9]
• Synchronous machine data conversion of H. W. Dommel [10]
• Underground cable models of L. Marti [11]
• New line constants program and improved frequency-dependent line model of J. R. Marti
[7]
EMTP with extensions to detailed HVDC controls was developed by Manitoba HVDC
Research Centre, Canada. It is a commercial version and is called EMTDC. TACS model is
not currently available in Microtran version of EMTP.
Caution:  Some of the power system software vendors in India also supply their own version
of digital computer program for transients and call it as EMTP. It must be mentioned
here that these programs suffer from unacceptable limitations and the validation methods
used for the results they produce may be questionable.
A note on terminology
Traditionally, the digital computer program for computation of electrical parameters of a
transmission line is called Line Constants Program. With the implementation of frequency
dependence of parameters it is more appropriate to call the program as Line Parameter Evaluation
Program. However, the usage of the name Line Constants Program still continues. In this book we
will use both the terminologies with the realization that they refer to one and the same program.

2.4  MODELING CAPABILITIES OF EMTP (MICROTRAN VERSION) [12]

In this section we provide basic explanation of models provided by Microtran. For specific details
such as input data requirement, etc., the reader should consult Reference 12. The readers who
are not familiar with line modeling, wave propagation and related topics should first go over
the theory behind line modeling presented in Chapter 3.

2.4.1  Network Branch Elements

2.4.1.1  Lumped Elements
2.4.1.1.1 Series R, L, C
This is to be used for the following types of branches without coupling. The branch can be a
“full” R-L-C series connected or with one or more of the elements missing.
• Lumped resistance is described by the equation
 EMTP: Modeling and Simulation Capabilities  9

v = R · i (2.1)
where v and i are the current voltage across and current through the resistance and R is
the value of the resistance.
• Lumped inductance is described by the equation
v = L di/dt (2.2)
where v and i are the current voltage across and current through the inductance and L
is the value of the inductance.
• Lumped capacitance is described by the equation
i = C dv/dt (2.3)
where v and i are the current voltage across and current through the capacitance and C
is the value of the capacitance.
The energy storage elements which are described by differential equations are actually
represented by their “discretized equivalents.” The discretization is done by application of the
trapezoidal rule of numerical integration and is explained in Chapter 10. The R-L combination
can be used to represent a short transmission line (<80 km).
2.4.1.1.2  Circuits
Single- and multiphase p circuit models are provided in EMTP to allow representation of
transmission lines and transformers when represented as mutually coupled circuits.
The single-phase p model can be used to represent a balanced (transposed) transmission
line, in which case the required data will be the positive sequence parameters, Rpos, Lpos and
Cpos. See Fig. 2.1.

Fig. 2.1 Single-phase p model.

The positive sequence parameters can be computed for a practical tower configuration using a
Line Parameter Evaluation Program (LPEP) or formulas suitable for hand computations. Theory
behind LPEP is covered in Chapter 3 and the detailed explanation in Chapter 5.
The multiphase p enables the user to represent the coupling between branches, for
example, interphase coupling of a three-phase transmission line or two three-phase circuits on
the same tower or two lines in the same right-of-way. The multiphase p model assumes that
the transmission line by design is constructed symmetrically. In this scalars Rpos, Lpos and Cpos
become coupled symmetric matrices [R], [L] and [C] in the phase domain, i.e., in abc coordinates.
Figure 2.2 shows the three-phase p model connected between two three-phase nodes k (k1, k2, k3)
and m (m1, m2, m3).
10  Computational Electromagnetic Transients

Fig. 2.2 Three-phase p model.

We can visualize the above three-phase multi-p as a “matrix connection” diagram shown
in Fig. 2.3.

Fig. 2.3  Matrix connection diagram of three-phase p model.

Each of the matrices is symmetrical and has the form as shown below for [R]
æR11 symmetric
[R] = R21 R22 ÷ (2.4)
÷
R31 R32 R33 ø

The differential equations describing the p section are given by
dikm
[vk] – [vm] = [R] [ikm] + [L]
dt
1 dvk
[ik] = [ikm] + [C ]
2 dt
1 dvk
[im] = [C ] – [ikm]
2 dt
(2.5)
The interpretations of the elements are as follows:
Rii + j w Lii = self impedance of branch i (impedance of loop formed by branch i and ground
return”)
 EMTP: Modeling and Simulation Capabilities  11

Rik + j w Lik = mutual impedance between branches i and k; Rik mutual resistance introduced
by the presence of ground and it expresses a phase shift between induced voltage
and inducing current. Rik π 0 with nonzero earth resistivity. [See Section 3.5. for
mutual impedance in the presence of ground.]
Cii = sum of all capacitances connected to the nodes at both ends of branch i;
Cik = negative value of capacitance from branch i to branch k
[vk] = [vk1 vk2 º]; [ik] = [ik1 ik2 º]; [ikm] = [ik1, m1 ik2, m2 º], [imk] is similar to [ikm]
Note that for a three-phase line the suffixes 1, 2 and 3 correspond to phases-a, b and c.
The impedance ([R] + jw [L]) and capacitance matrix elements for a three-phase line model
can be obtained from the corresponding sequence quantities from the following relationships:
1
Z11 = Z22 = Z33 = (Zzero + 2 Zpos )
3
1
Z12 = Z13 = Z23 = (Zzero + Zpos )
3
1
C11 = C22 = C33 = (Czero + 2 Cpos )
3
1
C12 = C13 = C23 = (Czero + Cpos )
3
(2.6)
where Zzero is the zero sequence impedance and Zpos is the positive sequence impedance which
is equal to negative sequence impedance for a static element. The derivation of the relationships
given in Eq. (2.6) is given in Chapter 3.
For computation of parameters of multiphase transmission lines or two lines in the same
right-of-way we resort to Line Constants Program, mtLine which is included in the Microtran
package.
For steady-state computations such as power flow or for EMTP steady-state solution, the
nominal p representation can be used. The validity of the nominal p model depends on the type
of branch element, overhead line or underground cable, and length. The guidelines are given
below [13]:
• For overhead lines, the length up to which nominal p representation is valid is given by
ℓ < 10000/f where f is the frequency, which works out to 200 km for 50 Hz and 170
km for 60 Hz.
• For underground cables, the length up to which nominal π representation is valid is given
by ℓ < 3000/f which works out to 60 km for 50 Hz and 50 km for 60 Hz.
Hence, lines with medium length (80 km to 200 km) can be represented by nominal p. For
longer lines, we should go in for equivalent p model or cascaded nominal p models.
For time-domain (transient) solution, a number of p sections have to be cascaded and the
number depends on the length of the line and maximum frequency that has to be “captured”
in the simulation [14]. The number of p sections n, can be estimated as follows:
v
n = (2.7)
4.44 f max
12  Computational Electromagnetic Transients

For example, if the lowest mode (ground mode or zero sequence mode) velocity is 2,59,000
km/second and if we are interested in capturing phenomena up to 3000 Hz, then the length
of the p section cannot exceed 19 km (approximately). Hence, to represent a 300 km line, we
require about 16 nominal p sections of 19 km each.
The p model can be used for steady state or frequency scan solutions and is not accurate
for time-domain (transient) solutions. The p model can give rise to spurious oscillations. For
transient solution, frequency-dependent distributed-parameter transmission line models should be
used. The following results from EMTP simulation [14], (Fig. 2.4) give an idea of inaccuracies
incurred by representing transmission lines in terms of p sections:

Fig. 2.4  Comparative response of a transmission line with p sections (8 and 32) and a model
with a constant parameter distributed parameter line.

Another important consideration in modeling transmission lines is the frequency dependence

of resistance and inductance parameters. The EMTP simulation using constant parameters at
power frequency can give rise to very inaccurate results. The theory behind inclusion of frequency
dependence is explained in Chapter 3 and the model of a transmission line incorporating
frequency dependence is given in Section 2.4.1.2.4.
Microtran provides a high input accuracy option for R, L and C which was required in the
case of transformer representation by means of inductances calculated from magnetizing and
short-circuit inductances. After the inverse inductance model of the transformer was implemented,
this option is no longer required.
The maximum number of phases that can be modeled is fifty.
2.4.1.1.3  Inductively and capacitively coupled branches
These can be modeled as limiting cases of multiphase p circuits.
14  Computational Electromagnetic Transients

4. You cannot model three ideal transformers in delta/delta connection because two of them
are sufficient to define all three line-to-line voltage transformations between phases A to
B, B to C, and C to A.

2.4.1.2 (Continuously) distributed-parameter lines

Refer Chapter 3 for theory behind distributed-parameter lines.
In an actual transmission line, the phase and interphase R, L and C parameters are
continuously distributed as against cascaded π circuits which are discretely distributed.
The voltage and current in a distributed-parameter line are described by partial differential
equations. The solutions of the partial differential equations indicate that the voltage and current
are actually waves travelling on the transmission line.
Imagine a transmission line energized at the sending end. The voltage will not appear
instantaneously at the receiving end since the voltage wave, which travels almost at the speed
of light, will take a finite amount of time to reach the receiving end. Hence, during the travel
time from the sending end to the receiving end, the ends remain electrically disconnected. This
characteristic distinguishes the continuously distributed-parameter line or simply distributed-
parameter line.
The voltages and currents on the N phases of a practical transmission line are functions of
location x and time t, and are related by the partial differential equations
¶v ¶v
– = [ L¢ ] + [ R¢] i
¶x ¶t
¶i ¶i
– = [C ¢ ] + [G ¢ ] v
¶x ¶t
(2.8)
The prime in R¢, L¢ and C ¢ is used to indicate distributed parameters, i.e., parameters per
unit length (for example in W/km), in contrast to lumped parameters. If the parameters R¢, L¢, C ¢
are assumed to be constant, i.e., not dependent on frequency, their values can be obtained from
Line Constants Program. The parameters for the frequency-dependent line model are obtained
from the auxiliary program fdData [15] included in the Microtran package.
2.4.1.2.1  Balanced transmission lines
By construction, most of the EHV lines are “unbalanced.” By that we mean that the mutual
couplings between the phases are not equal themselves. We can visualize the unbalance by
considering some of the common tower configurations given in Fig. 2.7.
The above configurations give rise to a series of impedance matrix of the line which will be
symmetric but not balanced. A balanced matrix is one in which the diagonals (self impedances)
are equal among themselves and the off-diagonals (mutual impedances) are equal among
themselves. Transposition of the line such that each phase conductor occupies all positions in
the tower will yield approximately balanced matrix when taken over one transposition cycle.
The remarks made for series impedance matrix also hold good for shunt capacitance matrices.
Transposition of transmission lines is dealt with in detail in Chapter 3.
 EMTP: Modeling and Simulation Capabilities  15

Fig. 2.7  Common tower configurations: (a) “flat” configuration and (b) triangular configuration.
[Reproduced (Fig. 2.7 (a)) from Reference 16 with permission.]

There are a number of transformations that decouple N mutually-coupled equations in phase

variables to give N uncoupled single-phase equations in modal variables. Out of the N equations
in the modal domain, one equation describes the ground return mode and the remaining N–1
equations describe line (sky or aerial) modes. The impedances of all N–1 line modes are identical
for common transformations used in power system analysis. In Microtran, the transformation
matrix used corresponds to 0ab transformation.
Note:
1. For a three-phase ac line, the ground return mode parameters are identical to zero sequence
parameters and the line mode parameters are the positive sequence parameters.
2. For a single-phase line, the ground return parameters are simply the single-phase parameters
and no line mode parameters exist.
Modes and modal transformations are explained in Chapter 3.
2.4.1.2.2  Untransposed transmission lines
For a general untransposed line the series impedance and shunt capacitance matrices are
symmetrical but not balanced. That is, all the diagonals are not equal among themselves and
so are all the off-diagonals. However, we can still create decoupled modes of voltages and
currents by making use of modal transformations that are specific to the tower-line geometry.
We cannot use standard transformation like 0αβ transformation to obtain decoupled modes.
To model a general untransposed distributed-parameter line in EMTP, the user must, therefore,
give the modal transformation matrix in addition to the line parameters in modal domain as
input data.
 EMTP: Modeling and Simulation Capabilities  17

Fig. 2.8  Variation of zero sequence parameters with frequency from Reference 17.
[Redrawn from Reference 17 with permission.]

Fig. 2.9  Variation of positive sequence parameters with frequency from Reference 17.
[Redrawn from Reference 17 with permission.]

Lumped Resistance Model

EMTP expects modewise line parameters for the constant-parameter line. Further, if the line is
untransposed, it expects the modal transformation matrix as well. Each set of modal parameters
constitutes a single-phase line. The losses in each of these lines in the lumped resistance model
are represented by total resistance of the line split as shown in Fig. 2.10.

Fig. 2.10  Lumped resistance model for accounting line losses.

In Fig. 2.10, Rtotal = R¢ x length of the line. The lumped resistance model is accurate as long
as Rtotal is small compared to the surge impedance √(L¢/C ¢ ). This is generally true for the line
18  Computational Electromagnetic Transients

mode. For the ground return mode, even with frequency dependence of line parameters taken
into account, lumped resistance model is normally used [12].
Distortionless Line Model
A distortionless line preserves the wave shape of voltage and current from the point of origin
of the wave (sending end) to the destination (receiving end).
Distortionless line models are seldom used in EMTP simulations because wave propagation
in power transmission lines is not distortionless. It can be shown that if the line is distortionless,
the following condition is satisfied:
R¢ G¢
= (2.11)
L¢ C¢
For overhead power lines G ¢ is very small and its value for distortionless line model is
artificially increased to make the line distortionless. This naturally increases the shunt loss. To
make the total line loss constant, the series resistance is correspondingly reduced. This then
implies that
R¢ G ¢ 1 R input
¢
= = (2.12)
L¢ C ¢ 2 L¢
where R input is the input value of line series resistance per unit length. The input value of the line
series resistance per unit length is automatically halved by EMTP and the halved value is taken
as the series resistance. The reciprocal of the remaining half is taken as the shunt conductance.
It can be easily seen that the attenuation constant a for the distortionless is given by
Rinput
¢ C¢
a = (2.13)
2 L¢
The corresponding decrement factor is e– al with l = length of line. The proof of Eq. (2.13)
is given in the Appendix to this chapter.
Lumped resistance or distortionless?
Following points will serve as guidelines for the user:
1. The lumped-resistance model (at three points) is accurate as long as the line total resistance
is small compared to the surge impedance.
2. The lumped-resistance model is the preferred model for constant-parameter line
representations.
3. Distortionless line models are seldom used, except for positive and negative sequence
line modes, but even there the lumped resistance line model is normally more accurate.
Note:  ac steady-state solutions at one or more frequencies can be performed by EMTP. These
are mainly done for the following purposes:
1. To provide initial values for subsequent transient solutions, tmax > 0, where tmax is the
maximum time for transient solution.
20  Computational Electromagnetic Transients

Equation (2.16) can be rewritten as

Vk – ZC Ikm = (Vm + ZC Imk) e (– g l) = (Vm + ZC Imk ) A(w) (2.18)
where we have denoted the propagation factor e– g l as (w). Equation (2.18) can be rearranged as

1 æVm
Ikm = Vk – + I mk ÷ A(w) (2.19)
ZC ZC ø
In the time domain, Eq. (2.19) corresponds to the following expression in
1 t max
ikm (t) = vk (t) – [vm (t – u) + ZC imk (t – u)] a(u) du (2.20)
ZC t min

Note that we now have two travel times, tmin and tmax, in the case of frequency-dependent
parameter line.
The essence of the J. Marti’s method is as follows:
1. Approximate A(w) and ZC (w) by rational functions in the frequency domain. The function
A(w) is of the form
( s + z1 ) ( s + z2 )  ( s + zn )
Aapprox (s) = e– stmin k (2.21)
( s + p1 ) ( s + p2 )  ( s + pm )
2. Expand that portion A(w) that does not include the exponential term as partial fractions.
3. The inverse Fourier transform after step 2 yields
aapprox (t) = k1 e – p1 ( t – tmin ) + k2 e – p2 ( t – tmin ) +  km e – pm ( t – tmin ) for t ≥ tmin
= 0 otherwise
4. Use the expression given in step 3 in place of a (u) in Eq. (2.20) and evaluate the integral.
This step gives the past history term for propagation.
5. The partial fraction expansion of ZC is interpreted as a set of series connected R-C networks
and a resistance between node k and ground. Integrate the capacitance equations using
trapezoidal rule to yield an equivalent resistance and a past history term one step behind.
6. Combine the past history terms from steps 4 and 5 and place it on the right-hand side [i]
and include equivalent conductance (reciprocal of equivalent resistance) in the conductance
matrix [G] to solve the set of equations
[G] [v] = [i] (2.22)
Refer to Chapter 8 for more details on the formation and solution of Eq. (2.22).
For modeling frequency dependence, the data for coefficients of partial fraction expansion
of the characteristic admittance and propagation function for each line mode, and the current
modal transformation matrix should be supplied by the user. This can be obtained from the
module fdData which produces this data in the form of a file that can be directly read by the
transients simulation module, mt. The modal transformation matrix will be computed at optional
user-specified frequency (or a default frequency) by fdData and it will be real and constant.
 EMTP: Modeling and Simulation Capabilities  21

In the frequency-dependent line model the line losses are continuously distributed. However,
in the case of overhead transmission lines in power systems, lumped versus distributed modeling
of the line losses is not normally a major factor determining the accuracy of the simulation
results.
2.4.1.2.5  Summary of data requirements for line modeling
The basic data requirements for the various models are given below. For exact format of the
input and other details the reader should consult Reference 12 which gives many illustrative
examples for various line models.
1. Balanced constant parameter lumped resistance or distortionless line model:
• Number of phases of the transmission line (maximum = 50).
• For a double-circuit three-phase line, this number would be 6.
• Resistance in ohms per unit length (R ¢) whose value can be zero plus
• One of the following data options (a), (b) or (c) for distributed-parameter lines
Option (a):  nonzero values of:
• Inductance in millihenries per unit length (L¢ ) or inductive reactance in ohms/
unit length (wL¢)
• Capacitance in microfarads per unit length (C ¢ ) or susceptance in microsiemens/
unit length (wC ¢ )
• Surge impedance (ZC) in ohms
Option (b): surge impedance and wave speed
Option (c): surge impedance and travel time (t) which must be larger than the step
width used for numerical integration (trapezoidal rule)
• Line length
Note:
(a) The per unit length resistance and data given as options must be specified for ground
and line modes (zero and positive sequence parameters for a three-phase line).
(b) Mixing of loss-handling methods (lumped resistance or distortionless) among different
modes is allowed for a multiphase line.
(c) Constant parameter does not imply that the parameters are at power frequency. In
fact the parameters, especially zero sequence, are to be specified at a much higher
frequency for more accurate simulation. The frequency to be specified depends on the
user judgment. A typical value would be around 1.2 kHz. The program mtLine/fdData
can be used to produce these values. In fact fdData produces a file that can be
directly read by the transients module mt. In this case, mt uses 0αβ transformation to
diagonalize the impedance matrix. The reader should refer to the reference manuals
for mtLine and fdData, Reference 15 for more details.
2. Unbalanced constant parameter lumped resistance or distortionless line model:
Data for case 1 plus
• Number of phases plus
22  Computational Electromagnetic Transients

• Transformation matrix. This matrix is available from fdData module computed at a

frequency specified by the user.
3. Balanced and untransposed frequency-dependent line parameter model:
Recall that for the frequency-dependent line model, the line parameters are considered to
be functions of frequency; that is, R ¢ = R ¢(w), L¢ = L¢ (w), G ¢ = G ¢ (w), and C ¢ = C ¢ (w). In
practice, G ¢ and C ¢ are usually constant. All the parameters, including the losses, are modeled
as continuously distributed along the line’s length. This model is based on the synthesis by
rational functions, in the frequency domain of the line propagation function and characteristic
impedance as explained earlier.
The line representation for these models is produced by the auxiliary program fdData in the
form of a model data file. This file contains the partial fractions expansion of the characteristic
admittance and propagation function for each line mode, followed by the transpose of the
currents modal transformation matrix.
For a perfectly transposed (and therefore, balanced) line, the modal parameters are exact
at all frequencies using a constant and real transformation matrix, e.g., Clarke’s transformation
matrix [16]. The transients simulation module mt recognizes the line as balanced and uses 0ab
transformation to convert between phase and modal quantities.

2.4.2 Switches
Any switch operation in a power system initiates transients between pre- and post-operation
steady states. The Microtran provides models for
• time-controlled and
• voltage-dependent switches.

2.4.2.1  Time-controlled switches

Following points are to be noted for time-controlled switches:
• Circuit breaker poles and similar types of switches whose opening and closing depends
upon time are modeled as time-controlled switches.
• The switch resistance is infinite in the open position and zero in the closed position.
• In this model, the user has to specify the closing (CT ) and opening (OT ) times. The
time-domain simulation begins with the switch open if CT ≥ 0 and with the switch closed
if CT < 0. The latter case indicates that the switch is closed during the steady state. The
actual switch closing within mt occurs at the time step closest to Tclose (same as CT).
• The opening of the switch is influenced by OT and another parameter called “current
margin (CM).”
• If CM = 0 then, within mt, the switch opens after the time step closest to Topen, as soon
as the switch current iswitch has gone through zero. If CM > 0 then, the switch opens after
the time step closest to Topen, as soon as the magnitude of the switch current (|iswitch|) has
become less than CM.
• The close-open sequence can be repeated as often as desired by specifying pairs of values
Tclose, Topen.
24  Computational Electromagnetic Transients

• The switch closes after the time step closest to closing time, Tclose as soon as the absolute
value of the voltage across the switch becomes greater than the breakdown voltage.
• There is another parameter to be specified by the user, Tdelay, delay time which determines
how long the switch remains closed (to simulate persistence of the arc). The switch will
open thereafter depending on conditions on current margin explained for time-controlled
switch.
• The sequence “closing, time-delay, opening” repeats itself after the first and subsequent
openings, as indicated in the sketch below (Fig. 2.11).

Fig. 2.11  The repeating sequence “closing, time-delay, opening”, points A, B and C correspond
to zero switch currents with current margin = 0.
[Redrawn from Reference 12 with permission.]

• The data requirements for the voltage-dependent switch are: closing time, breakdown
voltage, time-delay, the node from which time-dependent breakdown voltage has to be
picked up and current margin.
The readers are urged to go through many illustrative examples in Reference 12 which give
the input data for switches in the format required by Microtran.
2.4.3  Nonlinear Elements
By nonlinear elements we mean resistive and inductive elements that do not have a linear
voltage-current relationship. For example, if an input to a circuit element is a sinusoidal and
the output is also a sinusoidal wave of same frequency then the element is linear. It is nonlinear
otherwise. Microtran provides two methods of handling nonlinear elements:
• Piecewise linear approximation method
• Compensation method
2.4.3.1  Piecewise linear method (piecewise linear models)
2.4.3.1.1  Piecewise linear resistance
Surge arresters are also included in this category provided the characteristics of arrester disks
(silicon carbide or metal oxide) can be represented as a piecewise linear slope. The piecewise
linear characteristic to idealize a given voltage-current characteristic can exhibit two or more
than two slopes.
 EMTP: Modeling and Simulation Capabilities  25

2.4.3.1.1.1  Two-slope piecewise linear resistance and surge arrester characteristics

The voltage-current characteristic of this element is shown in Fig. 2.12(a). It is simulated by
the program with a series connection of voltage source, slope resistance R, and switch as shown
in Fig. 2.12(b).
The restriction regarding switch connection to nodes with no voltage source connected to
it applies to this case also. The piecewise linear resistance is ignored (disconnected) during
steady-state solution.

Fig. 2.12 (a) v–i characteristic, (b): element representation; k and m are the nodes between which
the element is connected and m is the internal node created by Microtran.
[Redrawn from Reference 12 with permission.]

Piecewise Linear Resistance

• Note that the element consists of a voltage source, a resistance and a switch all connected
in series. The operation of the element is straightforward.
• The voltage source represents the “saturation” voltage, i.e., the voltage at which the
characteristic changes slope from infinite value to finite value R.
• The switch opens (stops conducting) when the current goes to zero and stays open as
long as the voltage across the switch (v) is less than vSATURATION. Microtran also provides
the value of energy absorbed by the piecewise linear resistance each time the current
goes to zero.
• The user can also input the nonzero value of the switch resistance (Rswitch) if it is available.
Surge Arrester
• There is another input required from the user for this model, sparkover voltage vsparkover.
A nonzero value for vsparkover (sparkover voltage of the gapped arrester) results in an
approximate model for the arrester. A zero value for vsparkover implies a two-slope resistance
only.
• The approximation involved in surge arrester model is illustrated by a typical surge arrester
characteristic (Fig. 2.13(a)) and the simulated characteristic (Fig. 2.13(b)) shown below.
• For more details, the reader should consult Reference 12.
 EMTP: Modeling and Simulation Capabilities  27

As we can see that we have to specify two resistance values and two saturation voltages.
The first resistance value (60 W) and saturation voltage (150 volts) is input similar to the
two-slope case. The second of resistance (Rx) is computed as follows:

1 1 1
+ =
60 Rx 10
which gives Rx = 12 W. The saturation voltage (vSATURATION2) is read off from the graph.
The corresponding EMTP representation is shown in Fig. 2.15.

Fig. 2.15  EMTP representation of piecewise linear resistance characteristic shown in Fig. 2.14.
The piecewise linear resistance is connected between nodes 1 and 2.
[Redrawn from Reference 12 with permission.]

Similar representation can be obtained for piecewise linear resistance for more number of
slopes. In general, the parallel resistance (Rx – i) that will give the correct i th slope (Rslope - i ) is
given by

1 1 1
+ = (2.23)
Rslope – (i – 1) Rx – i Rslope – i

It is obvious that if there are n slopes in the piecewise linear characteristic then the total
number of resistors in the EMTP representation would be n and the number of parallel resistors
and corresponding saturation voltages will be n – 1.

Surge Arrester with Characteristic that has More than Two Slopes
The parallel resistance connection method can also be used to realize the surge arrester v – i
characteristic but the following restrictions apply:
1. The characteristic cannot contain a linear slope through the origin. This implies that the
characteristic shown in Fig. 2.14 is allowed and that shown in Fig. 2.16 is not allowed.
28  Computational Electromagnetic Transients

Fig. 2.16  A four-slope of piecewise linear resistance characteristic with

a linear slope through the origin.
[Redrawn from Reference 12 with permission.]

2. The sparkover voltage must be lower than the saturation voltage.

Note:
1. If the characteristic contains a negative slope, it is allowed provided a point of intersection
of the characteristic with the Thevenin characteristic of the system at the point of
connection of the piecewise linear resistance can be found. The reason for this will
become clear when we take up the compensation method of handling nonlinear elements.
The compensation method is explained briefly in Section 2.4.3.2. A negative resistance,
with operation in the first and third quadrant, could be a crude approximation of an arc
characteristic.
2. A forced negative resistance operation in the second and fourth quadrants is possible. For
more details the reader should consult Reference 12. Such a characteristic represents a
low-frequency corona model.
Table 2.1 summarizes the modeling options for piecewise linear resistance and surge arrester.
The data requirements will now include parallel resistance values and corresponding saturation
voltages.
2.4.3.1.2  Piecewise linear inductance
The simulation of piecewise linear inductance within EMTP is somewhat similar to piecewise
linear resistance with flux linkage – voltage and resistance – inductance being similar quantities.
A two-slope approximation for the flux linkage (y)-current (i) nonlinear characteristic is shown
in Fig. 2.17. The slope of the characteristic going through the origin represents inductance L1
and the next segment represents inductance L2.
The simulation of this characteristic within EMTP is illustrated in Fig. 2.18.
It is clear that for simulation of inductance L1, the switch in Fig. 2.18 must be open and
for simulation of L2, the switch should be closed with Lp given by
1 1 1
= + (2.24)
L2 L1 L p

30  Computational Electromagnetic Transients

Fig. 2.18  Simulation of two-slope of characteristic shown in Fig. 2.17 within EMTP.
[Redrawn from Reference 12 with permission.]

The parallel connection of inductances shown in Fig. 2.18 is special in the sense that the
flux in Lp is always calculated by integrating vk – vm (rather than vk – vm). The flux yp = y1
always, independent of the switch position [17]. This is because the parallel connection is just
to realize the slope L2 and in reality there is only one element.
How faithfully the simulation “tracks” the two-slope characteristic depends on the value
of step size of numerical integration, Dt. In some situations, there can be fictitious overshoots
when the slope changes from L1 to L2 and undershoots when slope changes from L2 to L1. This
is illustrated in Fig. 2.19.

Fig. 2.19  Overshoot and undershoot produced while simulating two-slope

characteristic (IFLUX = 0).
[Redrawn from Reference 12 with permission.]

The overshoots and undershoots can be controlled by specifying a parameter IFLUX to

EMTP. But the control is obtained at the expense of simulation on a shifted characteristic as
shown in Fig. 2.20.
Two more important points deserve mention regarding piecewise linear inductance model:
1. For transient simulations starting from automatic initialization of flux, the piecewise
inductance is modeled as the inductance in the linear region. The initial value of the flux
 EMTP: Modeling and Simulation Capabilities  31

Fig. 2.20  Simulation on a shifted characteristic, no over- or undershoots (IFLUX = 1).

[Redrawn from Reference 12 with permission.]

y(0) is computed using this inductance. EMTP considers only fundamental frequency
sources in the network while performing the steady-state solution for initialization purposes.
If the initial flux |y(0)| ≥ ysaturation, then a warning message given below is issued:
“* * * WARNING. ASSUMPTION THAT AC STEADY STATE HAS FUNDAMENTAL
FREQUENCY ONLY IS QUESTIONABLE WITH PRECEDING FLUX OUTSIDE
LINEAR REGION”.
2. The two-slope characteristic does not exhibit any hysteresis. Approximate simulation of
hysteresis in EMTP is possible. Consult Chapter 6 of Reference 17 for more details.
Data requirements
The initial flux yresidual, L1, L2, switch resistance Rswitch and ysaturation.
For piecewise linear inductance with more than two slopes, more parallel inductances are
switched in and out and their values have to be computed and given as input. Refer [12] for
more details and illustrative examples.
2.4.3.1.3  Power electronic devices
Microtran version of EMTP provides models for diodes, thyristors, triacs (anti-parallel thyristors)
and gate turn off devices. Functionally, these are voltage-controlled switches, turning ON when
the anode to cathode voltage is positive and OFF as soon as the current through the device
crosses zero from a positive to a negative value with possible additional delay and current
margin. Refer [12] for more details.

2.4.3.2  Compensation method

2.4.3.2.1 General
Consider a single nonlinearity, e.g., a gapless lightning arrester or nonlinear resistor connected
between a node and ground as shown in Fig. 2.21. We will briefly illustrate how this method
works deferring the detailed treatment to Chapter 8.
32  Computational Electromagnetic Transients

Fig. 2.21  Nonlinear element connected to node k.

Network is first solved without the nonlinear element to obtain the Thevenin equivalent at
node k. Then,
• As shown graphically in Fig. 2.22, the solution for the current through the nonlinear
element is obtained.

Fig. 2.22  Nonlinear element and network characteristics.

• The complete solution for the linear part including the “loading” effect of the nonlinear
element is found by superimposing the linear part solution with effect of the nonlinear
element as a current source i.
There are some important points to be noted regarding
• Number and location of the nonlinear elements
• Artificial disconnection of nonlinearities using “stub lines”
• Singularity check in case nonlinear element is the only branch element in the network
• Treatment of nonlinearities in the ac steady-state solution
These will be addressed in Chapter 8.
Following branch elements can be solved the using compensation method. The element
characteristic is specified point-by-point. If an analytical expression is available for the
characteristic, compute the dependent variable values at sufficiently closely-spaced independent
variable values and give that as the input to EMTP.
2.4.3.2.2  Nonlinear resistance and simple surge arrester
Figure 2.23 shows a typical discretized characteristic of a nonlinear resistor.
34  Computational Electromagnetic Transients

of the voltage across the time-varying resistance becomes greater than or equal to vSTART as
illustrated in Fig. 2.24. Before the characteristic becomes effective, the time-varying resistance
R(t) is taken to be infinite.

Fig. 2.24  Typical point-by-point characteristic, R(t) and illustration of when the R(t)
characteristic becomes effective.
[Redrawn from Reference 12 with permission.]

Data requirements
Time-varying resistance R(t) point-by-point as a piecewise linear characteristic (linear
interpolation will be used between points), magnitude of vSTART. Note that if vSTART is specified
as zero, the time-varying characteristic becomes effective at the beginning of transient study.
[See Reference 12 for more details.]
2.4.3.2.4  Nonlinear inductance
Here, the inductance depends on the current through it. This implies that the flux linkage (y)-
current (i) characteristic is not linear. Microtran requires y – i characteristic to be specified
point-by-point. Normally, the data is not readily available. For example, if we want to model
transformer core nonlinearity, the characteristic available will be a point-by-point description
of RMS voltage (VRMS) as a function of RMS current (IRMS). The Microtran package contains
a utility that converts VRMS – IRMS characteristic to y – i characteristic.
[See Reference 12 and related documents for more details.]
Data requirements
Steady-state values of current (iSTEADY) through and flux linkage (ySTEADY) of the nonlinear
inductance and point-by-point values of current (i) and flux linkage (y).
The values specified for iSTEADY and ySTEADY decide how the nonlinear inductance is to be
treated in ac steady-state solution if the transient study is to start from an internally computed ac
steady state. The automatic steady-state solution is indicated to EMTP by specifying a negative
starting value (TSTART) for the source data. The source description is given later in this chapter.
Following are the interpretations of the values of iSTEADY and ySTEADY within EMTP.
• iSTEADY > 0 and ySTEADY > 0: The nonlinear inductance will be represented in the steady-
state solution by the inductance of the linear region computed from iSTEADY and ySTEADY.
This is the normal case.
 EMTP: Modeling and Simulation Capabilities  35

iSTEADY = 0 and ySTEADY = 0: These values indicate that the user is not interested in ac
•
steady-state solution (TSTART ≥ 0) or the nonlinear inductance is to be ignored in the
steady-state solution.
iSTEADY = 0 and ySTEADY > 0: This implies that the nonlinear inductance is not conducting
•
during steady-state solution but it has a nonzero initial flux linkage. The nonzero initial
flux linkage will be automatically computed by the program from the voltage across the
nonlinear inductance.
iSTEADY > 0 and ySTEADY = 0: This case implies that there is a short circuit in steady state
•
and is not permitted. EMTP will terminate execution with error message.
2.4.3.2.5  Groups of metal-oxide surge arresters and other nonlinear elements
This branch type was in the earlier versions (for example, Reference 16) used for gapless metal-
oxide surge arresters with nonlinear characteristics i(v) of the form
a
|v|
i = sign v (2.25)
vref
Currently, Microtran offers this branch type along with the group that includes time-varying
resistances, as well as nonlinear resistances and nonlinear inductances. The nonlinear resistance
and inductance element characteristics are required in the piecewise linear form. The subroutine
CONNEC, in which each group of nonlinear elements is solved, can also be modified by the
user for user-developed models which require multiphase network connections.
In contrast to the nonlinear elements of Sections 2.4.3.2.2 to 2.4.3.2.4, the nonlinear
elements of this type can be connected to the network without requiring separation by travel
time. [See Reference 12 for more details.]

2.4.4  Voltage and Current Sources

2.4.4.1 General
Microtran provides for three types of source modeling:
• Frequently used standard functions for voltage or current sources that are built into the
program.
• Functions of arbitrary shape (up to ten) defined point-by-point at discrete time steps Dt,
2Dt, º. These values define the functional relationship with respect to time.
• User-defined functions in a user-supplied subroutine, as explained in Section 9.15 of
Reference 12.
Only salient features are discussed in this section. [See Reference 12 for more details.]
Following are the general points which the user must be aware of:
1. Generally, all sources are connected between node and local ground. However, there can
be occasions when sources are to be connected between nodes. These and other cases
are discussed below:
 EMTP: Modeling and Simulation Capabilities  37

Fig. 2.27  (a) Representation of voltage source with nonzero internal

resistance between two nodes.
[Redrawn from Reference 12 with permission.]
(ii) connect the voltage source between the node extra and ground
(iii) connect an ideal transformer with unity ratio between “extra” node and ground
on one side (say primary side) and connect the secondary side between nodes
k and m. See Fig. 2.27(b).

Fig. 2.27  (b) Representation of voltage source with zero internal resistance between two nodes.
[Redrawn from Reference 12 with permission.]

Voltage source has a nonzero internal inductance Li

A voltage source with nonzero internal inductance between two nodes can be
represented in a similar fashion with the internal inductance Li between nodes k and
m and current source between the nodes that depends on the initial conditions as
explained below:
Let the voltage be given by
v(t) = Vmax cos (wt + j) (2.26)
(i) If the case starts from zero initial conditions then the current is obtained by
integrating the voltage:
t
1
i(t) = v(t ) dt
Li 0

which gives
Vmax
i(t) = [sin(wt + j) – sin j] (2.27)
wLi
(ii) If the case starts from linear ac steady state then,
Vmax
i(t) = [cos(wt + j) – 90° (2.28)
wLi
38  Computational Electromagnetic Transients

The current source i(t) between nodes k and m is then represented as explained next.
(d) Current source between two nodes
The current source between two nodes is represented as two current sources one at
each node as shown in Fig. 2.28.

Fig. 2.28  Representation of current source between two nodes.

[Redrawn from Reference 12 with permission.]

Here, the current source between nodes A and B is represented as two current sources
one out of node A and the other into node B.
(e) Voltage and current source at the same node
If voltage and current sources are specified at the same node, the voltage sources
override, and the current sources are ignored. Current sources do not influence the
network in this case because they are directly short-circuited through the voltage
sources.
2. The source functions specified by the user must observe consistency in units, for example,
if the voltage sources are specified as volts V (or kilovolts) then current sources should
be amperes (or kiloamperes).
3. All built-in source functions have “start” and “stop” time parameters. Depending on the
source type, these parameters have certain restrictions and default values. These built-in
source functions are zero between 0 ≤ t ≤ TSTART and for t ≥ TSTOP. These two parameters
enable the user to specify a composite source function synthesized from the recipe of
built-in functions as illustrated in Fig. 2.29.

Fig. 2.29  A composite source function synthesized from two built-in source functions.
[Redrawn from Reference 12 with permission.]
40  Computational Electromagnetic Transients

2.4.4.2  Standard source functions in-built into EMTP

Following are the built-in functions in Microtran available to the user
1. Step Function
Function description:
a (t) = Amax for TSTART ≤ t ≤ TSTOP
= 0 otherwise.
(2.29)
This function can be used to
(i) approximate step functions in the case of zero initial conditions. The function is an
approximation because it has a finite rise time Δt / 2 with the CDA scheme [18, 19].
See Fig. 2.30(a).
(ii) represent dc source function with a (0) = Amax with nonzero initial conditions as shown
in Fig. 2.30(b). For the nonzero initial conditions, other sources and the branches should
also have correct dc initial conditions.

Fig. 2.30  Step function implementation in Microtran (a) zero initial conditions;
(b) nonzero initial conditions.
[Redrawn from Reference 12 with permission.]

The CDA scheme is explained in Chapter 10.

The steady-state computations will not be performed for dc sources and as mentioned
earlier, approximate dc steady state is obtained with a very low frequency ac sources.
Consequently, the parameters TSTART < 0 has no meaning for dc sources and is set to zero
by the program. Consult Reference 12 for more details.
2. Rectangular Ramp Function
Function description:
a (t) = 0  for 0 ≤ t ≤ TSTART
Amax
= (t – TSTART) for TSTART < t ≤ TSTART + T0
T0
= AMAX for TSTART + T0 < t ≤ TSTOP
(2.30)
In this case also TSTART < 0 has no meaning and is set to zero by the program. See
Fig. 2.31 for illustration.
 EMTP: Modeling and Simulation Capabilities  41

Fig. 2.31  Rectangular ramp function implementation in Microtran.

[Redrawn from Reference 12 with permission.]

3. Triangular Ramp Function

See Fig. 2.32 for illustration of the function.

Fig. 2.32  Triangular ramp function implementation in Microtran: (a) general case;
(b) special case.
[Redrawn from Reference 12 with permission.]
Function description for the general case:
a (t) = 0 for 0 ≤ t ≤ TSTART
æAMAX
= (t – TSTART) for TSTART < t ≤ TSTART + T0
T0 ÷ø
æAMAX – A1
= – (t – TSTART + T0 ) + AMAX for t > TSTART + T0
T1 – T0 ÷ø
(2.31)
In this case also TSTART < 0 has no meaning and is set to zero by the program.
4. Sinusoidal Function
See Fig. 2.33 for illustration.
Function description:  A general sinusoidal function should include a initial phase angle
with respect to a chosen reference. In Microtran, the initial phase angle can be provided
in terms of degrees or seconds. The latter case would imply that the initial phase angle is
w × initial phase angle in seconds. Here, w = 2pf where f = frequency of the sinusoidal
 EMTP: Modeling and Simulation Capabilities  43

It was later realized that the double-exponential surge function of Eq. (2.33) is not adequate
to represent short surges such as 8/20 ms. Also the maximum slope da/dt occurs as per the
measurements close to the peak value whereas, for the double-exponential it occurs at t = 0.
In order to accommodate the above and other requirements, Microtran provides a variety
of surge functions which can be categorized as follows:
Single- and Double-exponential Surge Function: The double-exponential surge function is
defined by Eq. (2.33) and restricted to be nonzero between TSTART and TSTOP as shown below:
a (t) = AMAX (e–at – e–bt) TSTART ≤ T ≤ TSTOP (2.33)

Surge Function of Standler:  This function is proposed basically to facilitate representation of

short surges as given in relevant IEC and IEEE standards.
• Surge function of CIGRE
• Surge function of Heidler
There are sub-categories of the first, second and fourth functions. We give below the sub-
categories of the first function. For others to refer [12].

Sub-categories of Single and Double-exponential Functions

1. Double-exponential surge function with input values for a and b.
2. Double-exponential surge function with input values for 30% / 90% “virtual front time”
and “virtual time to half-value”.
3. Double-exponential surge function with input values for 10% / 90% “virtual front time”
and “virtual time to half-value”.
4. Double-exponential surge function with input values for “rise time” and “full width at
half maximum.”
5. Double-exponential surge function with input values for “crest time” and “time to half-
value”.
For each sub-category a and b values can be read off from Table 2.2 [17].

Table 2.2

T1  / T2 (μs) 1/ (μs) 1/β (μs) AMAX to produce max {a(t)} = 1.0

1.2/5 3.48 0.8 2.014

1.2/50 68.2 0.405 1.037
1.2/200 284 0.381 1.010
250/2500 2877 104 1.175
250/2500 3155 62.5 1.104
44  Computational Electromagnetic Transients

Note:
(i) The first four entries correspond to lightning impulse and the last one to switching impulse.
The first column of the last entry is Tcr /T2.
(ii) For the first four entries, T1= virtual front time, typically 1.2 ms ± 30%, T2 = virtual time
to half, typically 50 ms ± 20%.
(iii) For the fifth entry, Tcr = time to crest, typically 250 ms ± 20%, T2 = virtual time to half,
typically 2500 ms ± 60%.

2.4.4.3  Current controlled dc voltage source

Brief introduction to HVDC link control
• Voltage regulation and current regulation is kept distinct and assigned to separate terminals.
• Normal operation: rectifier on constant current (CC) control and inverter on constant
extinction angle control (CEA) with adequate commutation margin. This is normally
referred to as mode 1 in HVDC literature.
• Rectifier and inverter control characteristics with quantities measured at the rectifier end
are shown in Fig. 2.35.

Fig. 2.35  Rectifier and inverter characteristics with voltage and current measured at the
rectifier end. Vdc ≈ VRECTIFIER = – VINVERTER.

• Rectifier characteristic can be moved horizontally by varying current order (IORDER) which
is the set point for link current. The measured link current is maintained at IORDER by
varying ignition angle (also known as firing angle) a. See Fig. 2.36 for illustration of
IORDER.
• Inverter characteristic can be raised or lowered by changing the transformer tap position
at the inverter end. A change in transformer tap causes a change in inverter extinction
angle g. The constant extinction angle regulator at the inverter brings back g to the original
value.
• The direct current which changes due to voltage change (due to tap change) is quickly
restored by constant current regulator at the rectifier by adjusting a.
• The rectifier tap changer acts to bring a within the desired range (typically, amin= 10° to
amax= 20°).
46  Computational Electromagnetic Transients

The current controlled dc source implemented in Microtran consists of a current regulator

shown in Fig. 2.37.

Fig. 2.37  Model of the current controller implemented in Microtran.

[Redrawn from References 12 and 17 with permission.]

The transfer function G(s) is given by

K (1 + sT2 )
G (s) = (2.35)
(1 + sT1 ) (1 + sT3 )
with limits placed on the output ea in accordance with rectifier minimum firing angle and
inverter minimum extinction angle.
The dc voltage Vdc is a function of ea and is given by
Vdc = k1 + k2 ea (2.36)
See Fig. 2.38. On comparison of Eq. (2.36) with Eq. (2.34) we see that ea is related to
a through a cosine relationship. Actually, ea minus a bias value is proportional to cos a and
as shown in Fig. 2.38. Observe that k1 is negative. The maximum and minimum limits on dc
voltage is due to the limits on ea.

Fig. 2.38  Variation of Vdc with ea.

From the current controller model of Fig. 2.37 and the transfer function given by Eq. (2.35)
we can derive the following differential equation for ea:
dea d ( I BIAS – i ) d 2 ea
(T1 + T3) = K ( I BIAS – i ) + KT2 – T1 T3 – ea (2.37)
dt dt dt 2
 EMTP: Modeling and Simulation Capabilities  47

The limiter on the transfer function indicates that it is a limiter without windup. Hence,
if the converter operates at the maximum limit ea max (or at the minimum limit ea min), either in
initial steady state or later during the transient simulation, it will back off the limit as soon as
the derivative dea/dt becomes negative (or positive) in Eq. (2.37). The second derivative of ea
is zero at the limit.
Computation of steady-state dc initial conditions
To compute the initial steady-state conditions, the user should specify, among other things,
initial dc voltage measured at some point in the link (could be one end) and dc current. We
will explain why dc current is required to be specified later.
With the specified value of Vdc (0), actually a very low frequency ac solution is performed
with Vdc = Vdc (0) cos 2p flow t with flow= 0.001 Hz. This trick produces results sufficiently close
to actual dc solution. The ac steady-state solution is explained in Chapter 8.
Initial steady-state dc current, Idc(0), is required to compute the dc voltage source between
two nodes where, one node is not ground. Actually, the steady-state module calculates the initial
condition i(0) for the dc current from the specified dc voltages Vdc(0) but its value is not readily
available in the program where it is first needed for representation of dc voltage source between
two nodes.
With the specified Idc(0), for the representation of the converter station, the voltage source
is computed from
Vdc (0) = Vsource – Requiv Idc (0) (2.38)
and then converted to current source in parallel with Requiv as given below:
Isource = Vsource / Requiv (2.39)
Requiv is the resistance of the link from the converter station to the point where the initial
dc quantities specified are measured. See Fig. 2.39.

Fig. 2.39  Representation of the converter station for steady-state solution.

[Redrawn from Reference 12 with permission.]

The set point IBIAS of the current controller is automatically computed within the program as
ea (0)
IBIAS = i (0) + (2.40)
K
which is obtained by setting the derivatives of ea to zero in Eq. (2.37).
 EMTP: Modeling and Simulation Capabilities  49

• Special types of power flow problems that involve untransposed lines single-circuit,
double-circuit, single-phase distribution lines connected to three-phase substations, etc.
[1]. Such cases cannot be solved with classical power flow programs, which assume that
the power system is completely “balanced” all voltages and currents symmetrical, and
represented by the single-phase positive sequence network only.
• Special types of short-circuit problems that require the phase shift delta-wye transformers
to be taken into account, simulation of simultaneous faults.

2.5.2  Simulation Capabilities of Time-Step Solution Module

The simulation capabilities of time-step solution module of EMTP can be categorized as follows:
• Transients associated with electric circuits
• Transients in power systems associated with lightning
• Transients in power systems associated with switching
• Electromagnetic transients involving synchronous and induction machines

2.5.2.1  Transients associated with electric circuits

Some of the problems associated with power system transients can be solved with a circuit-
model of the system consisting of lumped elements. Such models are amenable to closed-form
solution but a number of assumptions are required to make the problem tractable. This approach
is very well illustrated in Reference 20. It provides physical visualization of the system response
and develops the engineer’s ability to qualitatively predict the response. EMTP provides for this
level of modeling and it is the lowest level of modeling sophistication provided by it.

2.5.2.2  Transients in power systems associated with lightning

Computation of lightning transients is essential in evaluating surge arrester protection and
insulation coordination of air- and gas-insulated substations.

2.5.2.3  Transients in power systems associated with switching

Intended and inadvertent switching can cause very high overvoltages especially in extra high
voltage power systems. Computation of switching overvoltages is essential in designing the
countermeasures for them. They are also required for insulation coordination.

2.5.3  Electromagnetic Transients Involving Synchronous

and Induction Machines
The necessity of adding detailed synchronous machine models to EMTP was first recognized
when there were attempts to simulate the occurrence of subsynchronous resonance (SSR) in series
compensated lines radially connected to a generating station. The SSR caused the destruction
of shaft section of a tandem-compounded turbo-generator [21]. Transient SSR studies can be
conveniently carried out using Microtran.
Microtran also provides a detailed induction machine model which facilitates accurate
simulation of induction switching and reswitching transients in industrial systems and power
plant auxiliary systems.
 EMTP: Modeling and Simulation Capabilities  51

2.8 SUMMARY

This chapter begins with a list of computer programs available for simulation of electrical
transients in power systems and traces the history of EMTP development for approximately
past fifty years. It also provides the modeling capabilities of various EMTP versions in general
and Microtran in particular. The simulation capabilities and power system studies that can be
carried out using EMTP form the concluding part of this chapter.

REFERENCES
1. Hermann W. Dommel, Introduction to the Use of MicroTran and other EMTP Versions,
Department of Electrical and Computer Engineering, The University of British Columbia,
Vancouver, B. C., Canada, (WWW: http://www.ece.ubc.ca), 1988. Microtran is a registered
trademark of Microtran Power System Analysis Corporation, Canada.
2. Hermann W. Dommel, “Past, Present and the Future of Electric Power Systems from
my Perspective,” Zagreb, April 20, 2007.
3. D.J. Mader, “History of the Development Coordination Group,” EMTP REVIEW, Vol. 1,
No.1, University of Wisconsin, Madison, July 1987.
4. Can/Am EMTP News, Vol. 88-1, Editor: Thomas Grebe, USA.
5. L. Dubé and H.W. Dommel, “Simulation of control systems in an electromagnetic transients
program with TACS,” Proc. IEEE PICA Conf., pp. 266–271, May 1977.
6. K.C. Lee, Lightning Surge Propagation in Overhead Lines and Bus-Ducts and Cables,
Ph.D. Thesis, The University of British Columbia, Vancouver, Canada, July 1980.
7. J.R. Marti, “Accurate modelling of frequency-dependent transmission lines in
electromagnetic transients simulations,” IEEE Trans. Power App. Syst., Vol. PAS-101,
pp. 147-157, Jan. 1982.
8. V. Brandwajn, H.W. Dommel, and 1.1. Dommel, “Matrix representation of three-phase
N-winding transformers for steady state and transient studies,” IEEE Trans.,Vol. PAS-101,
pp. 1369-1378, June 1982.
9. V. Brandwajn, Synchronous Generator Models for Simulation of Electromagnetic Transients,
Ph.D. Thesis, The University of British Columbia, Vancouver, Canada, July 1977.
10. H.W. Dommel, “Data conversion of synchronous machine parameters,” EMTP
NEWSLETTER, Vol. 1, No. 3, 1980.
11. L. Marti, “Simulation of transients in underground cables with frequency-dependent modal
transformation matrices,” IEEE Trans. Vol. 3, pp. 1039-1110, July 1988.
12. Microtran Power System Analysis Corporation, MicroTran Reference Manual, Transients
Analysis Program for Power and Power Electronic Circuits, Last revision January 2002,
Canada.
13. P. Kundur, Power System Stability and Control, McGraw-Hill, 1994, USA.
52  Computational Electromagnetic Transients

14. J.C. Das, Transients in Electrical Systems Analysis, Recognition, and Mitigation, McGraw-
Hill Companies, New York 2010, p. 81.
15. Microtran Power System Analysis Corporation, MtLine & fdData Reference Manual;
Overhead line parameter and model generation support programs for Microtran
simulations, Published September 1992 Last revision, February 2005, Canada.
16. H.W. Dommel and I.I. Dommel, Transients Program User’s Manual, Copyright 1976,
Latest revision Aug. 1, 1978, The University of British Columbia, Vancouver, B.C., Canada
V6T 1W5. Last revision January 2002, Canada.
17. Hermann W. Dommel, EMTP THEORY BOOK, Second Edition, Microtran Power System
Analysis Corporation, Vancouver, British Columbia, Canada, May 1992 Last update:
April 1996.
18. J.R. Marti and J. Lin, “Suppression of numerical oscillations in the EMTP,” IEEE Trans.
Power Systems, Vol. 4, pp. 739–747, May 1989.
19. 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.
20. Allan Greenwood, Electrical Transients in Power Systems, John Wiley & Sons, 1991.
21. J.W. Ballance and S. Goldberg, “Subsynchronous Resonance in Series Compensated
Transmission lines,” IEEE Trans. Power App. and Syst., Vol. PAS-92, pp.1649–1658,
Sep./Oct. 1973.
22. EMTP DCG/EPRI, EMTP REVIEW, Vol. 4, No. 2, April 1990, University of Wisconsin,
Madison, USA.
23. J.M. Prousalidis et al., On Studying Ship Electric Propulsion Motor Driving Schemes,
Electrical and Computer Engineering Dept., National Technical University of Athens,
Iroon Politechniou 9, 15780, Zografou, Greece.
24. E.W. Kimbark, Electrical Transmission of Power and Signals, John Wiley & Sons, New
York, USA, 1949.

Appendix
A.1  Line of Low Attenuation and Low Distortion-Proof of Eq. (2.13) [24]
If the line is distortionless, the following condition is satisfied:
R¢ G¢
= (2.11)
L¢ C¢
For a low distortion and low attenuation,
R ¢ << wL ¢ and G ¢ << wC ¢ (A.1.1)
The series impedance per unit length is given by (see Fig. A.1.1)
R¢
Z ¢ = (A.1.2)
–1

R ¢ 2 + w 2 L¢ 2 e j cot
wL ¢
54  Computational Electromagnetic Transients

p 1 æG ¢ R¢
– + ÷
j 2 2 wC ¢ wL ¢ ø
= w L¢C ¢ e
Let
1 æG ¢ R¢
j = + ÷
2 wC ¢ wL ¢ ø
The attenuation constant is given by
æp
a = Re {g } = w L¢C ¢ cos – j÷
2 ø
Expanding the cosine in the above equation and using approximations for line of low
attenuation and distortion we get
1 æG ¢ R¢
a = w L¢C ¢ + ÷
2 wC ¢ wL ¢ ø
or
G¢ L¢ R¢ C¢
a = + (A.1.8)
2 C¢ 2 L¢
Dropping the first term in comparison with the second and setting R ¢ = R input /2 we get
Eq. (2.13).