1

Wide-Band Line Model Implementation in

Matlab for EMT Analysis
O. Ramos-Leaños, Student Member,IEEE, R. Iracheta, Student Member, IEEE

Abstract- The Wide Band (WB) or Universal (ULM) Line Model The WB Line model is the most general of the mentioned
currently is the most general line model available in EMTP for models. It is capable of representing accurately highly
simulating transient responses of power transmission lines and asymmetric aerial lines, as well as underground and submarine
cables. Since this model is still subject to further improvements, power transmission cables. WB line model is the most
an experimenting prototype of it has been developed in Matlab
recently developed one. Its robustness, accuracy and
environment. This development is reported here and the paper
also offers a comprehensive description of the WB model. Two computational efficiency still have to be improved. In
application examples are provided here, one involving an aerial addition, the technical publications describing this model are
line and the other one an underground cable. directed to a highly specialized public. This paper has the
following two main objectives: 1) to report the development of
Index Terms— EMTP, frequency dependent parameters, an experimental Matlab version of the WB model and 2) to
frequency domain, MATLAB, time domain, WB Line Model. provide a comprehensive description of this model.

I. INTRODUCTION II. BACKGROUND

T HE design and operation of power systems as well as of

power apparatuses, each time depends more on accurate
simulations of Electromagnetic Transients (EMTs). Essential
In practical cases, line parameters usually present strong
variations with frequency and this feature must be taken in
consideration for transient studies. Early attempts to include
to this is to count on advanced models for representing power frequency dependence relied on performing straight numeric
transmission lines and cables. The Electromagnetic Transients convolutions [3]. Two major problems were encountered with
Program (EMTP) is the software package most used for EMT this approach: 1) the requirement of an excessive computation
analysis. EMTP-RV, its latest version, offers various line time and 2) large error accumulation. These problems were
models. Among these are the Constant Parameters Line model solved with the introduction of the recursive convolution
(CP Line), the Frequency Dependent Line Model (FD Line) technique [5]. Subsequent realization that recursive
convolution is a form of State Space Analysis has brought
[1] and the Wide Band (WB) or Universal (ULM) Line Model
considerable progress to the specialty of power–system
[2].
transient analysis [5].
CP Line is the simplest model. It considers that line
The first model that succeeded at including frequency
parameters are constant and requires a very small amount of
dependence for a large class of aerial lines is the FD Line
computations. At the same time this is the least accurate one
model [1]. In addition to reference [1], a comprehensive
and it is recommended only for modeling lines on zones
description and update of the FD Line model is provided in a
distant to an area where a transient event occurs. The FD Line
companion paper [6].
model evaluates multi–conductor line propagation in modal
However, highly asymmetric lines, as well as cables,
domain and takes into account effects due to frequency
usually are not accurately represented by the FD model. These
dependence of the modal propagation parameters.
cases also require considering the frequency variations of
Nevertheless, as the transformations between the modal and
Modal matrices. Some of the proposed models maintain the
the phase domains are approximated by real and constant
basic idea of working internally in modal domain and the
matrices, its accuracy is limited to cases of aerial lines that are
modal–to–phase transformations are carried out by matrix–to–
symmetric or nearly–symmetric. The ULM or WB Line model
vector convolutions based on rationally synthesized modal
has its basis on Modal Analysis; nevertheless, as this model
matrices. A problem with these models is that often modal
resolves the multiconductor propagation relations in phase
matrices cannot be synthesized with sufficient accuracy and
domain, it avoids simplifying assumptions regarding modal–
guaranteed numerical stability. One of these models is the
to–phase transformations.
FDQ [7]. A second model of this kind is proposed in [17]. It is
_________________________________
The authors gratefully acknowledge the provision of scholarships by the
based on modal decomposition and Vector Fitting (VF). A
Mexican Science and Technology Council (CONACYT) to conduct graduate third model that overcomes the stability problem with the
studies. modal–matrix realizations is the one of Idempotents [8].
O. Ramos-Leaños is currently pursuing his Ph. D. degree in Electrical The problems with modal–matrix realizations and modal–
Engineering at École Polytechnique de Montréal.
E-mail address: [email protected]
matrix convolutions are overcome by phase domain based
R. Iracheta is currently doing a stage at École Polytechnique de Montréal. He models. The most prominent of these is the ULM or WB
is a Ph. D. candidate at CINVESTAV, campus Guadalajara, Mexico. model that resolves multi–conductor line propagation in the
E-mail address: [email protected] phase domain [1]. The WB model is the central subject of this

Authorized licensed use limited to: KU Leuven Libraries. Downloaded on April 15,2024 at 07:18:12 UTC from IEEE Xplore. Restrictions apply.
 2

paper. In addition to the phase domain option, there are other

alternatives being tested for the developed of full frequency
dependent line models. A very promising one is reported in
[12]. This alternative is based on the Theory of Characteristics
of Partial Differential Equations [13] and, in some way, it is
similar to the FD Line model.
Time Domain (TD) line models that are capable of
Fig. 1. Diagram of a simple transmission line.
including full frequency dependent effects are still being
developed. On the other hand, there are equivalent models in
Now, (3) and (4) are evaluated at = :
frequency domain that are very accurate and have been
available for some time [9]. As experimental reports of tests ()= = . (10)
involving transients on lines and cables are scant, frequency
()= = (11)
domain line models constitute a very valuable resource for the
improvement of time domain line models. Note in (10), as well in Fig. 1, that the definition of
follows the convention of currents entering a line being taken
III. BASIC TRANSMISSION LINE ANALYSIS as positive.
A. Transmission Line Equations From (10) and (11)
Electromagnetic behavior of transmission lines and cables is = 2 , (12)
described by the Modified Telegrapher Equations, which in and from (8):
frequency domain are expressed as follows: = ( ) (13)
with
= (1) = (14)
Matrix is called here the propagation factors matrix.
= (2) Expression (13) constitutes a model for the line terminal at
= . The following is a companion expression providing the
where is the vector of line phase voltages, is the vector of model for the terminal at = 0:
phase currents, is the series impedance matrix in per unit = ( ) (15)
length and is the shunt admittance matrix also in per unit
length. Uppercase letters are used here to indicate Frequency Expression (15) could have been derived in a similar form
Domain variables, lower case letters are for representing Time as (14), or by appealing to the symmetry of the line, Fig. 1. It
Domain variables and Boldface letters represent matrix and follows from (8) that can be interpreted as a vector of
vectors quantities. currents reflected at = 0 and traveling towards the terminal
The general solution of (1) and (2) for the vector of currents at = . To stress this fact is denoted as in (12):
is [15]: = 2 ; (16)
( )= (3)
similarly, in (9) represents a vector of currents reflected at
where and are vectors of integration constants and = and traveling towards = 0. On denoting by in
=√ is the so called propagation matrix. A solution for (15):
the voltage in terms of the same constants is derived replacing = 2 . (17)
(3) in (2):
( )=( ), (4) (16) and (17) define the line model depicted in Fig. 2. The WB
Line model is based on these expressions which are valid for
where, is the matrix of Characteristic Admittances: both, aerial lines and underground/submarine cables. The
relation of this model with the one in Nodal two-port form,
= (√ ) , (5) commonly used in frequency domain transient analysis [3], is
established next.
B. Transmission Line Model B. Nodal Form of Line Model
Consider now a line segment of length = , as the one Expressions (13) and (15) can be rearranged and grouped as
shown in Fig. 1. On evaluating (3) and (4) at = 0: follows:
(0) = = (6) = (18)
(0) = = . (7)
where is the unit matrix. The modal representation of the
Integration constant vectors and can thus be line is readily derived from (18) as follows:
expressed in terms of and as follows:
( 2 ) ( )
=( ) . (19)
=( )⁄ (8) ( ) ( 2 )
=( )⁄ (9) This representation is equivalent to the following standard
nodal form [3].

Authorized licensed use limited to: KU Leuven Libraries. Downloaded on April 15,2024 at 07:18:12 UTC from IEEE Xplore. Restrictions apply.
 3

currents or of voltages with the property that, the relative

proportions among its elements are maintained as it travels
[14]. In other words, all the components of a mode are
attenuated with the same rate and are propagated with the
same speed.

Fig. 2. Norton equivalent of a transmission system. IV. WB LINE MODEL IN STATE SPACE
csch( ) coth( ) The State Space (SS) form of the WB line model defined by
= . (20) expressions (16) and (17) arises when matrices defined in
coth( ) csch( )
(5) and defined in (14) are approximated by rational
Note that ( ) ( ) = coth( ) and ( functions.
)−12 =csch( ).
A. Rational Fitting
C. Modal Analysis It is clear from (26) and (27) that and are
The WB line model, given by (16) and (17), requires transcendental functions i.e., polynomials of infinite order.
evaluation of the following matrix functions: Nevertheless, a premise in practical EMT analysis of lines is
that one can attain accurate approximations of these functions
=√ (21)
by quotients of finite order polynomials in the variable =
and
; i.e., the purely imaginary frequency. Hence, can be
(22)
approximated as:
This is conveniently done by using Modal Analysis [14].
The usual assumption that the product always is , (28)
diagonalizable [16] is adopted here; thus:
= , (23) where K is the order of the approximation, is the k-th
residue, is the k-th pole and is a constant term or residue
where is the column-eigenvector matrix of and is its at = ∞.
diagonal matrix of eigenvalues. According to Modal Analysis The rational approximation of requires the extraction of
any analytic function of can be obtained as follows [4]: time delays from all its elements. This is essential to attain a
( ) 0 0 low order in the approximations [1]. A detailed explanation on
0 ( ) 0 how to do the time delay extraction can be found in [6].
( )= . (24) On appealing to the Spectral Theorem of diagonalizable
0 0 ( ) matrices, given by (27) can be expressed as the sum of its
Consider now: idempotent matrices , , … , , each one multiplied by its
=√ = , (25) correspondent eigenvalue:

where = √ = ( , , … , ) and = . The = . (29)

diagonal elements of are the modal propagation constants
γ
for each mode of the line. The real parts are the mode Consider the factors in (29) as:
attenuations and the imaginary parts are the modal phase γ
= .
shift rates. It follows from (25) that is given by:
If these factors where minimum phase functions, would
= √ = determine , according to Bode’s theorem from Network
0 0 Synthesis [18]. Transmission lines are distributed elements
(26) and their modal phase shifts are composed by a minimum
= 0 0 . phase term ′ and a time delay related term ′′ . Nevertheless,
0 0 can still be factorized as the sum of minimum phase
functions ′ , each one multiplied by its correspondent pure
It follows from (24) that matrix is given by: time delay term. This factorization is highly convenient for
modeling purposes:
√
= = (27) (30)
=
0 0
= 0 0 . where is a time delay shift correspondent to the k-th modal
propagation function [16]:
0 0
′′
Multi-conductor line theory provides clear physical = / .
meaning to eigenvalues and eigenvectors of the line matrix Minimum phase functions ′
are approximated by rational
product . Each eigenvector, or column of , represents a functions as follows:
mode of currents [16]. A mode is defined as a vector of

Authorized licensed use limited to: KU Leuven Libraries. Downloaded on April 15,2024 at 07:18:12 UTC from IEEE Xplore. Restrictions apply.
 4

where the following state variables have been introduced:

′ ,
(31)
, , = 1,2, … ,
, = , , (40)
On introducing (31) into (30) , = 1,2, … ,

with , the residues matrix, , the poles and , is a

,
(32) reflected current coming from the m side of the line.
,
By a similar process as the one applied to (36), the
Often modal delays come close to one another. In these following TD form of (40) is obtained
cases, the modes can be lumped in propagation groups of , = 1,2, … ,
similar velocities this reduces the computational burden of the = , , , , ( ). (41)
= 1,2, … ,
line model. The grouping of modes is done by taking the
arithmetical mean and the smallest time delay of the modes Note that the phase shifts of (30) have become time delays
that compose the group. The standard criterion for lumping applied to the remote node currents.
modes is that their total phase-shifts should not differ in more Finally the TD form of (39) is as follows:
than 10 degrees. From here on, instead of the number of N
modes in (32), the number of groups will be used.
, 2 , . (42)
Rational fitting is done through the Vector Fitting (VF) [19]
or the Weighted Vector Fitting (WVF) [20]. A general
description of these techniques is provided in the appendix. Expressions (41) and (42) constitute the State Space
equivalents of the convolutions required to evaluate , .
B. Continuous State Space
Consider expression (17). As the line model is to be C. Discrete State Space
interconnected with a main solver the 0 and sub-indices are Line model implementation in digital processors requires
renamed as and , respectively. Now, the following the previously obtained state space relations be transformed
definitions are introduced. Notice that only side k of the line is from Continuous Time to Discrete Time State Space (DSS).
taken into account since equations for side m can be obtained This is accomplished replacing time derivatives of the
by symmetry. Continuous Time expressions by a Numerical Differentiation
, = (33) rule; for example: Backward Euler, Midpoint differentiation,
and Gear, etc. The Midpoint rule is equivalent to the trapezoidal
, =2 , . (34) rule of integration which is the most used in EMTP. Hence,
the midpoint (trapezoidal) rule is applied as follows in (37):
By replacing (28) in (33):
= ( ) ( ). (43)
∆ 2 2
, . (35) = 1,2, … ,
where primed variables represent their values delayed one ∆
Note that the following state variables have been introduced time step. After algebraic process, (43) can be expressed in the
in (35): following DSS form [6]
= , = 1,2, … , (36) = ′ ( ), = 1,2, … , (44)
where
After some algebraic manipulation, and by applying the = (2 ∆ )⁄(2 ∆ ) (45)
Inverse Laplace Transform, its TD equivalent is obtained as and
follows: ⁄(2 (46)
=Δ Δ ) .
= , = 1,2, … , (37)
Expression (44) contains present input values; nevertheless,
what is required for a sequential simulation is that the states be
the TD form of (35) is simply:
calculated from past inputs only. Thus, introducing the next
variable:
(38) ′
, . = (47)
Equation (44) can be rewritten as:
′
Expressions (37) and (38) are the State Space equivalent to = , = 1,2, … , (48)
the convolution that would have to be carried out in time- where
domain to evaluate shunt currents , . =( 1) (49)
From this process, the following DSS form of (38) is
To evaluate auxiliary currents , in State Space,
obtained:
expression (32) is introduced in (34) as follows:
N
, (50)
, 2 , , (39)
where

Authorized licensed use limited to: KU Leuven Libraries. Downloaded on April 15,2024 at 07:18:12 UTC from IEEE Xplore. Restrictions apply.
 5

= (51)

is the line admittance.

Expressions (48) and (51) constitute the DSS relations to
evaluate recursively and dynamically shunt line currents , .
The DSS forms for evaluating , are obtained by Fig. 3. Norton equivalent of a transmission line in Time Domain
applying the midpoint rule to (41)
In general will not be an integer multiple of Δt. Since the
, = , , , ( , ( ) , ( )) (52) reflected currents are only available at discrete time points, it
= 1,2, … , , = 1,2, … , will be necessary to interpolate the input between the sample
where points. For a given time step the number of history values that
, = (2 ∆ , )⁄(2 ∆ , ) (53) need to be stored in memory can be calculated by:
and
, =∆ , ⁄(2 ∆ , ). (54) 1. =
(63)
Δ
Expression (42) maintains the same following form In a multiphase line there are usually several propagation
N
groups, each with its own travel time. One approach to save
, 2 , . (55) the history is to create a buffer for each propagation group.
Here, a single buffer is created to store the data of all the
propagation groups and the length of the history buffer is
V. MODEL IMPLEMENTATION chosen as in (63) using the biggest .
The known history values are stored at ( ), where
A. History Currents
is given by
Replacing (33) and (34) in (17) the model for the line side max ( )
in TD is expressed as: = Δ (64)
Δ
= . . . (56) given that
Δ ( 1)Δ (65)
Expression (56) provides the circuit depicted in Fig. 3.
Replacing (50) and (55) in (56) where is an integer; then
N ( )Δ (66)
(57) with 0 1. Thus
= 2 , .
, ( Δ ) , ( Δ Δ )
( ( Δ ) ( Δ Δ )) (67)
Finally by grouping all the terms from one time step behind , ,
into a history current . , (57) can be rewritten as: For the first time step the reflected current is saved into the
= (58) first position of the buffer and for each subsequent time step
,
where the reflected current continues to be saved in the subsequent
N spaces of the buffer. When the limit is reached the values are
overwritten see Fig. 4.
, =2 , (59)
VI. APPLICATION EXAMPLES
Expression (58) provides the branch currents in terms of In this section simulation results corresponding to the
history values. Given that the line is considered symmetrical implementation of the model in MATLAB are presented. The
the current of the node is expressed as: results are compared against the results obtained from the WB
= (60) model of EMTP-RV and a comparison against the FD Line
,
Model is shown. Two examples will be presented one for lines
B. Convolutions and History Maintenance and one for cables.
Consider the following convolution A. Aerial Line
, = . ( ). (61) For the first test case the line layout is as shown in Fig. 5.
The circuit is energized at the beginning of the simulation
At a given time step, it can be computed as:
(t=0) with a resistive three-phase sinusoidal voltage source
( ∆ ) = , ( ) , ( ) with a value of 230 kv rms line to line. The receiving end of
(62)
( ∆ ) = , ( ∆ ) , , ( ) the line is left open, the resistance of the source is 0.13 ohms,
the length of the line is 193.1 km, earth resistivity is 100
where ( ∆ ) is the input at present time, ( ∆ ) is the ohms/m, resistance of ground wires 3.75Ω/ , diameter of
output, , , , and , are DSS , ( ) is a reflected ground wires 0.95 cm, resistance of phase conductors 954
current. MCM ACSR: 0.0701 Ω/ , diameter of phase conductors
3.058cm, and bundle spacing 40cm.

Authorized licensed use limited to: KU Leuven Libraries. Downloaded on April 15,2024 at 07:18:12 UTC from IEEE Xplore. Restrictions apply.
 6

Time step 1 2 n
Fig. 4. History buffer management.

The simulation was carried for 32 ms with a time step of

1us. Fig. 6 shows the voltages at the receiving end of the line
calculated with EMTP-RV (solid line) and MATLAB (dashed
line). The results are overlapped. Fig. 7 shows the differences
between both results. Note that these differences are in the
order of 10 .
B. Underground Cable
Fig. 5. Aerial line configuration.
For the second test case the layout of the circuit is given in
Fig. 8. The length of the cable is 25km. The cores of the
circuit are energized at t=0 with an inductive three-phase
sinusoidal voltage source with a peak value of 1 p. u. The
cores at the receiving end of the cable are left open, the
inductance of the source is 0.1 mH, the sheaths are grounded
at both sides of the line trough a resistance of 15 Ω. The
electrical and physical data of the cable are given in table I.
The simulation is carried out for 10 ms with a time step of 1
us.
Fig. 10 shows the core voltage waveforms at the receiving
end of the cable. The values obtained with the MATLAB
model are superposed to those obtained with EMTP-RV. For
this case the differences with EMTP-RV are in the order of
10 . Fig. 11 shows a comparison between the WB and the
FD model. This is to demonstrate that FD line model is not
suitable for cables.

VII. CONCLUSIONS
For their design and operation, present-day power systems and
apparatuses each time require the accurate simulation of their Fig. 6. Voltage waveform at the receiving end of the line.
electromagnetic transient performance. An important aspect of
these simulations is the realistic representation of transmission
lines and power cables by digital computer models. The WB
line model is the most general simulation model available in
EMTP. By being of relatively recent creation this model is still
subject to substantial improvements in accuracy, stability and
computational efficiency. It has been postulated in this paper
that the best form to carry out these improvements is by
developing a version of the model in an interpretive
environment, like Matlab. This paper presents the
development of such prototype. In passing, a comprehensive
description of the WB model has been provided.
The prototype has been applied in the simulation of
transients involving aerial lines and power cables. It has been
validated with the WB line model implemented in EMTP-RV,
as well as with the results obtained through the frequency
domain techniques for transient analysis that are reported in
[9].
Finally, the limitations of another model intended for aerial
lines (the FD model) when applied to underground cables have
been shown by comparisons with the WB prototype described
here.
Fig. 7. Voltage waveforms differences. Insertion x axis zoom.

Authorized licensed use limited to: KU Leuven Libraries. Downloaded on April 15,2024 at 07:18:12 UTC from IEEE Xplore. Restrictions apply.
 7

TABLE I imply the use of least square methods. Additionally, the

CABLE’S ELECTRICAL AND PHYSICAL DATA
ambiguity is removed from ( ) forcing the unity in high
Component Value frequencies.
The zero-pole transfer function of ( ) is obtained from
Core Inner radius 3.175mm
Core Outer radius 12.54mm
(A.2) as:
Sheath Inner radius 22.735mm
Sheath Outer radius 26.225mm ∑
Insulation Outer radius 29.335mm ∏ ( )
Core resistivity 2.1x10-8 Ohm/m ( )= = (A.3)
∑ 1 ∏ ( )
Sheath resistivity 1.7x10-8 Ohm/m
Earth resistivity 250 Ohm/m
Earth relative permeability 1
Insulator relative permeability 3.5
Core insulator relative permittivity 2.0
Shield insulator relative permittivity 2.0
Insulation loss factor 0.001
Cable length 25km

Fig. 8. Cable system layout.

Fig. 10. Core voltage wave forms at the receiving end, comparison
with EMTP-RV.

Fig. 9. Cable example’s circuit.

VIII. APPENDIX
A. Description of VF

Step #1: Pole identification

The function ( ), which is intended to be fitted, is

multiplied by an unknown rational fitting ( ) of order N.
( ) ( ) = ( ) ( ) (A.1)
Fig.11. Core voltage wave forms at the receiving end, comparison
A common set of poles ( ) is used for the rational fitting with FD line model.
of the unknown function ( ) and the product ( ) ( ). In
this way the expression to be fit can be written in terms of the
From (A.3) it can be see that starting poles are removed
initial poles ( ) and residues ( , , ) as is shown next:
from the algebraic process and the zeros of ( ) are converted
into the new poles of the function ( ). These new poles are
= 1 ( ) (A.2) used in the second step of VF known as residue identification.

Step #2: Residue identification

Writing (A.2) for a vector of frequency samples gives an The poles calculated from the previous step are used for the
overdetermined linear problem of the form Ax = b and also,

Authorized licensed use limited to: KU Leuven Libraries. Downloaded on April 15,2024 at 07:18:12 UTC from IEEE Xplore. Restrictions apply.
 8

identification of residues ( ) by solving the least-square [8]. F. Castellanos, J. R. Marti, F. Marcano, “Phase-domain multiphase
problem of type = . transmission line models,” Elsevier, International Journal of Electric
Power & Energy Systems, vol. 19, No. 4,pp. 241-28, May. 1997.
Steps 1 and 2 are repeated several times with the new poles [9]. F. A. Uribe, J. L. Naredo, P. Moreno, L. Guardado, “Electromagnetic
as starting poles, until the method converges. As convergence Transients in Underground Transmission Systems Through the
is achieved, ( ) approaches unity because residues ( ) Numerical Laplace Transform,” Elsevier, International Journal of
become to zero. VF requires less than 5 iterations to assure Electrical Power and Energy Systems, ISSN:0142-0615, Vol. 24/3, pp
convergence. Even though, VF does not have a criterion to 215-221, Mar. 2002.
specify the optimal number of iterations. [10]. P. Moreno, R. de la Rosa, J. L. Naredo, "Frequency Domain
Computations of Transmission Line Closing Transients", IEEE
B. WVF Transactions on Power Delivery, ISSN 0885-8977, Vol. PWRD-6, No.
According to I.Kocar [20] the identification problem must 1, pp 57-63, Jan.1991.
be stated as: [11]. P. Moreno, P. Gómez, J. L. Naredo, J. L. Guardado, “Frequency
domain transient analysis of electrical networks including non-linear
conditions”, Elsevier, International Journal of Electrical Power &
1
1 ( ) Energy Systems, ISSN:0142-0615, Vol. 27, Issue 2, Pages 139-146
∑ 1 Feb. 2005.
[12]. A. Ramírez, J. Luis Naredo, P. Moreno, “Full Frequency Dependent
(A.4) Line Model for Electromagnetic Transient Simulation Including
Apparently the term outside the square bracket is a Lumped and Distributed Sources”, IEEE Transactions on Power
weighting term and its absence is the cause of ill-conditioning Delivery, ISSN 0885-8977, Vol. 20, No. 1, pp 292-299, Jan. 2005.
[13]. J. L. Naredo, A. C. Soudack, J. R. Martí, "Simulation of Transients on
in VF and poor fitting due to over emphasized low frequency
Transmission Lines with Corona Via the Method of Characteristics", in
samples [20]. WVF proposes to maintain this factor through
Proc IEE, Part C, pp. 81-87, Jan. 1995.
an iterative process and called it. [14]. L. M. Wedepohl, “Application of matrix methods to the solution of
travelling-wave phenomena in poly phase systems”, in Proc. IEE,
( )= (A.5) 1963, pp. 2200-2212.
[15]. L. M. Wedepohl, “Electrical characteristics of polyphase transmission
systems with special reference to boundary-values calculations at
The first iteration of ( ) is made by approximating the power-line carrier frequencies”, in Proc. IEE, 1965, pp. 2103-2112.
frequency domain response of with a first order partial [16]. L. M. Wedephol, lecture notes, “Frequency domain analysis of wave
expression which apparently produces acceptable starting propagation in multi-conductor transmission systems”, Dept. of
values. For subsequent iterations the weighting function can electrical engineering, The University of British Columbia, Vancouver,
be calculated with (A.5). And the system of equations British Columbia,1993.
[17]. B. Gustavsen, A. Semlyen, "Simulation of transmission line transients
becomes [20] = . The primed versions of and are
using vector fitting and modal decomposition," Power Delivery, IEEE
found by dividing each frequency row by its corresponding
Transactions on , vol.13, no.2, pp.605-614, Apr 1998.
( ) factor. WVF renders better conditioning and lower order [18]. H.W. Bode, “Network analysis and feedback amplifier design”, D.
realizations than does VF. VanNostrand, New York, 1945.
[19]. B. Gustavsen, A. semlyen, “Rational approximation of frequency
IX. REFERENCES domain responses by vector fitting”, IEEE Trans. On Power Delivery,
[1]. J. R. Marti, “Accurate modeling of frequency dependent transmission vol. 14, no. 3, pp. 1629-1635, Jul 1999.
lines in electromagnetic transient simulations,” IEEE Trans. On Power [20]. I. Kocar, J. Masheredjian, G. Olivier, “Weighting method for transient
Apparatus and Systems, vol. PAS-101, no. 1, pp. 147-155, Jan. 1982. analysis of underground cables”, IEEE Trans. On Power Delivery, vol.
[2]. A. Morched, B. Gustavsen, M. Tartibi, “A universal model for accurate 23, no. 3, pp. 1629-1635, Jul 2008.
calculation of electromagnetic transients on overhead lines and
underground cables,” IEEE Trans. On Power Delivery, vol. 14, no. 3, X. BIOGRAPHIES
pp. 1032-1038, Jul. 1999. O.Ramos-Leaños (S’08) received the M.Sc. degree in Electrical Engineering
[3]. H. W. Dommel, “Digital computer solution of electromagnetic from CINVESTAV, campus Guadalajara, Mexico, in 2008. He has worked for
transients in single and multiphase networks,” IEEE Trans. On Power the industry on the power quality field. He is currently pursuing the Ph. D.
App. Syst., vol. PAS-88, pp. 388, 1969. degree in the department of Electrical Engineering at École Polytechnique de
[4]. W. S. Meyer, H. W. Dommel, “Numerical modeling of frequency – Montréal.
dependent Transmission–Line parameters in Electromagnetic
Transients Program,” IEEE Trans. On Power App. Syst., vol. PAS-93, Reynaldo Iracheta. (S’10) received the M. Sc. Degree in Electrical
Engineering from the Autonomous University of Nuevo Leon (UANL), in
pp. 1401-1409, Sept. 1974.
2007. Now, he is currently pursuing the Ph. D. Degree at CINVESTAV,
[5]. A. Semlyen, A. Dabuleanu, “Fast and accurate switching transient campus Guadalajara, Mexico. His main interest is the Real-Time simulations
calculations on transmission lines with ground return using recursive of electromagnetic transient in power systems.
convolutions”, IEEE Trans. On Power Apparatus and Systems, vol. 94,
no. 2, pp. 561-571, Apr. 1975.
[6]. R. Iracheta, O. Ramos-Leaños, “Improving computational efficiency of
FD line model for real-time simulations of EMT’s,” Submitted to the
42th North American Power Symposium, Arlington, TX, USA, 2010.
[7]. L. Marti, “Simulation of electromagnetic transients in underground
cables using the emtp,” IEE 2nd International Conference on Advances
in Power System Control, Operation and Management 1993, p. 147.

Authorized licensed use limited to: KU Leuven Libraries. Downloaded on April 15,2024 at 07:18:12 UTC from IEEE Xplore. Restrictions apply.