1868 IEEE TRANSACTIONS ON POWER DELIVERY, VOL. 29, NO.

4, AUGUST 2014

Wideband Modeling of Large Grounding Systems to

Interface With Electromagnetic Transient Solvers
Keyhan Sheshyekani, Senior Member, IEEE, Majed Akbari, Student Member, IEEE, Barzan Tabei, and
Reza Kazemi, Student Member, IEEE

Abstract—This paper presents a new technique for wideband modeled in these popular tools. In fact, it has been common
characterization and modeling of multiport large grounding practice to disregard the exact model of grounding systems
systems in the form of an accurate Electromagnetic Transients in the transient analysis of electrical networks simulated in
Program-compatible model. This helps to consider the exact
behavior of grounding systems in the electromagnetic transient time-domain software packages, such as Alternative Transients
analysis of power systems. The methodology involves three stages. Program-Electromagnetic Transients Program (ATP-EMTP)
First, a general electromagnetic approach based on the method of [1], EMTP-RV [2], and PSCAD/EMTDC [3]. It is known that
moments solution to Maxwell’s equations is utilized to obtain the when lightning strikes a power line (i.e., phase conductor, shield
grounding system admittance matrix over the frequency range wire, or the tower) or when it directly strikes the substation
of interest. In the next stage, a rational approximation of the
grounding system admittance matrix is inferred by making use of components or a wind turbine system, a noticeable portion of
a modified matrix pencil method (Modified-MPM) developed in the current is driven into the ground. Hence, grounding systems
this paper. The Modified-MPM is well suited to the matrix fitting play a crucial rule in the determination of consequent voltage
applications with a common set of poles selected for all matrix ele- stresses imposed on system components. These include the
ments. The obtained rational approximation is finally employed to overvoltages throughout the grounding system appearing at the
construct a multiport time-domain model of the grounding system
expressed in the form of state-space equations. The proposed neutral points of aforementioned ground components as well
modeling approach is applied to the lightning transient analysis as those overvoltages generated at equipment terminals both
of a real subtransmission (63/20 kV) substation considering its affecting the system insulations [4] and [5]. Moreover, transient
grounding system as an integrated part of the electrical system. voltage disturbances within grounding systems can reportedly
It is shown that the traditional method of using a simple resistive lead to the malfunction of small-signal instrumentations by
model for representing the grounding system results in a notice-
able underestimation in the calculation of lightning-generated drifting the voltage reference point.
overvoltages in electrical systems. Lightning impulse currents are usually characterized by
their wide frequency content from dc to several megahertz
Index Terms—Electromagnetic transients, grounding systems,
lightning overvoltages, modified-matrix pencil method (Modi- over which the grounding system reveals different behavior
fied-MPM), multiport systems, state space, wideband modeling. at different frequency intervals [6]–[8]. Most often, for the
inclusion in time-domain electromagnetic transient solvers,
grounding systems at each node are treated and modeled either
I. INTRODUCTION as a lumped resistor or as an equivalent circuit of lumped in-
ductance, resistor, and capacitance whose values are calculated

T IME-DOMAIN electromagnetic transient solvers are

widely accepted for the analysis of power systems due
to their capability to deal with complex networks, nonlinear
based upon the quasistatic assumptions. These models, how-
ever, fail to provide accurate results at high frequencies when
used for the calculation of lightning-generated overvoltages,
phenomena, and devices (such as corona and surge arresters) in particular, for large grounding systems where the equipo-
as well as changes in network configuration. These solvers tential assumption for ground conductors is violated. Several
are being updated on a regular basis in terms of their calcula- methods have been alternatively used to model and analyze
tion efficiency and accuracy. Moreover, in view of emerging the grounding systems in the past few years and to interface
energy technologies, new models are steadily incorporated them with time-domain circuit-based tools [9]–[12]. In [9], a
into these simulation tools. However, grounding systems as an method based on transmission-line theory has been proposed to
integrated part of electrical networks (such as substations, pow- interface grounding systems to the EMTP. This method, how-
erplants, and wind turbine systems) have not been accurately ever, neglects the inductive and capacitive couplings between
different line segments used for the modeling of grounding
system conductors. In another attempt, the asymptotic method
Manuscript received May 23, 2013; revised October 28, 2013; accepted has been successfully used for interfacing the grounding
February 22, 2014. Date of publication April 01, 2014; date of current ver-
system with EMTP. In this method, the self-impedances of the
sion July 21, 2014. This work was supported by Bakhtar Regional Electricity
Company, Iran. Paper no. TPWRD-00613-2013. grounding system are modeled by long transmission lines with
The authors are with the Electrical and Computer Engineering Department, frequency-dependent parameters, while mutual impedances are
Shahid Beheshti University, G.C., Velenjak, Tehran 1983963113, Iran.
modeled by different transform functions with different poles
Color versions of one or more of the figures in this paper are available online
at http://ieeexplore.ieee.org. [10], [11]. This method uses an iterative procedure and models
Digital Object Identifier 10.1109/TPWRD.2014.2310631 the self and the mutual impedances with a different set of poles

0885-8977 © 2014 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission.
See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
SHESHYEKANI et al.: WIDEBAND MODELING OF LARGE GROUNDING SYSTEMS TO INTERFACE WITH ELECTROMAGNETIC TRANSIENT SOLVERS 1869

which can make it computationally expensive. Furthermore, the A. Matrix Pencil Method in the Frequency Domain
passivity has not been enforced to the obtained rational model. Let us assume that is the input admittance of an arbitrary
In [12], a modified mesh current method has been developed for port of a grounding system calculated by the MoM over the fre-
the analysis of lightning electromagnetic transients on thin-wire quency range of interest. The MPM can be used to approximate
structures above and buried in lossy ground which considers as a sum of exponentials (SoE) [18] and [19]
different coupling mechanisms among conductors. Although
this method is constituted based on the quasistatic assumptions,
(1)
it can reasonably predict the performance of grounding systems
for the lightning frequency range. However, the efficiency of
the method in dealing with grounding systems in conjunction where and are complex-valued coefficients and is the
with power system components, such as transformers, has not order of approximation. Discretizing (1) with a sampling fre-
been discussed. Finite-difference time domain (FDTD) (see quency yields
[13] and [14]) and the transmission-line method (TLM) [15]
have been also applied to the analysis of grounding systems.
Nevertheless, there is no effort made for the inclusion of large
grounding systems modeled by the FDTD and the TLM into (2)
the electromagnetic transient analysis of power systems.
In this paper, an efficient approach for multiport wideband The aim is to find the best estimates for , , ,
modeling of grounding systems is proposed. By virtue of this . This problem is generally regarded as a
method, one can model the entire electrical system together with nonlinear problem and is treated by making use of the MPM
its respective grounding system in an integrated manner suit- to obtain the complex-valued coefficients (see the Appendix
able for conducting the electromagnetic transient studies. To for further details). Having obtained the SoE approximation of
this end, first, the impedance matrix of the grounding system , its time-domain representation can be found either
is calculated over the working frequency interval by making by the inverse Fourier transform or by the closed-form inverse
use of the method of moments (MoM) solution to Maxwell’s Fourier procedure described in [25].
equations in the frequency domain [16]. Then, a modified ma- B. Matrix Fitting With Common Poles
trix pencil method (Modified-MPM) is proposed to derive a ra-
tional approximation of the grounding system admittance ma- The procedure for obtaining the poles and residues involves
trix as a multiport electromagnetic system (see [17]–[19]). The applying the MPM on the time-domain representation of the
Modified-MPM is a direct solution and needs no starting poles frequency-domain response (see [17]) which yields
unlike the vector-fitting (VF) method [20]–[23]. Moreover, it
is well suited to the matrix fitting applications with a common (3)
set of poles [20]. The obtained rationally fitted matrix is then
employed to construct a multiport time-domain model of the For a linear time-invariant (LTI) system, the eigenfunctions
grounding system expressed in the form of state-space equa- of the transfer operator are of the form where are the
tions [24]. The obtained state-space representation is then im- poles of the system, stand for the system residues, and is
plemented in the EMTP-RV software for the transient analysis the number of poles. The exponential representation of can
of a real 63/20-kV substation against short-circuit and lightning be finally cast in the following form:
currents. It is worth noting that the proposed modeling approach
has no limitation in terms of the number of grounding system
(4)
ports and the maximum working frequency.
This paper is organized as follows. In Section II, the de-
velopment of the modified-MPM is proposed. Section III The inherent multiport configuration of grounding systems
outlines general requirements for the derivation of a physically combined with their frequency-dependent behavior requires
consistent model. The development of the state-space model suitable analysis by means of the matrix formulation of input
is described in Section IV. The calculation and the fitting impedances/admittances rather than a single admittance vector
of the grounding system admittance matrix are discussed in seen just from one port. For taking the mutual impedances
Section V. Finally, time-domain simulation of the considered into account, the proposed method is required to be suitably
substation is given in Section VI. A short discussion is provided extended for a matrix fitting application. To this aim, we follow
in Section VII. Concluding remarks are given in Section VIII. the same procedure described in [20], where a least squares
approximation for a matrix is calculated in such a way that
II. MODIFIED-MATRIX PENCIL METHOD all elements1 are fitted by identical poles. This is achieved
The theory of the MPM [18] has been recently customized by stacking the elements of the grounding system admittance
by Sheshyekani et al. [17] to infer a rational approximation of matrix into a single vector, which is fitted by the MPM gener-
frequency-domain responses. Although the detailed formulation ating the common set of poles. In the next step, a least squares
of the MPM can be found in [18]–[19], in this section, the theory problem is solved to calculate the individual residues of each
is summarized and modified concerning the specific case of a 1Here, each element refers to a self or a mutual admittance vector calculated
grounding system admittance matrix. over the working frequency interval.
1870 IEEE TRANSACTIONS ON POWER DELIVERY, VOL. 29, NO. 4, AUGUST 2014

vector in the frequency domain. However, some modifications tion of the grounding system admittance matrix elements with
are required to cast the fitting procedure in a suitable form to be a common set of poles, i.e.,
treated as a least-squares problem. To this aim, (4) is rewritten
as (5) for each frequency sample in which the residues are
unknown, i.e.,
(11)

where and are, respectively, the number of rows and columns

of , and is the number of predefined common poles.
(5)
III. REQUIREMENTS FOR PHYSICALLY CONSISTENT MODELING
Note that the conjugate symmetry condition of real systems
requires the complex conjugate poles to come with complex The grounding system we characterize is a linear time-in-
conjugate residues and the real poles to come with real residues. variant (LTI) system composed of ports. The requirements for
Thus, the fitting process is subjected to a conjugate symmetry physically consistent modeling of grounding systems are sta-
condition. As known from [17], the MPM fulfills this condi- bility, causality, and passivity.
tion when it deals with the fitting of one single vector. How-
ever, in case of matrix fitting, when a common set of poles is A. Stability
selected for the whole matrix elements, the fitting is likely to The stability of the system is guaranteed when the output of
violate the conjugate symmetry condition so that the poles and the system is bounded for all bounded inputs [26]. Note that in
residues might appear in real-complex or complex-real pairs. our modeling, the voltages at the grounding system ports and
This happens since the common poles are not physically true the currents flowing through the ports are the system inputs and
poles as they are intentionally attributed to each vector element. outputs, respectively. In the frequency domain, two conditions
To circumvent this problem, we resort to the technique devel- of Laplace transform have to be satisfied to meet the stability
oped in [20] described hereafter. of the system. First, the region of convergence (ROC) of its
Let us assume that and refer to a complex conjugate transfer matrix has to include the imaginary axis; second,
pair of poles, i.e., has to be bounded. These two conditions are practically
satisfied when the poles of the system lie in the left-hand side
(6) of the imaginary axis (i.e., poles with negative real parts) [26],
[27].
The conjugate symmetry of the system requires that the
residues come in the following form: B. Passivity and Causality
The original definition of passivity of a physical system is
(7) related to the system’s inability to generate energy. Also, an LTI
system is causal if for every input that vanishes for , the
To guarantee this issue, the two corresponding elements output also vanishes for 0. For LTI systems, passivity also
and are modified as [20] covers system causality [26].
For the state-space model used in this paper, the grounding
system model is passive if and only if its admittance matrix
transfer function is positive real [28]. This requires:
(8) 1) the admittance matrix transfer functions to have no poles
This forces the corresponding residues in the solution vector with positive real parts;
to become identical to and , respectively. 2) all eigenvalues of real part of matrix are positive [29],
Finally, (5) is extended to the whole frequency points as an [30].
overdetermined linear matrix equation [20] In fact, for each element of a rationally fitted matrix ,

(9)

Since the working frequency interval contains only positive

frequencies and in order to preserve the conjugate symmetry
property, (9) is formulated in terms of real quantities

(10) (12)

The aforementioned procedure is applied to the whole ma- Note that the passivity enforcement should guarantee the
trix elements in a similar way leading to a rational approxima- system stability and preserve the key dynamic properties of the
SHESHYEKANI et al.: WIDEBAND MODELING OF LARGE GROUNDING SYSTEMS TO INTERFACE WITH ELECTROMAGNETIC TRANSIENT SOLVERS 1871

system (pole distribution). In this paper, we have used the same The obtained state-space equations can be directly imple-
procedure described in [29] to enforce passivity to the model mented in EMTP-RV using the state-space block [33].
by means of the residue perturbation method (see [29], [31]
and [32]).
V. GROUNDING SYSTEM ADMITTANCE MATRIX
CALCULATION AND FITTING
IV. DEVELOPMENT OF THE STATE-SPACE MODEL
The grounding system is assumed as a series of connected
Having obtained a passive pole-residue model for the cylindrical metallic conductors with finite conductivity and
grounding system admittance matrix, a time-domain repre- arbitrary orientation divided into several small segments. The
sentation is needed to incorporate the frequency response of electromagnetic model proposed by Dawalibi and Grcev[16]
the grounding system in EMTP-RV. To this end, state-space is utilized to obtain an admittance matrix for the multiport
equations are formulated as follows [33]: grounding system in the frequency range of dc to 1 MHz. In
this approach, using the thin wire approximation, the governing
electric field integral equation (EFIE) is formulated for the
induced currents along the grounding conductor segments and
solved by making use of the MoM which gives the current
distribution along the segments (see [9] and [10]). The elec-
(13) tromagnetic fields within the space are then calculated. For
lossy ground at higher frequencies, the electric field is not
where is the state vector, denotes the number of system conservative and the integral of the electric field between any
ports, and is the total number of state variables which equals two points is path dependent. Owing to this fact and in order to
the product of the total number of poles and the total number of define a unique voltage, the gradient of the scalar potential is
ports (i.e., ). obtained along a unique straight path and continues to a remote
Although there are several possible time-domain realizations earth reference point [34]. The impedance/admittance matrix is
for the fitted impedance/admittance matrix in the form of state- then constructed in the frequency domain using the calculated
space equations, the Jordan-canonical form is chosen in this voltages and currents [35]. To this aim, each element of the
paper [24]. For an -port grounding system, we assume that the impedance matrix is computed by the ratio of the calculated
matrices , , and contain only real poles and their cor- voltage at port due to the current injected at port when the
responding residues, and , , and hold only the complex remaining ports are open circuit
poles and residues. The Jordan-canonical form of state-space
equations is as follows:
(17)

After computing all components of the impedance matrix

, the admittance matrix is obtained by inversion of the
impedance matrix
(14)

where the star sign denotes the conjugate pair. (18)

Since complex poles do not have a direct meaning in time
domain, the similarity transform is introduced by In this part, we first obtain the admittance matrix of the
grounding system of the Arak-2 substation which is a 63/20-kV
substation of Bakhtar Regional Electric Co., Iran. Then, a
(15) rational approximation of the grounding system admittance
matrix is inferred by making use of the modified-MPM de-
scribed in Section II. Fig. 1 shows the schematic diagram
Substituting from (14) and (15) in (13), gives the final state- of the Arak-2 substation, together with its grounding system
space equations layout. The grounding grid is an unequally spaced 64 64 m.
The depth of the grid is 1 m, and the conductors are of radius
10 mm buried in soil characterized by a resistivity of
100 m and a relative permittivity of 10. The aforementioned
ground components, namely, the transformers, the arresters,
and the auxiliary grounding transformers are connected to
the grounding system through nine ports which are numbered
from 1 to 9 as shown in Fig. 1. Hence, the grounding system
is modeled as a nine-port network. For simplicity reasons, the
(16) very close ports, such as those associated with the primary and
secondary sides of transformers, are considered as a single port.
1872 IEEE TRANSACTIONS ON POWER DELIVERY, VOL. 29, NO. 4, AUGUST 2014

Fig. 1. Schematic diagram of the Arak-2; a 63/20-kV substation together with its grounding system layout.

We first calculate the self and mutual impedances of the multi-

port grounding system which leads to a 9 9 impedance matrix.
Note that in our calculations, the effect of frequency dependence
of soil electrical parameters is neglected [7]. It is also assumed
that no ionization occurs due to the high impulse current injected
into the grounding system [35]. Fig. 2 shows the magnitude and
phase of some of the diagonal and offdiagonal elements of the
impedance matrix calculated for different ports. The modest in-
creasing trend of diagonal elements (i.e., self-impedances) con-
firms the previous findings reporting the inductive behavior of
grounding grid self-impedance [6], [7]. The positive phase angle
further supports this inductive behavior. It is seen that and
take the same value at low frequencies while they diverge
at higher frequencies. This is due to the different effective area
at nodes 1 and 5 [36]. For the case of mutual impedances (i.e.,
offdiagonal elements), however, it is seen that these impedances
show a decreasing trend at higher frequencies, indicating a re-
duction in the transient voltage (or current) transferred from Fig. 2. Self and mutual impedances of the grounding grid shown in Fig. 1 to-
one terminal to the other [37], [38]. Note that the mutual im- gether with their FEM counterparts: (a) magnitude and (b) phase.
pedances are affected by a combination of factors, such as the
distance between the ports and the soil conductivity, while the
reciprocity implies that . In our calculation, for soil
characterized by a resistivity of 100 m, mutual impedances be-
come negligible at frequencies above 1.5 MHz. Results obtained
by the finite-element solution of Maxwell’s equations using the
COMSOL Multiphysics [39] are also shown in Fig. 2 which fur-
ther support the accuracy of the self and mutual impedances ob-
tained by the MoM. As discussed earlier, once the impedance
matrix of the grounding grid is obtained, the admittance ma-
trix could be calculated by the inversion of the impedance ma- Fig. 3. Diagonal and offdiagonal elements of the admittance matrix together
trix. To constitute the state-space model of the grounding grid, with their MPM-fitted counterparts before and after passivity enforcement.
we need to infer a rational fitting of the entire elements of the
admittance matrix to provide its pole-residue characterization.
This is done by the method proposed in Section II in which the which clearly shows the ability of the proposed method to accu-
elements are rationally fitted by a common set of poles. The in- rately fit the whole matrix elements by a common set of poles.
herent passive characteristic of the grounding system requires The eigenvalues of the admittance matrix have been calculated
the fitted model to also comply with the passivity requirements over the working frequency interval before and after passivity
(i.e., a positive value for the eigenvalues). To enforce passivity, enforcement. Fig. 4 shows that the whole eigenvalues become
we use the residue perturbation method [29], [31]. Fig. 3 depicts positive when the passivity is enforced. It is worth noting that
the magnitude of some of the diagonal and offdiagonal elements the fitting is done by 14 common poles selected for each ele-
of the grounding system admittance matrix together with their ment of the admittance matrix which are listed in Table I. Note
MPM-fitted counterparts before and after passivity enforcement that the rms error was found to be .
SHESHYEKANI et al.: WIDEBAND MODELING OF LARGE GROUNDING SYSTEMS TO INTERFACE WITH ELECTROMAGNETIC TRANSIENT SOLVERS 1873

Fig. 4. Eigenvalues of the grounding system admittance matrix before and after
passivity enforcement.
Fig. 5. Voltage at healthy phases (phase B) due to a single-line-to-ground fault
occurring at phase A.
TABLE I
COMMON POLES GENERATED BY THE MPM

where and are the height and base radius of the tower in
meters, respectively. has a value of 132.9 for a tower height
of 16 m and a tower base radius of 2.2 m.
A typical -capacitance transformer model is chosen for
modeling the transformers consisting of the steady-state model
VI. TIME-DOMAIN SIMULATION
and high-frequency coupling capacitances [30], [43]. The
As shown in Fig. 1, the substation consists of three 63-kV values of the capacitance between the HV terminals and ground
feeders connected to three similar 30-MVA, 63/20-kV 1.11 nF, between the LV terminals and ground
transformers. The transformers at the primary side are solidly 3.43 nF, and between the HV and the LV terminals
grounded, while at the secondary side, they are grounded 4.54 nF are obtained from the manufacturer’s data sheet. This
through auxiliary transformers with Zig-Zag connections. model gives quite a good estimation of the voltages transferred
The 63-kV transmission-line tower is considered to have a from the medium-voltage (MV) to the LV side [30].
grounding resistance of 10 which is a typical value for tower
footing resistance in Iran. This assumption simply neglects A. Short-Circuit Study
the frequency dependence of the grounding system footing For a short-circuit study, a single-phase-to-ground (SLG)
impedance. Surge arresters are placed right after the substa- fault is assumed to occur at the LV side of the transformer
tion’s entrance portal support at the 63-kV side and at the (Phase A). According to the resistive behavior of the grounding
low-voltage (LV) side of the transformers. A span of 200 m system and since the short-circuit current frequency is 50 Hz,
is considered for 63-kV power lines entering the substation, we expect the same results for both resistive and wideband
and a long span of 30 km is assumed at the left side of the models of the grounding system. Fig. 5 illustrates the conse-
power lines to get rid of the unwanted reflections from the line quent voltage of phase B (healthy phase) obtained when two
terminations. The system is implemented in EMTP-RV [2]. For different models of the grounding systems are used. It is seen
the grounding system, we use two different models, namely, from Fig. 5 that the results for both models match perfectly
the simple resistive model and the wideband model obtained which also confirms that the wideband model of the grounding
as described in the preceding section. For the case of a simple system can accurately predict the low-frequency overvoltages.
resistive model, a 0.83- resistor is used at each node that is Note that due to the zero-sequence impedance of the auxiliary
equal to the static resistance of the grounding grid (see Fig. 2) Zig-Zag transformer, the system at the 20-kV side is not solidly
calculated by means of the MoM [16]. Line insulators are grounded and we expect a rise of voltage in healthy phases due
modeled by the flashover switch in EMTP.RV with a flashover to an SLG fault as shown in Fig. 5.
voltage of 350 kV which is a typical value for the 63-kV lines.
In fact, this device is modeled as a zero resistance when closed B. Lightning Study
and as an infinite resistance when open [2]. The LV and HV
The lightning is assumed to strike the tower adjacent to the
sides of the power transformers are protected by surge arresters
substation. A CIGRE shape model [44] is adopted for modeling
with an MCOV of 14.4 kV and 51 kV , respectively, and a
the lightning current which has a rise time of 2.1 s, maximum
single column of zinc oxide (ZnO) which is 0.5 m long. The
steepness of 37 kA s and a time to half value of 77.5 s with a
proposed model in [40] is used to model the ZnO surge arresters
peak value of 70 kA. Fig. 6 shows the lightning-generated over-
and to calculate the necessary parameters.
voltages at nodes 1, 2, and 3 for the resistive model and the
Transmission lines are modeled as distributed lines with fre-
wideband model used for modeling the grounding system. As
quency-dependent parameters [41].
seen, the overvoltages calculated using the former model give
The tower itself is usually made of steel and is assumed to
the same values (25 kV) at all grounding nodes while the latter
have a cone shape and is modeled by a surge impedance whose
model reports different voltages at different nodes. It is also ap-
value can be obtained by [42]
parent that the wideband model leads to higher overvoltages
compared to the resistive model. This owes to the fact that the
(19)
wideband model takes the frequency dependence of the input
1874 IEEE TRANSACTIONS ON POWER DELIVERY, VOL. 29, NO. 4, AUGUST 2014

Fig. 7. Overvoltages when lightning strikes the shield wire of the tower adja-
cent to the substation, tower-1: (a) overvoltage at phase A and (b) overvoltage
at phase B.

b) The proposed grounding system modeling enforces the

passivity to the model which has not been discussed in
[10].
c) The whole grounding system admittance vectors, in-
Fig. 6. Overvoltages when lightning strikes the shield wire of the tower adja-
cent to the substation: (a) node 1, (b) node 2, and (c) node 3. cluding the self and mutual impedances, are fitted using
a common set of poles which basically differs from the
approach adopted in [10].
impedances into account over the lightning frequency interval. d) Compared to [10], the proposed method provides a more
A very interesting observation in these results is the ability of accurate fitting for the grounding system admittance ma-
the proposed modeling technique to predict the wave propaga- trix with a deviation error on the order of .
tion delay. In fact, as seen from Fig. 6, in the wideband model, e) The proposed method has the ability to predict the wave
the voltage peaks at different nodes occur with a certain delay. propagation delay within the grounding system.
These delays are due to the wave-limited propagation speed
VIII. CONCLUSION
in lossy soil [45]. Note that the resistive model cannot predict
the wave propagation delay as seen in Fig. 6. A modified-MPM was developed for wideband characteriza-
To further evaluate the effect of the wideband modeling of tion and modeling of multiport large grounding systems. This
grounding systems on the electrical stress imposed on power modeling approach helps to incorporate the exact behavior of
system equipment, results associated with the phase conductor large grounding systems into electromagnetic transient solvers,
overvoltages at the HV side of the transformer due to the same such as EMTP-like tools.
lightning strike are also obtained and shown in Fig. 7. It is In this regard, the admittance matrix of the multiport
known that these overvoltages are mainly due to the coupling grounding system was first obtained by making use of the
mechanism between the shield wire and the phase conductors MoM solution to Maxwell’s equations. The grounding system
[37] and [46]. It is again seen that these voltages are markedly admittance matrix was then rationally fitted by means of the
affected by the exact wideband modeling of the grounding grid. proposed modified-MPM, which uses a common set of poles for
each matrix element. The obtained rational approximation was
finally employed to construct a multiport time-domain model
VII. DISCUSSION
of the grounding system expressed in the form of state-space
This paper can be considered as a continuation of the work de- equations. The proposed approach was used to model the
veloped by Heimbach and Grcev in [10]. The grounding system grounding system of a real 63/20-kV substation.
admittance matrix is obtained by following the similar proce- It was shown that using the wideband model for grounding
dure as in [10]. However, the new contributions of this paper, systems results in a more realistic estimation of the light-
which have not been already discussed, can be summarized as ning-generated overvoltages compared to the traditional resis-
follows. tive model. In fact, using a resistive model for the grounding
a) It presents a general methodology based on the FD-MPM system leads to a considerable underestimation in the computa-
for providing a rational model of the grounding system tion of lightning-generated overvoltages within the grounding
admittance matrix in a direct procedure. It is different grid appearing at neutral points of power system equipment
from the iterative methods, such as the VF method and connected to the grounding system as well as on their terminals
the asymptotic method [10]. connected to the phase conductors. In addition, the proposed
SHESHYEKANI et al.: WIDEBAND MODELING OF LARGE GROUNDING SYSTEMS TO INTERFACE WITH ELECTROMAGNETIC TRANSIENT SOLVERS 1875

modeling approach was found to predict the wave propagation The matrix equation (A9) is solved by using QR decomposition
delay within the grounding system buried in lossy soil while with householder transformations [19].
the resistive model fails in the prediction of this effect.
Future works mostly deal with providing the grounding ACKNOWLEDGMENT
system block for the time-domain electromagnetic transient The authors would like to acknowledge valuable discussions
solvers. with Prof. F. Rachidi from the Swiss Federal Institute of Tech-
nology (EPFL).
APPENDIX
MPM IN THE FREQUENCY DOMAIN REFERENCES
To obtain a discrete frequency series ((2)), two matrices [1] “Alternative Transients Program (ATP) Rule Book,” Can/Am EMTP
and are constructed using samples given in (2) User Group, Leuven EMTP Center, Belgium, 1987.
[2] J. Mahseredjian, S. Dennetiere, L. Dube, B. Khodabakhchian, and L.
Gerin-Lajoie, “On a new approach for the simulation of transients
in power systems,” Elect. Power Syst. Res., vol. 77, no. 11, pp.
1514–1514, Sep. 2007.
[3] “PSCAD/EMTDC User’s Manual: Ver.4.2,” Manitoba HVDC Re-
(A1) search Centre, 2005.
[4] IEEE Guide for Safety in AC Substation Grounding, IEEE Standard 80,
where for . On the other Jan. 2000.
hand, and can be written as [5] IEEE Recommended Practice for Grounding of Industrial and Com-
mercial Power Systems (IEEE Green Book), IEEE Standard 142, 1991.
[6] L. Grcev and M. Popov, “On high-frequency circuit equivalents of a
vertical ground rod,” IEEE Trans. Power Del., vol. 20, no. 2, pt. 2, pp.
(A2) 1598–1603, Apr. 2005.
[7] M. Akbari, K. Sheshyekani, and M. R. Alemi, “The effect of frequency
In (A1) and (A2), is the pencil parameter and is set to to dependence of soil electrical parameters on the lightning performance
of grounding systems,” IEEE Trans. Electromagn. Compat., vol. 55,
efficiently filter out the noisy data [19] and no. 4, pp. 739–746, Aug. 2013.
[8] K. Sheshyekani, S. H. H. Sadeghi, R. Moini, F. Rachidi, and M.
Paolone, “Analysis of transmission lines with arrester termination,
considering the frequency-dependence of grounding systems,” IEEE
(A3) Trans .Electromagn. Compat., vol. 51, no. 4, pp. 986–994, Nov. 2009.
[9] F. Menter and L. Grcev, “EMTP based model for grounding system
analysis,” IEEE Trans. Power Del., vol. 9, no. 4, pp. 1838–1849, Oct.
(A4) 1994.
[10] M. Heimbach and L. Grcev, “Grounding system analysis in transients
where is given in (2). The matrix pencil is a generalized eigen- programs applying electromagnetic field approach,” IEEE Trans.
Power Del., vol. 12, no. 1, pp. 186–193, Jan. 1997.
value problem that could be formulated as . Ap- [11] M. Heimbach, EMC Analysis for Grounding Structures in Power Sys-
plying (A2), the matrix pencil can be formulated as tems, vol. 6, no. 3, pp. 157–162, May/Jun. 1996.
[12] S. Wang, J. He, B. Zhang, and R. Zeng, “Time-domain simulation of
(A5) small thin-wire structures above and buried in lossy ground using gen-
eralized modified mesh current method,” IEEE Trans. Power Del., vol.
26, no. 1, pp. 369–377, Jan. 2011.
where is the identity matrix, and in (A5) [13] T. Noda and S. Yokoyama, “Thin wire representation in finite differ-
shows that represents the generalized eigenvalues of the ma- ence time domain surge simulation,” IEEE Trans. Power Del., vol. 17,
trix pencil. Since is a diagonal matrix, , which is equal to no. 3, pp. 840–847, Jul. 2002.
[14] Y. Baba, N. Nagaoka, and A. Ametani, “Modeling of thin wires in
, is found. Hence a lossy medium for FDTD simulations,” IEEE Trans. Electromagn.
Compat., vol. 47, no. 1, pp. 54–60, Feb. 2005.
(A6) [15] Y. Liu, N. Theethayi, and R. Thottappillil, “An engineering model for
transient analysis of grounding system under lightning strikes: Nonuni-
where is the generalized eigenvectors corresponding to form transmission-line approach,” IEEE Trans. Power Del., vol. 20, no.
or in the equivalent form 2, pp. 722–730, Apr. 2005.
[16] L. Grcev and F. Dawalibi, “An electromagnetic model for transients
in grounding systems,” IEEE Trans. Power Del., vol. 5, no. 4, pp.
(A7) 1773–1781, Nov. 1990.
[17] K. Sheshyekani, H. R. Karami, P. Dehkhoda, M. Paolone, and F.
where is the Moore–Penrose pseudoinverse of or Rachidi, “Application of the matrix pencil method to rational fitting of
frequency-domain responses,” IEEE Trans Power Del., vol. 27, no. 4,
(A8) pp. 2399–2408, Oct. 2012.
[18] Y. Hua and T. K. Sarkar, “Matrix pencil method for estimating param-
and the superscript denotes the conjugate transpose eters of exponentially damped/undamped sinusoids in noise,” IEEE
Trans. Acoust., Speech, Signal Process., vol. 38, no. 5, pp. 814–824,
operation. May 1990.
Thus, we can also obtain by calculating the eigen- [19] T. K. Sarkar and O. Pereira, “Using the matrix pencil method to es-
values of . timate the parameters of a sum of complex exponentials,” IEEE Ant.
Propag. Mag., vol. 37, no. 1, pp. 48–55, Feb. 1995.
Finally, the complex amplitude can be easily computed [20] B. Gustavsen and A. Semlyen, “Rational approximation of frequency
by solving a least-squares problem as domain responses by vector fitting,” IEEE Trans. Power Del., vol. 14,
no. 3, pp. 1052–1061, Jul. 1999.
[21] B. Gustavsen, “Computer code for rational approximation of frequency
dependent admittance matrices,” IEEE Trans Power Del., vol. 17, no.
(A9) 4, pp. 1093–1098, Oct. 2002.
1876 IEEE TRANSACTIONS ON POWER DELIVERY, VOL. 29, NO. 4, AUGUST 2014

[22] B. Gustavsen, “Improving the pole relocating properties of vector fit- [44] CIGRE Working Group 01 (lightning) of Study Committee 33 (Over-
ting,” IEEE Trans. Power Del., vol. 21, no. 3, pp. 1587–1592, Jul. 2006. voltages and Insulation Co-ordination), “Guide to procedure for esti-
[23] D. Deschrijver, M. Mrozowski, T. Dhaene, and D. De Zutter, “Macro- mating the lightning performance of transmission lines,” 1991, docu-
modeling of multiport systems using a fast implementation of the ment 63.
vector fitting method,” IEEE Microw. Wireless Compon. Lett., vol. 18, [45] C. A. Balanis, Advanced Engineering Electromagnetics. New York,
no. 6, pp. 383–385, Jun. 2008. USA: Wiley, 1989.
[24] E. P. Li et al., “A coupled efficient and systematic full-wave time-do- [46] Y. Baba and M. Ishii, “Numerical electromagnetic field analysis on
main macro modeling and circuit simulation method for signal integrity lightning surge response of tower with shield wire,” IEEE Trans. Power
analysis of high-speed interconnects,” IEEE Trans. Adv. Packag., vol. Del., vol. 15, no. 3, pp. 1010–1015, Jul. 2000.
27, no. 1, pp. 213–223, Feb. 2004.
[25] H. Shi, “A closed-form approach to the inverse Fourier transform and
its applications,” IEEE Trans. Electromagn. Compat., vol. 50, no. 3,
pp. 669–677, Aug. 2008.
[26] J. H. Chung, “Efficient and physically consistent electromagnetic
macromodeling of high-speed interconnects exhibiting geometric Keyhan Sheshyekani (SM’13) was born in Iran
uncertainties,” Ph.D. dissertation, Dept. Elect. Comput. Eng., Univ. in 1979. He received the B.S. degree in electrical
Illinois, Urbana, IL, USA, 2012. engineering from Tehran University, Tehran, Iran, in
[27] A. V. Oppenheim, A. S. Willsky, and S. H. Nawab, Signal and Systems, 2001, and the M.S. and Ph.D. degrees in electrical
2nd ed. Upper Saddle River, NJ, USA: Prentice-Hall, 1997. engineering from Amirkabir University of Tech-
[28] L. Weinberg and P. Slepian, “Realizability conditions on -port net- nology (Tehran Polytechnique), Tehran, in 2003 and
work,” Inst. Radio Eng. Trans. Circuit Theory, pp. 217–221, Sep. 1958. 2008, respectively.
[29] B. Gustavsen and A. Semlyen, “Enforcing passivity for admittance ma- He was with Ecole Polytechnique, Federale de
trices approximated by rational functions,” IEEE Trans Power Syst., Lausanne Lausanne, Switzerland, in 2007 as a
vol. 16, no. 1, pp. 97–104, Feb. 2001. Visiting Scientist and later as a Research Assistant.
[30] A. Borghetti, A. S. Morched, F. Napolitano, C. A. Nucci, and M. Currently, he is an Assistant Professor of Electrical
Paolone, “Lightning-induced overvoltages transferred through distri- Engineering with Shahid Beheshti University, Tehran. His research interests
bution power transformers,” IEEE Trans. Power Del., vol. 24, no. 1, include power system modeling and simulation, smart grid, microgrids, and
pp. 360–372, Jan. 2009. electromagnetic compatibility.
[31] B. Gustavsen, “Fast passivity enforcement for pole-residue models by
perturbation of residue matrix eigenvalues,” IEEE Trans. Power Del.,
vol. 23, no. 4, pp. 2278–2285, Oct. 2008.
[32] A. Semlyen and B. Gustavsen, “A half-size singularity test matrix for
Majed Akbari (S’13) was born in Iran in 1988. He
fast and reliable passivity of rational models,” IEEE Trans. Power Del.,
received the B.S. and M. S. degrees (Hons.) in elec-
vol. 24, no. 1, pp. 345–351, Jan. 2009.
trical engineering from Shahid Beheshti University,
[33] B. Gustavsen and A. Semlyen, “A robust approach for system identifi-
Tehran, Iran, in 2010 and 2012, respectively.
cation in the frequency domain,” IEEE Trans. Power Del., vol. 19, no.
Currently, he is a Senior Consultant, involved with
3, pp. 1167–1173, Jul. 2004.
Tehran’s subway system and a Research Assistant
[34] M. Akbari, K. Sheshyekani, A. Pirayesh, F. Rachidi, M. Paolone, A.
at Sharif University of Technology, Tehran. His
Borghetti, and C. A. Nucci, “Evaluation of lightning electromagnetic
research interests include power system transients,
fields and their induced voltages on overhead lines considering the fre-
transportation systems, sustainable energy resources
quency-dependence of soil electrical parameters,” IEEE Trans. Elec-
with a particular reference to solar-cell systems, elec-
tromagn. Compat., vol. 55, no. 6, pp. 1210–1219, Dec. 2013.
tromagnetic compatibility, and grounding systems.
[35] K. Sheshyekani, S. H. H. Sadeghi, R. Moini, and F. Rachidi, “Fre-
quency-domain analysis of ground electrodes buried in an ionized soil
when subjected to surge currents: A MoM-AOM approach,” Elect.
Power Syst. Res., vol. 81, no. 2, pp. 290–296, Feb. 2011.
[36] L. Grcev, “Lightning surge efficiency of grounding grids,” IEEE Trans. Barzan Tabei was born in Kurdistan, Iran, in 1988.
Power Del., vol. 26, no. 3, pp. 1692–1699, Jul. 2011. He received the B.S. degree in electrical engineering
[37] X. Legrand, A. Xémard, P. Auriol, C. A. Nucci, and C. Mouychard, from Tabriz University, Tabriz, Iran, in 2010, and the
“Modeling of substation grounding for fast front overvoltage studies,” M.S. degree in electrical engineering from Shahid
presented at the Int. Conf. Power Syst. Transients, Lyon, France, Jun. Beheshti University, Tehran, in 2014.
2007. His research interests include power system mod-
[38] X. Legrand, A. Xemard, C. A. Nucci, and P. Auriol, “A method to eling for electromagnetic transient studies.
interface electromagnetic models of grounding systems with transients
programs,” presented at the CIGRE Int. Colloq. Power Qual. Light.,
Sarajevo, Bosnia and Herzegovina, May 2012.
[39] COMSOL Multiphysics Software Package. 2011. [Online]. Available:
http://www.comsol.com
[40] “Modeling of metal oxide surge arresters,” IEEE Trans. Power Del.,
vol. 7, no. 1, pp. 302–310, Jan. 1992.
[41] J. R. Marti, “Accurate modelling of frequency-dependent transmission Reza Kazemi (S’08) was born in Iran in 1987. He
lines in electromagnetic transient simulations,” IEEE Trans. Power received the B.S. and M.S. degrees in electrical en-
App. Syst., vol. PAS-101, no. 1, pp. 147–157, Jan. 1982. gineering from Amirkabir University of Technology
[42] I. M. Dudurych, T. J. Gallagher, J. Corbett, and M. Val Escudero, (Tehran Polytechnique), Tehran, Iran, in 2009 and
“EMTP analysis of the lightning performance of a HV transmission 2012, respectively.
line,” Proc. Inst. Elect. Eng., Gen., Transm. Distrib., vol. 150, no. 4, His research interests include power system elec-
pp. 501–506, Jul. 2003. tromagnetic transients, lightning protection, integra-
[43] B. Badrzadeh, M. Hogdahl, and E. Isabegovic, “Transients in wind tion of renewable energy sources, and smart grids.
power plants – Part I: Modeling methodology and validation,” in Proc.
Ind. Appl. Soc. Annu. Meeting, Nov. 2011, pp. 1–11.