IEEE TRANSACTIONS ON MAGNETICS, VOL. 48, NO.

2, FEBRUARY 2012 207

Transient Calculation of Electromagnetic Field for Grounding System

Based on Consideration of Displacement Current
Mladen Trlep, Marko Jesenik, and Anton Hamler
Faculty of Electrical Engineering and Computer Science, University of Maribor, Maribor 2000, Slovenia

This paper shows the use of FEM for a second order time dependent electromagnetic field problem, around grounding systems (GS).
Twenty-node isoparametric quadratic 3D finite element, three-node quadratic 1D finite element and a spatial transformation of the “in-
finite space” into the finite space are all applied to achieve better accuracy. Time integration is conducted with the Newmark algorithm.
The applied program solution is suitable for any GS and isotropic/anisotropic soil properties as well as time-varying fault current.
Index Terms—Electromagnetic modeling, electromagnetic transient, finite element methods, grounding, numerical simulation.

I. INTRODUCTION Equation (1.b) defines a well-known relation: divergence of

the total current density that occurs in the conductive do-
main is equal to zero. consists of two components: conduc-
HE PRIMARY goal of GS is to ensure the safety of
T personnel and prevent damage of installations. Their
secondary goal is to provide a common reference voltage
tion current density
sity (1.d).
(1.c) and displacement current den-

for all interconnected electrical and electronic systems. The (1.c)

program tool that is able to simulate the transient performance
of grounding systems is fundamental, because it enables the
optimization of the GS design, as well as the minimization of (1.d)
the disturbance level in the protected area. For that very reason,
the goal of this research is to develop the methodology which The following equation can be written for the non-conductive
allows a complete three-dimensional transient calculation of domain:
electromagnetic field, including the displacement current.
So far, three basic concepts have been used to simulate the (2)
transient performance of grounding arrangements: the circuit
approach, the transmission line approach, and the electromag-
Where is the permeability, the electrical conductivity and
netic field approach [1]. In this paper, the solution to analyze the
is the permittivity. Equations (1.a) and (1.b) already contain the
transient behavior of grounding system is based on the electro-
Coulumb’s gauge to ensure the unique solution to the magnetic
magnetic field theory and on the implementation of FEM. The
vector potential , which is given in greater detail in [2]. By
validity of the suggested method of analysis has been verified
applying the finite elements procedure and weighted residual
by the comparison of obtained results with the numerical and
method [2], the following equations for conductive domain are
experimental results found in [7].
obtained (3a), (3b):

II. FEM MODEL OF TRANSIENT ELECTROMAGNETIC FIELD

WITH CONSIDERATION OF DISPLACEMENT CURRENT
The governing partial differential equation for transient prob- (3a)
lems of GS can be derived from Maxwell’s equations. When
magnetic vector potential , electric scalar potential , conduc-
tion and displacement current are introduced to the conductive (3b)
domain the following equations are obtained:
For the non-conductive domain the (4) is given:

(4)
(1.a)
The described problem is an open boundary problem. There-
(1.b) fore, the numerical model includes the spatial transformation
[3], which divides the total domain of the open boundary
problem onto a non-transformed inner domain and the trans-
formed outer (infinite) domain. In order to get an accurate
Manuscript received June 27, 2011; accepted October 10, 2011. Date of cur- field calculation, the soil and the air in the transformed and
rent version January 25, 2012. Corresponding author: M. Trlep (e-mail: mladen.
trlep@uni-mb.si).
non-transformed domain of the problem are discretized by 20
Color versions of one or more of the figures in this paper are available online nodes second order 3D finite elements. The conductors of the
at http://ieeexplore.ieee.org. grounding grid or rods are discretized by the 3 nodes second
Digital Object Identifier 10.1109/TMAG.2011.2172918 order 1D finite elements [6], [8]. The final FEM equation is
0018-9464/$31.00 © 2012 IEEE
208 IEEE TRANSACTIONS ON MAGNETICS, VOL. 48, NO. 2, FEBRUARY 2012

represented by a system of second order ordinary differential

(5), where (6) represents the modified electric scalar po-
tential [2], in order to ensure that matrices , and are
symmetrical.

(5)

(6)

The column vector of unknown nodal potentials in (5) is

. Next column vectors are the first and the second
derivative of nodal potentials and , , are corresponding
matrices, which are linked with the potentials and and
the Laplacian operator ; with the conduction current
and with the displacement current . Time integration (5)
can be conducted with different time-step algorithms such as
Newmark’s, Crank-Nicolson’s, Wilson’s and others [8]. With
an assumption of linear interpolation throughout time [8], and
with the use of the Newmark algorithm, the following recursive
(7) is obtained from (5):

Fig. 1. Time step algorithm based on: I. the known voltage waveform u(t), II.
The known current waveform i(t).
(7)

where parameters and are given by the following expres- corresponding current is the same as in case I. The
sions: and . The recursive (7) enables given current and calculated current must be equal after
the calculation of the potential, in the new time step , de- the iteration process is completed.
pending on the preceding time step (n). During the GS analysis, In order to achieve such equality the iteration loop is nec-
two cases can occur: essary. Inside the current iteration loop changes
I. Voltage waveform is given for the injection point until the acceptable difference (EPS%) is reached between the
(IP), where the current is injected. It is necessary to cal- given value of current and calculated value of current
culate . . The presented cases show that between three and
II. Injected current waveform is given and it is necessary six iterations, at the allowed error of have been
to calculate for the injection point. necessary to calculate the current within a single time interval.
Result for both cases is the same final (5) and the algorithm It means that the case II is more time consuming, since the
(7) is used for time integration. However, in case II, it is nec- calculation time is approximately , where N_iter
essary, within each and every time interval, to additionally use represents the average number of iterations for a single time
current iteration loop, which is shown in the Fig. 1. interval.
Voltage Waveform is Given: It can be seen from Fig. 1
that for each time moment , it is necessary to set up the III. APPLICATION AND RESULTS
value of the electric potential for the injection point The application of program solution for both approaches
that represents an initial condition. The value of the electric po- is shown for the grounding grid from [7]. The considered
tential can be obtained from voltage waveform : grounding grid is of a rectangular type with dimensions 10 m
. By applying (7), and 10 m. It is made out of round copper conductors that are 50
are calculated for the domain and afterwards the injected cur- in cross section. There are four meshes in a grid, each
rent is calculated as well . The procedure is repeated measuring 5 m 5 m, as shown in Fig. 2. The grounding
for all time moments, within time interval and . grid is buried 0,5 m horizontally below the earth surface into
Current Waveform is Given: The case II (Fig. 1) also a two-layer soil with the upper-layer resistivity (from earth
requires to set up the value of the electric potential surface to the depth of 0,6 m) of 50 and the bottom-layer
for the injection point for the time moment . The value of resistivity of 20 and relative permittivity .
the potential is now unknown. Although the is known, it The application of our FEM solution is presented to analyze
does not help with formulation and it does not enable to set the behaviour of the grounding grid when fed by an injected time
up directly the potential value for the injected point. variable voltage or current at the discharge point, and the ratio
It can be solved by applying the iteration approach. On the basis between the nominal time increase and time to half-value the
of the already calculated potential, in the previous time moment impulse waveform voltage front respec-
, as well as the known current value and , tively. The Fig. 3(a) shows the comparison between the calcu-
the values of potential can be assumed. Therefore, lated and measured current (case I). Fig. 3(b) shows a similar
the procedure to calculate and as well as the comparison for case II.
TRLEP et al.: TRANSIENT CALCULATION OF ELECTROMAGNETIC FIELD FOR GROUNDING SYSTEM
Fig. 2. Schematic view of the grounding grid under analysis.
Fig. 3. (a) Comparison of the measured and calculated transient current. (b)
Comparison of the measured and calculated transient potential.
Comparison between cases I and II shows that the difference
among calculated and measured values exists, which is the con-
sequence of time integration, input data accuracy (conductivity
and permittivity) and accuracy of the measured and
. It is important to be aware of the fact that in case I the
input data is and the current , is calculated.
The case II is reversed; the input data is and the po-
tential is calculated. In case II, the chosen value of
the EPS% additionally influences on the accuracy of the com-
putation. Both cases have been calculated with (7) for
and . However, the notable differences between the
time behavior of respectively have not been
detected. The consideration of the displacement current
during the transient calculation of the electromagnetic field of
the GS makes the analysis even more complex. The problem,
which is of the first order (without ), becomes the second
order time dependent problem (7). Therefore, the question arises
when is it necessary to consider the as well? If the consid-
ered calculation is transient and the injected current in a GS is a
pulse current, the answer to the previous question lies in the ratio
between the electric conductivity and permittivity of the soil and
in the ratio between the injected current slope and cur-
rent value. The case I, however, requires the ratio of the driven
voltage slope and voltage value. Let us assumed the
the specific conductivity of soil is between
and or even less, relative permittivity is
209
TABLE I
RATIO BETWEEN SOIL PROPERTIES  AND "
Fig. 4. (a) Transient potentials for the IP of the grounding grid. (b) Components
of the transient current of the grounding grid.
between and . Therefore, by applying
the given (1.c), (1.d), the ratio between and can be
expressed with (8), which provides the important information
when there is a possibility of the greatest influence of the dis-
placement current.
(8)
If the ratio is calculated for all combinations of and
(cases: a, b, c, d), the following values are obtained, Table I.
It can be noticed that the contribution of is the largest
in cases when soil has a low conductivity and high permittivity
(case d). The contribution is significantly lower when the con-
ductivity is high and the permittivity is low (case a). The com-
parison for all four combinations of and has been conducted
for the case of grounding grid (Fig. 2) and standardized lightning
current waveform [9] (IEC 1312-1, first stroke, ,
, ). The analysis has confirmed previous
conclusions that the displacement current is relatively speaking
the highest in case d. Only in that case the displacement cur-
rent is comparable to the conduction current but only within the
time interval when the potential and/or total current increases.
In all other cases (a, b, c) the displacement current is neglected.
210
Fig. 5. (a) Components of the transient current of the grounding grid. (b) Com-
ponents of the transient current of the grounding grid.
Fig. 4(a) shows the comparison between time behaviour of po-
tentials for the injection point. Fig. 4(b) shows comparison be-
tween conduction and displacement current for case d. The dis-
placement current can be neglected even in theworst case (case
d), because the percentage value varies below 1% of , except
at the beginning of the time response when it can even reach
30% of . substation grounding systems,” IEEE Trans. Magn., vol. 31, no. 3, pp.
The crucial role is also played by the slope and am-
plitude of the injected current. If the analysis is repeated under
lightning current waveform of higher slope [9] (IEC 1312-1,
subsequent stroke, , ,
the results are completely different. The displacement current is
more obvious as it is shown in the following graphs. Fig. 5(a)
shows the situation for case d, in which the contribution is the
largest, as it is the case in the previous example (first stroke). The
influence of the displacement current is crucial because, at the
beginning, it even dominates over the conduction current. After-
wards the values are approximately the same and when the total
current reaches the maximum, then conduction current domi-
nates over the displacement current.
IEEE TRANSACTIONS ON MAGNETICS, VOL. 48, NO. 2, FEBRUARY 2012
Fig. 5(b) shows new case c. In the previous case c the
displacement current was negligible. The circumstances are
now completely different and the displacement current is high
enough and cannot be neglected.
IV. CONCLUSION
The paper presents transient calculation of electromagnetic
field for grounding system based on consideration of displace-
ment current, which represents a time dependent problem of the
2nd order. Usually the contribution of displacement current is
taken into consideration in the case of low soil conductivity. The
displacement current can be neglected when the soil conduc-
tivity is high. The permittivity is also an important parameter,
however, due to a soil type, it does not change significantly. The
slope of current waveform is probably the most impor-
tant. Analysis of circumstances for boundary values of electric
conductivity and permittivity of normal soil, as well as for two
characteristics current waveform (IEC 1312) gives the following
conclusions. In the case of current waveform, comparable to the
first stroke, the contribution of the displacement current is very
limited. During the stroke with extremely high slope, compa-
rable to the IEC 1312, subsequent stroke, the contribution of the
displacement current must be taken into consideration, which
leads to an accurate calculation of electromagnetic field for the
grounding system.
REFERENCES
[1] L. Grcev and F. Dawalibi, “An electromagnetic model for transients
in grounding systems,” IEEE Trans. Power Delivery, vol. 5, no. 4, pp.
1773–1781, Oct. 1990.
[2] O. Biro and K. Preis, “On the use of the magnetic vector potential in
the finite element analysis of the three dimensional eddy current,” IEEE
Trans. Magn., vol. 25, no. 4, pp. 3145–3159, Jul. 1989.
[3] A. Stochniol, “A general transformation for open boundary finite ele-
ment method for electromagnetic problems,” IEEE Trans. Magn., vol.
28, no. 2, pp. 1679–1681, Mar. 1992.
[4] B. Nekhoul, C. Guerin, P. Labie, G. Meunier, and R. Feuillet, “A finite
element method for calculating the electromagnetic fields generated by
2150–2153, May 1995.
[5] B. Nekhoul, P. Labie, F. X. Zgainski, and G. Meunier, “Calculating the
impedance of a grounding system,” IEEE Trans. Magn., vol. 32, no. 3,
pp. 1509–1512, May 1996.
), [6] M. Trlep, A. Hamler, and B. Hribernik, “The analysis of complex
grounding systems by FEM,” IEEE Trans. Magn., vol. 34, pp.
2521–2524, Sep. 1998.
[7] Z. Stojković, M. S. Savić, J. M. Nahman, D. Salamon, and B. Buko-
rović, “Sensitivity analysis of experimentally determined grounding
grid impulse characteristics,” IEEE Trans. Power Delivery, vol. 13, no.
4, pp. 1136–1141, Oct. 1998.
[8] I. M. Smith and D. V. Griffiths, Programming the Finite Element
Methods, 4th ed. Hoboken: Wiley, 2004.
[9] Protection Against Lightning Electromagnetic Impulse. Part 1:
General Principles, IEC International Standard 1312-1, IEC, 1995,
Geneva, Switzerland.