GROUND2014 International Conference on Grounding and Earthing

&
& 6th International Conference on
Lightning Physics and Effects
6th LPE Manaus, Brazil May, 2014

A PROCEDURE FOR MODELING GROUNDING SYSTEMS CONSIDERING THE SOIL IONIZATION

EFFECT USING COMSOL MULTIPHYSICS

Eduardo H. Mafra, Rafael A. Coelho, Hugo A. D. Almaguer

Vilson L. Coelho* and Alexandre Piantini**
Regional University of Blumenau, FASATC Faculty SATC* and University of So Paulo**
Brazil

Abstract This paper aims at presenting a procedure
developed to improve the use of COMSOL Multiphysics for
simulating grounding systems, taking into account the soil
ionization phenomena. The approach has been validated by
comparing the numerical results with data reported in the
technical literature.
1 - INTRODUCTION
The soil ionization is an important phenomenon to be
considered in the analysis of grounding systems. When
lightning strikes a grounded structure, the electric field
associated with the high impulsive current injected into
the grounding conductors may cause the soil around
them to breakdown. This phenomenon causes a
decrease in the electric potential in the area of soil
ionization and consequently reduces the value of the
ground impedance. The transient characteristics of the
grounding systems during the soil ionization process are
typically non-linear [1 - 5].
problems, apparently it has not been used in applications
involving the transient behavior of grounding systems.
Then, another objective of this work is to demonstrate
that this software appears as an attractive alternative to
be used as a computational tool in studies related to this
subject matter.
2 - MODELING OF SOIL IONIZATION PHENOMENA
As mentioned previously, in this study the effect of soil
ionization on the transient behavior of grounding systems
will be considered from the use of DM and EBM models.
Both models, when implemented in a numerical method,
allow to simulate different types of grounding topologies,
regardless of their geometric characteristics. In this
section, brief descriptions are provided for the DM and
EBM.
2.1 Dynamic Model (DM)

Several mathematical models have been proposed to
take into account the dynamic effect of soil ionization on
grounding systems submitted to lightning currents. Some
of them, like the Dynamic Model (DM) [2, 4] and the
Energy Balance Model (EBM) [3, 6] make use of
analytical functions to represent the space-time variation
of the soil resistivity. In both models the resistivity of the
soil ionization areas decreases due to the discharge and
then gradually recovers its initial value when the
deionization process occurs. These approaches seem to
be closer to the physical phenomenon than the purely
geometric ones [5]. In order to implement computationally
the different methodologies for analyzing the transient
behavior of grounding systems, some numerical methods
such as e.g. the Finite Difference Time Domain (FDTD),
the Finite Element Method FEM, and the Transmission
Line Model (TLM) have been adapted and used.
The main goal of this work is to use the software
COMSOL Multiphysics [7] as a modeling tool for typical
The Dynamic Model proposed by Liew and Darweniza [4]
was initially applied to concentrated electrodes. The
method is based on experimental results and considers a
space-time variable function of soil resistivity for the
region surrounding the electrode.
When the actual electric field E (V/m) in the soil region
under analysis exceeds the critical electric field strength
Ec (V/m), the ionization begins and the following
expression is adopted for the resistivity [2, 4]:



=
 . e  (1)
where 0 (m) is the steady state resistivity, t (s) is the
time, and t1 (s) is the ionization time constant. In the
same manner, during the deionization process the time
variable resistivity behavior is expressed as [2, 4]:
 



=
 + 

 . 1    . 1
(2)
grounding topologies subjected to lightning currents, 
taking into account the soil ionization effects predicted by
the DM and EBM. For this purpose, a procedure has
been developed to implement efficiently the soil
ionization algorithms using LiveLink for MATLAB, an
interfacing module that allows integrating COMSOL with
MATLAB to extend the modeling possibilities with
scripting programming in the MATLAB environment.
Although COMSOL has been used successfully for many
years in solving a lot of physical and engineering
Session VI - P51 - 252
where
 (m) is the minimum value reached by the soil
resistivity during the ionization and t2 (s) is the
deionization time constant. The parameters t1 and t2
depend only on the surge waveform and must be
obtained experimentally [2, 4].
Equations (1) and (2) show that the soil resistivity decays
exponentially when the electric field on the soil region
exceeds the critical value. Afterwards, when the electric
field decreases, the deionization process starts. This
process stops when the initial stationary value of the soil
resistivity is restored.
2.2 Energy Balance Model (EBM)
Similarly to the DM, the Energy Balance Model describes
the soil ionization by changing the material properties of
the soil and not the geometry of the electrodes.
The model was derived from the premise that the soil
ionization process is similar to the arc phenomenon
occurring inside a circuit breaker during a current
interruption [3]. Therefore, the soil ionization could be
represented according to an energy balance [6] which
associates the input energy with the variation of the
resistivity.
3.1 DEFINING THE SOIL IONIZATION ZONES BY
EQUIPOTENTIAL SURFACES
It is well known that soil breakdown takes place when the
electric field overcomes the soil dielectric strength. So, it
is possible to divide the soil in various regions, according
to the geometry of the equipotential surfaces formed from
the injected electrode. Figure 1 shows the distribution of
the equipotential lines for a vertical rod. The electric field
values in each sub-division should be the same or very
close from each other to ensure that the computation
results of the ionization on each soil division will be very
close to those obtained from each individual region cell.

In this way, Sekioka [3] proposed a current-dependent

grounding resistance model based on the energy balance
of soil ionization. The general expression for the
resistivity is given by

  


=
 .  1 +   .  !" (3)
 


#= (4) (a)
$

where 0 (m) and % (Sm) are the initial soil resistivity

and conductance, respectively, & (J/m) is an energy
constant, i (A) is the discharge current, P (W/m) is the
power loss, and (s) is the time constant for the
resistivity to recover its initial value [3]. The constant &
must be obtained from experimentally.

3 - DESCRIPTION OF THE PROPOSED PROCEDURE

Several numerical methods are used by COMSOL, but

the emphasis is on the Finite Element Method. Automatic
meshing tools are also available. The meshes may be (b)
constituted by tetrahedral, hexahedral and pyramidal Figure 1 - Equipotential lines for a vertical rod. (a) Top view.
elements [7]. (b) Side view.

In a typical COMSOL mesh for modeling grounding As an example, Figure 2 shows the soil divisions for a
systems, a very large number of cells are used even for single vertical rod embedded in a homogeneous soil. The
very simple configurations such as a vertical rod. Then, rod is at the center of the discretized domain and the soil
during the simulations, the soil ionization algorithms regions are cylindrical and concentric to the electrode.
implemented in the MATLAB codes must be solved for During the COMSOL time domain simulation, the
each time step of the iterative process in each cell of the MATLAB codes run and compare, step by step, in each
mesh, in order to upgrade the values of the soil region, the calculated electric fields with the soil dielectric
resistivity. That means that the whole simulation process strength. These MATLAB codes have been developed
is computationally very expensive in terms of processing with the implementation of the time-variable soil resistivity
time and memory space required. approaches proposed by the DM [2] and EBM [3]
methods. Thus the soil zones can be ionized or not
In order to overcome this problem and at the same time depending on the local electric field intensity.
take into account correctly the nonlinear behavior of the
grounding system during the ionization, a new approach
is proposed in this work. Basically, the procedure
consists in dividing the soil in various regions or zones
from the excitation electrode. For each region the soil
ionization is assessed; there is no need to do this at each
node. Thus, a significant reduction in processing time
and memory storage is achieved. The proposed scheme
consists of three basic steps which will be briefly
described in this section.
Figure 2 Example of soil divisions for the ionization study in the
case of a single vertical rod.

Session VI - P51 - 253

3.2 DEFINITION OF THE MAXIMUM EXTENSION OF
THE IONIZED SOIL REGION

The process of soil ionization for concentrated electrodes

occurs mainly around the current injection point on the
ground surface. This means that, in a typical situation,
only a small portion of the study domain that is being
modeled suffers from this phenomenon. Then, in order to
improve the computational efficiency of the simulations,
the maximal radial distance where the ionization occurs
was obtained. Therefore, the soil division described
Figure 3 - Electric field intensity as function of the radial distance
previously will be defined according to this radius and not in the instant when the lightning current reaches its peak value.
in the whole domain to be modeled.
For this example and considering a typical electrical
Equations (5) to (11) were deduced in cylindrical strength (Ec = 300 kV/m), the maximum extension
coordinates for finding the final equation (12) which considered for the ionized soil region is rm = 53.05 cm.
represents the maximum radius of ionization. Assuming
that the current ' A injected into the top of the rod is 3.3 CALCULATION OF THE CURRENT FLOWING IN
distributed uniformly across its lateral surface and EACH SOIL ZONE
*
defining )  
as the current density, then
+
According to the Energy Balance Model [3], the time
(5)
variable soil resistivity is directly proportional to the
' = ) -.
injected current which penetrates into the soil divisions.
On the other hand, the AC/DC COMSOL module used in
3 2
' =   ) / 0 1 (6) this work allows computing the electric potential to
remote ground for all cells of the domain modeled.
' = 4 26 7 / (7) Therefore, in order to obtain this current, the Ohm's Law
is applied in each predefined soil zone, considering the
8
)=
 3 9 2
:; (8) difference between the electric potentials at the internal
and external radius of the region, valid for the cylindrical
symmetry of concentrated electrodes.
where ' A is the current flowing from the rod into the
soil, r m is the radial distance from the rod, L m is the D
length of the electrode and :; is the unit vector in the ' = (13),
E
radial direction. The radial distance can be computed
from the point form of Ohm's Law, where V V is the difference between the electrical
potentials of the two limits of the soil region under
) = @

(9) analysis and R  is the resistance of this division,
?
calculated according to the direction in which the current
is injected. However, two different situations must be
@=' :;

assessed to find R, as will be described next.
(10)
 3 9 2

/ = '

(11) 1. Rod's side regions, in which the current flows to the
 3  2
soil in the radial direction from the rod, as indicated in
where (m) is the soil electric resistivity and A (V/m) is Figure 4
the electric field.

The upper limit of the radius of the ionization zone is

obtained from (11) for the case of electric field equal to
the critical electric field value A (V/m) and current equal
to the peak value of the injected current 'B (A):

/+ = 'C (12).
 3  2

Although in the general case the current is not uniformly

distributed, the radius obtained from (12) can be used as
a reference during the modelling. To illustrate the
application of the procedure described above, a
simulation was done for the case of a vertical rod with
length and diameter of 1 m and 1.6 cm, respectively. The
soil resistivity is 100 m and the peak value of the
lightning current is equal to 10 kA. Figure 3 shows the
behaviour of the electric field intensity, as a function of
the radial distance from the rod.
Figure 4 Rod's side regions.

Session VI - P51 - 254

The potential difference between the radial limits of the The resistance is obtained from (22) as the ratio V(L) / I:
soil region is given by
2
J
S= (23)
I= K @ -;
3 J K 
(14),

where b m and a m are, respectively, the external and
internal radii of the soil region. Substituting (10) into (14)
and integrating:
I = ' NO P/R
 3 2
The resistance is obtained from the Ohm's Law, so that
S = NO P/R
 3 2
3.4 COMPUTED PROCEDURE FLOWCHART
A flowchart of the algorithm of the proposed procedure is
presented in Figure 6. It shows the exchange of
(15). information between the simulation in COMSOL and the
parallel algorithm in MATLAB, in which the simulations
using the ionization models DM and EBM are performed.
(16).

2. Regions below the bottom rod end, in which the

current flows orthogonally from the rod, as indicated in
Figure 5

The same development of the previous case was

followed to find an expression to compute the resistance
of the soil regions below the bottom rod end to the
current flowing into the soil. Now, the current flux is in the
z-direction.
2
I =  @ -T (17)

3 J
' =  K ) / 0 / (18)

' = 4 6P  R   (19)

Figure 6 - Flowchart of simulation procedure.
8
)= :T (20)
3 J K 

where :T is the unit vector in the axial direction. 4 - RESULTS

Substituting (20) into (9) and isolating E: 4.1 VALIDATION

The proposed procedure has been validated by
@= ' :T (21)
3 J K 
comparing the results of the simulations with data
reported in the technical literature related to concentrated
Substituting (21) into (17) and integrating: electrodes. Specifically in this paper, a case from [2] was
used for the analysis.
I7 = ' 7 (22)
3 J K 
In [2], a FDTD scheme was used to model a single rod,
61 cm long and with radius of 1.4 cm, embedded in
homogeneous soil ( = 50 m, r = 8, and Ec = 110
kV/m), using the Dynamic Model approach. The lightning
current injected into the top of the electrode has a peak
value equal to 3.5 kA and a double-exponential waveform
(5/16.5 s). The results voltage at the injected point to
the remote ground obtained from the simulations using
COMSOL with those presented in [2] (using FDTD) are
compared in Figure 7a. A comparison of the transient
ground impedances is shown in Figure 7b.

Figure 5 Regions below the bottom rod end.

Session VI - P51 - 255

 (a) (a)

(b)

(b) Figure 8 - Comparison between the DM (blue) and EBM (red)

results for a vertical rod. (a) Voltage at the injection point.
Figure 7 - Comparison between COMSOL results (DM (b) Transient grounding impedance.
approach) and FDTD simulations [2]. (a) Voltage at the injection
point. (b) Transient ground impedance.

4.2 FURTHER SIMULATIONS

In order to compare the results obtained with the EBM

and DM, a common grounding topology for electrical
distribution systems was simulated: a vertical steel
copperweld rod, 2.4 m long, and with diameter of 14 mm.
The soil resistivity was assumed to be 500 m and the
relative permittivity (r) was equal to 10. The critical
electric field was set to 300 kV/m. The lightning current
injected into the top of the electrode has a double-
exponential waveform (1.5/6 s) with peak value of
10 kA. The constants of the DM were 0.5 s (ionization
time constant) and 1 s (deionization time constant). The
Figure 9 - Geometric dimensions of the simulated
energy constant Qo for the EBM was 7000 J/m. The
concrete pole base.
voltage at the injection point and the transient grounding
impedance calculated with the two methods are shown in
The soil properties were the same of the previous case
Figure 8.
(single rod). For the concrete block, the following values
were adopted for the electric parameters: = 75 m and
A more complex structure was considered for the
r = 5. The conductive reinforcements were modeled by
evaluation of the COMSOL capabilities and the proposed
steel bars with length of 1.8 m and diameter of 10 mm.
procedure: a concrete pole base acting as a grounding
The field strength was set to 300 kV/m. The values
topology for overhead power distribution lines. This kind
adopted for the ionization and deionization time
of structure has an internal metal reinforcement mesh
constants for the DM were 4 s and 8 s, respectively.
and penetrates the soil to a depth of up to a few meters.
The energy constant Qo for the EBM was 5000 J/m. The
electrode located in the central part of the structure was
The geometric characteristics used to model the concrete
chosen as the excitation electrode. The injected current
pole base are presented in Figure 9. This case
surge has a double-exponential waveform (3/10 s) with
corresponds to a 12 m long "Double T" type concrete
peak value equal to 10.7 kA. Figure 10 shows the
pole with mechanic resistance of 300 daN and base
distribution of the equipotential lines at the instant when
1.8 m deep [8]. This pole base has six reinforcement
the current surge reaches its peak value.
rods.

Session VI - P51 - 256


Figure 10 Distribution of the equipotential lines for
a concrete pole base.
It may be noted that, although this is a more complex
structure, the potential distribution is mostly radial and
has a similar behavior to the single rod shown in Figure
1. This indicates that the pole base can be regarded as a
concentrated ground and the procedure proposed in this
paper can be applied.
Figure 11 shows the potential at the injection point and
the transient grounding impedance obtained for both
models of soil ionization.
(a)
(b)
Figure 11 - Comparison between DM (blue) and EBM (red)
results for a concrete pole base. (a) Voltage at the injection
point. (b) Transient ground impedance.
As shown in Figures 8 and 11, both ground structures
single vertical rod and pole base have similar
behaviors. That is, at the beginning of the process, when
the electric field is growing but still below the critical
value, the impedance curves tend to reach the steady
state values. When the soil ionization begins, it causes a
Session VI - P51 - 257
significant and fast decrease of the impedance values.
Thereafter, during the deionization stage, the transient
impedance attempts to return to the steady state value,
but more slowly. A comparison between the ionization
models shows that, for the DM, the deionization process
starts immediately when the instantaneous electric field
gets below the critical value. Then the impedance returns
to the steady state value faster than in the case of the
EBM, for which the deionization occurs only when the
accumulated energy is smaller than the accumulated
losses in the soil, which makes the process slower.
5 - CONCLUSIONS
This paper has proposed a procedure that allows
modeling grounding systems with COMSOL Multiphysics
and MATLAB taking into account the soil ionization
effects according to the Dynamic and Energy Balance
models. Comparisons between calculated results and
results reported in the technical literature show a
satisfactory agreement and validate the proposed
procedure. The behavior of both ionization models were
compared for some grounding topologies commonly used
in electric distribution systems. The next steps of the
investigation include the development of experiments to
enable comparison between measured and calculated
results.
6 ACKNOWLEDGEMENTS
This work was supported by the CPFL and RGE electric
power companies and developed within the framework of
the Brazilian Electricity Regulatory Agency (ANEEL) R&D
Program. Special thanks are due to Eng. Plinio L. Nosaki,
from RGE, for his important support.
7 REFERENCES
[1] V. Cooray, Lightning Protection, IET Power and Energy
Series, Vol. 58, 2010.
[2] G. Ala; P. L. Buccheri; P. Romano and F. Viola, Finite
difference time domain simulation of earth electrodes soil
ionization under lightning surge condition, IET Science,
Measurement and Technology, Vol. 2, No. 3, pp. 134-145, 2008.
[3] S. Sekioka et al., Current-Dependent Grounding Resistance
Model Based on Energy Balance of Soil Ionization, IEEE
Transactions on Power Delivery, Vol. 21, No. 1, pp. 194-201,
Jan. 2006.
[4] A. C. Liew and M. Darveniza, "Dynamic Model of impulse
characteristics of concentrated earths", Proc. IEE, Vol. 121,
No.2, pp. 123-135, Feb. 1974.
[5] L. Grcev, Modeling of grounding electrodes under lightning
currents, IEEE Transactions on Electromagnetic Compatibility,
Vol. 51, No. 3, Aug. 2009.
[6] O. Mayr, Contributions to the theory of static and dynamic
electric arc (in German), Archive for Electrical Engineering, Vol.
37, pp. 588-608, 1943.
[7] COMSOL Multiphysics. Available in: http://www.comsol.com/,
(Accessed: Dec./2013).
[8] H.A.D. Almaguer; R.A. Coelho; P. Luiz Nosaki; V.L. Coelho;
and A. Piantini, "A feasibility study on the use of concrete pole
bases as a grounding topology for distribution systems,"
International Symposium on Lightning Protection (XII SIPDA),
pp. 209-213, Oct. 2013.
Main author
Name: Vilson Luiz Coelho
Address: R. Tenente Silveira 570/Ap.308; CEP 88.010-301
Florianpolis SC.
Phone: (48)9980.6238; E-mail: vilson.coelho@vlc.eng.br