You are on page 1of 5

2013 Electrical Insulation Conference, Ottawa, Ontario, Canada, 2 to 5 June 2013

Comparison Between the Use of Surface and


Volume Conductivity to Compute Potential
Distribution along an Insulator in Presence of a
Thin Conductive Layer
C. Volat
University of Quebec at Chicoutimi
Chicoutimi, QC, Canada
cvolat@uqac.ca

Abstract— This paper describes a comparative study on the both polluted and ice-covered insulators. However, taking into
modeling of a thin conductive dielectric layer usually found on account the thickness of the layer leads to a drastic increase of
polluted or ice-covered insulators. Two different approaches such the size of the numerical model due to a large number of
as those offered by most FEM commercial software were studied. elements dedicated to the thin conductive layer modeling. This
The first one is a volume approach which takes into account the is due to the fact that a minimum size of mesh element is
thickness of the thin layer. The second is the surface approach required to model thin conductive layer in order to provide
where the thin layer is treated as a specific boundary condition. some reliable results. This requirement becomes important
Simulations were performed using the FEM commercial software when 3D simulations have to be performed, which
Comsol Multiphysics® that allows both volume and surface automatically leads to large problem size to solve.
approaches. Parameters such as conductivity and permittivity of
the thin layer as well as the number of elements used are studied The second approach is to consider the thin layer as a
both in 2D axisymmetric and 3D modeling. The results obtained conductive surface. This approach has been firstly introduced
demonstrated that the surface approach is the best solution as it by Andersen in a FEM program in order to calculate E-field
provides the same results as the volume approach but with 3 and potential distributions in presence of resistive coating on
times less elements required for the mesh. The surface approach the surface insulation [6]. Later, such approach was also
should then be considered for 3D complex problems where thin implemented in a BEM program to analyze conductive film
conductive dielectric layer is present. effects over the surface of an insulator [5]. Using this
approach, the thin conductive layer can be modeled as a
Keywords: FEM, thin conductive dielectric layer, polluted specific interface or boundary condition. Hence, this permits
insulator, potential distribution. to simplify the model as well as to decrease considerably the
I. INTRODUCTION number of elements required to mesh the layer. Such specific
interface or boundary condition is actually available in
E-field modeling using commercial software has become commercial FEM software which was used in previous works
an indispensable tool employed in different high voltage to model thin water film at the ice surface [7].
engineering fields and particularly in outdoor insulation
simulations. Indeed, E-field distribution calculation is largely This paper attempts to clarify some modeling aspects
employed for adequately designing and dimensioning related to the presence of a thin conductive dielectric layer on
insulators and corona rings as well as to study the potential insulator surface. A comparison between the volume and
and E-field distribution modifications under pollution or icing surface approach is performed. In fact, the influence of
conditions [1-4]. For these specific conditions, a thin different parameters related to the thin layer like permittivity,
conductive layer mimic a wetted pollution layer or thin water volume conductivity as well as the number of elements used
film at the ice surface must be considered. Modeling of a thin for the mesh were taken into account and discussed. The
conductive surface (few millimeters or hundred of simulations were performed with different models: a 2D
nanometers) versus the insulator length leads to an important model which analytical solution is known and a 2D
dimensional ratio is the principal difficulty. Depending of the axisymmetric and 3D uniformly polluted insulator.
used software, the thin conductive layer can be treated in two
ways. II. MODELING OF THIN CONDUCTIVE LAYER
A. Thin layer with volume conductivity (volume approach)
The first approach is to consider the conductivity volume
layer ; in this case the thickness of the layer in the numerical For most HV insulation equipment, the magnetic energy
model must be taken into account. Such approach has been can be negligible compare to the electric energy and electric
used in previous works based on Finite Element Method field can be decoupled from the magnetic field. In this
(FEM) [1-3] and Boundary Element Method (BEM) [4-5] for

978-978-1-4673-4744-0/13/$31.00 ©2013 IEEE 409


condition, the electric field can be JG expressed only as the TABLE I. α AS A FUNCTION OF THIN LAYER CONDUCTIVITY
gradient of the electric potential V ( E = −∇ ⋅V ).
Conductivity σV (S/m) 1e-9 1e-6 1e-3 1
Considering a conductive layer of thickness d, of volume
Ratio α
-3 3
3.7e 3.7 3.7e 3.7e6
conductivity σV (S/m) and permittivity ε (F/m) which are both
considered as isotropic, the conservation law that governs the
conductive layer, in harmonic regime, can be expressed as depending on pollution severity [8]. Table I present some α
follows [5-6]: values obtained for different conductive layer conductivities
JJG JG
( )
∇. J v + jω ∇. D = 0 (1) according to the frequency and permittivity values given
previously. The influence of the layer permittivity has to be
with JV (A/m2) being the volume current density, D (C/m2) the considered under a layer conductivity of around 10 µS/m.
electric flux density and ω (rad/s) the pulsation. Also, both conductivity and permittivity have to be considered
for a conductivity range between 0.1 nS/m to 10 µS/m. Finally
Equation (1) shows that the conductive layer is treated as a
for conductivity value above 10 µS/m, the permittivity effect
conductive dielectric material.
can be neglected.
B. Thin layer as conductive surface (surface approach)
III. COMPARISON BETWEEN VOLUME AND SURFACE
If the conductive layer is very thin, the conduction current
APPROACH
and the displacement current densities can be considered
uniform along its thickness d. In that case, the approximation A. Validation of volume and surface approach with 2D
of the layer as a conductive surface can be done. Based on model and its analytical formulation
these assumptions and as proposed by different authors [5-6], In order to provide a comparison between volume and
all the vector quantities of (1) can be projected on the surface approach, the commercial FEM software Comsol
conductive surface, which leads to the following equation: Multiphysics® was used for all the simulations done in this
study. This software has the advantage to be able to model thin
d(σV + jωε)∇ 2V+jωρS = 0 (2)
dielectric conductive surface layer as described by (2).
where: Therefore, comparison with volume approach is possible using
- V (V) is the electrical potential; the same software.
- d (m) the thickness of the layer;
- ρS (C/m2) the surface charge density In order to validate the influence of the layer conductivity,
the simple model shown in Fig. 1, extracted from [5], was
Equation (2) will be used as a boundary condition between used. This model represents a thin dielectric plate of thickness
the insulator or the ice layer and the air to model the presence h and permittivity εd covered by a conductive surface of length
of the conductive layer of thickness d, volume conductivity σV L of 20 cm and conductivity σs.
and permittivity ε.
C. Influence of layer permittivity
In harmonic regime, the behavior of a material is governed
by its electrical conductivity σV, permittivity ε or relative
permittivity εr and applied frequency or pulsation ω regarding
the value of ratio α given by the following equation [8]:
Figure 1. Model used to validate both surface and volume conductivity
σV (3)
α= approach [5]
ωε Considering a symmetric problem along the z axis (see Fig.
For high α values, the thin layer is considered as a good 1), the analytical expression of the real value of the electric
conductor and displacement current can be neglected. This potential along the conductive surface can be obtained as
corresponds to a resistive regime where potential and E-field followed [5]:
distributions are governed by σV and can be determined using
cosh ⎡⎣ k ( 2 L − x ) ⎤⎦ cos ( kx ) + cos ⎡⎣ k ( 2 L − x ) ⎤⎦ cosh ( kx )
conductive DC approximations [5]. On the other hand, low α V ( x) = V0 (4)
values mean that the conductive current can be neglected and 2 ⎡⎣cosh 2 ( kL ) − sin 2 ( kL ) ⎤⎦
that E-field and potential distributions are dominated by the where V0 is the applied voltage equal to 100 V and k is a
permittivity (capacitive regime) where electrostatic simulation coefficient equal to:
can be employed. For intermediate α values both the
conductivity and permittivity of the thin layer have an ω.ε d (5)
k=
influence on E-field and potential distributions. This 2hσ S
corresponds to capacitive-resistive regime as defined by [8] where:
and quasi-static approximation has to be employed. - εd (F/m) is the electrical permittivity of the dielectric
For polluted or ice-covered insulators, the service plate of thickness h;
frequency is 60 Hz and εr can be considered equal to 80 for - σS (S) is the conductivity of the conductive surface.
melting water film or wetted polluted layer [2, 4]. Finally, σV - ω (rd/s) is the pulsation of the applied voltage.
can cover a large range of value, from few nS/m to few S/m

410
Reference line

(b)

(a) (c)
Figure 2. Comparison of potential distributions along the thin layer for Figure 3. Numerical model of the dead-end 28 kV insulator (a) surface
surface conductivities equal to 1 nS and 0.1 mS approach (b) and volume approach (c)

The comparisons of analytical and numerical results


obtained for both volume and surface approach are presented Figure 4 presents a comparison of the potential distribution
obtained with the two approaches for a thin layer conductivity
on Fig. 2 for two following surface conductivities σS: 1nS and
of 10 nS/m and a relative permittivity of 15 and 80. The
100 nS. For the volume approach, a thickness d of the potential distribution was computed along the same reference
dielectric conductive layer was fixed at 1 mm, the same value line as presented on Fig. 3.
was used in the surface approach modeled as a boundary
surface condition in the commercial FEM software. The
volume conductivity σV in the model was in accordance to the
surface conductivity of the analytical model. The simulations
were performed in the quasi-static mode in order to take into
account the influence of thin layer conductivity and
permittivity.
εr = 80
The results obtained from Fig. 2 exhibit a good
concordance between analytical formulation and the two
numerical surface and volume approaches implemented with
the commercial FEM software.
εr = 15
B. Influence of layer conductivity and permittivity
In this part, the volume and surface approach was applied
to a uniform polluted insulator in order to compare both
approaches in terms of layer permittivity and conductivity
influence. The insulator used for this simulation is a 28-kV Figure 4. Comparison of potential distributions along the uniform
dead-end thermoplastic elastomeric (TPE) insulator and its polluted insulator for thin layer conductivity of 10 nS/m and relative
principal characteristics are presented in Table 2. permittivity of 15 and 80 respectively
The 2D axisymmetric numerical model of the insulator The results confirmed the previous results shown in Fig. 2
with both volume and surface modeling approaches are shown regarding the use of both volume and surface approach to
TABLE II. CHARACTERISTICS OF THE 28-KV DEAD-END TPE model a thin dielectric conductive layer which conduct to
INSULATOR almost the same potential distributions. Indeed, as illustrated
Section length (mm) 438 by the results of Fig. 4, the discrepancy obtained between
Dry arcing distance (mm) 285 surface and volume approaches is lower than 3 % for the
Leakage distance (mm) 675
maximum value and lower than 1% for the average value.
Shed quantity 6
Outer shed diameter (mm) 100 Additionally, the results obtained demonstrated that for the
Rod diameter (mm) 14 conductivity of 10 nS/m, the electrical permittivity of the thin
layer has a notable influence on the potential distribution as
in Fig. 3. For the volume approach, a thin layer having a 1 mm the ration α given by (1) is equal to 0.2 and 0.037 for relative
thickness was used. The simulation was done for layer permittivity values of 15 and 80 respectively. With such a
conductivities of 1 nS/m, 1 µS/m and 0.1 mS/m and for ratio value, potential distribution is governed by the
relative permittivity of 15 and 80. The voltage applied to the capacitive-resistive effect of the insulator and the thin layer at
insulator was fixed at 16.2 kVrms (the phase-to-ground service its surface which is adequately taken into account by both the
voltage) as the simulations were performed in the quasi-static volume and surface approach.
mode.

411
C. Comparison of volume and surface approach in 3D
modeling of the uniform polluted insulator
Tridimensional (3D) simulations of the uniform polluted
dead-end 28 kV insulator (Table II) were performed in order
to study the influence of the thin layer modeling approach on
the discrepancy of the potential distributions as well as on the
problem size (mesh element number). The thickness of the
layer was fixed at 1 mm, its volume conductivity at 1 µS/m
and its relative permittivity at 80. The 3D insulator model,
presented on Fig. 7, is simply based on the 2D axisymmetric
one (Fig. 3) which was rotated on 360°.
Figure 8 presents the comparison of potential distributions
obtained between the 2D axisymmetric surface approach and
Figure 5. Discrepancy obtained between potential distributions
the 3D volume and 3D surface approach. The corresponding
computed using surface and volume approach for thin layer conductivity discrepancy relative to 2D surface approach is presented on
of 10 nS/m and relative permittivity of 15 and 80 respectively Fig. 9 and the number of elements used for the mesh is
presented in Table III.
In order to study the effect of the thin layer conductivity
The results of Figs. 8 and 9 demonstrate a very good
and permittivity on the potential distribution, additional
concordance between 2D and 3D surface approach with an
simulations were performed using surface approach applied to
average discrepancy of 0.13 %. For the 3D volume approach,
the same dead-end 28 kV insulator. The results, presented in
Fig. 6, were obtained for conductivity values of 0 S/m (clean
insulator), 10nS/m, 1µS/m and 0.1mS/m and for permittivity
thin layer values of 15 and 80 respectively. Under a volume
conductivity of 1µS/m, the influence of the thin layer
permittivity increases as the conductivity decreases. For low
conductivity, the potential distribution obtained is closed to
the potential distribution of the clean insulator which is only
governed by the capacitive regime. The thin layer can be then
considered as a good dielectric, which is consistent with the
result of Table I. Additionally, as the conductivity increase,
the influence of the permittivity decreases and becomes
neglected for a volume conductivity greater than 1 µS/m. In
fact, the thin layer becomes a better conductive layer (resistive
regime) as verify by the ratio α given presented in Table I.
These results are in agreement with the results presented by
[2].
Figure 7. 3D modeling of the uniform polluted 28 kV PTE insulator

Figure 6. Comparison of potential distributions along the uniform


polluted insulator as a function of its volume conductivity and its Figure 8. Comparison of potential distributions obtained along the
relative permittivity uniform polluted insulator using 3D modeling for both surface and
volume approach (σV = 1 µS/m and εr = 80)

412
The results presented in this paper have permitted to
compare surface and volume approaches in terms of thin layer
conductivity and relative permittivity influence as well as in
terms of problem size. For that, a commercial FEM software,
Comsol Multiphysics was used which permit to implement the
two approaches.
Depending of the thin layer conductivity, both surface and
volume approach modeling have demonstrated that the
permittivity of the thin layer has an influence for values lower
than 1 µS/m. For higher values, only the thin layer conductivity
dominates. So with both approaches, capacitive-resistive
regime of the problem can be taken into account.
Both surface and volume approach provide the same
results in terms of potential distribution in 2D axisymmetric as
well as in 3D. However, in 3D, the volume approach requires
Figure 9. Comparison of discrepancy obtained between 2D
a large number of elements as more than 25 % of these
axisymmetric and 3D potential distributions for both surface and volume elements are dedicated to the thin layer. This demonstrated
approach (σV = 1 µS/m and εr = 80) that for 3D modeling, the surface approach should be
a good concordance is also obtained but with an average prioritized.
discrepancy of 1.74 % which decreases to 0.58 % with mesh Additionally, in the surface approach, the polluted layer is
refining. This confirms that the two approaches give some not drawn as it is considered as a specific boundary condition.
very good results both in 2D and 3D. This permits to greatly simplify the construction of the 3D
model which is a task to be considered in the case of complex
problems where insulator, tower, grading ring and support
system must be taken into account.
REFERENCES
[1] .J. Phillips et al., “Electric Fields on AC Transmission Line Insulators”,
IEEE Trans. On Power Delivery, vol. 23, no. 2, 2008, pp. 823-830.
[2] S. Ilhan, A. Ozdemir, S. H. Jayaram and E. A. Cherney, “AC and
Transient Electric Field Distributions along a 380 kV V-String
Insulator”, Conference records of the 2012 IEEE Symposium on
Electrical Insulation (ISEI), 2012, pp. 399-403.
However, the use of the surface approach in 3D modeling [3] W. A Sima, Q. Yang “Potential and Electric-field Calculation along an
is advantageous in terms of number of elements. As revealed Ice-covered Composite Insulator with Finite Element Method”, IEE
by the information of Table III, the 3D volume approach Proceedings Generation, Transmission and Distribution, vol. 153, no. 3,
requires a larger number of elements (more than 6.5 times for 2006, pp. 343-349.
the refined mesh) than the 3D surface approach. This [4] C. Volat and M. Farzaneh, “3-D Modeling of Potential and Electric
Field Distributions along an EHV Ceramic Post Insulator Covered with
difference comes principally from the number of elements Ice - Part I: Simulations During a Melting Period”, IEEE Trans. on
used to model the polluted layer which represent about 25 % Power Delivery, vol. 20, no. 3, , July 2005, pp. 2006-2013.
of the total number of elements used to mesh the entire model. [5] J.L. Rasolonjanahary, L. Kräenbühl and A. Nicolas, “Computation of
Using the surface approach is an economical strategy to Electric Fields and Potential on Polluted Insulators using a Boundary
decrease the size of 3D modeling in order to obtain adequate Element Method”, IEEE Trans. on Magnetics, vol. 28, no. 2, March
1992, pp. 1473-1476.
solution when thin conductive dielectric layer is present.
[6] Q.W. Anderson, “Finite Element Solution of Complex Potential
Electric Fields”, IEEE Trans. on Power Apparatus and Systems, vol.
IV. CONCLUSION 96, no. 4, July/August 1977, pp. 1156-1160.
The performance and functionalities of current FEM [7] C. Volat, S.M.A. Emran and M. Farzaneh, “Dynamic Evolution of
commercial software’s are increasingly exploited to study the Potential and Electric Field Distributions Around an Ice-Covered
influence of pollution or ice layer of outdoor insulator electrical Insulator During Ice Accretion”, Proceedings of the 10th International
Workshop on Atmospheric Icing of Structures, IWAIS 2002, Brno,
performance. The main difficulty in such problem lies in the Czech Republic, June 2002, 6-8.
modeling of the thin conductive dielectric layer present at the [8] E. Asenjo S., N. Morales O and A. Valdenegro E., “Solution of Low
insulator or ice surface. This thin layer can then be modeled Frequency Complex Fiels in Polluted Insulators by Means of the Finite
using a volume approach which takes into account the Element Method”, IEEE Trans. on Dielectric and Electrical Insulation,
thickness of the layer or using a surface approach where the vol.4, no. 1. February 1997, pp. 10-16.
thin layer is approximated by a specific boundary condition. [9] M. Farzaneh, W. A. Chisholm, “Insulators for Icing and Polluted
Environments”, IEEE Press Series on Power Engineering, Oct. 2009

413

You might also like