2021 3rd International Conference on High Voltage Engineering and Power Systems (ICHVEPS)

The Study of Electric Field and Partial Discharges

2021 3rd International Conference on High Voltage Engineering and Power Systems (ICHVEPS) | 978-1-6654-4354-8/21/$31.00 ©2021 IEEE | DOI: 10.1109/ICHVEPS53178.2021.9600924

on XLPE Insulation under DC High Voltage using

COMSOL Multiphysics
Apoorva Sahu Rakesh Sahoo Subrata Karmakar
Department of Electrical Engineering Department of Electrical Engineering Department of Electrical Engineering
National Institute of Technology National Institute of Technology National Institute of Technology
Rourkela, India Rourkela, India Rourkela, India
[email protected] [email protected] [email protected]

Abstract—Partial discharges measurement and analysis is one The main causes for partial discharges are cavities or voids
of the prime factors when it comes to the condition assessment which are present within the dielectrics. The sources of these
of power cables. The presences of microcavities in the manufac-
cavities are due to manufacturing defects or poor workman-
turing process are the major cause of field enhancement when
subjected to high voltage stress. These are the precursor of PD ship during installation. Modeling of PD in cavities within
activity, if it continued for longer duration; there may be a chance dielectric insulating materials has received a lot of attention.
of insulation breakdown. This work mainly emphasizes the Modeling PD activities has the advantage of allowing the
distribution of electric potential, electric field, and PD activities characteristics impacting PD activity under various conditions
in the presence of artificially created cavity defects of an XLPE
to be easily analyzed. To analyze the effect of void in a
insulation sample using the finite element method by COMSOL
Multiphysics software. The simulation has been carried out in dielectric material various simulation models are built using
both normal and defective samples by varying elliptical and finite element method by COMSOL, ANSYS software, and the
spherical cavity geometry, cavity location, and supply voltage. results are also compared with the experimental set-up [2], [3].
The characteristics of the initial behavior of applied DC voltage The presence of various cavities enhances the electric field in
are also analyzed before it reached a steady-state value. PD
the insulation surface which is mainly occurring due to the
inception voltage and 2-dimensional time resolve pattern are also
analyzed by integrating COMSOL Multiphysics with MATLAB non-uniformity of relative permittivity. The local electric field
environment. distribution is influenced by the cavity shape and position,
Index Terms—Partial discharge, XLPE insulation, DC high which impacts PD properties [4], [5]. The simulation of the
voltage, COMSOL Multiphysics electric field and potential distribution within the insulation
can provide a simple tool to locate the electrical stress that
I. I NTRODUCTION could be harmful to the system over the long term of operation
The insulation in high voltage components is subjected to [6]. PD activity initiates from these highly stressed regions in
high stress due to the high applied electric fields. The failure the presence of the availability of free electrons. Electrical
of insulation may result in significant disruption of a power conductivity is an important parameter in PD activity and
outage. In insulation diagnostics, partial discharges (PD) are is directly affected by electrical distribution and temperature
very significant which emphasizes localized electric discharge [7]. Space charges are more significant under dc conditions
that can occur within the insulation. Since the discharges do than ac conditions. Therefore simulation studies are performed
not span the entire distance between the electrodes, there is no to obtain the field strength, PD inception voltage, and phase
direct breakdown of the insulation. Partial discharges, on the resolve pattern, which might be helpful to identify the presence
other hand, are widely accepted as a sign of insulation system of various types of cavities inside the insulation sample.
defects and deterioration. Discharges can cause breakdown The three capacitance models, also known as the “a-b-c”
depending on the type and position of the discharges, as well model, Pedersen’s model, Niemeyer’s model, and Forssen’s
as the design of the insulation system. The trend in rapid models are well-known PD models that use the Finite Element
industrialization development played an important role in the Analysis (FEA) approach.
field of high voltage applications such as XLPE insulated cable In this work, a Finite Element simulation model using
networks. The performance of solid insulation is affected by a COMSOL Multiphysics has been implemented to investigate
variety of aspects, including temperature changes, load cycle the distribution of electric field inside the elliptical and spher-
changes, moisture intrusion, and mechanical stress, etc [1]. ical cavity by considering electric current and heat transfer
Even though the insulation passed all required tests before model in the presence of HV DC supply. PD inception voltage
installation and usage, it may not maintain the same propor- and 2-dimensional time resolve pattern are also analyzed by
tional functioning quality throughout its lifetime. Degradation integrating COMSOL Multiphysics with MATLAB environ-
phenomena caused by PD have been a major source of worry. ment.

978-1-6654-4354-8/21/$31.00 ©2021 IEEE 309

Authorized licensed use limited to: NATIONAL INSTITUTE OF TECHNOLOGY ROURKELA. Downloaded on December 09,2021 at 11:12:33 UTC from IEEE Xplore. Restrictions apply.
 II. M ODEL C ONFIGURATION Where E is electric field in V /m. V represents electric
In this section, the FEA approach is used to explain the potential in volts. J defines the current density in A/m2 . ϵ0 is
two-dimensional cavity model geometry, various equations the permittivity of vacuum and ϵr is the relative permittivity.
associated with the model, and the flow chart of the MATLAB σ is the conductivity of the material in S/m.
TABLE I. PARAMETERS OF THE S IMULINK MODEL
code. Simulation of electric field and potential distribution
across the void is analyzed by defining parameters, drawing Material Relative permittivity Electrical Conductivity (S/m)
XLPE 2.3 1 × 10(−13)
the geometry of XLPE sample, selection of material, providing Cavity surface 2.3 1 × 10(−13)
appropriate boundary condition, and meshing using COMSOL Air filled Cavity 1 1 × 10(−15)
Multiphysics. FEA model is created using COMSOL soft-
ware integrated with MATLAB. Partial Differential Equations High voltage DC is applied across the electrode, from which
(PDE) are used to solve the electric potential and temperature electric field distribution and voltage distribution curve can be
in the model. Ordinary Differential Equations (ODEs) are obtained from the model. The field model which is shown
generated from PDE equations, and ODEs are solved. Since in equation 6 describes electric voltage distribution in the
dielectric.
−∇.(σ∇V ) = ∇.(ϵE) = ∇.(ϵ∇V ) (6)
The temperature distribution can be computed using the
FEA software’s ‘Heat Transfer Module by Conduction’. The
PDE equation that governs the situation is
∂T
ρC( ) − ∇.(k∇T ) = Q (7)
∂t
III. PARTIAL DISCHARGE PATTERN ANALYSIS
PD can’t be simulated only in COMSOL software, for this
reason, it’s integrated with MATLAB. Formation of PDs inside
the void required two necessary conditions. As a result, the
first condition is to ensure that the voltage in the cavity (Vc )
is greater than the voltage required to start the PD event
(Vinc ), i.e (Vc > Vinc ). The second prerequisite is to ensure
that there are enough free electrons at the cavity. To reduce
Fig. 1 Schematic diagram for 2D simulation model the need for physical characteristics of material qualities, a
there are fine mesh elements used by applying 2D model simple term is employed in the model to express the electron
but the time required for simulation is high. The model’s generation rate equation. For PD occurrence the field in the
problem is solved using two application modes: ’Electric cavity center should exceed in the inception field with the
current (ec)’ and ’Heat Transfer in Solid (ht)’. Table I show the condition the electron should be available in that instantaneous
parameter like sample geometry, relative permittivity, electrical time t. Electron generation rate (EGR) is used to model the
conductivity which is used in the simulation. Fig. 1 shows the free-electron availability. Net (t) is denoted as the number of
2-D simulation model of XLPE dielectric having the width of electrons generated at time t. Nes (t) is denoted as the electron
6 mm wide and 2.5 mm height containing air-filled void. A generation rate due to surface emission and Nev (t) electron
cavity is built of elliptical shape with the major axis 0.7 mm generation rate due to volume ionization.
and minor axis 0.5 mm. DC high voltage is connected to the
Net (t) = Nes (t) + Nev (t) (8)
model upper surface electrode and ground is connected to the
lower surface. To make simple model electron generation is mostly con-
When supply voltage is provided to the simulation model, sidered due to surface emission. By considering initial electric
electric field distribution occur according to equation 1 to field Einc0 and electron generation rate Nes0 than the number
equation (5). of free electron generation at time t due to previous PD is
E = −∇V (1) shown in equation 9.
 
−(t − tP D ) Ecav (tP D ) Tmat
J = σE (2) NP D = NP D exp exp . (9)
τdec Einc0 Tamb
∇.ϵ0 ϵr E = ρ (3) A probability is utilized to assess the possibility of a PD
occurring where PD depend on Net (t) electron generation rate
∂ρ and time duration ∆t as per equation 10.
∇.J = − (4)
∂t
P (t) = Net (t).∆t (10)
∂
J = (σ + ϵ0 ϵr )E (5) Vc
p
= 24.41(ρair d) + 6.73 ρair d (11)
∂t i

310

Authorized licensed use limited to: NATIONAL INSTITUTE OF TECHNOLOGY ROURKELA. Downloaded on December 09,2021 at 11:12:33 UTC from IEEE Xplore. Restrictions apply.
 occurring due to the non-uniformity of relative permittivity
between the dielectric material and air.

Fig. 3 Potential distribution in the presence of air filled void

Fig. 2 Flowchart of MATLAB program for PD activity

Fig.2 illustrated the flowchart which is used for initiation

of PD and pattern analysis. At first MATLAB workspace is
cleared after setting up the subdomain, boundary condition, Fig. 4 Electric field distribution in the presence of air filled void
meshing and model runs with initial conditions. The electric
The initial applied voltage distribution waveform pattern
field across the cavity surface was extracted. If the field across
across the cavity is given in Fig. 5. Voltage is varied from
the cavity is greater than the inception value and free-electron
10 kV to 22 kV. According to the observation, the voltage
present then the partial discharge takes place. Free electron
becomes steady after few microseconds. Here the shown
availability is stochastic. Partial discharge is a random process
waveform indicated towards the initial behavior of voltage
therefore for randomness P variable is added. P is compared
pulse up to. Cavity behaves as a capacitor due to lower
with random number R. R is varied between 0 and 1. If P
permittivity. The change in conductivity shows little influence
is greater than R that means partial discharge takes place,
in pure DC voltage.
electrical conductivity will set to higher value due to more
availability of free electron after the discharge conductivity is
again set to lower value and if P is less than R there would be
no partial discharge and conductivity value will remain same.
Conductivity value is changing in each iteration depending
upon the relation of P and R. For the PD condition when
cavity voltage exceeds the inception voltage, cavity breakdown
occurs and inception voltage is calculated by equation(11).
Where Vc i is the inception voltage in kV, ρair is the normalize
air density, and d is the height of the cavity.
IV. R ESULTS AND D ISCUSSION
In this simulation model, the upper electrode of the insulat-
ing material is connected to the high voltage DC source and
the lower electrode is grounded. An air-filled elliptical cavity
with a major axis of 0.7 mm and a minor axis of 0.5 mm Fig. 5 Initial behaviour of voltage across the cavity
within the XLPE dielectric material is exposed to variable dc The variation of electric field distribution due to change in
voltage. Fig. 3 and Fig. 4 pointed towards the electric potential applied voltage across the void is shown in Fig. 6. During
and electric field distribution for 14 kV applied DC voltage the discharge, process charges accumulate in the cavity wall
respectively. The presence of an air field cavity enhances that is why the electric field intensity is more intense in
the electric field in the insulation surface which is mainly the void. Field magnitude is not uniform along the y-axis

311

Authorized licensed use limited to: NATIONAL INSTITUTE OF TECHNOLOGY ROURKELA. Downloaded on December 09,2021 at 11:12:33 UTC from IEEE Xplore. Restrictions apply.
 field is slightly distorted in the boundary of the cavity and the
dielectric because of the permittivity of insulation and charge
accumulation on the cavity surface. For elliptical cavity the
field distortion near the boundary of the cavity surface is more
as compared to spherical cavity which is shown in Fig. 6 (a)
and (b) respectively.
The maximum field stress within the void varies depending
on the size, shape, and position of the void within the insula-
tion material. Fig. 7 (a) shows the electrical field distribution as
a function of distance varied from the central axis, for variable
elliptical cavities in the axial and radial distance, whereas Fig.7
(b) shows the field distribution by variation of spherical cavity
radius which are located at the center of dielectric material.
Fig. 8 represents the voltage across cavity and PD phe-
nomenon at 14 kV DC supply. Vc i obtained for this model
using equation 11 is 5.2 kV. When the applied voltage is
greater than the inception voltage PD takes place in the
cavity. During the discharge process after satisfying the voltage
condition, there are chances when there aren’t enough free
electrons to launch the electron avalanche. As a consequence,
there is a time lag between the first and second PD. PDs are
moved forward in phase as a result of the time lag, and they
occur at voltages greater than the inception voltage level. This
process is stochastic. The expected value of referred to as
statistical time lag. This phenomenon can be seen in Fig. 8
Fig. 6 Distribution of Electric field for (a) elliptical and (b) spherical void by (a) at 0.078 s for spherical and 0.133 s for elliptical cavity
variation of applied voltage
when the voltage is higher than the inception voltage.

Fig. 7 Distribution of Electric field for (a) elliptical and (b) spherical cavity Fig. 8 (a) PD behavior and (b) PD initiation at 14 kV DC supply
by variation of dimension
The most common representation of partial discharge pat-
because the cavity size is quite large compared to material tern in time domain format is shown in Fig.9. Discharge
thickness. But it is symmetrical along the y-axis. The electric magnitude is plotted against a small time window. The pattern

312

Authorized licensed use limited to: NATIONAL INSTITUTE OF TECHNOLOGY ROURKELA. Downloaded on December 09,2021 at 11:12:33 UTC from IEEE Xplore. Restrictions apply.
 [2] O. E. Gouda, A. A. ElFarskoury, and A. A. Farag, “Investigating the
effect of cavity size within medium-voltage power cable on partial
discharge behavior,” IET Generation,Transmission & Distribution, vol.
12, no.5, pp. 1190-1197, 2017.
[3] J. Joseph, S. Mohan, and S. T. Krishnan, “Numerical modelling, sim-
ulation and experimental validation of partial discharge in cross-linked
polyethylene cables,” IET Science, Measurement and Technology, vol.
13, no. 2, pp. 309-317, 2018.
[4] H. A. Illias, G. Chen, and Paul L. Lewin, “Modeling of partial discharge
activity in spherical cavities within a dielectric material,” IEEE Electrical
Insulation Magazine, vol. 27, no. 1, pp. 38-45, 2011.
[5] H. A. Illias, M. A. Tunio, A. H. A. Bakar, H. Mokhlis, and G. Chen,
“Partial discharge phenomena within an artificial void in cable insulation
geometry: experimental validation and simulation,” IEEE Transactions
on Dielectrics and Electrical Insulation, vol. 23, no. 1, pp. 451-459,
2016.
[6] X. Gu, Y. Xu, Y. Yan and P. Zhao, “Partial discharge in a cylindrical
cavity in XLPE under DC voltage: Simulation and experiment,” IEEE
Transactions on Dielectrics and Electrical Insulation, vol. 26, no. 6, pp.
1831-1839, 2019.
[7] P. H. F. Morshuis and J. J. Smit, “Partial discharges at DC voltage: their
mechanism, detection and analysis,” IEEE Transactions on Dielectrics
and Electrical Insulation, vol. 12, no. 2, pp. 328-340, 2005
[8] R. N. Hampton, “Some of the considerations for materials operating
under high-voltage, direct-current stresses,” IEEE Electrical Insulation
Magazine, vol. 24, no. 1, pp. 5-13, 2008.

Fig. 9 PD pattern (a) elliptical and (b) spherical cavity for 14 kV applied DC
voltage

is recorded up to 10 seconds for both elliptical and spherical

cavity at 14 kV DC supply which looks like a train of PD pulse
where the magnitude of discharges depending upon time lag
between consecutive discharges.

V. C ONCLUSION
Electric field distribution, initiation of PD, and pattern
analysis are the major parameters for analyzing the condition
of insulating material. For modeling the finite element method-
based cavity geometry, COMSOL Multiphysics software is
used in this work. The 2-dimensional cavity in XLPE di-
electric material is modeled in various conditions. Electric
field distribution can be determined using this model which
provides insight mechanism of partial discharge. The effect
of variable applied voltages has been studied. Statistical
time lag, electron generation rate, and electrical conductivity
significantly influence the partial discharge behavior. Partial
discharge pattern has been plotted and patterns occur as a
train of pulse which observed in simulation. In the future,
the hardware implementation of the work can be performed
to analyzed the accuracy and error rate by varying different
geometry.

R EFERENCES
[1] J. M. Rodrı́guez-Serna, R. Albarracin-Sanchez, and A. A. Mas’ ud,
“Finite-element-analysis models for numerical simulation of partial
discharges in spherical cavities within solid dielectrics: a review and
a novel method,” High Voltage, vol. 5, no. 5, pp. 556-568, 2020.

313

Authorized licensed use limited to: NATIONAL INSTITUTE OF TECHNOLOGY ROURKELA. Downloaded on December 09,2021 at 11:12:33 UTC from IEEE Xplore. Restrictions apply.