A|T|C|T

Application and Theory of Computer Technology

Conference Paper

Evaluation of a Numerical Model Using COMSOL

MULTIPHYSIS Package
Adil Abdelwahab Sharawi 1, El-Amin Hamoda 2 and Abdelrahman Ali Karrar 2,*
1 Faculty of Engineering-Alimam Elmahadi University, shash2022@gmail.com
2 Electrical& Electronic Dept.- Khartoum university, akarrar@uofk.edu

Received: 27-September-2016; Accepted: 24-November-2016; Published: 03-December-2016

Abstract: A number of computer program packages have been developed for the solution of the
numerical methods. The main objectives of the paper were: to explore the benefit of computer
program packages that have been developed for the solution of problems based on numerical
methods in solving mathematical models, to solve the fundamental equations Laplace`s and
Poisson`s equations which are known in mathematics as boundary value problems, and plot the
voltage stress over a porcelain insulator which represents a mathematical model and to predict the
power frequency breakdown voltages. Use was made of COMSOL MULTIPHYSIS package based
on finite element method. It is found that, the simulation and experiment results recorded in many
studies are in close agreement. Thus, it may be concluded that the finite-element packages can be
used with a fair degree of accuracy to design most of numerical models.

Keywords: Simulation, COMSOL MULTIPHYSIS Dirichlet boundary condition, Neumann

boundary condition, finite element method, electric field.

1. Introduction

The techniques for using computers to imitate, or simulate operation of various kinds of real-
world facilities or processes is usually called a system ; which is defined by Schmidt and Taylor as
a collection of entities , that act and interact together toward the accomplishment of some logical
end. In order to study this system scientifically, we often have to make a set of assumptions about
how it works. These assumptions which usually take the form of mathematical or logical
relationships constitute a model that is used to gain some understanding of how the corresponding
system behaves. If the relationships that compose the model are simple enough, it may be possible
to use mathematical methods (such as algebra, calculus, or probability theory) to obtain exact
information to questions of interest; this is called an analytical solution. However, most systems are
too complex to allow realistic models to be evaluated analytically, and these models must be studied
by means of simulation. In simulation we use a computer to evaluate a model numerically, and
data are gathered in order to estimate the desired true characteristic of the model [1].
One of the most important decisions a modeler or analyst must make in performing a
simulation study concerns the choice of software. If the selected software is not flexible enough or
is too difficult to use, then the simulation project may produce erroneous results or may not even
be complete. The simulation packages can be classified into two types; the general purpose
simulation package which can be used for any simulation application and an application-oriented-

17 | P a g e
 A|T|C|T
Application and Theory of Computer Technology

simulation package which is designed to be used for a certain class of application. There are some
numerous features to consider when selecting simulation software. These features may be
categorized as being in one of the flowing groups [1]:

(a) General capabilities (including molding flexibility and ease of use)

(b) Hardware and software considerations.

(c) Animation.

(d) Statistical features

(e) Customer support and documentation

(f) Output reports and plots.

2. Numerical Methods
Before any engineering analysis or design may be under-taken for any system, it is necessary to
abstract from the physical objects mathematical models. Mathematical models are descriptions in the
form of algebraic equations or useful computer programs. The first step for developing a
mathematical model is to formulate differential equations that describe all interactions in the system.
These equations are then solved in specific regions with a set of boundary conditions for the desired
variables [2].
Analytical solution in one, two or three-dimensions exist only for very simple configurations.
Analytical methods give the exact results by solving mathematical models of physical situation.
Difficulties arise for practical problems particularly dealing with the boundary and initial conditions.
In some cases, the problems become impossible to represent by analytical method unless it is linear
with simpler boundary condition [3]. For more practical geometries numerical methods, using
successive approximations are usually used. Numerical methods include finite difference method
(FDM), finite element method (FEM), charge simulation method (CSM) and boundary element
method (BEM). These methods are used to solve Laplace’s or Poisson’s equations for the general
two- or three-dimensional field. Most engineering problems are solved by using numerical
methods. At present numerical methods are the most powerful tools with highly developed computer
programs [3].

3. Numerical Packages

A number of computer program packages have been developed for the solution of problems
using numerical methods. Some of the programs have been developed in such a manner that the
same program can be used for solution of problems belonging to different branches of engineering
with little modification. Commercial software are now days available with various possibilities
(ANSYS , ESTAT, ALGOR, and COMSOL MULTIPHYSIS etc. [5].

4. FEM Analysis

Finite Element methods (FEM) are the most efficient and popular computational methods for
solving partial differential equations (PDES), and this is due to the fact that FEM have many
advantages over many other numerical methods; Finite element methods can work on complex
irregular geometries, they use non-uniform meshes to reflect solution gradation and they can
construct high-ordered approximations. A finite element method utilizes a variational problem that
18 | P a g e
 A|T|C|T
Application and Theory of Computer Technology

involves an integral of the differential equations over the problem domain; the domain of the problem
is divided into sub-domains called finite elements and the solution of the problems is approximated
by simple polynomials functions on each element, these polynomials have to be pieced together so
that the resulting approximate solution would have an appropriate degree of smoothness over the
entire domain, then the variational integral is evaluated as the sum of the contributions from each
finite element in an approximate solution given by piece-wise polynomials defined throughout the
domain. Certain steps in formulating a finite element analysis of a physical problem are common to
all such analyses, whether structural, heat transfer, fluid flow, electrostatic field or some other
problem. These steps are embodied in commercial finite element software package. These steps are
described as follows [2]:

1-Pre-processing

The pre-processing step is, quite generally, described as defining the model and

Includes:

1. Define the geometric domain of the problem.

2. Define the element type(s) to be used
3. Define the material properties of the elements.
4. Define the geometric properties of the elements (length, area, and the like).
5. Define the element connectivity (mesh the model)
6. Define the physical constraints (boundary condition
7. Conditions).
8. Define the loadings

2-Solution

During the solution phase, finite element software assembles the governing algebraic
equations in matrix form and computes the unknown values of the primary field variable(s). The
computed values are then used by back substitution to compute additional, derived variables, such
as relative permittivity, reaction forces, and heat flow.

3-Post-processing

Analysis and evaluation of the solution is referred to as post-processing. Post-processing software

contains sophisticated routines used for sorting, printing and plotting selected results from a finite
element solution.

5. Field Equations

5.1 Electrical Field and Potential Distributions Calculation

For applications of AC having extra low frequency as 50/60 or DC voltages, problems may be
considered as an electrostatic field problem and therefore the electric and magnetic field components
may be considered independently of each other, and calculations made on the basis of static field
concepts. In the case of electrostatic and (also quasi-static) fields only Maxwell electric field equations
(flux divergence) are considered. One simple way for electric field calculation is to calculate electric
potential distribution. Then the electric field distribution is directly obtained by the negative gradient

19 | P a g e
 A|T|C|T
Application and Theory of Computer Technology

of electric potential distribution [7]. Obtaining Poisson’s equation is exceedingly simple, for from the
point form of Gauss’s law [8]:

. D   ………………………………………..…1

The definition of D

D=  E …………………………….……..……….…2

And the gradient relation

E  V …………………………………..…….….…..3

By substitution we have

. D    .( E)  - .(  V)   ...............4

Or

 .( V )    ………………………………….…5

  0,
Without space charge Poisson’s equation becomes Laplace’s equation

 .( V )  0 ………………………… ……………….6

   0 r  0
Where  is resistivity  m.
,  is material dielectric constant. , is free space dielectric
12
constant (8.854  10 F/m)
,  r is relative dielectric constant of dielectric material. Placing equation
(1) into equation (2) Poisson’s equation is obtained.

5-2 FEM analysis of the electric field distribution

The two- dimensional problem for which Laplace’s or Poisson’s equations applies is:

V 2  2V 0 ................( Laplacian region)

V 2    ..................................7
x 2 y 2  F ( x, y ).......( Poissonian region)

Where V is electric potential and F( x,y) =  /  for electrostatic fields within a medium of
permittivity  and containing distributed charge of density  ( x, y ). The field problem is given
within an x-y plane, the area of which has to be limited by given boundary conditions, by contours
on which some field quantities are known.

Poisson’s equation is more general than Laplace’s equation, and Poisson’s equation in terms of the
Cartesian coordinate system is defined as :

2 2
2 u  u
u    g(x, y) .....................................8
2 2
x y

20 | P a g e
 A|T|C|T
Application and Theory of Computer Technology

For a two–dimensional problem the boundary conditions are

u
q
uu and n ……………………...............................9

Where u is the electrostatic potential in the x-y plane and u and q are known voltage and flux
variables defining boundary conditions, and n is outward normal unit vector at the boundary [5].
Supposing that the domain under consideration does not contain any space and surface charges, a
two-dimensional functional F(u) in the Cartesian system of coordinates can be formed as follows [6]
:

1   du  2  du  
2

F (u ) 
2 
D
 x     y 
  dx 
  dxdy ...............................10
 dy  


Where  x and y are the x and y components of dielectric constant in the Cartesian system of
coordinates and u is the electric potential. In case of isotropic permittivity distribution

( =
x= y ) equation (8) can be reformed as:

1   du  2  du  2 
F (u )          dxdy....................11

 
2 D
dx   dy   

Inside each sub-domain De a linear variation of the electric potential is assumed as described in (7)

u e ( x , y )   e1   e 2   e 3 (e  1,2,3,....n ) .........12

Where ue(x,y) is the electric potential of any arbitrary point inside each sub-domain De.
 e1 ,  e 2 ,  e3 represent the computational coefficients for a triangle element e, n is total number
of triangle elements.

The calculation of the electric potential at every node in the total network composed of many triangle
elements was carried out by minimizing the functional F(u) , that is :

F ( u ) i
 0 ; i  1,2, np ............13
u i

Where, np stands for the total number of nodes in the network. Then a compact matrix expression

Where, np stands for the total number of nodes in the network. Then a compact matrix expression

    
S
ji
 u T , i, j  1, 2, ....np ..............................14
i j

S  
u
Where ji the matrix of coefficients is, i is the vector of unknown potentials at the nodes and

T
j

is the vector of free terms. By solving equation (14) the unknown potential can be found.

21 | P a g e
 A|T|C|T
Application and Theory of Computer Technology

For electrical point view, one of the following conditions may be satisfied on the boundary of the
first type [11]:

Dirichlet boundary condition: is a type of boundary condition, named after Johann Peter
Gustav Dirichlet , which when imposed on an ordinary or a partial differential equation, specifies the
values a solution needs to take on the boundary of the domain. If the electric potential of all points
are known, and this is possible when the available conductor is connected to a fixed potential source,
the boundary condition is a Dirichlet one.

Neumann boundary condition: is a type of boundary condition, named after Carl Neumann,
which when imposed on an ordinary or a partial differential equation, specifies the value that the
derivative of the solution is to take on the boundary of the domain, for example if the whole charge
on the boundary surface is known, while the electric potential over different points is unknown. This
is a case when the available conductor is not connected to a fixed potential (floating). Normally in
such a case the whole charge is equal to zero, the boundary is Neumann one[12].

6. COMSOL MULTIPHYSIS

In this paper, COMSOL/ M 3.4 was used for analysis the electric field and potential distributions
for a porcelain insulator. COMSOL/M [13] is an interactive commercial software package for
modelling and simulating scientific and engineering problems based on partial differential equations.
The modelling features of COMSOL/ M allows to simultaneously modelling any of combination of
phenomena in two and three-dimensional, for all fields of engineering and sciences. COMSOL/M
graphical user interface includes functions for CAD modelling, physics or equation definitions,
automatic mesh generation, equation solving, visualization, and post-processing.

Wei Xiong study the applications of COMSOL/M software in heat transfer processes in the three
mechanisms of heat transfer: conduction, convection and radiation, COMSOL/M was used to
simulate radiation heat transfer process in ordered to calculate the heat transfer problems
[14].Michael Chase used COMSOL/M to simulate fluid flow and heat transfer in fixed beds with tube
to particle diameter ratios of N=5 and N= 5.86. The results obtained from COMSOL/M was compared
with results published used similar simulation using CFD package [15], the two results were in closed
agreement. The two thesis presented that COMSOL/M is a useful tool for simulation of wide range
of physical models.

7. Physical Problem

7.1 Insulator Geometry

A 33kv porcelain insulator was selected to simulate electric field and potential distributions; the
basic design of the insulator is as follows: porcelain shed with different diameters having a relative
dielectric constant of 6.0, cement having relative dielectric constant of 2.33. , and surrounding of the
insulator is air having a relative dielectric constant of 1. A conductor is directly connected to the
voltage source and run through the top groove; the porcelain is grounded by a metal rod screwed
into the lower metallic thread. The total creepage distance is 685 mm. Creepage distance is known as
the shortest distance between the metal wires at the two ends of the insulator. When following the
contours of the insulator, excluding intermediate metal fittings. This distance is easily measured by
sticking masking tape to the insulator surface.
Figure 1 show the actual shape, Figure 2 show the cross section area of the insulator.

22 | P a g e
 A|T|C|T
Application and Theory of Computer Technology

Figure 1 Insulator Figure 2 Cross section

7.2 Simulation Using COMSOL MULTIPHYSIS

The physical problem which was studied and simulated is: For the 33 kV porcelain insulator ,
which is commonly use in the Sudanese National Grid .The potential, electrical field and equipotential
surfaces for the two dimensional field is to be determined for normal voltage and flashover voltage.
Flash over is defined as a disruptive discharge through air or over the surface of solid insulation,
between parts of different pollution or polarity by the application of voltage wherein the breakdown
path become sufficiently ionized to maintain an electric arc [14]. Even the flashover, caused by air
breakdown generally, do not cause physical damage to the insulators but the system may breakdown,
this causes many problems to the stability of the network.

The solution of equation (7) depends on boundary conditions of the problem. When the problem is
solved by FEM, all boundary conditions of the problem must be known, an since the porcelain is
unbounded, an artificial boundary is assumed as the boundary at infinity as shown in figure 3
according to this, boundary conditions are taken as :

V= V0 volt on the conductor at the upper part of the porcelain and this is a Dirichlet boundary
condition.

V=0 (ground) volt, on the lower part of the porcelain, and this also a Dirichlet boundary condition.

V
 0
n , on all the other outer boundary, and this is a Neumann boundary condition.

n.(D1-D2) = 0 on the surfaces of the dielectric of the insulator as continuity . The normal vector n ,
points from medium (2) to medium (1).

23 | P a g e
 A|T|C|T
Application and Theory of Computer Technology

Figure 3 Geometry of the sample

7.3. Results

The results obtained from FEM using COMOSL/M is laid out in figures 4-12 in two cases :

(a) A 27 kV Phase voltage is applied first; to specify the normal voltage. Figures 4-9 show the mesh,
potential mapping, electric field and line diagram along the porcelain for the first case.

(b) A 50-120 kV Phase voltage is applied second, to specify the flashover voltage. If the electric
stress in air at atmospheric pressure exceeds 2.6 kV/mm (measured value), ionization can occur.
Depending on the gap configuration, flashover may follow. Figures (10-15) show mesh, potential
mapping, electric field and line diagram when the voltage reaches 105 kV which is
approximately 3 times the normal voltage .Figures (8,14) show clearly that the electric field (which is
the gradient of the voltage) is smaller when the voltage lines are more spread out; in the conductor
there is negligible voltage drop across its length, therefore there no electric field within it.

(a) 27 kV Phase voltage

Figure (4) Mesh (27KV) Figure (5) equipotential mapping (27KV)

24 | P a g e
 A|T|C|T
Application and Theory of Computer Technology

Figure (6) Arc – length for potential (27 kV) Figure (7) Electric field (27 kV)

Figure (8) Arc – length for potential (27 kV) Figure (9) Electric field (27 kV)

(b) 105 kV Phase voltage

Figure (10) Mesh (105 k) Figure( 11) equipotential mapping (105 kV)

25 | P a g e
 A|T|C|T
Application and Theory of Computer Technology

Figure (12) Arc – length for potential (105 kV) Figure (13) Electric field (105 kV)

Figure (14) Arc – length for electric field (105 kV) Figure (15) Equipotential mapping and electric field 105 KV)

5. Conclusions

FEM with conjunction with a computer program packages called COMSOL/M was used to solve
the fundamental equations, Laplace`s and Poisson`s equations in two dimension and plot the voltage
stress over a 33KV Pin-type porcelain insulator, which is consider as mathematical model, with the
aim of validating the results obtained from simulation with experimental data published, which
show that flashover may occurs in any clean insulations if the voltage is 3 to 4 times the normal
voltage. It may be concluded that FEM with conjunction with a computer program packages
COMSOL/M is a very powerful tool for the design of high voltage equipment and for the design of
various engineering configurations models.

References
1. Averill M. Law. “ Simulation modeling and analysis ” Fourth edition McGraw–Hill, 2007; pp.3

26 | P a g e
 A|T|C|T
Application and Theory of Computer Technology

2. E.Kuffel; et al., “High Voltage Engineering Fundemental Bessel” second edition, Butterworth- Heinemann,
200, pp. 3-5.
3. Vassiliki, T., Kontargyri, Ioannis F. and Gonoes, A. Stathopulos “Measurement and simulation of electric
field of high voltage suspension insulators” European transaction of electrical power, published online 5
march 2008 in Wily Inter-Science.
4. Ian Gladwell, “Boundary value problem” Department of Mathematics, Southern Methodist University,
Dallas, TX. Published online at www. Scholarpedia, 2008.
5. Young W. Kwon, and Hyochoog Bag “The finite element method using matlab” CRO press London 1996.
6. Hutton, David V., “Fundamentals of finite element anlysis,” 1ed, McGRAM-HILL, 2004; pp. 1-12.
7. A. Kara; et al., “Effect of Dielectric Barriers to the Electric Field of Rod-plane AirGap” COMSOL Usera
Conference, 2006, Praque.
8. William H. Hayt, JR., “Engineerining Electromagetics,” 4ed, McGRAM-HILL, 1981, pp.206.
9. Young W. Kwon, Hyochoong Bang, The finite element method using Matlap, CRC press 1996; pp.83.
10. B.Marungsri; et al., “Analysis of Electric Field and Potenial Distributions along Surface of Silicone Rubber
Insulators under Various Contamination Conditions Using Finite Elment Method” World Academy of
Science, Engineering and Technology, 2009; vol. 53, PP.13355
11. Jawad Faiz, M.Ojaghi, “Instructive Review of Computation of Electric Fields usin Different Numerical
Techniques Method,” Int. J. Engng Ed.Vol.18,NO 3 pp , Great Britain, 2002.
12. Cheng A., and D. T. Cheng, “Engineering Analysis with Boundary Elements,” pp.268, 200.
13. http://www.comsol.com
14. Wei Xiong “Application of COMSOL MULTIPHSICS to Heat Transfer Processes” Arcada University of
Applied Sciences, May 2010 Helsinki.
15. Michael Chase “Simulation of Flow and Heat Transfer Through Packed Beds of Spheres,” Worcester
Polytechnic Institute 2011.
16. Tanushri Doshi, “Performance Anlysis of Composite Insulator up to 1200kv using Electric Field
Calculation,” Master thesis, Arozona Uinversity, December 2010; pp.9.

Authors’ information

Adil abdelwahab Sharawi is an assistant professor and the head of the Department of Electrical Engineering
at University of Engineering and technical studies at Alimam Elmahadi universty in Kosti, Sudan. He received a
doctorate degree in Electrical Engineering (Power Systems) from U.K,

2014, He received a B.SC. degree in Electrical and Computer Engineering from the Omdurman Islamic
Universty. He received a M.SC degree in Renewable Energy from the University of Khartoum, Sudan. His
research activities include the Renewable Energy, high voltage engineering. His teaching includes Renewable
Energy, control system and fundemental of electrical engineering.

El-Amin Hamoda is an associate professor in the Department of Electrical Engineering at University of

Khartoum, Sudan. He received a doctorate degree in power systems from Queens University, Belfast, Nothern
Ireland. His research interests include high voltage systems, power system fast transients and switching.

Abdelrahman Ali Karrar is an assistant professor in the Department of Electrical Engineering at University of
Khartoum, Sudan. He received a doctorate degree in Electrical Engineering (Power Systems) from of Loughborough
University of Technology, 1992, the U.K. He is an IEEE member and his reasearch interests involve power system
operation and control, and power system stability.

© 2016 by the authors. Submitted for possible open access publication under the
terms and conditions of the Creative Commons Attribution (CC-BY) license
(http://creativecommons.org/licenses/by/4.0/).
28 | P a g e