Report Main

ELEC6089 High Voltage Insulation Systems Assignment 1
HV AC 275kV Bushing Design
Thomas J. Smith, David Mahmoodi, Brendan Hickman, Patrick P. L. Fong
University of Southampton
March 25, 2014
Abstract
The complexities involved in working with high voltages is of great importance to bushing design.
Transformer bushings are used to insulate high voltage conductors as they pass through the grounded
exterior of the transformer. Bushings are required to avoid various electrical breakdown mechanisms
caused by the potential difference from high potential equipment to the surrounding earth objects.
This report presents a comprehensive investigation of transformer bushing design, considering the
different types of breakdown in high voltage systems, varying stress control methods for DC systems
and different insulation material.
Beyond this, the bulk of the report is a comparison of different simulation models effectiveness.
Solid bushing and capacitive grading bushing, the act of inserting conducting foils within the insu-
lation; non-graded, radial and axial bushing were all simulated and the resulting fields tested for
many different failure criteria. Axial grading, by controlling the axial field responsible for surface
flashover, was the most effective model. A further effort was made to improve the foil interfaces
by attaching semiconductor material to the tip as well as smoothing the foil edge profile. These
improved peak field values considerably for the radial design; proof that within bushing design there
is still necessary research to be done.
1 Introduction
The design of electrical equipment always involves an aspect of insulation design. For the safe and
efficient operation of electrical equipment it is necessary to have an electrical circuit and a means of
isolating this circuit from the surrounding environment [1]. Power systems contain a complex structure
of generators, transmission lines, transformers, switchgear and other equipment. All of these different
devices require an appropriately selected insulation material in order to isolate the mechanical casings
and support structures from the high voltage components [2].
The purpose of this report is to describe the design and simulation of a high voltage bushing. Bushings
are an integral part of power system insulation. IEEE standard C57.19.00 describes a bushing as
“an insulating structure, including a through conductor or providing a central passage for such a
conductor, with provision for mounting on a barrier, conducting or otherwise, for the purpose of
insulating the conductor from the barrier and conducting current from one side of the barrier to
the other.”[3]. Bushings are required for situations such as connecting the external conductor to the
internal windings of a transformer through the walls of the metal oil tank. The walls of the transformer
housing will be grounded, but need to be shielded from the incoming high voltage conductor, hence
1
ELEC6089 High Voltage Insulation Systems Assignment 1 2
the use of an insulating bushing [1]. An example of this application can be seen in figure 1.1, as 400kV
grid conductors enter an oil filled transformer casing. The shedding on the outer cylinder can be seen
in figure 1.1 which helps increase electrical strength in wet conditions [1].
(a) Transformer wall connection (b) Wide view
Figure 1.1: High Voltage Bushings on the 400kV Transformers at Staythorpe CCGT Power Station,
Newark, UK (Photos: Thomas J. Smith)
2 Overview of Insulation Failure
There are some consequences that need to be taken into account when designing systems for operation
at high voltage. High voltage systems generate a higher electric potential than surrounding objects,
which are usually at earth potential. These arise large electric potential gradients or electric fields.
The values of high field regions within the electric field may cause electrical breakdown and partial
discharges leading to failure of the system [4]. It is therefore important to design systems to minimise
the chance of these events.
Electrical breakdown is the action of electrical conduction across an insulating medium, usually a
dielectric, following the voltage across this medium exceeding the breakdown voltage of the specific
material. This usually happens when the potential difference is extremely high and arcing can be seen
in gaseous insulating mediums. This can cause changes to the compounds in the insulating medium
and also cause damage to equipment in the form of treeing as shown in figure 2.1.
There are different types of electrical breakdown to consider. The main types of electrical breakdown
that are involved in high voltage insulation systems are general breakdown in the system, surface
flashover, partial discharge in the dielectric insulation as well as corona discharge in air.
2.1 General Breakdown
In the case of a high voltage transformer bushing, there is a combination of both high electric fields and
a close proximity to the grounded surroundings. As mentioned in section 1, the purpose of a bushing is
to insulate the conductor from ground. The most basic form of bushing is a non-condenser bushing [6].
It is the radial component of the tangential electric field that causes this typical breakdown directly
from the conductor to the grounded flange. Insulation must have a very high dielectric constant
Figure 2.1: Electrical tree tracing the path of damaged insulation caused by electrical breakdown [5]
to withstand high voltage conditions. Typical materials used for bushing manufacture include Resin-
impregnated paper (RIP) and Oil-impregnated paper (OIP). The importance of this type of bushing is
that as the electric strength of the insulation increases, radial thickness may be reduced [6]. Typically
the general breakdown strength of bushings is very high relative to PD inception and is only a
secondary concern.
2.2 Surface Flashover
The partnering electric breakdown effect for solid, non-condenser bushing is surface flashover; break-
down caused by the electric field travelling from the conductor to the surroundings via the surface of
the bushing. The properties governing the axial height of the bushing are both the axial electric field
and the surrounding medium [6]. Typically the criteria for surface flashover breakdown is much lower
than general breakdown, and improvements to the bushing have primarily been in shaping the axial
field distribution. This investigation considers an air-to-oil bushing and as oil is more than twice as
strong dielectrically as air at atmospheric pressure the air end must be approximately twice as long
[7]. Creepage distance is a critical factor here, as defined in the IEC 60137 standard as the “shortest
distance along the surface of an insulator between two conductive parts” [8]. For industrial standard
bushings, ceramic shedding is used to extend the length of the surface of the bushing. While this
is beyond the scope of this report and therefore not modelled, it is important to note that it will
typically increase the effective axial length of the bushing by a factor of 4.
2.3 Partial Discharge
Partial Discharge (PD) is defined by the British Standards 60270:2001 as a localized electrical discharge
that only partially bridges the insulation between conductors and which can, or can not, occur adjacent
to a conductor and as a consequence of local electrical stress concentrations in the insulation [9]. The
most common cause of partial discharge is a void, originating from manufacturing imperfections, within
the dielectric material where the void contains material of a lower electrical breakdown strength (gas
or air) than the dielectric material. Under high electric fields across the dielectric insulation these
voids will experience local electrical breakdown as the rate of charge increase is greater than decay and
inception voltage is exceeded. Due to these conditions, PDs typically occur under intense tangential
electric field resulting in electron emission [10] and as the discharges spread across the surface of the
solid dielectric breakdown will occur [4]. In order to ensure the insulation system can sufficiently
reduce the chance of PD, the inception voltage of the insulation must be relatively high compare to
the operating voltage of the conductor.
Partial discharge is a primary cause of ageing within High Voltage systems, this is because each
successive discharge applies electro-mechanical forces to the void itself and the insulation progressively
deteriorates [11]. Therefore reducing tangential electric fields and measurement of PDs is of high
importance.
2.4 Corona Discharge
Corona discharge is the ionisation of fluid around a conductor of high electric potential creating a high
electric field ionisation region [12]. The ionisation of the surrounding fluid is due to the discharge from
the conductor. In high voltage systems, corona discharge occurs most commonly from the conductor
to air surrounding the conductor as illustrated in figure 2.2. Corona discharge occurs at an area of
intense electric field, but not sufficiently intense to cause arcing. This type of breakdown might not
cause fatal damage to the insulation, but would shorten the life time of the insulation system. It also
represents power losses to the surrounding, so it is important to design insulation system minimising
the effect of corona discharge.
Figure 2.2: Corona discharge appearing on the insulation surface of high voltage conductor
3 Grading Methods
Electric field stress control is important in the design of many power system elements, especially
cable terminations and bushings [2]. Failure of a bushing can damage the power transformer it is
protecting, which can be an expensive mistake [1]. Bushings are required to withstand Electrical,
Mechanical and Thermal stresses as defined in the IEEE standard C57.19.00 [3]. The design of the
bushing is largely determined by the insulation material chosen and the resolution of these conflicting
sources of stress. A good bushing design has insulation that can withstand the applied voltage and
thermal characteristics appropriate for the current carried by the conductor [7].
HV Conductor
Air
Bushing
X X
Transformer
Casing
X X
Transformer Oil
Figure 3.1: The Bushing Problem
The problem grading methods attempt to resolve is laid out in figure 3.1. The grounded transformer
casing is shown in light grey which is perpendicular to the the bushing insulation shown in dark grey
and the high voltage conductor in white. The top of the bushing is exposed to air, while the other
side is exposed to transformer oil. Conducting a numerical analysis or simulation would show that
the conductor surface within the plane of the transformer casing and at the points marked by red
crosses would experience high electric field stress. The bushing insulation is designed to withstand the
high electric field between the conductor and the transformer casing, however at the points marked
with crosses the interface between the solid insulation and the air/transformer oil would cause surface
discharge leading to relatively low flashover voltages [4]. It is therefore necessary to develop methods
of reducing electric field stress to a more uniform distribution for both functional purposes and the
economic use of space and materials [2].
3.1 Overview of Low Voltage and DC Solutions
There are several methods that can be used dependent upon the application. Low voltage solutions
include internal and external screening electrodes, while resistive stress control can be used for DC
applications. Sometimes these solutions are used in combination to achieve an acceptable result.
3.1.1 External Screening Electrode
External screening electrodes are parts outside the conductor that are not electrically connected to
the conductor. They are made of metal conductors such as aluminium. Corona ring designs are
intended to reduce the electric field strength around the bushing terminal, hence reducing the chance
of corona or partial discharge. Grading ring designs are intended to reduce the potential gradient of
insulator, hence reducing the chance of electrical breakdown. These screening electrodes come with
various shapes according to the different designs. The main types of design take the shape of sphere,
toroid or ring. These are shapes which prevent regions of intense electric field strength by varying the
electric potential distribution and help contain electric field as much as possible. The reduction in
corona discharge not only reduces the power loss, it also suppresses the ageing speed of the insulator.
An example of these can be seen on figure 1.1(a) and figure 1.1(b), where they are placed at the
top of the bushings. The diameters of these designs are closely related to the electric field strength
around the electrode, so the diameters of these designs must be carefully considered in order to avoid
electrical breakdown.
3.1.2 Internal Screening Electrode
Internal screening electrodes are also used to control the electric potential distribution so the electric
field strength is within acceptable level to reduce the chance of breakdown. They are placed inside
the insulator and usually in a pressurised gas. Some existing designs have the shape of the internal
electrode as a 3D cone and is referred to as a deflector. There are also designs which are disk shape.
An example of the field distribution with the cone shape deflector inserted is demonstrated in figure
3.2.
Conductor
Deflector
Insulation layer
Figure 3.2: Field distribution with deflector
3.2 Capacitive Grading
Capacitive grading was first proposed by R.Nagel of Siemens in a German paper published in 1906 [7].
The value of this type of arrangement was quickly recognised, and is now industry standard practice
for AC bushing designs for 25kV - 1500kV applications [2]. The general concept of the design is
illustrated in figure 3.3, showing the isolated foils inserted inside the solid bushing insulation. Shown
in red in figure 3.3 is the potential field with no grading, and in blue with the isolated conductive foils
inserted. It shows that the whole dielectric is much more evenly stressed with the capacitive grading
method.
The insulation is stressed in both radial and axial directions, which sum to give the tangential field.
The radial component Er can cause breakdown of the insulating material, while the axial component
Ez can cause surface discharge along the boundary [13]. Attention must be paid to the design and
shape of the boundary, so that the critical value for inception voltage for surface discharge is not
75%
50%
Ez 25%
75%
50%
25%
Transformer
Casing
Er
Figure 3.3: Field Distribution both without capacitive grading (shown in red) and with capacitive
grading (shown in blue), modified from [2]
exceeded [14]. These can be seen in green in figure 3.3. These sum to give the tangential field Et at
any point in the design.
Before proceeding, it is first necessary to introduce some terms. Firstly, the radius of the foil is
Ln
b ln b rn
sn
rn rn-1
r0
Figure 3.4: Symbols for calculating capacitive grading, modified from [13]
referenced from the centre of the conductor, and termed rn . The spacing between each foil is defined
in equation 1.
Sn = rn − rn−1 (1)
Additionally, the length of each foil is referred to as Ln and the difference in length on the right
and left side between each foil is termed bln and brn . Symmetric double sided capacitive grading is
achieved when bln = brn [13]. The total number of foils in the system is N . Also note that subscript
n denotes the outermost foil.
Inserting isolated conducting foils forms a set of coaxial capacitor units [4]. The equation for the
capacitance of one of these capacitors depends on the radial displacement rn and length of each foil
Ln , as in equation 2.
2πLn
Cn = rn (2)
ln( )
rn−1
The most widely used method to choose the dimensions and locations of the foils is double sided
capacitive grading, of which there are two variants; radial grading and axial grading [13]. The aim of
capacitive grading is to evenly distribute the electric field between the foils. To achieve this, an even
voltage difference between foils is required as in equation 3, where V is the total voltage difference
between the conductor and the casing, N is the number of foils required and ∆V is the voltage
difference between each consecutive pair of foils [15].
V
∆V = (3)
N
For the voltage between each foil to be constant, as in equation 3, the capacitance between each
consecutive pair of foils must also be constant. This is expressed as Cn = Cn−1 = · · · = C0
3.2.1 Radial Grading
The radial spacing and dimension of each foil is determined in the following derivation, which has
been verified and modified from [4]. In radial grading, the radial component of the electric field Er is
kept constant between all consecutive foils. The radial electric field is related to the voltage difference
and the spacing between each foil, as in equation 4. ∆V is already defined as a constant from equation
3, and so to have equal field the foil spacing Sn should also be constant.
∆V
Er = = Constant (4)
Sn
Given this condition and equation 2 for coaxial capacitance, the length of each foil is required to change
from foil to foil. The lengths and radii of consecutive foils can be calculated from the relationship in
equation 5.
2πLn 2πLn−1 2πL1
Cn = rn = Cn−1 = rn−1 = · · · = C1 = r1 (5)
ln( ) ln( ) ln( )
rn−1 rn−2 r0
The common factor of 2π cancels from equation 5 giving a simple equation linking the lengths and
radial displacements of consecutive foils, as in equation 6.
Ln Ln−1 L1
rn = rn−1 = · · · = r1 (6)
ln( ) ln( ) ln( )
rn−1 rn−2 r0
An approximate solution for thin foils can then be found. Under the thin foil assumption, rn =
Sn
rn−1 + Sn and << 1 even for the smallest radii of the inner foil. This is shown in equation 8.
rn
rn 1 Sn
ln( ) = ln ≈ (7)
rn−1 Sn rn
1−( )
rn
Ln rn ≈ Ln−1 rn−1 ≈ · · · ≈ L1 r1 (8)
Equation 6 can then be used to determine an exact solution while equation 8 can be used to find an
approximate solution in conjunction with initial data regarding the length and radial displacement
of the first foil and the spacing of the foils to calculate the parameters of all the other foils in the
bushing. Nevertheless, it should be noted, at this stage, that r0 refers to the surface of the conductor.
3.2.2 Axial Grading
In axial grading, the axial component of the electric field Ez is kept constant between all of the foils.
The following equations prove that the length of each foil must decay by a constant value for each
consecutive foil, and the radius at which it is placed is determined by a simple iterative formula.
The axial electric field is related to the voltage difference and the length change between each con-
secutive foil as in equation 9. Under symmetric capacitive grading, bn = bln = brn with reference to
figure 3.4. ∆V is already defined as a constant from equation 3, and so to have equal field, the change
in foil length bn should also be constant.
∆V
Ez = = Constant (9)
bn
The relationship between Ln and bn is defined in figure 3.4, as explained in equation 10.
Ln = Ln−1 − 2bn (10)
Since equation 9 requires the change in foil length bn to be constant, equation 2 for coaxial capacitance
requires the radius of each foil to change from foil to foil. This can be simplified to a similar form
as equation 6, except that the initial information required, in this case, is different. The necessary
parameters are: the length of the first foil(L1 ), the radius of conductor and first foil(r0 , r1 ). However,
for initial calculation of Ln (n = 1, 2, ...N ) the size of the constant difference in length between each
of the foils should be known (bn ). All other lengths and radii can then be calculated.
In case of axial grading where different material is used on each side of the bushing (bln 6= brn ), a
similar calculation is carried out for each side. The total length of each foil is found by adding Lln and
Lrn . Furthermore, position of each foil could be calculated using the recursive formula in equation 11
Ln r1
rn = r(n−1) exp ln( ) (11)
L1 r0
4 Design Details
The reference model for this project is shown in figure 4.1. The reference design is a paper impregnated
with oil bushing with 21 aluminium foils of 100µm thickness. One side of the bushing is exposed to
air, the other to oil, similar to a transformer bushing. The diameter of the conductor is 100mm, the
bushing diameter is 300mm. The length of the first foil is 5000mm long, and fixed 2mm into the
bushing at the conductor voltage. The outer foil is also set 2mm inside the bushing and is directly
connected to the earthed flange. The conductor is used at 275kV AC voltage, and the design was
taken from a bushing that was in operation for around 30 years.
Figure 4.1: The reference problem taken from [16]
4.1 Design Issues
In section 3.2 the initial information required for both radial and axial grading includes the length
and radial displacement of the innermost foil. In the reference design the following initial information
is given in table 4.1.
Table 4.1: Initial Information for Reference Design
Initial Information Value

Conductor Diameter (ID) 100mm
First Foil Length L1 5000mm
First Foil Radius r1 52mm
Outer Bushing Diameter (OD) 300mm
Outer Foil Radius r21 148mm
This information intuitively fits radial grading best, since there is no requirement to assume the length
of the outermost foil. However, there is a discrepancy between the standard literature problem and
the reference design. The first foil is connected to the high voltage conductor, and the last foil is
connected to the earthed flange. This is to eliminate the electric field on the boundary interface as
far as possible on both sides of the bushing, so that the voltage drop occurs exclusively inside the
bushing insulation.
This has an impact on the calculations described in section 3.2. Since the innermost foil is at the
same voltage as the conductor, there is no capacitance between them, as shown in figure 4.2. The
derivation of the iterative equations assumes a capacitance between the current foil and the previous
foil or conductor (rn , rn−1 ). The first foil in figure 4.1 is therefore indexed as 0 and not 1.
L2 = ?
L1 = ?
L0 = 5000mm 275kV
C 0 = 0F r2 r1 r0
Conductor 275kV
Figure 4.2: Diagram to explain the assumptions required
This means that there is not sufficient initial information to proceed with ether radial or axial grading,
since the first non-connected foil length L1 is not given. The iterative equations require the first non-
connected foil length L1 for radial ( L1 & r1 for axial) grading to be known as shown in equation 12
and 13. In the radial grading case, all radii variables are known due to the even spacings under radial
grading. In the axial grading case, the length of foils are known due to known parameters bln and brn .
r2
ln( )
r1
L2 = L1 r1 (Radial grading) (12)
ln( )
r0
L2 r1
r2 = r1 exp ln( ) (Axial grading) (13)
L1 r0
If this is not taken into account then a flawed design will be produced in both cases. Equations 14 and
15 show a wrongly described first iteration of the radial and axial grading formula. For radial grading
the resulted design is shown in figure 4.3. This shows that the length of second foil is much bigger
than the first foil. This is clearly wrong, and does not give the hyperbolic shape from the beginning
of the foils.
56.8
ln( )
L1 = 5000 52 = 11256mm (Radial grading) (14)
52
ln( )
50
49418 52
r2 = 52 exp ln( ) = 54.05mm =⇒ r21 = 103.52mm (Axial grading) (15)
5000 50
In order to proceed with the calculations there must be an assumption of the length of the first
unconnected foil. A reasonable assumption is that this follows the hyperbolic shape of the other foils
in radial grading. In axial grading the foil lengths are expected to decay with the axial field strengh.
Both of these expectations are realised when the first foil is assumed to be isolated for calculation
purposes only. These assumptions in both cases help to evenly distribute the electric field radially or
axially accordingly. To achieve this design mathematically, the following assumptions are made for
the calculations only:
1. Foil 0 is not connected to the HV conductor for both cases.
2. The conductor surface is spaced a distance of Sn from foil 0 in radial grading.
3. The conductor surface is spaced an adjustable distance from foil 0 in axial grading.
Flawed Radial Graing Design

200
150
100
50
Radius (mm)
0
−50
−100
−150
−200
−6000 −4000 −2000 0 2000 4000 6000
Length (mm)
Figure 4.3: Flawed Radial Grading Design
The reason of each assumption according to the design constrains are explained as following:
• Assumption 1 is required to be able to use the capacitor derived iterative formula on foil 0.
• Assumption 2 is required so that the radial spacing is kept constant.
• Assumption 3 is required so that axial grading could be calculated with a varying parameter
value for r0 . This makes it possible to adjust the initial gap so that the last foil will be placed
exactly at 148mm.
The first iteration has been calculated under these assumptions, giving the result in equations 16 and
17 which are expected values. The remainder of foil parameters can then be calculated using the
iterative formulas in each grading.
56.8
ln( )
L1 = 5000 52 = 4558mm (Radial grading) (16)
52
ln( )
47.2
49418 52
r2 = 52 exp ln( ) = 55.15mm =⇒ r21 ' 148mm (Axial grading) (17)
5000 50 − 1.007
5 Matlab Calculations
Two Matlab scripts were developed for computation of radial and axial grading. These scripts were
built to be easily customisable for any number of foils and any initial values, to cater for the calculation
of improved designs. They also automatically output data in a form for direct input into the COMSOL
model, auto-update a LATEXfile containing the data to form results table and displays the results in
both 2D and 3D plots for quick design verification.
5.1 Matlab Radial Grading
In the case of radial grading the code takes a required number of foils, and the inner and outer
dimensions of the bushing, to calculate the radial location and length of each foil using the radial
grading method as described in section 3.2 and also using the assumptions made for radial grading
design on 4.1.
For current design with specified parameters, the script plots the calculated foil positions in a 3D
graph shown in figure 5.1. Also the 2D plot of this design is shown in figure 5.1(b). This figure
illustrates the hyperbolic shape which was expected for radial grading. These figures allows a quick
verification of the scripts accuracy before proceeding to simulation.
(a) 3D Representation of foil radial position and length
Radial Graing
150
100
50
Radius (mm)
−50
−100
−150
−2500 −2000 −1500 −1000 −500 0 500 1000 1500 2000 2500
Length (mm)
(b) 2D Representation of foil radial position and length
Figure 5.1: Matlab Generated Plots of Geometric Radial Design
Table 5.1 shows the values obtained for radial grading.
5.2 Matlab Axial Grading
The axial grading script is similar to the radial grading script in design. However, as it is explain in
reference [15], the constant difference of foil length in oil and air side of bushing (bair , boil ) is calculated
Foil N.O Radius(mm) Length(mm) Foil N.O Radius(mm) Length(mm)

1 52.00 5000.00 12 104.80 2420.43
2 56.80 4558.22 13 109.60 2312.01
3 61.60 4188.21 14 114.40 2212.90
4 66.40 3873.79 15 119.20 2121.93
5 71.20 3603.30 16 124.00 2038.15
6 76.00 3368.13 17 128.80 1960.73
7 80.80 3161.78 18 133.60 1888.98
8 85.60 2979.27 19 138.40 1822.29
9 90.40 2816.68 20 143.20 1760.16
10 95.20 2670.92 21 148.00 1702.12
11 100.00 2539.51 22 150.00 0.00
Table 5.1: Radial Grading Calculations Results
by considering the value of flash over distance Lair and Loil . These values are calculated by using the
fact that average electric field strength along the boundary surface in oil side should be 3 to 4 Kv/cm
and nearly 9 to 12 Kv/cm in air side of the bushing. These values are calculated using equation 18.
∆V ∆V
bair = boil = (18)
900 v /mm 300 v /mm
Additionally, as shown in equation 19, in the calculation of r2 a parameter called Rparameter is sub-
tracted from r0 to change the position of assumed conductor surface. The radius of the last foil
(r21 ) could be correctly adjusted by making changes in this parameter and executing the script.
Rparameter = 1.007mm was the best value found for this design.
L2 r1
r2 = r1 exp ln( ) (19)
L1 r0 − Rparameter
The script plots 2D and 3D plots of axial grading to the user. Figure 5.2(a) illustrates the 3D
configuration of the design’s axial bushing. The 2D plot of this design, which is shown in figure
5.2(b), shows a nearly straight line from the edges of the first to last foil. In fact, the foil length
reduction is constant on both sides (boil , bair ). But, as the Matlab script is using the exact calculation
(non-linear) rather than the approximate method (linear), the foil endings are not placed exactly on
a straight line . It also illustrates that bair > boil as it was expected due to the relative permittivity
of each material. Finally the calculated values for axial grading, using this design method, are shown
in 5.2. These values along the given parameters in table 5.3 were used for COMSOL simulations of
this axial grading design.
The final information required to be able to proceed to the simulation phase is the relative permittivity
of each material. This was gathered from [13] and is shown in table 5.3. This data is used in all
COMSOL models.
(a) 3D Representation of foil radial position and length
Axial Graing
150
100
50
Radius (mm)
−50
−100
−150
−2500 −2000 −1500 −1000 −500 0 500 1000 1500 2000 2500
Length (mm)
(b) 2D Representation of foil radial position and length
Figure 5.2: Matlab Generated Plots of Geometric Axial Design
6 Modelling Results
The following simulations were completed using the COMSOL multiphysics software package. COM-
SOL is a professional finite element simulation package able to model a variety of physical features.
The following models are created using the AC/DC module, which is used to simulate electric and
magnetic fields [17]. Specifically, the electrostatics interface is used. This solves a charge conservation
equation for a given voltage and spacial distribution of charge [17].
6.1 Finite Element Methods (FEM)
There are inherent difficulties in solving the partial differential equations that govern many practical
engineering problems [4]. Despite knowing the equations and appropriate boundary conditions that
govern a problem, many are complicated by irregular geometries or other discontinuities. Numerical
methods allow approximate solutions to be obtained for problems intractable by analytic methods
Foil N.O Radius(mm) Length(mm) Foil N.O Radius(mm) Length(mm)

1 52.00 5000.00 12 95.65 4359.84
2 55.15 4941.80 13 100.68 4301.64
3 58.46 4883.61 14 105.90 4243.44
4 61.92 4825.41 15 111.32 4185.25
5 65.53 4767.21 16 116.93 4127.05
6 69.32 4709.02 17 122.73 4068.85
7 73.26 4650.82 18 128.74 4010.66
8 77.39 4592.62 19 134.95 3952.46
9 81.68 4534.43 20 141.35 3894.26
10 86.15 4476.23 21 147.96 3836.07
11 90.81 4418.03 22 149.96 0.00
Table 5.2: Axial Grading Calculations Results
Material Relative Permittivity (r )

Air 1
Oil 2.2
Paper Impregnated with Oil 4
Aluminium 108
Table 5.3: Relative Permittivity of Materials
[18]. In an analytic solution, the whole system is governed by a mathematical equation valid for the
entire region of interest. Although these differential equations are often mathematically compact, it is
difficult to obtain an answer unless the system is unreasonably simplified [18]. In FEMs, the complex
geometry is broken into a series of much smaller and simpler geometries [4]. These geometries can
be squares, rectangles or triangles in 2D or the 3D equivalent shapes. These simpler shapes form
interconnected subregions for which an approximate function, usually a high order polynomial, can
be used to represent the actual function. If the complex geometry is split into an adequate number
of simple shapes, these approximate functions closely matches the exact solution [18].
By default COMSOL uses a triangular discretisation to split up a complex geometry in a process called
meshing. This forms an unstructured grid of triangles, allowing the mapping of complex or curved
geometries. Other numerical methods such as Finite Difference Methods require a structured grid,
hence FEMs are more flexible with regards to geometry [4]. Meshing requires an initial understanding
of the expected outcomes of the problem, so that the mesh can be refined in areas of interest. Each
triangular element is approximated by a linear interpolation of the potential at the vertices of the
triangle. A set of linear algebraic equations are formed by minimising the error between the actual
solution and a set of approximate linear trial functions [18].
6.2 Equation Derivation and Boundary Conditions
The electrostatics interface of the AC/DC COMSOL module uses the electric potential V to calculate
static electric fields. A Poisson type partial differential equation is derived using classical electrostatics
and Gauss’ Law [19].
By taking Gauss’s Law:

∇.D = ρv (20)
the equation for electric flux density D:
D = 0 r E (21)
this can be combined with the equation for a static electric field:
E = −∇V (22)
to give by substitution:
∇.D = ∇.(0 r E) = −∇.(0 r ∇V ) = ρv (23)
which is more usually written:
ρv
∇2 V = − (24)
0 r
where 0 is the permittivity of free space, r is the relative permittivity of the material, E is the
electric field strength and ρv is the volume charge density.
In the special case where there is zero volume charge density, that is ρv = 0 then the equation simplifies
to Laplace’s Equation:
∇2 V = 0 (25)
The models used in this paper are 2D axisymmetric, meaning that a 2D model is used to describe
a 3D object that can be rotated 360o about a central point r = 0 to give a 3D geometry. This
assumes that not only is the geometry the same in the angular ϕ direction, but also that the electric
potential is constant. In this case, Poissons equation can be rewritten in cylindrical coordinates for a
2D axisymmetric model, it is multiplied by r to ensure there are no singularities at r = 0 [18].
∂ T
   
∂V
 ∂r   ∂r  rρv
 ∂  .(r  ∂V ) = − (26)
0 r
∂z ∂z
The boundary conditions are defined as the following:
1. All boundaries with the conductor and with foil 0 are set to V = V0 = 275kV
2. All boundaries with the transformer wall and with the outermost foil are set to V = 0
3. The interface between two insulator sub-domains is defined by n.(D1 − D2 ) = ρv and n × (E1 −
E2 ) = 0. These equations specify that the normal component of the electric flux density is
discontinuous at the interface between two dielectrics, and that the tangential component of
the electric field is continuous across the dielectric interface. n is the normal outward vector
pointing from dielectric 2 to dielectric 1.
6.3 Workflow
In order to simulate the electric field distribution within our bushing design, 2D axisymmetric models
were created. The general workflow to achieve this is:
1. Build a geometry representing the physical structure of the bushing.

2. Assign each geometric domain a material. The material selection determines the relative per-
mittivity r of each domain.
3. Define the charge conservation equation and all initial conditions. This includes setting which
boundaries are at ground and conductor potential and setting boundary conditions.
4. Design a mesh. The geometry is split into smaller elements in order to compute the charge
conservation equation. For designs with foils, special meshing parameters are required to speed
up the process.
5. Carry out the study. This stage is the actual computation of the solution.
6. Post-processing. Display the results in a number of formats including 3D, 2D and 1D plots, or
export the data for post-processing in Matlab.
6.4 Baseline Model
In order to minimise the computation time required for each model, it was necessary to determine the
areas of interest in the model. A bushing geometry was built with no foils inserted. A high quality
mesh was produced and the system was solved to find the electric field distribution throughout the
bushing and the surrounding area.
Figure 6.1: Baseline Model Geometry
Figure 6.2: Extra Fine Meshing for Baseline Model

Figure 6.3: Normal Electric Field (V/m) for Baseline Model
The air and oil extends radially 2m from the end of the bushing and 1m in the axial direction. This is
to understand the anticipated area of interest in the model. By considering figure 6.3 it is clear that
there is very little happening further than 500mm radially from the bushing surface and there is very
little of interest further than 200mm in the axial direction. Therefore all further models will adhere
to this geometry, ensuring that the area of interest is captured, while decreasing simulation times to
a minimum.
In figure 6.3, the normal electric field throughout the whole system is displayed. Electric field is
usually taken to be a vector quantity, that is a magnitude and direction. For example, the electric
field in the radial direction is termed Er and in the axial direction is Ez . The normal electric field is
not a normal vector, but an absolute value
√ of the electric field strength at a given point. This can be
described mathematically as normE = E.E. It gives a reasonable interpretation of the electric field
strength in either vector direction in the model, and will be used to compare all other models.
6.5 No Foils
In order to illustrate the requirement for capacitive grading within AC bushings, a simulation of a
bushing with no foils was conducted. Figure 6.4 shows the geometry and materials used for each
section. The conductor length is 6000mm and has a width of half the inner diameter, 50mm. The
paper impregnated with oil insulation is 5500mm long in order to accommodate the first foil length
of 5000mm with sufficient clearance. The transformer wall is modelled as a 50mm high aluminium
block that is 500mm wide, and is placed vertically in the middle of the design. The surrounding oil
and air are 500mm wide at the centre of the bushing, and extend to the length of the conductor. The
paper impregnated with oil insulation is sloped to avoid sharp corners and high field strengths.
Figure 6.4: Geometry and Materials for No Foils Bushing
A very fine mesh was used in this model as shown in figure 6.5. This allows for the maximum level of
accuracy in the results produced, and is possible since the geometry of this model is not over-complex.
It was necessary to ensure the mesh in the plane of the transformer wall was very fine in order to
sufficiently capture areas of very high field.
Figure 6.5: Extra Fine Mesh for No Foils Bushing
Figure 6.6: Normal Electric Field (V/m) for No Foils Bushing
The motivation for capacitive grading is clear in figure 6.6. The central area of stress in the plane
of the transformer wall is shown in figure 6.7. The highest stress is shown at the corners of the
transformer wall and in the plane between the transformer wall and the conductor. This reaches field
strengths greater than 4.5kV /mm which can cause the inception of partial discharge. This simulation
bears close resemblance to figure 3.1 in section 3 of this report, which was used to describe the electric
field problem. In order for this bushing to function within the design constraints, capacitive grading
must be used.
Figure 6.7: Area of very high electric stress
6.6 No Grading
In order to rectify the high electric stress identified in the no foils model, isolated foils are introduced
to capacitively grade the electric field. To prove the requirement for grading the lengths or radial
displacement of the foils, a simulation where the foils are maintained at the same length was performed.
This model was expected to fail the design criteria, proving that introducing isolated foils into the
bushing achieves very little and it is the radial or axial grading of the field that is critical. The bushing
has 21 foils spaced at an even radial spacing, but with no change in length of each foil. Each foil is
1756mm in length, and is centred in the within the bushing.
Figure 6.8: Geometry of the No Grading Model
Producing a finite element mesh for this model caused issues with long computation time. Even using
the coarsest default setting, the meshing time was of the order of days. This is due to the difference in
size of the aspects of the geometry. The foils are just 0.1mm thick, which requires a very small set of
triangles in order to mesh this area. The other domains are up to 6000mm long, which is considerably
larger and requires a different size of triangles.
Figure 6.9: User defined meshing for no grading simulation - whole view
In order to solve the issue, a set of meshing rules were produced. Firstly, the mesh within the conductor
and within the foils can be very coarse. Within a conducting material, no electric field is expected,
hence there is no reason to finely mesh that area. Areas of interest or areas where the field changes
rapidly need to have a very fine mesh to ensure the accuracy of the results is sufficient. The meshing
rules dictated a minimum number of points at the tip of each foil, and along the surface of the bushing.
This resulted in a refined mesh shown in figures 6.9 and 6.10.
Figure 6.10: User defined meshing for no grading simulation - foil tip view
Once the geometry has been meshed appropriately, the system can be solved to determine the electric
field strength throughout the model, as shown in figure 6.11.
(a) Normal Electric Field (V/m) for the No-Grading Model
(b) Normal Electric Field (V/m) in the Plane of the (c) Normal Electric Field (V/m) at the tips of the Foils
Transformer Wall
Figure 6.11: Examination of the Electric Field in the Non-Grading Model
Figure 6.11(b) shows the normal electric field in the plane of the transformer wall. The areas of very
high field at the corners of the transformer wall have been eliminated by the use of the capacitive foils.
However, the field close to the conductor is still very high. An additional issue is introduced by using
capacitive foils. Considering figure 6.11(c), the tips of the foils cause some considerable disturbance of
the electric field. This model achieves very little improvement, if at all, over the model with no foils.
Now that the requirement for grading has been shown in section 6.6, and the requirement to use a
formal axial or radial grading method in this section, the radial and axial designs can be compared.
6.7 Radial Grading
The COMSOL geometry for the radial grading model is shown figure 6.12. Each foil is an equal radial
distance according to the radial grading method. The lengths of each foil are taken from table 5.1.
The remainder of the geometry is identical to the No Foils and No Grading models.
As with the No Grading model, the geometry in this design is very complex with a variety of scales.
In order to reduce computation time but maintain a reasonable level of accuracy in areas of interest,
the mesh was produced to the rules outlined for the No Grading model. The meshing for this model
is shown in figure 6.13 and 6.14.
The model was then solved to find the normal electric field throughout the model. It can be seen in
figure 6.21(a) that the field between the foils nearest the conductor is reduced below 4.5kV /mm. This
Figure 6.12: Geometry of the Radial Model
Figure 6.13: Meshing of the Radial Model
Figure 6.14: Meshing of the Radial Model Focused on a Foil Tip
is now evenly spread in the radial direction, as expected due to the radial grading of the electric field.
The electric field in the radial direction is much improved.
However, the axial electric field is not considered in the radial grading design method. The field at
the foil tips is largely due to the axial field strength. It is necessary to inspect the strength of the
electric fields at the tips of the foils. It can be seen in figure 6.21(b) that the electric field at the tips
of the foil exceeds 4.5kV /mm in some places. Section 7 will take this observation further, to establish
if it meets the design criteria.
Figure 6.15: Normal Electric Field (V/m) for the Radial - Whole view
(a) Normal Electric Field (V/m) in the Plane of the Trans- (b) Normal Electric Field (V/m) at the tips of the Foils
former Wall
Figure 6.16: Examination of the Electric Field in the Radial Model
6.8 Axial Grading
The asymmetric axial grading method results in an uneven geometry about the centre of the bushing.
The lengths of each foil is determined in table 5.2. This requires the paper insulation to take a different
profile than the other models in order to adequately cover the foils, as shown in figure 6.17. The air
side has a much longer surface area due to the higher creepage distance of air.
Figure 6.17: Geometry of the Axial Model
The meshing for this model follows the design rules, in order to produce the finest mesh over the key
points of interest, shown in figure 6.18. Particularly figure 6.19 shows the fine mesh around the tip
and surface of a foil, which is a key area of interest to determine if the PD inception condition is
maintained.
Figure 6.18: Meshing of the Axial Model
The electric field problem can now be solved over the whole model. This produces the surface plots
shown in figures 6.20 and 6.21. It can be seen that the field in the plane of the transformer wall is
not well graded, as it is neglected in the axial design method. The field at the foil tips for which the
main contributor is the axial field has a smaller area exposed to high field strengths than the radial
Figure 6.19: Meshing of the Axial Model Focused on a Foil Tip and Surface
method. However, it remains an issue to be considered. This will be discussed and contrasted with
the other designs in the discussion in section 7.
Figure 6.20: Normal Electric Field (V/m) for the Axial Model - Whole view
(a) Normal Electric Field (V/m) in the Plane of the Trans- (b) Normal Electric Field (V/m) at the tips of the Foils
former Wall
Figure 6.21: Examination of the Electric Field in the Axial Model
.
7 Discussion
As has been outlined in this report so far, there are problems inherent within bushing design and
multiple solutions to these problems. A full discussion and comparison will be made to determine the
most accurate and reliable model for HV bushing design. Additionally, reasoning and justification for
the incremental changes will outline the development of bushing and possible further improvements
that could be made in design.
7.1 Success Criteria
In section 2 four criteria were mentioned to determine the efficacy of the design: general/intrinsic
breakdown defined by the radial field, surface discharge defined by the axial field, partial discharge
within the material defined by areas of high tangential field and corona discharge within air.
Regarding general breakdown it is the largest breakdown strength of the model and so the least likely
to fail. High Voltage Engineering Fundamentals has it defined as well in excess of 1MV/cm [4]. The
area of interest for this intrinsic breakdown is the line from the conductor straight to the flange, and
a measurement of radial electric field.
Surface discharge is not as readily defined for typical graded designs, however a comparison can be
made from shedding specifications. Shedding increases the effective creepage length of the bushing
by a factor of 4 and for a typical design 40mm/kV is given as the required length for the operational
voltage [6]. The IEC 60815 standard defines the creepage distance for ceramic insulators according
to levels of pollution in the environment [8]. For bushings less than 300mm in diameter the creepage
distance in very heavy pollution levels is 31mm/kV, which is lower than the extreme value quoted
in [6]. The creepage distance is multiplied by the highest r.m.s phase-to-phase voltage within the
equipment.
Fortunately and reasonably given its significance, partial discharge inception voltage is well defined
for all types of insulation material. For resin-impregnated paper the value is 36kV/cm and for oil-
impregnated paper the value is 45kV/cm, justifying the more common use of OIP [6]. Due to its
nature, a simple investigation of trends is not sufficient to estimate areas of possible PD origins and
areas of interest must be found and tested against the inception voltage. They are typically located
at the edges of foils or near the conductor itself. If a design passes testing for PD, then it is assumed
to pass the intrinsic breakdown requirement since the PD threshold is significantly lower than the
intrinsic breakdown threshold.
Corona discharge can occur in both air and oil, however, since the inception voltage is much lower
for air and the effects of weather and pollution affect the air side of the bushing much greater it is of
higher importance. For the discussion, it will be studied as a subsection of surface discharge.
7.2 No Foils
Nowadays, solid bushing is only used for voltages below 25kV. This is due to a combination of all
factors as shown in the previous figure 6.6. When simulated with a conductor ten times its rated
voltage its flaws are clear. As mentioned previously, the primary criteria to solve are high electric
fields inducing partial discharges and large axial fields causing flashover.
This is best illustrated with figure 7.1(b), not only are the fields present at the interface to the flange
well in excess of the PD inception voltage, the location of these high electric fields also mean flashover
x 10
6 Normal Electric Field in the No Foils Model
10
Conductor Mid Wall
9 Air Interface
Normal Electric Field Strength (V/m)

Oil Interface
Air
8 PD Inception Voltage
Bushing
7
6
Transformer
Casing 5
A B
A B
A B 4
1
Transformer Oil
0
50 100 150 200 250
Not to scale A Radial Distance from Conductor Centre (mm) B
(a) Location of Cut (b) The No Foils Model Fails PD Criteria
Plane
Figure 7.1: The Normal Electric Field in the No Foils Model
is a serious risk due to the lack of shaping of the axial electric field distribution. A result of such
localised electric stress is that because the field is greater than the PD inception voltage the voids
will generate exponentially more discharges, leading to faster ageing and a much higher chance of
breakdown. The solution, assuming keeping voltage constant, would be to increase the bushing height
and width.
Conductor
Air
0mm
Bushing
Transformer
Casing
2750 mm
2750mm
0mm
Transformer Oil
Not to scale
(a) Location of Cut (b) The Axial Electric field along the bushing surface
Plane(s)
Figure 7.2: Axial electric field along the surface of the bushing
It has already been established that the axial electric field is the primary component for surface
discharge. 7.2(b) shows how the axial field varies along the surface of the bushing. Of interest is that
there is very little field throughout the surface apart from at the flange junction; the most sensitive
area. The short distances and large fields show that corona discharge is possible and will most likely
lead to total failure of the system.
7.3 No Grading
The proposal to study a model of condenser bushing without any thought to grading exposes the
fundamentals to grading design. Figure 7.3(b) shows that, for the most part the electric field is below
the PD inception voltage, at least at the central interface. However looking at figures 6.11(b) and
6.11(c) show that the foil placement did not effectively shape the field at the bushing interface at the
tips. Beyond analysing the concentrations of electric field it is important to note that the distribution
is non-linear; in comparison to the ideal goal of a linear drop of field across the capacitors [4]. The
placement of the outermost foil serves as a location of high field density and although located far from
the flange itself it would likely be a source of both PDs and corona discharges.
Conductor x 10
6 Normal Electric Field in the No Grading Model
10
Mid Wall
Air 9 PD Inception Voltage
Air Interface
Bushing 8 Oil Interface
Transformer 6
Casing
A B 5
A B
A B 4
1
Transformer Oil
0
50 100 150 200 250
Not to scale Radial Distance from Conductor Centre (mm)
(a) Location of Cut (b) The No Grading Model Fails PD Criteria

Plane
Figure 7.3: The Normal Electric Field in the No Grading Model
The most noticeable change that still displays fault characteristics for the non-graded model is the
axial electric field plot along the surface. 6.11(c) is the relevant Electric field plot demonstrating the
effect of non-graded foils. Although installing non-graded capacitors did move the large electric fields
away from the junction, it clearly failed in reducing the magnitudes of such fields. This is due to a
lack of gradient within the foils, allowing a concentration to gather on the edge of the insulation. The
resulting effect would be a great deal of PDs and corona discharge and possible surface discharge.
The fact that corona discharges not only degrade the medium but lead to a loss of power prove this
to be an unsatisfactory and incorrect solution.
One of the goals of capacitive grading, beyond generating uniform fields is to have an electric potential
that decreases uniformly from the conductor to flange. The voltage gradient is of course related to
electric field and is a possible indicator of bushing performance. However as is shown, the radial
electric field through the centre is not of primary concern in regards to testing for failure.
Conductor
Air
0mm
Bushing
Transformer
Casing
2750 mm
2750mm
0mm
Transformer Oil
Not to scale
(a) Location of Cut (b) The Axial Electric field along the bushing surface for no-grading
Plane(s) model
Figure 7.4: Axial electric field along the surface of the bushing
Conductor 5 Electric potential in the No Grading Model

x 10
3
Air Mid Wall
Bushing 2.5
Electric Potential (V)
2
Transformer
Casing
1.5
A B
0.5
Transformer Oil
0
50 100 150
Not to scale Radial Distance from Conductor Centre (mm)
(a) Location of Cut (b) The Electric potential across the midwall
Plane
Figure 7.5: The electric potential within the no grading model
7.4 Comparison of Axial and Radial Grading Solutions
The design of both types of capacitive grading method were discussed in the previous sections. The
difference in performance of the types in general is the radial grading will improve the intensity of
the radial electric field and the axial grading will improve the intensity of the axial electric field.
By improvement of intensity of electric field, this means the electric field according to the type of
grading method is evenly distributed in a particular direction. Choosing the right design for different
application is essential as this will reduce the chances of various types of electrical breakdown and
hence the bushing design would be more durable.
There are two components of electric field cause different problems. The radial electric field is mainly
responsible to the breakdown of insulating material, for example, electrical breakdown between foils.
On the other hand, the axial electric field is mainly responsible to the surface discharges along the
boundary of the insulation, for example, corona discharge to the surrounding.
7.4.1 Radial Electric Field
The maximum intensity of the radial electric field for the radial design is expected to be lower than the
maximum of the axial design. This means the radial grading design would perform better to suppress
the intensity of radial electric field. The radial electric field across the radial direction of both axial
and radial grading designs are shown in figure 7.6(b).
6 Radial Electric Field of Designs

x 10
Conductor 5
Axial Grading
Radial Grading
4.5
Air
4
Bushing
3.5
Electric Field (V/m)
3
Transformer
Casing 2.5
A B
2
1.5
0.5
Transformer Oil 0
0 50 100 150
Radial Position (mm)
Not to scale
(a) Location of Cut (b) Radial electric field across the mid-point of the two designs of bushing.
Plane
Figure 7.6: The Normal Electric Field compared in the Axial and Radial models
Figure 7.7: 2D surface plot showing the constant distribution of electric field in the axial direction
from mid-point to the air end of bushing in the radial design
The result clearly shows the peak value of electric field for the axial design (≈ 4.5 × 103 kV m−1 ) is
greater than the peak value of electric field for the radial design (≈ 3.5 × 103 kV m−1 ). Also, from the
figure 7.6(b), the radial electric field of the radial grading design is more evenly distributed. This is
the effect of grading, hence similar effect should be seen in the axial electric for axial grading design.
The electric field distribution across any perpendicular cut lines to the foils would result in almost
identical results, because the foils behave similar to capacitors in series. Electric field being constant
across a capacitor and this is true across any two neighbouring foils. Figure 7.7 shows the electric field
between consecutive foils from the mid-point to the foil tips. It can be seen that the field is constant
in the axial direction between two consecutive foils, except at the tips, which is also true in the axial
case.
7.4.2 Axial Electric Field
The axial component of electric field is responsible for the surface discharges along the surface of the
bushing design. Axial component of the field contributes to these surfaces, because the axial electric
field at the edges of the foil are the most intense. Figure 7.8 shows the magnitude of electric field at
each foil edge.
The result shows the axial design has a lower peak value of electric field at these edges of foil. This is
the expected result, because axial grading changes the axial field. Also, a clearer maximum peak of
electric field should be seen at the outermost foil of the radial graded design due to the sharp turn in
the shape of the bushing design. Although the result shows a maximum peak at the last outermost
foil, it is not significant. Generally, the electric field at the foil tips in both designs exceeds the partial
discharge inception voltage for oil-impregnated paper insulation. This will result in partial discharge.
However, the points which are taken in to consideration are the highest values of electric field locally
which drops off rapidly moving away from the point. The sum of the area exceeding the PD threshold
is approximately 0.03% of the volume of the overall bushing design, hence, this can be consider as an
affordable problem in the bushing design. Also, in the model, all the foils are considered to terminate
with a 90◦ bend. In reality, designs may use various various techniques to suppress the effect of this
bend. Some design uses semi-conducting material at the ends of the foil [4].
7.4.3 Surface Flashover
Surface flashover is the partial discharge along the surface of the insulation, this is cause by a intense
electric field. The parameter which takes into account of surface flashover is the creepage distance as
mention previously. The typical value for calculating the creepage distance is 35mm/kV [8]. and for
more polluted environment, a higher value of creepage distance might be considered.
In bushing designs, the creepage distance is increased by the addition of a shed around the bushing.
These shed designs increase the creepage distance of insulation designs by increasing the surface
length. Designs of shed can have varying ratios between creepage distance to axial length [20], designs
for more polluted environment have bigger ratios.
According to the typical value to calculate the creepage distance, for a 275kV design, the creepage
distance required is 9625mm. Where the designs of bushing in this report has axial length of approx-
imately 3m, the ratio of minimum nominal creepage distance to axial length required is 3.2:1. This
is satisfied by many shedding designs, including the ABB GOE design with a ratio of 4.0:1 and the
GOE2 design with a ratio of 3.94:1 [20].
These shed provides a greater creepage distance to reduce the chance of surface flashover, however,
they have little affect on the electric field of the bushing. Hence, addition of shed does not reduce
corona discharge.
x 10
7 Peak Axial Electric Field at Each Edge of Foil(Air Side)
2
1.8
1.6
1.4
1.2
0.8
0.6
0.4
0.2
Radial Graded Design
Axial Graded Design
0
0 5 10 15 20 25
nth foil (n)
(a) Air side axial electric field
x 10
6 Peak Axial Electric Field at Each Edge of Foil (Oil Side)
18
16
14
12
10
2
Radial Graded Design
Axial Graded Design
0
0 5 10 15 20 25
nth foil (n)
(b) Oil side axial electric field
Figure 7.8: Electric field at the edge of each foil.
7.5 Design Improvements
The high electric field at the tips of the foils has been identified as a cause for concern. The field
strength at this point exceeds the PD inception voltage by double in the axial design, and around
four times in the radial design. It is therefore necessary to improve the base design.
The first aspect to consider is the accuracy and relevance of the COMSOL model. The very thin foils
are modelled as perfect rectangles with sharp 90o corners. This approximation was made to make the
geometry of the model simple for input to COMSOL. An improvement is to smooth the corners of
the foil to reduce electrical tensions and optimise electric field intensity [21]. By removing the sharp
edges of the foils, responsible for the localised electrical fields, the electric field distribution will be
more uniform and better fit the base theory. A refined model was required, to investigate the effect
of smoothing the foil tips.
The radial model was taken forward, since it already passes the radial field criteria. If a way to reduce
the field at the foil tips could be found, the radial design would pass all criteria. The end of each of
the foils was filleted, so that the ends formed a semi-circle. While this is still not the optimum design
for an electrode, it is certainly an improvement over the sharp angles.
(a) Curved Foil Tips (b) Angled Foil Tips
Figure 7.9: Normal Electric Field at the Tips Improvement
The curved foil tip is clearly visible in figure 7.9(a). There is an improvement, the curved design has
a slightly smaller area of field strengths higher than the threshold. However, it has not solved the
issue completely. Further design modifications are required.
Literature on bushing design refer to the high field at the tips problem [4]. Beside the grading methods,
some other additions to the bushing design have been put forward to reduce the intensification of the
electric field at the edges of the foils. Foils made from semiconducting materials can be used to adapt
the potentials at the foil edges to a less concentrated field distribution [4], though installation is not
mentioned.
This has been modelled in COMSOL by adding a square to the edge of the foil 0.1mm in dimension.
The outer edges of the square are filleted to a curve in the same way the tips were previously. These
squares were allocated the relative permittivity of r = 11.7, which is the permittivity of silicon. The
actual manufacture processes are not in the scope of this report, and different semiconductors may be
used in industry. However, adding this to the COMSOL will at least prove the concept.
The semiconductor block with curved edges can be seen in figure 7.10(a). It is shown alongside the
original angled design for comparison in figure 7.10(b). Once again, there is a subtle improvement in
the area exposed to very high field concentrations. However, it remains well above the 4.5kV /mm
threshold value.
(a) Semiconductor Foil Tips (b) Angled Foil Tips
Figure 7.10: Normal Electric Field at the Tips Improvement
The normal electric field values have been extracted and plotted in figure 7.11 to enable a more direct,
quantitative comparison. It is clear that the radial semiconductor tipped model has reduced the field
strength by almost half, and is nearly at the same level as the axial field design. The curved foil design
is an improvement, but the semiconductor design has a significant impact on the field distribution.
x 10
7 Normal Electric Field at foil tips Air side
2
1.8
1.6
1.4
1.2
0.8
0.6
0.4 Axial Original

Radial Original
0.2 Radial Curved
Radial Semiconductor
0
0 5 10 15 20 25
Foil Number (n)
Figure 7.11: Scatter showing the normal electric field at the tips of each foil in the axial, radial, radial
curved and radial semiconductor models.
8 Conclusions
Electrical bushings have applications across a range of high voltage engineering and is fundamentally
used to suppress the intensity of electric field throughout the system. High concentrations and large
magnitudes of electrical fields cause a variety of breakdown within the bushing. Focusing on high
voltage AC bushing design, capacitive grading methods are used extensively to control the electric
field distribution along the desired boundaries. This control of electric field is based on the condition
of constant electric field across desired points in the bushing and this condition is the basis of both
the radial and axial grading methods; derived from this condition and modelled within this report.
The resultant electric field distribution in the graded direction in the simulations was found to be not
strictly linear, but close to a linear line. By controlling the electric field, an evenly distributed electric
field can be achieved. Also, depending on the type of grading method used, suppression of the peak
field within specific regions is achieved.
From the results shown in the previous sections, the graph of variation in radial electric field shows
the peak radial electric field for the axial graded design, the limit of 4.5kV/mm is just exceeded. The
radial electric in the radial grading design is consistently below the 4.5kV/mm limit.
However, the axial electric field in both designs exceeds the threshold consistently. More importantly,
the axial electric field for the radial graded design is approximately twice as much as the axial graded
design, although this was greatly improved using semiconductor tips. Hence, the results show the
radial design suppresses more in radial electric field, while the axial electric field is suppressed more
in the axial graded design as expected.
The choice of design depends on the effect of the axial and the radial electric field. According to the
weakness for bushing designs, the axial electric field usually cause more problems as the breakdown
within a foil would require a higher electric field than surface discharge along the boundary layer.
Using the semiconductor coated foil tips as proven in the radial design could further improve the axial
field in the axial design. The choice of bushing would therefore be an axial grading design, using
semiconductor coated tips.
REFERENCES 36
References
[1] D.F. Warne. Newnes Electrical Power Engineer’s Handbook. Elsevier Science, 2005.
[2] R.E. James, Q. Su, and Institution of Engineering and Technology. Condition Assessment of
High Voltage Insulation in Power System Equipment. IET power and energy series. Institution
of Engineering and Technology, 2008.
[3] Ieee standard general requirements and test procedure for power apparatus bushings. IEEE Std
C57.19.00-2004 (Revision of IEEE Std C57.19.00-1991), pages 1–17, 2005.
[4] J. Kuffel, E. Kuffel, and W.S. Zaengl. High Voltage Engineering Fundamentals. Elsevier Science,
2000.
[5] Captured lightning - lichtenberg figures. [Online] Avaialbe: http://damnamazing.blogspot.

co.uk/2008/11/captured-lightning-lichtenberg-figures.html, last viewed on 25.3.2014,
2008.
[6] J.S. Graham. High-Voltage Engineering and Testing, 3rd Edition. IET Power and Energy Series,
2013.
[7] J.H. Harlow. Electric Power Transformer Engineering. The Electric Power Engineering Hbk,
Second Edition. Taylor & Francis, 2004.
[8] International Electrotechnical Commission. Insulated Bushings for Alternating Voltages Above
1000 V (IEC 60137, Ed. 5.0 (2003) MOD). 2008.
[9] High-voltage test techniques partial discharge measurements. BS EN 60270:2001, 2001.
[10] Hulya Kirkici. Surface flashover characteristics of dielectric thin films in vacuum: A review.
ELECO’99 International Conference on Electrical and Electronics Engineering, page 46, 1999.
[11] Ghosh P.S. Ghani A. Thayoob, Y.H.M. Frequency spectral analysis of electrical partial discharge
signals in xlpe cable under various soil conditions. In Power and Energy Conference, pages
1528–1531, 2008.
[12] Goldman A. Sigmond R.S. Goldman, M. The corona discharge, its properties and specific uses.
Pure & Appi. Chem., Vol. 57, No. 9, pages 1353–1362, 1985.
[13] Zeeshan Ahmed. Analysis of partial discharge in OIP bushing models. Master’s thesis, Royal
Institute of Technology (KTH), September 2011.
[14] H Rodrigo, L Graber, DS Kwag, DG Crook, and B Trociewitz. Comparative study of high voltage
bushing designs suitable for apparatus containing cryogenic helium gas. Cryogenics, 57:12–17,
2013.
[15] Fachgebiet Hochspannungstechnik Technische Universitt Darmstadt. Measures for field and po-
tential control. [Online] Available: http://www.hst.tu-darmstadt.de/uploads/media/hvt2_
v_02.pdf, last viewed on 23.12.2014.
[16] George Chen. ELEC6089 High Voltage Insulation Systems Assignment 1 HV AC 275kV Bushing
Design Briefing Notes, February 2014. The University of Southampton.
[17] COMSOL. Introduction to AC/DC Module - COMSOL Version 4.3. Part No. CM020104, 2012.
[18] Mohammad R Meshkatoddini. Study of the electric field intensity in bushing integrated zno surge
arresters by means of finite element analysis. In COSMOL Users Conference, 2006.
REFERENCES 37
[19] W.H. Hayt and J.A. Buck. Engineering Electromagnetics. MCGRAW-HILL Higher Education,
2012.
[20] Transformer bushings, type goe and goe(2), technical guide. [Online] Available: http://www05.
abb.com/global/scot/scot252.nsf/veritydisplay/60f3cd5c2edf5ff4c125771b002dbef1/
$file/1ZSE%202750-105%20en%20Rev%207%20(GOE).pdf, last viewed on 23.03.2014, 2010.
[21] Hafezinasab H. Karimi Madahi S.S. Majid Alambeygi Hassani, M. Optimization of electric field
in 230kv power transformers bushing. J. Basic. Appl. Sci. Res., Vol. 2, No. 5, pages 4939–4951,
2012.
REFERENCES 38
A Individual Contributions
Team Member Contribution

Thomas J. Smith COMSOL modelling of all designs. Sole author of section 1, 6 and 7.5.
23914254 Co-author of section 3 and 4. Project manager.
David Mahmoodi Derivation of equations from the texts. Matlab script author. Sole author
24241407 of section 5 and co-author of 3 and 4.
Brendan Hickman
Sole author 7.1, 7.2, 7.3. Co-author of sections 2, 8.
24230596
Patrick P. L. Fong
Sole author of 3.1 & 7.4. Co-author of section 2, 7, 8.
23875771
REFERENCES 39
B Code Listings
Radial Grading Matlab Code

1 %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
2 % ELEC6089 High V o l a t a g e I n s u l a t i o n B us hi ng D e s i g n
3 % Radial grading c a l c u l a t i o n
4 % Author − David Mahmoodi
5 % Date − 1 4 / 0 2 / 2 0 1 4
6 %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
7 clc ; close all ; clear all
8 %% D e c l a r e i n g Given V a r i a b l e s
9 Voltage = 275000;
10 Inner diameter = 100;
11 Outer diameter = 300;
12 F i r s t f o i l l e n g t h = 5000;
13 N = 21;
14 Foil Thickness = 0.1;
15 First Gap = 2;
16 Last Gap = 2 ;
17 %% D e f i n i n g new v a r i a b l e s
18 D e l V o l t a g e = V o l t a g e /N−1; %V o l t a g e between e a c h f o i l
19 D e l R a d i u s = ( ( O u t e r d i a m e t e r − I n n e r d i a m e t e r ) −2∗( F i r s t G a p+Last Gap ) ) / ( 2 ∗ (N−1) ) ; % S p a c i n g between e a c h foil
20 % I n i t i a l i s e v e c t o r s ( 2 2 f o r 3D p l o t t i n g )
21 L = z e r o s ( 1 ,N+1) ;
22 R a d i u s = z e r o s ( 1 ,N+1) ;
23 %% C a l c u l a t i o n
24 L ( 1 )= F i r s t f o i l l e n g t h ; %The f i r s t f o i l i s 5000mm and c o n n e c t e d t o t h e c o n d u c t o r , no c a p a c i t a n c e between
c o n d u c t o r and t h i s f o i l
25 r 0 = I n n e r d i a m e t e r / 2 ; %R a d i u s o f t h e c o n d u c t o r
26 R a d i u s ( 1 )= I n n e r d i a m e t e r /2 + F i r s t G a p ; % R a d i a l p o s i t i o n o f f i r s t f o i l = 52mm
27 %C a l c u l a t e t h e r a d i a l p o s i t i o n s o f a l l f o i l s
28 f o r i =2:N
29 R a d i u s ( i )=R a d i u s ( i −1) + D e l R a d i u s ;
30 end
31 % R e f e r t o S e c t i o n 2 . 2 f o r an e x p l a n a t i o n o f t h i s a s s u m p t i o n
32 L ( 2 )= l o g ( R a d i u s ( 2 ) / R a d i u s ( 1 ) ) ∗ L ( 1 ) / l o g ( R a d i u s ( 1 ) / ( R a d i u s ( 1 )−D e l R a d i u s ) ) ;
33 %F o l l o w t h e i t e r a t i v e f o r m u l a
34 f o r i =3:N
35 L ( i )= l o g ( R a d i u s ( i −1)/ R a d i u s ( i ) ) ∗ L ( i −1) / l o g ( R a d i u s ( i −2)/ R a d i u s ( i −1) ) ;
36 end
37 %For p l o t t i n g − add t h e o u t e r s h e l l
38 L (N+1)=L (N) −.5∗L (N) ;
39 R a d i u s (N+1)=R a d i u s (N)+ Last Gap ;
40 %% P l o t i n g
41 figure
42 p l o t ( R a d i u s ( 1 : end −1) , ( ( 6 0 0 0 + L ( 1 : end −1) ) / 2 ) , ’ o ’ ) ; h o l d on
43 p l o t ( R a d i u s ( 1 : end −1) , ( ( 6 0 0 0 −L ( 1 : end −1) ) / 2 ) , ’ o ’ ) ;%a x i s e q u a l
44
45 x=z e r o s ( 1 , 2 ∗ (N+1) ) ;
46 y=x ; j =1;
47 f o r i = 1 : 2 : 2 ∗ (N+1)
48 x ( i )=L ( j ) / 2 ;
49 x ( i +1)=−L ( j ) / 2 ;
50 y ( i )=R a d i u s ( j ) ;
51 y ( i +1)=R a d i u s ( j ) ;
52 j=j +1;
53 end
54 y2=−y ;
55 % 2D P l o t
56 figure
57 axes ( ’ FontSize ’ ,16 , ’ fontWeight ’ , ’ bold ’ )
58 r e c t H = r e c t a n g l e ( ’ P o s i t i o n ’ , [ − 1 . 1 . ∗ x ( 1 ) ,− r0 , 2 . 2 ∗ x ( 1 ) , 2∗ r 0 ] ) ;
59 s e t ( rect H , ’ FaceColor ’ , ’ r ’ )
60
61 for i = 1 : 2 : 2 ∗ (N)
62 h o l d on
63 l i n e ( x ( i : i +1) , y ( i : i +1) , ’ LineWidth ’ , 2 )
64 l i n e ( x ( i : i +1) , y2 ( i : i +1) , ’ LineWidth ’ , 2 )
65 %a x i s e q u a l
66 end
67 r e c t H = r e c t a n g l e ( ’ P o s i t i o n ’ , [−L (N+1) / 2 , O u t e r d i a m e t e r / 2 , L (N+1) , 1 0 ] ) ;
68 s e t ( rect H , ’ FaceColor ’ , [ 0 , 0 , 0 ] )
69 r e c t H = r e c t a n g l e ( ’ P o s i t i o n ’ , [−L (N+1)/2,−10− O u t e r d i a m e t e r / 2 , L (N+1) , 1 0 ] ) ;
70 s e t ( rect H , ’ FaceColor ’ , [ 0 , 0 , 0 ] ) ;
71 t i t l e ( ’ R a d i a l G r a i n g ’ , ’ FontName ’ , ’ Times New Roman ’ , ’ F o n t S i z e ’ , 3 4 , ’ f o n t W e i g h t ’ , ’ bold ’ ) ;
72 x l a b e l ( ’ Length (mm) ’ , ’ FontName ’ , ’ Times New Roman ’ , ’ F o n t S i z e ’ , 2 4 , ’ f o n t W e i g h t ’ , ’ bold ’ )
73 y l a b e l ( ’ R a d i u s (mm) ’ , ’ FontName ’ , ’ Times New Roman ’ , ’ F o n t S i z e ’ , 2 4 , ’ f o n t W e i g h t ’ , ’ bold ’ )
74 x l i m ( [ x ( 2 ) x ( 1 ) ] ) ; y l i m ( [ y2 ( end ) −10 y ( end ) + 1 0 ] )
75 % 3D P l o t
76 K=50; s c l = . 1 ; % Z d i r e c t i o n s c a l l i n g v a l u e f o r p l o t t i n g
77 p= 6∗N/3+1; %a d j u s t i n g t h e Cut i n t h e 3D shap
78 figure
79 axes ( ’ FontSize ’ ,16 , ’ fontWeight ’ , ’ bold ’ ) ;
80 R=[ r 0 r 0 ] ;
81 [ X, Y, Z ] = c y l i n d e r (R, 5 ∗K) ;
82 Z ( 2 , : ) =( L ( 1 )+ . 1 ∗ L ( 1 ) ) ∗ s c l ;
83 Z(1 , : ) = − Z(2 , : ) ;
84 s u r f (X, Y, Z , ’ F a c e C o l o r ’ , [ 1 , 0 , 0 ] , ’ E d g e C o l o r ’ , [ 1 , 0 , 0 ] ) ;
85 f o r i =1:N
86 h o l d on
87 R=[ R a d i u s ( i ) R a d i u s ( i ) ] ;
88 [ X, Y, Z ] = c y l i n d e r (R,K) ;
89 Z ( 2 , : )= L ( i ) ;
REFERENCES 40
90 Z ( 1 , : )= −L ( i ) ;
91 X = X( : , 1 : p) ;
92 Y = Y( : , 1 : p) ;
93 Z = Z ( : , 1 : p)∗ s c l ;
94 t e s t s u b j e c t = s u r f (X, Y, Z ) ;
95 s e t ( t e s t s u b j e c t , ’ FaceAlpha ’ , 0 . 6 , ’ E d g e C o l o r ’ , ’ b ’ )
96 axis equal
97 end
98 Ground=R a d i u s (N+1) −1;
99 f o r i =1:30
100 R=[ Ground+i Ground+i ] ;
101 [ X, Y, Z ] = c y l i n d e r (R,K) ;
102 Z ( 2 , : ) = L (N+1)∗ s c l ;
103 Z(1 , : ) = − Z(2 , : ) ;
104 X = X( : , 1 : p) ;
105 Y = Y( : , 1 : p) ;
106 Z = Z ( : , 1 : p)∗ s c l ;
108 end
109 camlight
110 l i g h t i n g gouraud
111 t i t l e ( ’ R a d i a l G r a i n g ’ , ’ FontName ’ , ’ Times New Roman ’ , ’ F o n t S i z e ’ , 2 4 , ’ f o n t W e i g h t ’ , ’ b o l d ’ ) ;
112 x l a b e l ( ’R(mm) ’ , ’ FontName ’ , ’ Times New Roman ’ , ’ F o n t S i z e ’ , 1 6 , ’ f o n t W e i g h t ’ , ’ b o l d ’ , ’ R o t a t i o n ’ , 9 0 , ’
HorizontalAlignment ’ , ’ r i g h t ’ )
113 z l a b e l ( ’ L (cm) ’ , ’ FontName ’ , ’ Times New Roman ’ , ’ F o n t S i z e ’ , 1 6 , ’ f o n t W e i g h t ’ , ’ b o l d ’ , ’ H o r i z o n t a l A l i g n m e n t ’ , ’ r i g h t ’ )
114 %% S a v i n g r e s u l t s t o f i l e
115 FID = f o p e n ( ’ R a d i a l V a l s 2 1 . t e x ’ , ’w ’ ) ;
116 f p r i n t f ( FID , ’ \\ b e g i n { t a b l e } [ ! htb ] \ n ’ ) ;
117 f p r i n t f ( FID , ’ \\ b e g i n { c e n t e r }\n ’ ) ;
118 f p r i n t f ( FID , ’ \\ b e g i n { t a b u l a r } { | c | | c c | c | | c c | } \ n ’ ) ;
119 f p r i n t f ( FID , ’ \\ t o p r u l e \n ’ ) ;
120 f p r i n t f ( FID , ’ \\ t e x t b f { F o i l N .O} & \\ t e x t b f { R a d i u s (mm) } & \\ t e x t b f { Length (mm) } & \\ t e x t b f { F o i l N .O} & \\
t e x t b f { R a d i u s (mm) } & \\ t e x t b f { Length (mm) }\\\\ \\ t o p r u l e \n ’ ) ;
121 Max = f l o o r ( (N+1) / 2 ) ; L (N+1)= 0 0 ;
122 f o r i =1:Max
123 f p r i n t f ( FID , ’ %. f & %4.2 f & %4.2 f & %. f & %4.2 f & %4.2 f \\\\\ n ’ , i , R a d i u s ( i ) , L ( i ) , ( i+Max) , R a d i u s ( i+Max) , L
( i+Max) ) ;
124 end
125 f p r i n t f ( FID , ’ \\ b o t t o m r u l e \n ’ ) ;
126 f p r i n t f ( FID , ’ \\ end { t a b u l a r }\n ’ ) ;
127 f p r i n t f ( FID , ’ \\ end { c e n t e r }\n ’ ) ;
128 f p r i n t f ( FID , ’ \\ c a p t i o n { R a d i a l Grading C a l c u l a t i o n s R e s u l t s }\n ’ ) ;
129 f p r i n t f ( FID , ’ \\ l a b e l { t a b l e : r a d i a l v a l s }\n ’ ) ;
130 f p r i n t f ( FID , ’ \\ end { t a b l e }\n ’ ) ;
131 f c l o s e ( FID ) ;
132
133 SID = f o p e n ( ’ R a d i a l −P o i n t s −A i r . t x t ’ , ’w ’ ) ;
134 f o r i =1:N
135 f p r i n t f ( FID , ’%f \ t%f \ t \ r \n ’ , R a d i u s ( i ) , 3 0 0 0 . 0 1 + ( L ( i ) / 2 ) ) ;
136 end
137 f c l o s e ( SID ) ;
138 SID = f o p e n ( ’ R a d i a l −P o i n t s −O i l . t x t ’ , ’w ’ ) ;
139 f o r i =1:N
140 f p r i n t f ( FID , ’%f \ t%f \ t \ r \n ’ , R a d i u s ( i ) , 2 9 9 9 . 9 9 − ( L ( i ) / 2 ) ) ;
141 end
142 f c l o s e ( SID ) ;
Axial Grading Matlab Code

1 %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
2 % ELEC6089 High V o l a t a g e I n s u l a t i o n B us hi ng D e s i g n
3 % Axial grading c a l c u l a t i o n
4 % Author − David Mahmoodi
5 % Date − 1 4 / 0 2 / 2 0 1 4
6 %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
7 clc ; close all ; clear all
8 %% D e c l a r e i n g Given V a r i a b l e s
9 Voltage = 275000; %A p p l i e d v o l t a g e ( v o l t )
10 Inner diameter = 100; %mm
11 Outer diameter = 300; %mm
12 F i r s t f o i l l e n g t h = 5 0 0 0 ; %mm
13 N = 21;
14 Foil Thickness = 0.1; %mm
15 First Gap = 2; %mm
16 Last Gap = 2 ; %mm
17 E boundary surface Air = 300; %v o l t /mm
18 E b o u n d a r y s u r f a c e O i l = 3 ∗ 3 0 0 ; %v o l t /mm
19 %% D e f i n i n g new v a r i a b l e s
20 D e l V o l t a g e = V o l t a g e /N−1; %V o l t a g e between e a c h f o i l
21 b Air = Del Voltage / E boundary surface Air ;
22 b Oil = Del Voltage / E boundary surface Oil ;
23 % I n i t i a l i s e v e c t o r s ( 2 2 f o r 3D p l o t t i n g )
24 L = z e r o s ( 1 ,N+1) ;
25 L A i r = z e r o s ( 1 ,N+1) ;
26 L O i l = z e r o s ( 1 ,N+1) ;
27 R a d i u s = z e r o s ( 1 ,N+1) ;
28 R parameter =1.007; % Parameter f o r a d j e s t i n g assumption v a l u e o f r0
29 %% C a l c u l a t i o n
30 L ( 1 )= F i r s t f o i l l e n g t h ; %The f i r s t f o i l i s 5000mm and c o n n e c t e d t o t h e c o n d u c t o r , no c a p a c i t a n c e between
c o n d u c t o r and t h i s f o i l
31 r 0 = I n n e r d i a m e t e r /2 ; %R a d i u s o f t h e c o n d u c t o r
32 R a d i u s ( 1 )= I n n e r d i a m e t e r /2 + F i r s t G a p ; % R a d i a l p o s i t i o n o f f i r s t f o i l = 52mm
33 L A i r ( 1 )= L ( 1 ) / 2 ;
34 L O i l ( 1 )= L ( 1 ) / 2 ;
REFERENCES 41
35 %C a l c u l a t e t h e r a d i a l p o s i t i o n s o f a l l f o i l s
36 f o r i =2:N
37 L A i r ( i )=L A i r ( i −1)− b A i r ;
38 L O i l ( i )=L O i l ( i −1)− b O i l ;
39 L ( i )= L A i r ( i )+L O i l ( i ) ;
40 end
41 % R e f e r t o S e c t i o n 2 . 2 f o r an e x p l a n a t i o n o f t h i s a s s u m p t i o n
42 R a d i u s ( 2 )= R a d i u s ( 1 ) ∗ exp ( ( L ( 2 ) /L ( 1 ) ) ∗ l o g ( R a d i u s ( 1 ) / ( r0−R p a r a m e t e r ) ) ) ;
43 %F o l l o w t h e i t e r a t i v e f o r m u l a
44 f o r i =3:N
45 R a d i u s ( i )= R a d i u s ( i −1)∗ exp ( ( L ( i ) /L ( i −1) ) ∗ l o g ( R a d i u s ( i −1)/ R a d i u s ( i −2) ) ) ;
46 end
47 %For p l o t t i n g − add t h e o u t e r s h e l l
48 L (N+1)=L (N)− . 5 ∗ L (N) ;
49 R a d i u s (N+1)=R a d i u s (N)+ Last Gap ;
50
51 %% P l o t i n g
52 figure
53 p l o t ( R a d i u s ( 1 : end −1) ,3000+ L A i r ( 1 : end −1) , ’ o ’ ) ; h o l d on%a x i s e q u a l
54 p l o t ( R a d i u s ( 1 : end −1) ,3000 − L O i l ( 1 : end −1) , ’ o ’ ) ;%a x i s e q u a l
55 x=z e r o s ( 1 , 2 ∗ (N+1) ) ;
56 y=x ; j =1;
57 f o r i = 1 : 2 : 2 ∗ (N+1)
58 x ( i )=L A i r ( j ) ;
59 x ( i +1)=−L O i l ( j ) ;
60 y ( i )=R a d i u s ( j ) ;
61 y ( i +1)=R a d i u s ( j ) ;
62 j=j +1;
63 end
64 y2=−y ;
65 % 2D P l o t
66 figure
68 r e c t H = r e c t a n g l e ( ’ P o s i t i o n ’ , [ − 1 . 1 . ∗ x ( 1 ) ,− r0 , 2 . 2 ∗ x ( 1 ) , 2∗ r 0 ] ) ;
69 s e t ( rect H , ’ FaceColor ’ , ’ r ’ )
70 f o r i = 1 : 2 : 2 ∗ (N)
71 h o l d on
72 l i n e (−x ( i : i +1) , −y ( i : i +1) , ’ LineWidth ’ , 2 )
73 l i n e (−x ( i : i +1) , −y2 ( i : i +1) , ’ LineWidth ’ , 2 )
74 %a x i s e q u a l
75 end
76 r e c t H = r e c t a n g l e ( ’ P o s i t i o n ’ , [ L (N+1) / 3 , O u t e r d i a m e t e r / 2 , L (N+1) / 2 , 1 0 0 0 ] ) ;
78 r e c t H = r e c t a n g l e ( ’ P o s i t i o n ’ , [ L (N+1)/3,−10− O u t e r d i a m e t e r / 2 , L (N+1) / 2 , 1 0 ] ) ;
80 t i t l e ( ’ A x i a l G r a i n g ’ , ’ FontName ’ , ’ Times New Roman ’ , ’ F o n t S i z e ’ , 3 4 , ’ f o n t W e i g h t ’ , ’ b o l d ’ ) ;
81 x l a b e l ( ’ Length (mm) ’ , ’ FontName ’ , ’ Times New Roman ’ , ’ F o n t S i z e ’ , 2 4 , ’ f o n t W e i g h t ’ , ’ b o l d ’ )
82 y l a b e l ( ’ R a d i u s (mm) ’ , ’ FontName ’ , ’ Times New Roman ’ , ’ F o n t S i z e ’ , 2 4 , ’ f o n t W e i g h t ’ , ’ b o l d ’ )
83 x l i m ( [ x ( 2 ) x ( 1 ) ] ) ; y l i m ( [ y2 ( end ) −10 y ( end ) + 1 0 ] )
84 % 3D P l o t
85 K=50; s c l = . 1 ; % Z d i r e c t i o n s c a l l i n g v a l u e f o r p l o t t i n g
86 p= 6∗N/3+1; %a d j u s t i n g t h e Cut i n t h e 3D shap
87 figure
89 R=[ r 0 r 0 ] ;
90 [ X, Y, Z ] = c y l i n d e r (R, 5 ∗K) ;
91 Z ( 2 , : ) = ( L A i r ( 1 )+ . 1 ∗ L A i r ( 1 ) ) ∗ s c l ;
92 Z(1 , : ) = − Z(2 , : ) ;
94 f o r i =1:N
95 h o l d on
96 R=[ R a d i u s ( i ) R a d i u s ( i ) ] ;
97 [ X, Y, Z ] = c y l i n d e r (R,K) ;
98 Z ( 2 , : )= L A i r ( i ) ;
99 Z ( 1 , : )= −L O i l ( i ) ;
100 X = X( : , 1 : p) ;
101 Y = Y( : , 1 : p) ;
102 Z = Z ( : , 1 : p)∗ s c l ;
103 t e s t s u b j e c t = s u r f (X, Y, Z ) ;
104 s e t ( t e s t s u b j e c t , ’ FaceAlpha ’ , 0 . 8 , ’ E d g e C o l o r ’ , ’ b ’ )
105 axis equal
106 end
107 Ground=R a d i u s (N+1) −1;
108 f o r i =1:30
109 R=[ Ground+i Ground+i ] ;
110 [ X, Y, Z ] = c y l i n d e r (R,K) ;
111 Z ( 2 , : ) = L (N+1) ∗ s c l ;
112 Z(1 , : ) = − Z(2 , : ) ;
113 X = X( : , 1 : p) ;
114 Y = Y( : , 1 : p) ;
115 Z = Z ( : , 1 : p ) ∗ s c l −100;
117 end
118 camlight
119 l i g h t i n g gouraud
120 t i t l e ( ’ A x i a l G r a i n g ’ , ’ FontName ’ , ’ Times New Roman ’ , ’ F o n t S i z e ’ , 2 4 , ’ f o n t W e i g h t ’ , ’ b o l d ’ ) ;
121 x l a b e l ( ’R(mm) ’ , ’ FontName ’ , ’ Times New Roman ’ , ’ F o n t S i z e ’ , 1 6 , ’ f o n t W e i g h t ’ , ’ b o l d ’ , ’ R o t a t i o n ’ , 9 0 , ’
HorizontalAlignment ’ , ’ r i g h t ’ )
122 z l a b e l ( ’ L (cm) ’ , ’ FontName ’ , ’ Times New Roman ’ , ’ F o n t S i z e ’ , 1 6 , ’ f o n t W e i g h t ’ , ’ b o l d ’ , ’ H o r i z o n t a l A l i g n m e n t ’ , ’ r i g h t ’ )
123 %% S a v i n g r e s u l t s t o f i l e
124 FID = f o p e n ( ’ A x i a l V a l s 2 1 . t e x ’ , ’w ’ ) ;
125 f p r i n t f ( FID , ’ \\ b e g i n { t a b l e } [ ! htb ] \ n ’ ) ;
126 f p r i n t f ( FID , ’ \\ b e g i n { c e n t e r }\n ’ ) ;
127 f p r i n t f ( FID , ’ \\ b e g i n { t a b u l a r } { | c | | c c | c | | c c | } \ n ’ ) ;
128 f p r i n t f ( FID , ’ \\ t o p r u l e \n ’ ) ;
129 f p r i n t f ( FID , ’ \\ t e x t b f { F o i l N .O} & \\ t e x t b f { R a d i u s (mm) } & \\ t e x t b f { Length (mm) } & \\ t e x t b f { F o i l N .O} & \\
t e x t b f { R a d i u s (mm) } & \\ t e x t b f { Length (mm) }\\\\ \\ t o p r u l e \n ’ ) ;
REFERENCES 42
130 Max = f l o o r ( (N+1) / 2 ) ; L (N+1)= 0 0 ;

131 f o r i =1:Max
132 f p r i n t f ( FID , ’ %. f & %4.2 f & %4.2 f & %. f & %4.2 f & %4.2 f \\\\\ n ’ , i , R a d i u s ( i ) , L ( i ) , ( i+Max) , R a d i u s ( i+Max) , L
( i+Max) ) ;
133 end
134 f p r i n t f ( FID , ’ \\ b o t t o m r u l e \n ’ ) ;
135 f p r i n t f ( FID , ’ \\ end { t a b u l a r }\n ’ ) ;
136 f p r i n t f ( FID , ’ \\ end { c e n t e r }\n ’ ) ;
137 f p r i n t f ( FID , ’ \\ c a p t i o n { A x i a l Grading C a l c u l a t i o n s R e s u l t s }\n ’ ) ;
138 f p r i n t f ( FID , ’ \\ l a b e l { t a b l e : a x i a l v a l s }\n ’ ) ;
139 f p r i n t f ( FID , ’ \\ end { t a b l e }\n ’ ) ;
140 f c l o s e ( FID ) ;
141
142 SID = f o p e n ( ’ A x i a l −P o i n t s −A i r . t x t ’ , ’w ’ ) ;
143 f o r i =1:N
144 f p r i n t f ( FID , ’%f \ t%f \ t \ r \n ’ , R a d i u s ( i ) ,3000.1+ L Air ( i ) ) ;
145 end
146 f c l o s e ( SID ) ;
147 SID = f o p e n ( ’ A x i a l −P o i n t s −O i l . t x t ’ , ’w ’ ) ;
148 f o r i =1:N
149 f p r i n t f ( FID , ’%f \ t%f \ t \ r \n ’ , R a d i u s ( i ) ,2999.9 − L O i l ( i ) ) ;
150 end
151 f c l o s e ( SID ) ;

Report Main

Uploaded by

Copyright:

Available Formats

Report Main

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Report Main

Uploaded by

Copyright:

Available Formats

ELEC6089 High Voltage Insulation Systems Assignment 1

HV AC 275kV Bushing Design

Thomas J. Smith, David Mahmoodi, Brendan Hickman, Patrick P. L. Fong

March 25, 2014

(a) Transformer wall connection (b) Wide view

2 Overview of Insulation Failure

2.1 General Breakdown

2.2 Surface Flashover

2.3 Partial Discharge

2.4 Corona Discharge

Figure 3.1: The Bushing Problem

3.1 Overview of Low Voltage and DC Solutions

3.1.1 External Screening Electrode

3.1.2 Internal Screening Electrode

3.2 Capacitive Grading

3.2.1 Radial Grading

Ln rn ≈ Ln−1 rn−1 ≈ · · · ≈ L1 r1 (8)

3.2.2 Axial Grading

Ln = Ln−1 − 2bn (10)

Figure 4.1: The reference problem taken from [16]

4.1 Design Issues

Table 4.1: Initial Information for Reference Design

Initial Information Value

Figure 4.2: Diagram to explain the assumptions required

1. Foil 0 is not connected to the HV conductor for both cases.

2. The conductor surface is spaced a distance of Sn from foil 0 in radial grading.

Flawed Radial Graing Design

Figure 4.3: Flawed Radial Grading Design

• Assumption 2 is required so that the radial spacing is kept constant.

5.1 Matlab Radial Grading

(a) 3D Representation of foil radial position and length

(b) 2D Representation of foil radial position and length

Figure 5.1: Matlab Generated Plots of Geometric Radial Design

Table 5.1 shows the values obtained for radial grading.

5.2 Matlab Axial Grading

Foil N.O Radius(mm) Length(mm) Foil N.O Radius(mm) Length(mm)

Table 5.1: Radial Grading Calculations Results

(a) 3D Representation of foil radial position and length

(b) 2D Representation of foil radial position and length

Figure 5.2: Matlab Generated Plots of Geometric Axial Design

6.1 Finite Element Methods (FEM)

Foil N.O Radius(mm) Length(mm) Foil N.O Radius(mm) Length(mm)

Table 5.2: Axial Grading Calculations Results

Material Relative Permittivity (r )

Table 5.3: Relative Permittivity of Materials

6.2 Equation Derivation and Boundary Conditions

By taking Gauss’s Law:

The boundary conditions are defined as the following:

1. Build a geometry representing the physical structure of the bushing.

6.4 Baseline Model

Figure 6.1: Baseline Model Geometry

Figure 6.2: Extra Fine Meshing for Baseline Model

Figure 6.3: Normal Electric Field (V/m) for Baseline Model

Figure 6.4: Geometry and Materials for No Foils Bushing

Figure 6.5: Extra Fine Mesh for No Foils Bushing

Figure 6.6: Normal Electric Field (V/m) for No Foils Bushing

Figure 6.7: Area of very high electric stress

Figure 6.8: Geometry of the No Grading Model

(a) Normal Electric Field (V/m) for the No-Grading Model

Material Relative Permittivity (r )