Haritha Thesis
Haritha Thesis
Haritha Thesis
Certificate
It is certified that the work contained in this thesis, titled Thermal Modeling of Electrical Utility
Transformers by Haritha V V S S has been carried out under our supervision and is not
submitted elsewhere for a degree.
Acknowledgements
I would like to thank my advisers Dr. Amit Jain and Dr. M. Ramamoorty for their guidance
and support during the entire course of my research. The regular discussions with them and
their constant feedback helped me immensely in completing this thesis work satisfactorily.
I would like to thank Vijai Electricals, Hyderabad, INDIA for their technical support at several
stages of the work.
I thank my friends in the Power Systems Research Center, for their constant encouragement
and joyous company. I would like to thank my family and friends back home. Without their
support and encouragement this thesis would not have seen the light of day.
My acknowledgment would not be complete without mentioning my friends from MS and M.
Tech (2007 batch). Their company throughout my stay in IIIT, sharing the joys during the
highs and providing comfort during the lows is unforgettable. I thank them for this humbling
experience.
Abstract
The importance of transformers, with their role in transmission and distribution of electrical
power and with the effect of their performance on the system, is an obvious axiom in the
modern days power systems. In addition to their momentous share in the capital investment
of a power system, transformer outages have a considerable economic impact on the operation
of the power systems.
In the course of continuous efforts to make the existing power network smarter and efficient,
thermal modeling and monitoring of transformers has become important in the field of
transformer engineering. With all the advances in the design techniques as well as material
engineering, it is the transformer thermal limitations that decide the loading and designing of
the transformer from the purview of user as well as manufacturer. With the research in power
systems on the whole progressing towards development of a smart grid, which infers that
each of the equipment should be smart, that includes that the monitoring of each individual
equipment should be intelligent, accurate as well as fast and economical, the problem of
thermal Modeling of transformers has been gaining momentum all the more.
The maximum temperature in the transformer interior is a significant parameter governing a
transformers performance and life expectancy. Though the temperature rise in the
transformer interior by itself may not have immediate effects, it does trigger other undesirable
consequences like excessive deterioration of insulation, which in the long run will reduce the
life of the transformer, thus affecting the economics of the power system. Thus the possible
maximum temperature rise in the transformer for certain kind of loading needs to be estimated
so as to be able to decide on the operational conditions as well as estimate the remaining life of
the transformer and plan accordingly. In the perspective of the user, temperatures in a
transformer are important to determine the amount and duration of over load it can sustain,
and to estimate the effects on the life of the transformer by operation at various temperatures.
Table of Contents
INTRODUCTION ........................................................................................................................................ 1
1.1
1.2
1.3
2.1.1
Heat Dissipation............................................................................................................................ 4
2.1.2
2.2
2.3
2.3.1
2.4
2.4.1
2.4.2
2.4.3
2.5
2.5.1
3
3.1
3.2
3.3
3.4
3.5
3.6
3.7
4.2
4.2.1
4.2.2
4.3
4.3.1
4.3.2
4.3.3
4.3.4
Modeling of Radiators................................................................................................................. 39
4.3.5
4.3.6
4.4
5
5.2
5.3
5.4
5.5
DISCUSSIONS .......................................................................................................................................... 83
CONCLUSIONS......................................................................................................................................... 84
6.1
CONCLUSIONS ........................................................................................................................................ 84
6.2
APPENDIX.................................................................................................................................................. 88
7.1
7.2
PUBLICATIONS ............................................................................................................................................... 90
REFERENCES................................................................................................................................................... 91
ii
LIST OF FIGURES
Fig. 2.1: Heating Curves 11
Fig. 2.2: Cooling Curves 12
Fig. 2.3: Arrhenius Law of Insulation Ageing 17
Fig. 4.1: Thermal Model of a single element.. 28
Fig. 4.2: Steady state thermal model showing interconnection of elements. 29
Fig. 4.3: Calculation of resistances Thermal Model for single element.38
Fig. 4.4: Modeling the convection Modified thermal model of oil element.. 42
Fig. 4.5: Modeling the convection Thermal model of oil element with diode..43
Fig. 5.1: Transformer 1: Geometry... 47
Fig. 5.2: Transformer 1: Elemental Division.. 48
Fig. 5.3: Transformer 1: FEA Implementation in NISA. 49
Fig. 5.4: Transformer 1: Flux Density distribution. 50
Fig. 5.5: Transformer 1: Loss Distribution.... 51
Fig. 5.6: Transformer 1: Numbering of Elements.. 52
Fig. 5.7: Transformer 1: Thermal Model... 57
Fig. 5.8: Transformer 1: Thermal Profile... 58
Fig. 5.9: Transformer 2: Geometry.... 60
Fig. 5.10: Transformer 2: Elemental Division.... 61
Fig. 5.11: Transformer 2: FEA Implementation in NISA.. 62
iii
iv
LIST OF TABLES
Table 1: Thermal Electrical Analogy. 35
Table 2: Dimensions of the considered Transformer 1.. 46
Table 3: Material Properties: FEA Implementation of Transformer 1.49
Table 4: Material Thermal Properties: Thermal Modeling of Transformer 1.. 51
Table 5: Calculation of Thermal Model Parameters for Transformer 1.. 52
Table 6: Calculation of Tank to Ambient Resistances for Transformer 155
Table 7: Thermal Model Implementation-Comparison with test values for Transformer 1. 59
Table 8: Dimensions of the considered Transformer 2.. 59
Table 9: Material Properties: FEA Implementation of Transformer 2.62
Table 10: Material Thermal Properties: Thermal Modeling of Transformer 2. 64
Table 11: Thermal Model Implementation-Comparison with test values for Transformer 2... 67
Table 12: Dimensions of the considered Transformer 3.67
Table 13: Material Properties: FEA Implementation of Transformer 3.70
Table 14: Material Thermal Properties: Thermal Modeling of Transformer 3.72
Table 15: Thermal Model Implementation-Comparison with test values for Transformer 3... 75
Table 16: Dimensions of the considered Transformer 4.... 75
Table 17: Material Properties: FEA Implementation of Transformer 4.78
Table 18: Material Thermal Properties: Thermal Modeling of Transformer 4. 80
Table 19: Thermal Model Implementation-Comparison with test values for Transformer 4... 83
Chapter 1
1
1.1
INTRODUCTION
Transformer An Overview
will not cause the transformer interior temperature exceed the thermal limits of insulation and
accordingly design the amount of insulation such that the worst case of operation also will not
cause the interior temperatures to exceed the thermal limits of materials. From the user side,
given a transformer, it has to be ensured that the loading on the transformer doesnt cause the
interior temperatures go beyond the threshold limits. In either case, calculating interior
temperatures of a transformer for a particular loading condition, which is termed as thermal
modeling is important with respect to transformer.
1.2
Thesis Contribution
With the significance of thermal modeling stated, it is important to have a good thermal model
that can calculate the temperatures at desired location inside the transformer. Earlier works in
this area concentrated on developing models that calculate the winding hotspot temperature as
well as top oil temperature using the design values of transformer. While the importance of
hotspot temperature and top oil temperature in indicating the transformer thermal status is
maintained, the current work primarily focuses on developing a thermal model that can
calculate the temperature profile across the transformer geometry, thus giving a better picture
of thermal status of transformer interior. With such a model, the temperature at desired
location in the transformer interior can be calculated, besides obtaining the general thermal
profile. In addition, besides obtaining the maximum temperatures, the model can provide the
location of the maximum temperature. This would certainly help in better understanding of
the thermal status of transformer and would also help in the optimization of insulation ratings
used inside the transformer. With further modifications and improvement this model can also
be used for thermal monitoring of transformers.
1.3
Organization of Thesis
This thesis presents the development of an electrical equivalent model simulating the thermal
behavior of the transformer based on the thermal-electrical analogy with the application of
2
Finite Element Analysis to the transformer geometry to derive the distribution of losses. The
thesis is organized in to six chapters. The first chapter gives an introduction to the problem
and explains the significance of the problem along with the contribution of the work to
research in the field of transformer thermal modeling. The second chapter provides the
thermal basics. It is very important to understand how heat is generated and dissipated and
also the different modes of heat dissipation in a solid body in general, so as to understand the
problem with respect to transformers. The axioms guiding the heating and cooling processes
in the transformer, that is, the Newtons laws of cooling are explained. The consequences of
heat buildup and deterioration of insulation are also discussed. The third chapter gives a wide
literature survey of the previous research attempts made in this area. Various techniques are
classified broadly and presented accordingly. The fourth chapter presents the proposed
method. Obtaining the loss distribution with the application of finite element analysis
technique to the transformer geometry is discussed first followed by the development of the
thermal model. The various assumptions and modeling issues are discussed. The fifth chapter
discusses the implementation of this method on four different standard transformer designs.
An implementation of this method to a three phase transformer design is also presented along
with obtained results. Observations and conclusions from previous chapters are given in
chapter six. This chapter provides a summary of the work done for the thesis and gives the
possible scope for future research in this area.
Chapter 2
2 THERMAL BASICS HEAT BUILDUP IN A
TRANSFORMER
2.1
This section presents a few thermal concepts related to generation and dissipation of heat. This
is very important to analyze the thermal behavior of transformers, or for that matter, any
electrical device. Heat dissipation is a pronounced problem in transformers because
transformers are enclosed devices and because transformers have no rotating parts which
provide inherent ventilation. Any solid body with losses occurring in it generates heat
inherently, because, any form of energy loss in a body is dissipated in the form of heat. The
loss can be either magnetic loss, due to interaction of fluxes or hysteresis loss, or ohmic loss,
due to the current passing through a conductor and the resistance being offered by it or
mechanical loss, due to moving parts and the friction between the surfaces in contact.
2.1.1
Heat Dissipation
The heat generated in any body is dissipated to the surrounding media and finally to the
ambient. The usage of the term ambient medium or ambient temperature is similar to
reference point in an electrical circuit. As far as radiation is concerned, the ambient
temperature is the temperature of sky and the ground to which the heat is radiated by the hot
bodies. For the other forms of heat dissipation like conduction, natural and artificial
convections, the ambient temperature is the temperature of the bulk of the air at a distance too
remote to be affected by the thermal field of the heated body. The use of air temperature as
ambient temperature is justified only where most of the heat dissipated is by convection, as in
the case of transformers. In any case, the heat generated in a body is dissipated to the
surrounding medium through one or more of the possible modes of heat dissipation, namely
conduction, convection and radiation.
Convection: Convection can be defined as the heat transfer in a gas or liquid by the
circulation of fluid currents from one region to another.
Radiation: Radiation of heat may be defined as the emission and propagation of energy in
the form of rays or waves.
2.1.2
Conduction: This mode of dissipation of heat is important in the case of solid parts of
machine like copper, iron and insulation. The equation of heat flow by conduction
between two surfaces separated by a heat conducting medium is given in equation 2.1.
Qcon =
(1 2 )
R
( 2.1)
be solved by methods of calculation similar to those used in electrical circuits. The thermal
resistance, like electrical resistance can be written as in equation 2.2.
R =
t
S
( 2.2 )
Convection: Convection can be either natural with the heat transfer taking place by means
of natural fluid currents or forced with the heat transfer taking place by means of induced
or forced fluid currents.
a) Natural Convection: Liquid and gas particles near the heater body become lighter and
rise, giving place to cooler particles, which in turn get heated and rise. This natural process,
due to changes in fluid density, is known as natural convection. The heat dissipated per unit
surface by natural convection is given by equation 2.3.
Qconv = K c (1 0 )
( 2.3 )
b) Forced Convection: In modern machines where there is too much of heat generation,
artificial circulation of the cooling medium is done to enhance the convection and hence the
increase in heat dissipation. For example, a transformer tank may be cooled by blasting air on
it or a turboalternator may be cooled by circulating hydrogen. This is known as cooling by
artificial convection. The problem of calculation of heat dissipation by artificial convection is
even more complex as it mainly depends upon the constructional features of the machine.
These constructional details are different for every machine and so no exact relationship can
be given for artificial convection.
Radiation: The heat dissipated by radiation from a surface depends upon its temperature
and its other characteristics like color, roughness etc. For the case of a very small spherical
radiating surface inside a large and or black spherical shell, the heat radiated per unit of
surface is given by Stefan Boltzmann law given in equation 2.4.
( 2.4 )
Where, Qrad = Heat dissipated by radiation per unit area, J per unit area.
T1, T0 = Absolute temperatures of the emitting surface and the ambient medium
respectively, K
e = Coefficient of emissivity; 1 for perfect black bodies, and is always less than unity
for others.
In general, the heat transfer in any body is by conduction and convection assisted by radiation.
The above equations help in quantizing the heat transferred by each of these modes of heat
transfer. [2]
2.2
Losses are produced in various parts of electrical machines due to which the machine
temperature rises. After sometime, the machine attains a steady temperature rise, at which, the
heat produced in the machine is equal to the heat leaving its surface by convection and
radiation. This is known as Newtons law of cooling. It is another form of energy conservation
principle.
The equation 2.4 for heat dissipated by radiation can be simplified as given in equation 2.5
with suitable assumptions.
Qrad = rad S
( 2.5 )
Qconv = conv S
( 2.6 )
( 2.7 )
2.3
The temperature of a machine rises when it is loaded, starting from cold condition. The
temperature, in the beginning, increases at a rate determined by the power loss. As the
temperature rises, the active parts of the machines dissipate heat partly by conduction, partly
by radiation and in most cases, largely by means of air cooling. The higher the temperature
rise, the greater would be the effect of these methods of cooling. Therefore, as the temperature
rises, the rate at which the temperature rises falls off owing to better heat dissipating
conditions. The temperature of any part of a machine, not only depends on the heat produced
in it, but also on the heat produced in other parts. This is because there is always a heat flow
from one part to another. For example, the heat produced in the part of the winding
embedded in the slot flows partially through the insulation to the laminations and partially to
the end windings. Thus the end windings have to transfer to the air, not only the heat
produced in them but also a part of the heat produced in the slot portion of the winding. That
is why the determination of temperatures is difficult, particularly in compactly enclosed devices
like transformers. Electrical machines are not homogenous bodies and their parts are made up
of different materials like copper, iron and insulation which have different thermal resistivities.
Due to this, it is rather difficult to calculate the temperature of a part of the machine.
However, it is worthwhile taking the theory of heating of homogeneous bodies as the basis for
analyzing the process of machine heating. The results obtained from such a theory are
applicable to a certain degree, to the different parts of the machine as a whole.
9
The equation of temperature rise with time of any body is given by equation 2.8.
t
t
= m 1 e T + i e T
h
( 2.8 )
Th = G h
...(2.9)
= m 1 e
Th
The cooling mechanism is also guided by a similar equation given as equation 2.11
10
...( 2 .1 0 )
= i e
Tc
...( 2 .1 1)
The heating and cooling curves are graphical representation of equations 2.10 and 2.11 and are
illustrated in figures 2.1 and 2.2 respectively.
11
2.4
Heat in a transformer
The process of energy transfer in the case of transformers involves currents in the conductors
and fluxes in the ferromagnetic parts. Thus, there are I2R losses in the windings and core losses
in the ferromagnetic cores. The core losses include the hysteresis losses which are due to the
magnetic inertia of the core material and eddy current losses, which are due to the circulating
currents developed in the flux carrying parts of the transformer, mainly the core. The losses
taking place in the transformer cores and windings, during conversion of energy from one
voltage to another voltage level, are converted into thermal energy and cause heating of the
corresponding transformer parts. In addition to this, losses occur in tank walls and end plates
on account of leakage flux. All these losses appear as heat and the temperature of every
12
affected part of the machine rises above the temperature of the ambient medium, which is
normally the surrounding air. The heat generated inside the transformer must be dissipated
without allowing the windings to reach a temperature, which will cause excessive deterioration
of the insulation. So, care should be taken that the heat generated in a machine is properly
dissipated without being accumulated. The heat generated in any equipment is dissipated into
its surroundings by convection and conduction assisted by radiation.
2.4.2
The heat generated in various parts of the transformer is dissipated to the ambient in various
stages. The heat generation, majorly occurs in the core and coils and it is this heat that has to
be transferred to the external ambient in the first place. The heat transfer within the core and
coils and from the core and coils to the oil is by conduction. The heat transfer within the oil is
by convection and the heat transfer between the hot oil and the heat exchanger is through
conduction. When the insulating oil is involved in the heat transfer, since perfect contact with
the heated surface is rare, the heat transfer is mainly dependent upon the fluid flow conditions,
i.e., whether the flow is stream line or turbulent and upon the condition of the surface. The
thermal conductivity of the coolant is much smaller than that of the metals. Apart from the
heat being dissipated by means of heat exchanger surfaces, the transformer tank also partly
dissipates the heat generated by radiation. Radiation in transformers, does not normally occur
by itself and in almost every case, it is accompanied by convection. Since dull metallic paints
cause more radiation, all the electrical machines are painted with dull metallic paints usually
grey in color in order to have large heat dissipation from radiation and thus reduce
temperature rise.
The transformer oil, in the process of heat transfer gets heated up and its temperature rises
which is detrimental to its operation. Through external means, this heat in the oil should be
dissipated and then cooled oil should be circulated back into the transformer. Here, no
moisture, or gases should enter the transformer. This makes the cooling in the transformer a
13
major issue. Different cooling methods, with suitable heat exchangers are used depending
upon the quantity of heat to be handled.
Oil Natural Air Natural (ONAN) Cooling: Heat transfer in oil occurs because of the
natural thermal head generated due to convection in the oil. The hot oil at the top of the
tank is circulated back to the bottom of the tank through tubes, passing through which the
oil gets cooled because of cooler ambient around the tubes. It can be enhanced by the
usage of radiators and fins.
Oil Natural Air Forced (ONAF) Cooling: For larger transformers, where the amount
of heat carried by the oil is huge, the heat exchange between the hot oil and cooler ambient
via the radiator tubes can be augmented by means of an air blast achieved through the
usage of fans. They blow air through the hollow spaces, drive the hot air out and suck
cooler oil in and thus cause better heat exchange between oil and external air.
Oil Natural Water Forced (ONWF) Cooling: The radiator tubes can be cooled by
water instead of using an air blast as water is more effective in transferring heat, but only
when a natural water head is already available. This method proves cheaper and efficient
for transformers at hydro power stations
Forced and Directed Oil Cooling: In large transformers, the natural circulation of oil is
insufficient for cooling the transformer and forced circulation is employed. Also, guiding
vanes are used to direct the oil flow in the cooling ducts in paths that ensure quicker and
efficient heat transfer from the coils to the oil.
The choice of cooling system is made depending on the loading of the transformer.
Accordingly, a transformer is given different ratings with different cooling methods.
2.4.3
With all the losses being generated in the transformer and the cooling system efficiently
transferring this heat to external ambient, the problem of heat buildup in the transformer still
14
remains. A major reason for this is that with all the guiding vanes and cooling ducts, there
might be still some areas in the transformer where the coolant might not reach and the heat
might not be dissipated and hence gets accumulated. Apart from this, the transformer may be
operating at worse conditions of operations with load and frequency fluctuations, which might
cause increased losses that the cooling system may not dissipate effectively. It is a very usual
condition that the transformer is over loaded frequently for shorter durations or continuously
for longer durations. The losses, based on which the cooling system performance is
anticipated, may be practically different from the calculated values. Iron loss may change
because of change in grain orientation due to punching, clamping of laminations and the
pressures during these actions. Due to all or any of the above reasons, heat does get built up in
the transformer despite the existence of the cooling system.
On the top of all these, ambient temperatures do decide the rate of heat dissipation, since heat
dissipation is linearly proportional to the temperature difference between the transformer and
external ambient. Hence, it is obvious that the transformer cannot be loaded as much on
hotter days as on cooler days and on hotter days even lesser temperature rise of the
transformer oil may not be properly dissipated and might result in heat accumulation.
2.5
It may affect the chemical properties of oil causing its dissociation and generation of
moisture and gases
This may increase the pressure in the tank, and in the worst case, this might cause
the explosion of the tank.
This may change the thermal and electrical properties of the windings and core
15
Though most of the consequences seem to have very little possibility of occurrence, they
cannot be neglected because, transformers are usually installed in remote locations unlike
other machines, where frequent maintenance is not always possible. Also, regular monitoring
of the device is not possible. The above consequences are not sudden in nature; they are
gradual, cumulative and related to each other. Also, it is difficult to identify any of these
happening unless an apparent damage takes place. So, if one of the above consequences
happens, it eventually causes the others to happen, and result in a permanent and irreparable
damage. [3]
2.5.1
In all the consequences that follow the heat buildup in the transformer, the major one is the
insulation deterioration. It is the first thing that happens with the excessive heat buildup and it
is the one that triggers other undesirable consequences/damages. Also, insulation is very
costly and is a major contributor in the cost of the transformer. The insulation of the
transformer tends to age and deteriorate when heated. The higher is the temperature, the
faster is the insulation deterioration.
During periods of subnormal operating temperature, the loss of life of the insulation will be
less than normal. But when, the operating temperatures are greater than normal, the loss of
life will be higher than normal. Consequently, the transformer may be safely operated for a
time at above normal temperatures provided the loss of insulation life during this period is
adequately compensated for, by operation for a sufficiently long time at temperatures lower
than normal. This deterioration of life of insulation with temperature is mathematically given
by Arrhenius law of Insulation Ageing, which is a non linear relation, given in equation 2.12.
B
A+
T
L ife = e
...( 2 .1 2 )
Where A and B are constants, derived by experiment and T is the absolute temperature.
16
In the range of 80oC to 120oC, which is the usual winding hot spot temperatures, this law
can be expressed in a more convenient form called Montsinger relation as given in equation
2.13.
Life = e p
...(2.13)
17
Chapter 3
3 TRANSFORMER THERMAL MODELING
LITERATURE SURVEY
3.1
There are broadly three ways to measure transformer interior temperatures, namely, the
usage of empirical formulae (IEEE and IEC standard formulations), direct measurement
using fiber optic sensors and usage of suitable computer based mathematical thermal
models. Empirical formulae to calculate maximum interior temperatures assume the heating
process of oil and winding to be similar to the charging and discharging process of a
capacitor. Using the measurements obtained from the transformers heat run test, the
formulae calculate Hot Spot Temperature (HST) and Top Oil Temperature (TOT) at rated
load and predict the value of HST and TOT for any loading condition and see if the
transformer thermal limits are observed, thus ensuring the safe feasibility of such a loading
condition. While IEEE formulae give precise formula to calculate the value of HST, IEC
standards stipulate a factor to be multiplied with the measured average temperature rise over
TOT to calculate the winding hottest spot rise. On the other hand, direct measurement
involves embedding the fiber optic sensors at various locations inside the transformer and
then by running the transformer at desired loading conditions, the value of HST is directly
recorded from the sensor measurements. The HST is checked with the thermal limits of the
transformer to see if the particular loading condition is safe for the transformer.
Both of these methods have their own limitations. Empirical formulae are derived based on
certain assumptions which are made to generalize the formulae to suit to any transformer.
The other side of the coin is that, since the transformer design, construction and loading
18
conditions change the transformer thermal response in a non linear way, generalizing the
formulae makes the formulae unspecific for a transformer and hence the transformer
thermal response is poorly tracked. Moreover, including too many constraints and
coefficients make the formulae too complex to be solved. However, they are cost efficient
techniques. In contrast, direct measurement techniques using sensors are quite accurate but
costly. The sensors themselves are costly and embedding them inside the transformer for
testing and then removing them after the testing is done are cost involving processes.
Moreover, placement locations for sensors also play their part as even after placing many
sensors, one may miss the hottest location. So, it can be understood that both the
conventional methods are not the optimal answers to the problem of transformer thermal
monitoring. Constant improvement of technology in the field of computer science and its
applications in wide range of research areas enables the problem to be solved with the
application of suitable software techniques. So, software based simulations and modeling
prove to be cost effective as well as efficient techniques to answer this problem.
3.2
The IEEE formulae give the possible value of hot spot temperature inside the transformer
using the value of top oil temperature measured from the heat run test so as to guide the
user (or the designer) to decide the safe loading conditions of the transformer. Applications
of loads in excess of nameplate rating involve some degree of risk. IEEE formulae are
designed to identify these risks and to establish limitations and guidelines, the application of
which will minimize the risks to an acceptable level [4]. The basic equation for the
calculation of the hottest-spot temperature is as given in equation 3.1.
H = A + TO + H
Where, H is the winding hottest spot temperature, C
A is average ambient temperature during the load cycle to be studied, C
19
...(3.1)
TO = A + TO
...(3.2)
The top-oil temperature rise at a time after a step load change is given by the following
exponential expression containing an oil time constant.
TO = ( TO ,U
T 0
TO , I ) 1 e
+ TO , I
...(3.3)
Where, TO,U is the ultimate top oil temperature rise over the ambient for load L, C
TO,I is the initial top oil temperature rise over the ambient, C
T0 is the oil thermal time constant, sec.
The winding hot spot rise over the top oil temperature H is calculated using similar
exponential expression as in equation 3.3 while the value of rated hotspot temperature rise
above top oil temperature is given by equation 3.4.
H , R = H / A, R TO , R
...(3.4)
Where, H/A,R is the rated hot spot temperature rise above ambient, oC
TO,R is the rated top oil temperature rise above ambient, oC
H/A,R can be measured by embedded detectors or is usually supplied by the manufacturer
on test report and TO,R is usually supplied by the manufacturer on test report [4].
While the IEEE formulae seem to give a general and simple formula that can be applied to
any transformer, on the other hand, it can be seen that the constants used in the formulae
are not that easily available. The accuracy of the testing procedure to determine the various
constants used in the formulae also affect the accuracy of the final estimated hot spot
20
temperature value. A more accurate but iterative method is also given in the IEEE Loading
guide, but that is too elaborative and time consuming and requires lot number of
measurement data [5]. Also the IEEE model, while it accounts for the thermodynamic effect
of load on Hot Spot Temperature (HST), it does not accurately account for the effects of
Top Oil Temperature (TOT) variations on HST. From the formulae, it can be seen that if
TOT
changes
instantaneously,
then
HST
will
also
change
instantaneously.
3.3
Fiber optic sensors use optical fiber as the sensing element. Optical fibers can be used as
sensors to measure strain, temperature, pressure and other quantities by modifying a fiber so
that the quantity to be measured modulates the intensity, phase, polarization, wave length or
transit time of light in the fiber. Temperature can be measured by an optical fiber using
its evanescent loss that varies with temperature. Thus they were an answer to the increased
need of accuracy in temperature measurements and fiber optic sensors for effective
measurement of high temperatures were designed. However, in the case of transformers,
survival of these sensors from the interior stresses was a problem due to their fragility in the
initial days. In response to this important need, fiber optic sensors have significantly
improved to the point that direct measurement of winding temperature is now becoming the
preferred method to measure the Hot Spot Temperature (HST) than using standard
empirical formulae. Compatibility of fragile fiber optic sensor with transformer factory
environment which had been a problem in the past, is now resolved with sturdy fiber jackets,
proper spooling of sensor during factory work and simplified through-wall connections.
21
But the fiber optic sensors are costly and so this has effect on the final cost of transformer.
However, the demand for quality and accuracy of measurements is more prominent in the
market, which indicates that the measurement of temperatures through methodologies that
have possibilities of giving better quality results are being looked forward than using only the
present IEEE and IEC standards in the industry.
3.4
References [5] [24] present the various attempts made to develop the thermal model for a
transformer based on thermal electrical analogy that can calculate the hot spot temperatures
better than what the IEEE thermal model does. A basic thermal model that uses thermal
electrical analogy, which states that the thermal parameters of heat flow, temperature, heat
storage and heat dissipation are analogous respectively to the electrical parameters of current,
potential difference (voltage), capacitance and resistance, is used as a basis to build lumped
parameter models that simulate the thermal behavior of the transformer that is exponential
[9], [10]. One model represents the winding to oil heat transfer and the other represents oil
to external air heat transfer. All the heat generated inside the transformer due to the losses is
represented by current sources and all the heat storage inside the transformer is represented
by a single capacitance, claiming that the single capacitance as a lump represents the total
heat storage inside the transformer. The resistance connected in each of the models
represents the heat dissipation from winding to oil or oil to external ambient as the case may
be. This resistance is modeled to be non linear because of the fact that heat transfer from the
winding to transformer oil as well as from the transformer oil to external ambient depends
upon whether the transformer oil is directed (or forced) or natural and whether the external
air is natural or forced respectively.
The required parameters like rated load top oil temperature rise over the ambient, the top oil
time constant and the value of exponent that determines the nonlinearity of the resistance
were found from the technique of minimization of the integral-squared-error over one day
22
of test data. Such a method of determination of thermal model parameters would be quite
practical for an on-line parameter determination calculation that automatically and
continually estimates the key thermal model parameters. A sudden change in the value of
one of the parameters such as the rated top oil rise could indicate that the fans or pumps
have failed. A gradual increase in the calculated rated top oil rise might indicate that the
radiators are becoming increasingly fouled in some way. Therefore this method helps to
make the transformer monitoring online [10].
As this model doesnt consider the oil viscosity as well as temperature dependence on the
transformer oil thermal parameters, ref. [6] presents an improvisation with the inclusion of a
nonlinear thermal resistance in the model, which takes into account changes in the
transformer oil thermal characteristics and viscosity with temperature using an
experimentally determined constant. Ref. [7] presents a further improvisation of this model
which includes the effect of temperature dependence of the losses generated inside the
transformer, using a loss correction factor as well as the specific design of the transformer
windings and their influence on the oil circulation and the temperature gradient at the top of
the winding stack. Also, the oil viscosity changes with temperature are also modeled. The
method of calculating the transformer time constants and the changes in their values is
refined. Ref. [11] suggests the simultaneous solving of the two exponential equations derived
from the winding to oil and oil to ambient models. An online transformer monitoring
system based on this model is presented in ref. [12]. Ref [13] gives similar online transformer
monitoring model, but, instead of including the effect of external cooling on the transformer
by means of non linearity in the resistor, it has been separately dealt as an equation.
All the models discussed hitherto, use top oil temperature as reference to calculate the
hotspot temperature. Instead, since bottom oil temperature which is the temperature at the
winding oil ducts just below the winding, might properly track the temperature of oil
adjacent to the winding, especially in the case of transient over loads, usage of bottom oil
temperature to calculate the hotspot temperatures might improve the accuracy of
23
3.5
With the advent of computers in every domain of engineering sciences and with the
development of computer simulations, solving of complex problems has become easy along
with increased accuracy and controllability over the solution process. In the field of power
system engineering, complex multi variable problems which demanded either a compromise
in the solution accuracy or solution speed or the cost of solution have been properly catered
by these computer simulations. In the field of thermal modeling of transformers to calculate
the interior temperatures of the transformers, some work has been done to simulate the
transformer interiors using standard software packages. A technique used in these software
24
packages for such problems is Finite Element Analysis (FEA). In the recent past, some work
has been done in using the computer simulations and FEM packages in the transformer
analysis and are presented in [25] [28].
Ref. [25] describes a computer model, which can predict hot-spot temperatures for different
types of cooling regimes and transformer winding geometries coded using FORTRAN. Ref.
[19] presents a model that uses Finite Element Analysis for the accurate determination of
stray losses stating the fact that underestimation of losses inside the transformer particularly
the stray loss is one of the possible reasons for the hotspot inside the transformer. Ref [26]
presents a 3 dimensional model that aims at accurate estimation of the stray losses in a
power transformer using an Integral Equation Method (IEM) and Finite Element Method
(FEM) owing to the fact that stray losses in a large rated transformer can be around 20% to
25% of the total losses.
3.6
The rationalism and reasoning which have been human assets are now being attempted to be
imbibed to the machines with the progressing development in the field of computation. A
few recent research attempts to answer the problem of transformer thermal modeling
through the use of artificial intelligence methods of fuzzy logic, genetic algorithms and
neural networks, are discussed in [29] [32]. Ref [29] presents an equivalent heat circuit
based thermal model of an oil-immersed power transformer and a methodology for model
construction using intelligent learning, based on real-world data. A genetic algorithm is used
as a search method, based on a few on-site measurements, to determine the thermal model
parameters. The proposed thermal model can continuously calculate temperatures of the
main parts of an ONAN or OFAF cooled power transformer under various ambient and
load conditions. Ref [30] presents a further simplified thermal-electric analogous thermal
model of an oil-immersed power transformer rooted on the principles of heat exchange and
electric circuit laws. Ref. [31] gives the application of neural networks to the transformer
25
analysis as well as fault diagnosis and ref. [32] gives the application of fuzzy logic algorithms
as well as expert systems.
3.7
Economic and operation motivations have been the reason for the research for accurate
thermal modeling of transformers and the possible approaches are to measure the hot-spot
temperature using the sensors or to calculate it using standard empirical formulae or a thermal
model. In order to ensure that the conventional IEEE and IEC standards cope up with the
improvements in the design technology as well as to improve their accuracy in the modeling of
overloads etc. they are constantly being revised with more accurate and advanced models
aiming at a better representation of oil temperature inside the winding, considering variations
in the winding resistance, oil viscosity and oil inertia. Direct measurement of winding
temperature with fiber optic sensor provides a good alternative, but they are costly.
This makes evident the fact that neither the IEEE standards nor the fiber optic sensors could
be an optimal solution to address the problem of transformer thermal modeling and
transformer interior temperature calculations, to be specific, the hot spot temperature
calculations. With the improved capabilities of computer automations, constant efforts are
being made to derive software based thermal models for accurate measurement of hot spot
temperatures. Improvisation is needed not only in the model for measurement of temperatures
but also in the methods of accurate determination of eddy current losses, stray losses, cooling
mechanism etc.
A thermal model for a transformer can be created to deliver either the temperature distribution
across the transformer geometry, or the characteristic temperatures (HST, TOT) that indicate
the thermal status of the transformer interior. The current market competency demands more
research insight into the problem of transformer thermal modeling and calls for even more
precise, intelligent, accurate and above all economic solutions. The past research focused on
26
the development of lumped parameter thermal models that calculated only the HST and TOT
and there is no evident attempt on obtaining a temperature distribution profile across the
geometry of transformer. An attempt has been made in this work to calculate the temperature
distribution across the transformer geometry which can give a better understanding of the
thermal conditions inside the transformer.
27
Chapter 4
4 TRANSFORMER THERMAL MODELING USING LOSS
DISTRIBUTION
4.1
The proposed method is based on thermal electrical analogy and the basic guideline is the fact
that the losses in the transformer are distributed across the geometry and not concentrated at a
single point and hence the model uses distributed values of losses instead of lumping them.
The principle is that each point-element in the transformer generates heat because of the loss
in it. The heat transferred to that element or from that element depends upon the temperature
of neighboring elements. The element stores a little amount of heat, which is the cause of
temperature rise of that element and dissipates the rest into surrounding medium as long as its
temperature is greater than that of the surrounding medium. The temperature of each pointelement, therefore, depends on heat generation in that element and also the temperature of the
surrounding elements. Thus, each point element is considered to have a loss causing heat
generation in it, a heat conductor dissipating heat to neighboring elements and a heat retaining
element that causes heat storage inside it.
28
As stated, there exists an analogy between the thermal and electrical parameters and the
thermal parameters of heat flow, temperature, heat storage and heat dissipation are
analogous respectively to the electrical parameters of current, potential difference (voltage),
capacitance and resistance. Based on this analogy, the behavior of every point element in the
transformer can be represented in the form of an electrical circuit as shown in fig. 4.1. The
dissipation of heat to neighboring elements in all the directions is modeled making use of two
resistors representing the heat flow in either direction. If the loading on the transformer is
steady with no major variations, the temperature rise would follow an exponential pattern and
gets constant in the steady state. In the case of such a condition, the capacitance can be
ignored and the model can be simulated. In this way, every point element can be modeled and
can be connected to neighboring elements modeled similarly as shown in fig. 4.2.
Implementation wise, since modeling of each point element is difficult practically, the
transformer geometry is divided into finite number of sections or elements (as referred to
hereafter) and each element is modeled accordingly as given in fig. 4.1. Each element can be a
part of core, winding, tank or oil, geometry wise with dimensions different from the
surrounding elements and hence the value of parameters of the model changes for each
element. The division of the geometry into elements can be based on the change in the loss
29
concentration in the part of the geometry and smaller the size of the each element, higher is
the complexity in the model and finer is the thermal profile obtained. The current source
represents the heat generated in the element because of electrical loss occurring in it. If the
element is a part of core, the hysteresis loss and eddy current loss summed up forms the value
of current source in the model of that element and if the element is a part of winding, the
ohmic loss occurring in that part of winding is used as the value of current source in the model
of that element. Oil elements will not have current sources as there are no losses occurring in
the oil. The resistors used in the model to represent the heat dissipation of an element to
neighboring elements are calculated using the values of thermal conductance of the material,
which is the property of the material. The capacitor value, which represents the heat storage
inside the transformer, if used, is calculated using the value of specific heat capacity of the
material of the element.
As straightforward as the model might look, the complexity lies in determining the loss in each
element. Though it is an obvious fact that the loss is not uniformly distributed across the
geometry of the heat generating elements (core and windings), determination of the loss
distribution pattern and the gross loss in each of element division is not easy. Copper loss is
more or less uniformly distributed and hence the loss in each element division of the winding
geometry can be proportionally a fraction of the total ohmic loss in the winding. The problem
of obtaining the loss distribution across the core is a more complex issue and has to be
addressed using high end computational techniques. In the current work, the technique of
Finite Element Analysis (FEA) is applied to obtain the flux distribution, which is non uniform
across the core and calculate the core losses from the values of flux density grossed across
each element division.
Each element is thus modeled accordingly and the individually modeled elements are
interconnected to form the total geometry of the transformer. The electrical mesh circuit, thus
30
formed is the thermal model of transformer and solving the model gives the temperature at
each node, thus giving the thermal profile for the transformer geometry.
4.2
The ohmic loss in the winding can be grossed as the product of the square of the current
passing through the winding with the resistance offered by the winding (Copper) and the loss
in each element division of the winding can be proportionally a fraction of the ohmic loss in
the entire winding. The core loss in the transformer is the sum of hysteresis loss as well as
eddy current loss. The hysteresis loss is because of the magnetic inertia of the magnetic
material, or in technical terms, due to the remnant magnetization left over in the magnetic
dipoles during alternate cycles of magnetization and demagnetization caused by sinusoidal
alternating current. Similarly, as the magnetic material also would have got certain inherent
electrical conductivity, circulating alternating currents are generated in the magnetic material
due to alternating magnetic flux in it. As less as the electrical conductivity of the magnetic
material may be, still these circulating currents termed as eddy currents would exist and the
ohmic losses due to these currents are termed as eddy current losses.
For a given transformer core material, given frequency of excitation, the core loss is dependent
on the flux density in the core cross section defined mathematically. So, if the flux density is
known, the core loss can be determined. Hence, if the flux density distribution across the core
as well as in each element division of its geometry is known, the core loss can be determined
by means of mathematical substitutions and calculations. It is here that the Finite Element
Method (FEM) is made use of. The Finite Element Analysis (FEA), sometimes referred to as
finite element method (FEM), is a computational technique used to obtain approximate
solutions of boundary value problems in engineering.
31
4.2.1
The concept of FEM deals with applying the differential equations over smaller sub domains
of the entire large domain area, and then building up the solution of next layer of sub domains
using the already calculated values of field variables of the neighboring domains as boundary
conditions and this is extended to the current problem. Flux density in the core is associated
with the development of magnetic field in and around the core due to the alternating current
generating an alternating electric field. The values of flux densities at various points in the
transformer are governed by Maxwells differential equations given in equations 4.1 to 4.3. In
addition, equations 4.4 to 4.6 give the necessary constitutive relations. These equations give the
value of flux density in terms of current densities. Depending on the value of current densities
in the element and the value of boundary conditions on the element, the FEA tool calculates
the flux densities in each element. The boundary conditions for the elements on the outer end
of the geometry would be the user defined limiting value of the field variable on the boundary
of the geometry and here assumed to be the value of field variable (magnetic field) being zero.
(Dirichlet boundary condition)
B
E =
t
D
H = J +
t
D =
...(4.2)
D =E
...(4.4)
B = H
...(4.5)
J =E
...(4.6)
...(4.1)
...(4.3)
The loss in the transformer is majorly in core and windings. The losses in the other parts are
negligible and can be ignored with no loss of accuracy of the results.
Loss distribution in the windings: The losses in the windings are the ohmic losses due
to current flowing in the conductors. As superior as the electrical conductivity of the copper
may be, still, there would be good amount of ohmic loss in the windings owing to the high
values of currents flowing through the windings. The general expression for the calculation of
ohmic loss is given in equation 4.7.
W
Cu
2 l
= ( J A)
...(4.7)
Loss distribution in the core: The finite element analysis of the transformer geometry is
performed and the flux density distribution across the transformer geometry as well as the
average flux density across each element is obtained. Now, the core losses are to be calculated
using the flux density values. The mathematical expressions for the core losses, which
comprise of hysteresis and eddy current loss are given in the equations 4.8 and 4.9.
Wh = K h Bm1.6 f
Watts / kg
We = K e K 2f Bm2 f 2t 2 n
Watts / kg
Where, Wh and We are hysteresis and eddy current loss respectively, Watts
Ke and Kh are material constants that are found out experimentally
f is the frequency of the alternating flux, Hz
Bm is the maximum value of operating flux density, Wb/m2
Kf is the form factor of the ac wave form
34
...(4.8)
...(4.9)
The principle behind the thermal model is explained in section 4.1 and the development of
thermal model for the entire transformer is on the same lines. The section elaborates the
thermal modeling process.
4.3.1
The analogy between the thermal and electrical parameters is given in table 1.
Sl. No
Element
Thermal parameter
Electrical Parameter
Through Variable
Current, i ampere
Across Variable
Temperature, oC
Voltage, v volts
Dissipation element
Storage element
35
The development of thermal model and solving of it to obtain the thermal profile is based on
the analogy between the thermal and electrical parameters of a material and the similarity
between the development of electrical potential in the conductor with the flow of charge and
the temperature rise in the body because of the flow of heat. The two important thermal
parameters of a substance or a material are its specific heat capacity and its thermal
conductivity. These are general properties which can be made specific to a specific volume of
the material in which case they become thermal capacitance and thermal resistance of that
particular volume of the material. This holds good for metals, but for fluids (transformer oil),
additional modeling considerations and coefficients are needed to model the heat transfer. This
is because, besides being guided by the thermal conductance and the temperature difference,
the heat transfer in liquids is affected majorly by heat convection and fluid viscosity.
4.3.2
The basic model for each element, whatever the division of element might be, is explained in
the section 4.1. All the individually modeled elements are to be connected together to form an
electrical mesh circuit which is the thermal model for the entire transformer geometry. This
connection is needed because, the temperatures and the dissipation of heat in each element are
not independent and these depend on the temperature and heat dissipation of all the
neighboring elements. But while integrating individual elements geometrically to form the
thermal model, few modeling constraints would have to be observed.
The loss generated in each of these elements is independent of the neighboring blocks. So
the current source of every element, which is the heat generated in that element must be
separately connected to the ambient.
The oil sections will not have source since there are no losses in the oil sections. In the
transformer, oil functions as insulating as well as cooling liquid and is used as a medium of
heat transfer. It has no effect on heat generation.
36
The outer tank section also dissipates the heat to the ambient, which is modeled by
thermal resistances connected between each element on the tank boundary to the ambient. In
this way, ambient is modeled separately.
Ambient serves the purpose of ground in conventional electrical circuits. But the
difference here is that in conventional electrical circuits, ground is at zero potential, whereas
ambient in this electrical equivalent for simulating the thermal model is the surrounding
atmosphere which is not at zero temperature. The ambient temperature is in the range of 20oC
or 30 oC depending on the surrounding atmosphere temperature. It can be as high as 45 oC - 50
o
C and as low as 5 oC - 10 oC. Hence the temperature values obtained from the thermal model
is the temperature rise of the corresponding locations above the ambient and not the
temperatures themselves.
Symmetry is observed in the circuit for heat dissipation. The heat generating elements in
the middle of geometry dissipate in both the directions, while the elements on either side
dissipate in the respective direction only. This does not restrict in any way the heat flow in any
direction; it is just that the circuit is made symmetrical.
Since the model is two dimensional, in a way, we are studying the heat dissipation in two
directions, while the losses that are generated in the transformer dissipate in all the three
directions equally. So, since we limit the model to two-directional (X-Y) heat dissipation, the
heat to be dissipated must also be proportionally reduced, i.e., made two thirds of the total
heat generation.
In the figure 4.1, there are two resistors. The resistor in the horizontal direction represents
the heat flow along X-direction and the resistor in the vertical direction represents the heat
flow in the Y-direction.
37
4.3.3
Calculation of Sources: For the core elements, the value of core loss in each element will
form the current source and for the winding elements, the value of ohmic loss will be the value
of current source. For the other elements, there will be no current sources as there is no heat
generation in the other parts of the transformer. The leakage fluxes and the loss due to them
may help improve the accuracy and help in transformer design analysis, but the contribution to
heat generation by them is not so significant.
opposition offered by the element to the flow of heat through it. This resistance is offered in
both X and Y directions for either directions of heat flow and hence the thermal resistance of
each element is represented by two resistances corresponding to the opposition offered in each
direction of heat flow. The resistance along the X direction represents the opposition offered
by the element to the heat transfer in horizontal direction to neighboring elements. Similarly
the resistance placed in the Y-direction is the opposition offered to the heat flow in vertical
direction. The thermal resistance of each element depends on the material property of thermal
conductivity (Watts/mK), specific to the material of the element. Consider a single element as
shown in the figure 4.3.
38
If x and y are the dimensions of the element in X and Y directions respectively, z is the Zdirection depth and is the thermal conductivity of the material of the element, the thermal
resistance offered by the element to the heat flow in either direction is given by the equations
4.10 and 4.11.
Rhorizontal =
Rvertical =
x
yz
y
xz
...(4.10)
...(4.11)
circuit, if the transient thermal performance is of interest or if the load is so non uniform that
the thermal transients might affect the transformer. In any case, the method to calculate the
thermal capacitance values is presented. The thermal capacitance of an element represents its
capacity to store the heat, which causes a rise in its temperature. This is determined by the
material property of specific heat capacity. The thermal capacitance of each element is
calculated by multiplying the value of specific heat capacity of the material of the element with
the mass of the block. So, if an element of dimensions x, y and z in X, Y and Z directions
respectively as in figure 4.3 is considered, whose specific heat capacity is s J/kg0C and density is
d kg/m3, then the value of thermal capacitance of the material is given by equation 4.12.
4.3.4
...(4.12)
Modeling of Radiators
Radiators are heat exchangers used to transfer thermal energy from one medium to another
for the purpose of cooling and heating. In the transformer, radiators help draw out heat from
the hot oil in the top and re-circulate the cooled oil back into the tank from the bottom. They
are commonly used for ONAN, ONAF/ONAN and OFAF/ONAF/ONAN types of
cooling for slightly larger transformers, while for smaller transformers, corrugations on the
39
tank surface itself serve as radiators. Radiators are tubes or fins through which the oil from the
top of the tank is circulated through, to the bottom of tank and while the oil is circulated the
heat from the oil is drawn out. They have heat exchanging elements joined to top and bottom
headers which are connected to the transformer tank by welding and consist of previously
rolled and pressed thin steel sheets to form a number of channels or flutes through which the
oil flows. The surface area available for heat dissipation is increased by the use of radiators. As
the oil passes downwards, due to natural circulation heat is carried away by the surrounding
atmospheric air. This cooling of oil is augmented by blowing air onto the radiator tubes and
this is what happens in forced air cooling.
While modeling, the effect of radiator is modeled by means of thermal conductance path
provided from the top and bottom oil to the ambient, which is represented by means of
resistors connected to the top and bottom oil on either side of the tank. The dissipation occurs
through the radiator surface and hence the thermal conductivity of the radiator metal (same as
the tank metal) is to be used in the calculation of radiator resistance. The diameter of the
radiator tube would be the length of the heat transfer path and the surface area of the radiators
would be the surface area of dissipation
4.3.5
The heat transfer process in the transformer is by conduction, convection and sometimes by
radiation. The heat transfer within inside the core and windings is by conduction, which is a
linear process depending on the thermal conductivity of the material as well as the temperature
difference between the surfaces in contact for heat conduction. The heat transfer from the
core and windings to the oil is also conduction since they are designed to be in good contact
and there exists a temperature difference between them. But when it comes to heat transfer
within the oil (heat gets transmitted within the oil to reach the outside cooling system from the
transformer interiors) the process is not entirely linear. Due to fluid currents in the oil, the heat
transfer within the oil is due to convection as well as conduction. Convection is a complicated
40
phenomenon depending upon many variables such as density and viscosity of the liquid,
temperature difference between the heated surface and coolant, thermal resistivity of the fluid
and gravitational constants. However, it is to be noted that the convection takes place only in
the oil and only in the vertical direction (Y-direction) and hence doesnt affect the heat transfer
in horizontal direction (X-direction).
In order to model the convection, the effect of convection is to be understood. Convection in
the transformer oil has effect heat transfer in a manner such that the heat tends to rise up and
thus top portion of oil is at a relatively higher temperature than the bottom portion. As the oil
gets heated, the molecules become lighter and hence rise up giving place to cooler and hence
heavier oil molecules which, by gravity come down. Hence, within the oil, the heat transfer in
the upward direction is more than the heat transfer in the downward direction and hence the
top oil is always at a higher temperature than the bottom oil. The amount of convection fluid
currents also depends upon the viscosity of the fluids, here, the oil. More the viscosity of the
oil, lesser is the convection. The viscosity of a fluid is in turn dependent on the temperature of
the fluid and this makes convection a complex process to explain mathematically. Hence the
increase in the heat transfer due to convection is to be observed practically and a multiplicative
factor to the vertical resistance in the thermal model of oil elements is to be decided to
simulate this increase in heat transfer.
As oil is a viscous liquid, there cannot be too much convection as in the case with non viscous
liquids. The convection, as observed experimentally could increase the heat transfer by 100%
and not much more than that in any case and hence the heat transfer within the oil in the
vertical direction now becomes twice of the original value. Hence, the resistances in the
vertical direction (Y-direction) of oil blocks are modified, i.e., halved to include this convection
effect. Thus, an increase in heat transfer is achieved by reducing the resistance which accounts
for increased heat transfer due to convection.
Therefore, if x, y and z are the dimensions of an oil block with as the thermal conductivity,
the horizontal resistance Rh and vertical resistance Rv would be as in equations 4.13 and 4.14.
41
x
yz
...(4.13)
y
2 xz
...(4.14)
Rh =
Rv =
The tendency is of the hotter oil to go up against gravity, leaving the cooler oil at the bottom
of the tank is to be simulated. So, in other words, it can be stated that the convection in oil
doesnt allow heat flow downwards as far as convection effect is concerned. It is not that the
bottom portion of oil does not have any temperature rise, but relatively, the heat concentration
is towards the upper portion of oil. This phenomenon also takes place only in the vertical
direction and is modeled by partly obstructing the oil flow in downward direction. This is
achieved by using diodes in the thermal model, because the resistors will not simulate this
behavior and this obstruction has to be partial only. In order to ensure the convenience of
modeling, the oil blocks are modified as given in figure 4.4. Here Rh represents the horizontal
direction thermal resistance while Rv represents the vertical direction thermal resistance after
the multiplicative factor accounting for the increase in the heat transfer because of convection
fluid currents being incorporated.
FIG 4.4: MODELING THE CONVECTION MODIFIED THERMAL MODEL OF OIL ELEMENTS
42
Since the two vertical resistors equivalent to one single vertical direction thermal resistance
stand parallel to each other, their value has to be double the value of a single vertical resistor
that can equivalently replace them. In order to present partial obstruction to the downward
flow of heat, thus modeling convection, a diode is included in one of the arms of the model
shown in fig. 4.4. Diode, by nature, allows only unidirectional flow of charge, which by
analogy, means heat and hence the downward flow of heat is obstructed. The other arm does
not contain any diode, thus allowing the bidirectional flow of heat. In this way, the summed
effect of viscosity as well as fluid convection is pronounced in the model. A single oil element
modeled in this way is shown in fig. 4.5.
FIG 4.5: MODELING THE CONVECTION THERMAL MODEL OF OIL ELEMENTS WITH DIODE
All the oil blocks are modeled in the same way. This totally would present the summed effect
of viscosity as well as convection.
4.3.6
The heat transfer from the surface of the tank to ambient is both by convection (air currents)
and radiation (tank radiates heat to ambient). In transformers, radiation does not normally
occur by itself and in almost every case, it is accompanied by convection. The surface of the
tank radiates heat to the ambient and acts like a heat sink dissipating the heat to the open
ambient. With the completion of modeling of all the blocks and the total thermal model being
43
assembled, the ambient has got to be modeled. The heat dissipation from the tank surface to
the ambient, i.e., the heat sink behavior of the tank surface is modeled by resistors connecting
the elements on the outer surface of the tank elements to ground (ambient).
The resistance offered by heat sink is given in terms of ohm-m2. So, for each element on the
outer surface of the tank to which the resistor modeling the heat dissipation from the tank is
connected to, if ohm-m2 is the heat sink resistance and A is the area of surface through
which it dissipates heat to the ambient in m2, (A= lz, where l and z are the dimensions of the
element to which the resistance is connected to; l can be either x or y depending on the
location of the element), then the resistance offered by that element of heat sink is given by
equation 4.15.
R=
ohms =
lz
ohms
...(4.15)
While calculating the ambient resistances to be connected to each element of the tank, if the
tank element is situated on the vertical section of the tank, the dimension l would be the Ydimension, where as if it is on the horizontal portion of the tank geometry, then l would be the
X-dimension of the element. Accordingly, since, the corner blocks have two dissipating
surfaces, there will be resistances on both X and Y direction. Thus, the values of ambient
resistors can be determined.
4.4
The entire transformer geometry thus modeled as an electrical mesh circuit consisting of
current sources and resistors (and capacitors) has to be solved for potentials at each node,
which, by analogy are the temperature rises of the corresponding points on the transformer
geometry above the ambient. Summing the ambient temperature value to each of these values
of temperature rise, the value of temperatures at each corresponding location on the
transformer can be obtained.
44
Since the transformer thermal model would be too big to be solved by hand, suitable software
can be made use of to solve the electrical circuit to yield potentials at each node. MULTISIM
is the software package used in the current work.
45
Chapter 5
5 IMPLEMENTATION ON DIFFERENT TRANSFORMER
DESIGNS
5.1
The considered transformer is 15 KVA, 11KV/250V, single phase shell type (3-limb two
winding) transformer with the winding wound on the central limb. The current density in the
LV and HV winding is 2A/mm2 and 0.95A/mm2 respectively. The dimensions of this
transformer (mentioned as transformer 1 now onward in this thesis work) are given in table 2.
Sl. No.
Description
Dimension
230 mm
230 mm
90 mm
90 mm
46
190 mm
240 mm
5 mm
540 mm
580 mm
10
10 mm
11
10 mm
12
Thickness of LV winding
15 mm
13
Thickness of HV winding
30 mm
47
The considered geometry here represents the two dimensional cross section of the transformer
in the XY plane. As the Z-direction depth would be needed for calculations and taking the Z
direction depth as unity, which means 1 meter, is not a likely case, a better consideration of the
Z-direction depth of the considered plane is maintained that is same as the core thickness. So,
the plane that is being modeled through the thermal modeling process is that of core thickness
transformer cross section in the XY plane. The transformer geometry is divided into finite
number of elements and this virtual elemental division with the dimensions is as shown in
figure 5.2.
On performing the finite element analysis on the transformer using the FEA Software tool
following the same elemental division shown in fig. 5.2, the flux density distribution across the
transformer geometry is obtained. Figure 5.3 shows the transformer modeled in the FEA tool
48
NISA and figure 5.4 shows the flux density distribution obtained after performing Finite
Element Analysis on the considered transformer geometry.
The material properties used in the analysis in the implementation are as given in table 3.
Sl. No.
MATERIAL
MUXX(1/or) SIXX(1/)
Windings (Copper)
795800
58000000
Transformer oil
795800
400
4000000
800
4000000
49
From the flux density distribution and the values of flux density grossed over each element, as
generated by the tool, the values of core losses are calculated in each element division of the
core as explained in the equations 4.8 and 4.9. The values of Kh and Ke of the core are 0.0062
and 3.48 respectively. The value of form factor Kf is taken as 1.1 while the thickness of
lamination t was 0.27mm. The loss in the each element division of the winding is calculated
from the ohmic loss calculations given in equation 4.7. The electrical conductivity of copper is
58x106 mho/m. The loss distribution across the considered transformer geometry is as shown
in figure 5.5.
50
The values given are the losses in the corresponding blocks in rate of heat transfer W/sec.
These values of losses are used for current sources in the model. In the calculation of the
parameters of the thermal model the values of thermal conductivity as well as specific heat
capacity of the different materials used in the transformer, as given in table 4, are used.
Thermal conductivity
Specific Heat
(W/mK)
Capacity (J/kgoC)
Sl.
No.
Material
Description
CRGO steel
Core
26
450
Copper
Windings
400
386
Mineral Oil
0.72
2060
Structural Steel
Tank
45
400
Using the values of thermal conductivity as well as specific heat capacity, the values of
resistances as well as capacitances to be used for modeling of each element are calculated. The
51
value of z wherever required is used as 0.09m and x and y would be the geometrical
dimensions of the element in the XY plane. The modeling discussed in the current work aims
at steady state temperature profile for a uniform loading on the transformer and so the
capacitors are eliminated from the thermal model. The numbering of elements for reference is
as shown in figure 5.6 and the corresponding calculation of resistances is shown in Table 5.
Sl. no.
MATERIAL
ELEMENTS
RH()
RV()
Current
Source (A)
LV Winding
1, 2, 3, 4
0.004
0.213
6.6
HV Winding
5, 6, 7, 8
0.0073
0.107
0.43
0.43
0.601
0.167
1.1
0.434
0.143
1.29
0.376
0.048
3.87
0.1255
3
4
Core outer limbs
52
0.072
2.58
0.187
0.048
3.87
0.1258
89, 114
0.43
0.43
1.141
10
0.774
0.24
0.61
11
0.337
0.55
1.39
97, 106
0.774
0.24
2.172
100, 103
0.337
0.55
5.07
1.62
212.4
1.62
212.4
16
9, 18, 31, 40
12.96
26.4
17
11.124
30.84
3.7
92.4
19
5.55
61.68
20
3.7
92.4
21
1.4
15.18
22
18.51
18.51
23
10.2
33.312
18.51
18.51
25
33.33
10.3
26
12.96
26.4
12
Core inner limbs
13
14
15
18
Inner Oil sections
24
53
27
11.124
30.84
28
3.7
92.4
5.55
61.68
30
3.7
92.4
31
50, 71
33.24
10.2
32
8.1
33.33
33
0.043
4.3
34
0.024
7.74
35
0.215
0.86
36
0.43
0.43
37
0.86
0.215
38
7.74
0.024
39
3.01
0.06
40
2.58
0.072
41
0.86
0.215
1.29
0.143
43
0.86
0.215
44
132, 157
3.87
0.048
45
0.02
9.9
29
Outer Oil Section
Tank
42
Tank
54
Now, the tank to ambient heat dissipation and the heat sink behavior of tank surface is to be
modeled. The heat sink resistivity of the transformer tank is taken as 0.05 ohm /m2. And the
correspondingly calculated value of resistance connected to each element is as given in table 6.
Tank element to ambient resistance
Sl. No.
Tank Element
11.4
6.335
57
114
57
6.335
21.285
19
57
10
38
11
57
12
132, 157
12.66
13
4.955
value ()
55
Regarding the modeling of radiator resistances, the radiator tubes are made of pressed steel
and the cross sectional thickness is low so as to allow quicker and efficient transfer of heat to
the external ambient. The thermal conductivity of radiator tube material is taken as
0.12W/mK. The diameter of the radiator tube, which is the length of heat transfer path, is
0.01m. The surface area of the tank to which the radiators are connected is equal to the height
of the tank lz multiplied by ly, which is 0.54 x 0.09 m2. As explained in the section 4.3.4, the
value of radiator resistance would be as given in equation 5.1.
Rrad =
1
0.01
= 1.7
0.12 0.54 0.09
...(5.1)
This resistance has to be connected both to the top oil as well as bottom oil and hence the
value of resistance connected at each point (top oil and bottom oil) will be 3.4 , as the total
resistance has to be split into two.
The different modeling constraints discussed in section 4.3.2 are taken care of and the
convection in the oil is also modeled as explained in section 4.3.5 and the final thermal model
of the transformer is constructed as shown in figure 5.7.
This electrical equivalent mesh circuit has to be solved for potentials at each node, which are
the temperature rise of the corresponding point on the transformer geometry above the
ambient. MULTISIM has been used for the purpose of solving the network. An introduction
to the software is provided in the appendix. The problem geometry is modeled in the
schematic GUI of the software and the simulations are run, which gives the value of potential
at desired node or the potential at every node as desired in our study. The values of
temperature rise thus noted and superposed on the geometry of the transformer for enhancing
the understanding is given in figure 5.8, which is the Thermal Profile across the considered
geometry of the transformer
56
57
58
The critical points have been tabulated and compared against the practical test values as given
in table 7.
Sl. No.
Parameter
Test data
Winding
temperature
550C
42.660C
Top Oil
Temperature
400C
31.20C
Table 7. Thermal Model Implementation - Comparison with test values for transformer 1
5.2
The considered transformer is 25 KVA, 33KV/250V rating, single phase, core type (2-limb
two winding) transformer with the winding wound on the both the limbs one over the other.
The dimensions of this transformer (mentioned as transformer 2 now onward in this thesis
work) are given in table 8.
Sl. No.
Description
Dimension
250 mm
210 mm
100 mm
100 mm
260 mm
300 mm
5 mm
59
612 mm
610 mm
10
4.5 mm
11
17 mm
12
Thickness of LV winding
13 mm
13
Thickness of HV winding
36.5 mm
60
The transformer geometry is divided into finite number of elements and this virtual elemental
division with the dimensions is as shown in figure 5.10 (elements are shown as same size in
figure only for the clarity purpose though in the simulation they are considered according to
the actual dimensions).
On performing the finite element analysis on the transformer using the FEA Software tool
following the same elemental division shown in fig. 5.10, the flux density distribution across
the transformer geometry is obtained. It is not necessary to limit the elements to minimum
number and if required, more number of divisions can be made. Figure 5.11 shows the
transformer modeled in the FEA tool NISA and figure 5.12 shows the flux density distribution
obtained after performing Finite Element Analysis on the considered transformer geometry.
61
The material properties used in the analysis in the implementation are as given in table 9.
Sl. No.
MATERIAL
MUXX(1/or) SIXX(1/)
Windings (Copper)
795800
58000000
Transformer oil
795800
400
4000000
800
4000000
62
From the flux density distribution and the values of flux density grossed over each element, as
generated by the tool, the values of core losses are calculated in each element division of the
core as explained in the equations 4.8 and 4.9. The values of coefficients Kh and Ke of the core
are 0.005 and 2.523 respectively. The value of form factor Kf is 1.1 while the thickness of
lamination t was 0.27mm. The loss in the each element division of the winding is calculated
from the ohmic loss calculations given in equation 4.7. The electrical conductivity of copper is
58x106 mho/m. The values of thermal conductivity as well as specific heat capacity of the
different materials are given in table 10.
63
Sl.
No.
Material
Description
Specific Heat
Thermal conductivity
(W/mK)
Capacity (J/kgoC)
CRGO steel
Core
26
450
Copper
Windings
400
386
Mineral Oil
0.72
2060
Structural Steel
Tank
45
400
Using the values of thermal conductivity as well as specific heat capacity, the values of
resistances as well as capacitances to be used for modeling of each element are calculated. The
value of z wherever required is used as 0.1m. The tank to ambient heat dissipation and the heat
sink behavior of tank surface is modeled using the heat sink resistivity of the transformer tank
as 0.05 ohm /m2. The radiators are modeled and the value of each resistance to be connected
both to the top oil as well as bottom oil is calculated as 2.732 . The different modeling
constraints discussed in section 4.3.2 are taken care of and the convection in the oil is also
modeled as explained in section 4.3.5 and the final thermal model of the transformer is
constructed as shown in figure 5.13 and the obtained Thermal Profile across the considered
geometry of the transformer is given in figure 5.14.
64
65
66
The critical points have been tabulated and compared against the practical test values as given
in table 11.
Sl. No.
Parameter
Test data
Winding
temperature
600C
69.940C
Top Oil
Temperature
550C
47.150C
Table 11. Thermal Model Implementation - Comparison with test values for Transformer 2
5.3
The considered transformer is 16 KVA, 11KV/433V rating, single phase, shell type (3-limb
two winding) transformer with the winding wound on the both the limbs one over the other.
The current density in the LV winding is 3.071 A/mm2 and the current density in the HV
winding is 2.795 A/mm2.The dimensions of this transformer (mentioned as transformer 3 now
onward in this thesis work) are given in table 12.
Sl. No.
Description
Dimension
163 mm
135 mm
84 mm
84 mm
5
6
365 mm
175 mm
5 mm
600 mm
530 mm
10
3.5 mm
11
11 mm
12
Thickness of LV winding
10.5 mm
13
Thickness of HV winding
19 mm
68
The transformer geometry is divided into finite number of elements and this virtual elemental
division with the dimensions is as shown in figure 5.16 (elements are shown as same size in
figure only for the clarity purpose though in the simulation they are considered according to
the actual dimensions).
On performing the finite element analysis on the transformer using the FEA Software tool
following the same elemental division shown in fig. 5.16, the flux density distribution across
the transformer geometry is obtained. It is not necessary to limit the elements to minimum
number and if required, more number of divisions can be made. Figure 5.17 shows the
transformer modeled in the FEA tool NISA and figure 5.18 shows the flux density distribution
obtained after performing Finite Element Analysis on the considered transformer geometry.
69
The material properties used in the analysis in the implementation are as given in table 13.
Sl. No.
MATERIAL
MUXX(1/or) SIXX(1/)
Windings (Copper)
795800
58000000
Transformer oil
795800
400
4000000
800
4000000
70
From the flux density distribution and the values of flux density grossed over each element, as
generated by the tool, the values of core losses are calculated in each element division of the
core as explained in the equations 4.8 and 4.9. The values of coefficients Kh and Ke of the core
are 0.0052 and 2.82 respectively. The value of form factor Kf is 1.1 while the thickness of
lamination t was 0.27mm. The loss in the each element division of the winding is calculated
from the ohmic loss calculations given in equation 4.7. The electrical conductivity of copper is
71
58x106 mho/m. The values of thermal conductivity as well as specific heat capacity of the
different materials are given in table 14.
Sl.
No.
Material
Description
Thermal conductivity
Specific Heat
(W/mK)
Capacity (J/kgoC)
CRGO steel
Core
26
450
Copper
Windings
400
386
Mineral Oil
0.72
2060
Structural Steel
Tank
45
400
Using the values of thermal conductivity as well as specific heat capacity, the values of
resistances as well as capacitances to be used for modeling of each element are calculated. The
value of z wherever required is used as 0.084m. The tank to ambient heat dissipation and the
heat sink behavior of tank surface is modeled using the heat sink resistivity of the transformer
tank as 0.05 ohm /m2. The radiators are modeled and the value of each resistance to be
connected both to the top oil as well as bottom oil is calculated as 3.8 . The different
modeling constraints discussed in section 4.3.2 are taken care of and the convection in the oil
is also modeled as explained in section 4.3.5 and the final thermal model of the transformer is
constructed as shown in figure 5.19 and the obtained Thermal Profile across the considered
geometry of the transformer is given in figure 5.20.
72
73
74
The critical points have been tabulated and compared against the practical test values as given
in table 15.
Sl. No.
Parameter
Test data
Winding temperature
550C
42.380C
500C
36.760C
Table 15. Thermal Model Implementation - Comparison with test values for Transformer 3
5.4
The considered transformer is 45 KVA, 11 KV/250 V rating, three phase, core type (3-limb
three winding) transformer with each phase of the winding wound on each of the limbs one
over the other. The current density in the LV winding is 2 A/mm2 and the current density in
the HV winding is 0.95 A/mm2.The dimensions of this transformer (mentioned as
transformer 4 now onward in this thesis work) are given in table 16.
Sl. No.
Description
Dimension
230 mm
230 mm
90 mm
90 mm
255 mm
240 mm
5 mm
75
800 mm
580 mm
10
10 mm
11
10 mm
12
Thickness of LV winding
15 mm
13
Thickness of HV winding
30 mm
The transformer geometry is divided into finite number of elements and this virtual elemental
division with the dimensions is as shown in figure 5.22 (elements are shown as same size in
76
figure only for the clarity purpose though in the simulation they are considered according to
the actual dimensions).
On performing the finite element analysis on the transformer using the FEA Software tool
following the same elemental division shown in fig. 5.22, the flux density distribution across
the transformer geometry is obtained. It is not necessary to limit the elements to minimum
number and if required, more number of divisions can be made. Figure 5.23 shows the
transformer modeled in the FEA tool NISA and figure 5.24 shows the flux density distribution
obtained after performing Finite Element Analysis on the considered transformer geometry.
77
The material properties used in the analysis in the implementation are as given in table 17.
Sl. No.
MATERIAL
MUXX(1/or) SIXX(1/)
Windings (Copper)
795800
58000000
Transformer oil
795800
400
4000000
800
4000000
78
From the flux density distribution and the values of flux density grossed over each element, as
generated by the tool, the values of core losses are calculated in each element division of the
core as explained in the equations 4.8 and 4.9. The values of coefficients Kh and Ke of the core
are 0.0062 and 3.48 respectively. The value of form factor Kf is 1.1 while the thickness of
lamination t was 0.27mm. The loss in the each element division of the winding is calculated
from the ohmic loss calculations given in equation 4.7. The electrical conductivity of copper is
58x106 mho/m. The values of thermal conductivity as well as specific heat capacity of the
different materials are given in table 18.
79
Sl.
No.
Material
Description
Specific Heat
Thermal conductivity
(W/mK)
Capacity (J/kgoC)
CRGO steel
Core
26
450
Copper
Windings
400
386
Mineral Oil
0.72
2060
Structural Steel
Tank
45
400
Using the values of thermal conductivity as well as specific heat capacity, the values of
resistances as well as capacitances to be used for modeling of each element are calculated. The
value of z wherever required is used as 0.09m. The tank to ambient heat dissipation and the
heat sink behavior of tank surface is modeled using the heat sink resistivity of the transformer
tank as 0.05 ohm /m2. The radiators are modeled and the value of each resistance to be
connected both to the top oil as well as bottom oil is calculated as 3.464 . The different
modeling constraints discussed in section 4.3.2 are taken care of and the convection in the oil
is also modeled as explained in section 4.3.5 and the final thermal model of the transformer is
constructed as shown in figure 5.25 and the obtained Thermal Profile across the considered
geometry of the transformer is given in figure 5.26.
80
81
82
Parameter
Winding temperature
42.380C
36.760C
Table 19. Thermal Model Implementation - Comparison with test values for Transformer 4
5.5
Discussions
An observation of the obtained results shows that the obtained results are near to the
measured values, which shows that the model is definitely reliable. Though there is some
deviation of the obtained results with the actual measurements, it is owed to the some
limitations of modeling and the simplifications assumed while modeling. The credibility of the
model is proved in its application to different types of transformers and yet yielding reliable
results.
An observation of winding temperatures shows that the values obtained from the thermal
model for the shell type of transformers is less than the values obtained from direct
measurements. In the case of core type transformers, the temperatures obtained from the
thermal model are greater than the direct measurement values. This difference in trend
perhaps comes because of the two dimensional modeling involved. When the model is
assumed to be in the two dimensional plane, the third dimensional width does not get
modeled, the core type design allows lesser oil to get included as the design in 2-d plane
occupies a lot of space in the plane. However, in practical implementation, the volume of oil is
maintained the same irrespective of the design for a given rating of transformer. Owing to the
limitations of the two dimensional modeling over the realistic three dimensional modeling, the
model show difference in trends in temperature results for the core type transformers and shell
type transformers.
83
Chapter 6
6 CONCLUSIONS
This chapter summarizes the proposed modeling and gives the conclusions. The future
potential of the work is also discussed.
6.1
Conclusions
limit on the temperature, one can understand that the loading on the temperature on hotter
days cannot be as high as on cooler days. The parameters of the thermal model are based on
the dimensions of the transformer and properties of the materials. So, for a transformer with
different dimensions and properties, the parameter values used in the thermal model would be
different. As our interest was more towards the steady state thermal behavior, our designed
model gives steady state response. If required, we can modify the circuit and use that model to
simulate the transient behavior too. The work has been presented in six different chapters and
this chapter summarizes some of the observations and conclusions from each of the previous
chapters.
Chapter 1 provides the basic introduction to the problem by discussing the contribution of
thesis to the problem of thermal modeling of transformers. An overview of the transformer is
provided wherein the cause and effects of the heat accumulation and temperature rise inside
the transformer is explained.
Chapter 2 presents the thermal concepts required to understand the heating process, the heat
generation and dissipation inside the transformer. The heat generation and dissipation and the
different modes of heat dissipation is presented in general for a solid body and then explained
with respect to the transformer. The various axioms that guide the heating and cooling process
are explained. Then, the process of heat accumulation in the transformer is explained with the
reasons for its happening. The consequences of heat accumulation are given with emphasis
laid on the insulation ageing and its undesirable effects.
Chapter 3 presents the literature survey of the transformer thermal modeling techniques. The
techniques of measuring transformer interior temperatures are classified into the empirical
formulae which include the IEEE formulations to calculate the winding hotspot temperatures
for a given loading, the usage of fiber optic sensors which is a costly and the mathematical
thermal models. The mathematical thermal models are further classified as basic thermal
models, thermal models that use computer based simulations, thermal models using advanced
techniques like artificial intelligence techniques. The chapter provides the concluding remarks
85
discussing the drawbacks of these techniques, emphasizing the need for wider thermal profile,
thus providing a platform based on which the proposed thermal model technique was
developed.
Chapter 4 forms the crux of this thesis, where the thermal modeling of transformer using the
loss distribution has been presented. The proposed model with necessary theory has been
explained along with elaboration on various modeling issues. The technique is dealt in two
parts, namely, finding out the loss distribution in the transformer and developing the thermal
model for the transformer with the calculated loss distribution. Summarizing the methodology
of the thermal model, the losses in the core and coils for a particular loading condition are
found out. The issue of non uniform core loss distribution is handled by using Finite Element
Analysis. The values of losses along with the values of resistances and capacitance are used to
construct the thermal model for that particular loading condition, which gives the
corresponding steady state thermal profile. The model, which has been explained in theory is
explained with examples in the subsequent chapters
Chapter 5 deals with the implementation of the proposed model on four different transformer
designs. Two single phase shell type transformer designs, a single phase core type transformer
design and a three phase core type transformer designs are taken and the model is
implemented on each of them. To ease the understanding of the proposed thermal model,
presented in chapter 4, the implementation of the model on a practical transformer design is
explained in detail. The modeling constraints, the values of different parameters and properties
and their calculations are presented and the simulation results which yield the thermal profile
for the considered transformer geometry are presented. Three more different transformer
designs are taken and the modeling as well as the thermal model with the obtained thermal
profile results has been presented. A comparison with the transformer heat run test results is
presented in all the cases for the purpose of assessment of the results. The results, which are
not far away from the test cases, do give the credibility to the model and its application to
86
different transformers. This model is a general model for any type of transformer and hence
can be executed to any kind of transformer construction as well as rating.
6.2
87
7 APPENDIX
The appendix gives the necessary details and introduction about the software used in the
current work.
7.1
NISA provides an integrated and comprehensive suite of general and special purpose
programs for computer aided engineering (CAE). The NISA Suite of FEA Software covers a
wide spectrum of engineering applications, e.g., linear and nonlinear structural and heat
transfer analysis, structural and shape optimization, electromagnetic analysis, fatigue analysis,
fluid flow analysis and printed circuit board stress and heat transfer analysis. It consists of
three phases of programming, all of which are interfaced with the parent module. They are:
Pre Processing: Deals with the creation of model, finite element modeling and defining the
Analysis: The actual analysis chosen in the Pre processing module is applied to the
Post processing: The viewing of results in display and further conversions of code from
one programming format to the other etc, come under this phase of programming.
The two major analysis types in NISA/EMAG are the electric field analysis (EFIELD) and the
magnetic field analysis (MFIELD). Magnetic field analysis is used in the project, since magnetic
fields are being dealt with. NISA is used to find the flux densities across various elements in
the designed transformer geometry, which is further used to calculate the losses in various
elements and thereby proceed to the thermal model. The analysis chosen is 2D magneto
dynamic analysis (MGDN) as it deals with ac sinusoidal excitation and the output desired
being the Magnetic flux density. MGDN analysis provides for Magnetic field calculations in
88
magnetic and conducting materials due to sinusoidal (ac) current excitation. The outputs
provided would be Magnetic vector potential distribution, Magnetic Flux density distribution,
Magnetic field distribution, Eddy current density distribution, Total current density
distribution, Electric field distribution due to Eddy currents, Total Electric field distribution,
Power loss density distribution, Stored magnetic energy density for each element, Total stored
magnetic energy, Total power loss [37].
7.2
Introduction to MULTISIM
A number of schematic based and SPICE based tools are available for solving simulating
electrical networks. They solve the electrical networks using circuit theorems and analyses.
SPICE based tools require programming, while schematic based tools are more user friendly.
A few important tools which are SPICE based, schematic based, and tools integrating both,
are HSPICE, PSPCIE, Tanner tool (LTSPICE), MATLAB (SIMULINK), MULTISIM etc. Of
these MULTISIM is chosen, since it is more user-friendly. MULTISIM has the ability to
calculate the potentials at different nodes of an electrical circuit. Thus, by using MULTISIM
the potentials at every node of the thermal model are being calculated, which gives the
temperature at those points on the transformer geometry using the values of losses calculated
from the flux densities available through simulation in NISA. The measurements of interest
are recorded by connecting the oscilloscopes (or any desired measuring instrument available
from the library) and then running the simulation.
89
PUBLICATIONS
[1]
[2]
90
REFERENCES
[1]
[2]
[3]
[4]
IEEE Guide for Loading Mineral Oil Immersed Transformers, IEEE Std C57.911995 (Revision of IEEE Std C57.91-1981, IEEE Std C57.92-1981, and IEEE Std
C57.115-1991)
[5]
[6]
[7]
[8]
[9]
91
[10]
Glenn Swift, Tom S. Molsinki, Waldemar Le Rene Bray, and Robert Menzies, A
Fundamental Approach to Transformer Thermal Modeling Part II: Field
Verification, IEEE Transactions on Power Delivery, April 2001.
[11]
[12]
Asaad Elmoudi, Jussi Palola, Matti Lehtonen, A Transformer Thermal Model for
use in an on-line Monitoring and Diagnostic System, Power Systems Conference
and Exposition, 2006. PSCE '06.
[13]
G. L. Alegi, W. Z. Black, Real Time Thermal Model for an Oil Immersed Forced
Air Cooled Transformer, IEEE Transactions on Power Delivery, April 1990.
[14]
[15]
[16]
Zoran Radakovic, Kurt Feser, A New Method for Calculation of the Hot-Spot
Temperature in Power Transformers, With ONAN Cooling, IEEE Transactions on
Power Delivery, October 2003
[17]
[18]
92
[19]
[20]
[21]
[22]
[23]
[24]
[25]
Dr. Jan Declercq, Wim Van Der Veken Accurate Hot Spot Modeling in A Power
Transformer Leading to Improved Design and Performance, Transmission and
Distribution Conference, IEEE 1999.
[26]
93
[27]
[28]
Douglas J. Nelson and J. Patrick Jessee, A Coupled Thermal Magnetic Model for
High Frequency Transformers: Part I - Model Formulation and Material Properties,
VOL. 15, NO. 5, IEEE Transactions On Components Hybrids, and Manufacturing
Technology, October 1992.
[29]
Tang W. H., Wu Q. H., Richardson Z. J., Equivalent heat circuit based power
transformer thermal model, IEE Proceedings on Electrical Power Applications,
IET Journal 2002.
[30]
[31]
[32]
[33]
[34]
94
[35]
[36]
[37]
[38]
[39]
Richard Bean, Transformers for the Electric Power Industry, Tata McGraw Hill
Publications
[40]
95