energies

Article
A Simplified Top-Oil Temperature Model for
Transformers Based on the Pathway of Energy
Transfer Concept and the Thermal-Electrical Analogy
Muhammad Hakirin Roslan 1,2 , Norhafiz Azis 2,3, *, Mohd Zainal Abidin Ab Kadir 2 ,
Jasronita Jasni 2 , Zulkifli Ibrahim 2,4 and Azalan Ahmad 5
1 Faculty of Engineering, Universiti Pertahanan Nasional Malaysia, Kem Sg Besi 57000, Malaysia;
hakirinroslan@gmail.com
2 Centre for Electromagnetic and Lightning Protection Research, Universiti Putra Malaysia,
Serdang 43400, Malaysia; mzk@upm.edu.my (M.Z.A.A.K.); jas@upm.edu.my (J.J.); zulib78@gmail.com (Z.I.)
3 Institute of Advanced Technology (ITMA), Universiti Putra Malaysia, Serdang 43400, Selangor, Malaysia
4 Faculty of Engineering Technology, Universiti Teknikal Malaysia Melaka, Durian Tunggal 76100, Malaysia
5 Malaysia Transformer Manufacturing Sdn. Bhd, Ampang 54200, Selangor, Malaysia; azalan@mtm.com.my
* Correspondence: norhafiz@upm.edu.my; Tel.: +60-3-8946-4373

Received: 28 September 2017; Accepted: 24 October 2017; Published: 11 November 2017

Abstract: This paper presents an alternative approach to determine the simplified top-oil temperature
(TOT) based on the pathway of energy transfer and thermal-electrical analogy concepts. The main
contribution of this study is the redefinition of the nonlinear thermal resistance based on these
concepts. An alternative approximation of convection coefficient, h, based on heat transfer theory
was proposed which eliminated the requirement of viscosity. In addition, the lumped capacitance
method was applied to the thermal-electrical analogy to derive the TOT thermal equivalent equation
in differential form. The TOT thermal model was evaluated based on the measured TOT of seven
transformers with either oil natural air natural (ONAN) or oil natural air forced (ONAF) cooling
modes obtained from temperature rise tests. In addition, the performance of the TOT thermal model
was tested on step-loading of a transformer with an ONAF cooling mode obtained from previous
studies. A comparison between the TOT thermal model and the existing TOT Thermal-Electrical,
Exponential (IEC 60076-7), and Clause 7 (IEEE C57.91-1995) models was also carried out. It was
found that the measured TOT of seven transformers are well represented by the TOT thermal model
where the highest maximum and root mean square (RMS) errors are 6.66 ◦ C and 2.76 ◦ C, respectively.
Based on the maximum and RMS errors, the TOT thermal model performs better than Exponential
and Clause 7 models and it is comparable with the Thermal-Electrical 1 (TE1) and Thermal-Electrical
2 (TE2) models. The same pattern is found for the TOT thermal model under step-loading where the
maximum and RMS errors are 5.77 ◦ C and 2.02 ◦ C.

Keywords: top-oil temperature thermal model; pathway of energy transfer; thermal-electrical

analogy; nonlinear oil thermal resistance; transformers

1. Introduction
Thermal modeling is one of the important studies for the estimation of the top-oil temperature
(TOT) and hot-spot temperature (HST) in transformers. TOT is defined as the average of the tank outlet
oil temperature and the oil pocket temperature [1]. Meanwhile, HST consists of ambient temperature
(AT), TOT rise, and HST rise [1]. One of the common approaches to carry out thermal modeling of
transformers is through a numerical network thermal model. Currently, there are two standards that
utilize this approach to determine HST and TOT, known as IEC 60076-7 and IEEE C57.91-1995 [1,2].

Energies 2017, 10, 1843; doi:10.3390/en10111843 www.mdpi.com/journal/energies

Energies 2017, 10, 1843 2 of 15

These standards use the exponential function for constant loading and the differential function for
dynamic loading for determination of HST and TOT [1,2]. A number of studies were also carried out
to develop alternative thermal models for HST and TOT [3–18]. The majority of these studies were
carried out based on the thermal-electrical analogy that represents the heat transfer equivalent circuit
in the form of a resistor capacitor (RC) circuit [3]. Based on this concept, the current source analogy
represented the heat input due to losses, whilst the nonlinear resistance analogy represented the effect
of air or oil cooling convection currents [3].
TOT is one of the important components to determine HST in transformers. A model based on
the thermal-electrical analogy was developed [5,6] to determine the TOT thermal model. In these
studies, the nonlinear thermal resistance was expanded through inclusion of the Nusselt, Prandtle,
and Grashof numbers. The loss variation, viscosity changes with temperature, equivalent thermal
capacitances, oil time constants for different designs, and oil circulation were also considered in [5–7].
The TOT obtained by TOT thermal model in [5] was quite close to the measured TOT, as compared
to Annex G model in IEEE C57.91-1995 for transformers with oil natural air forced (ONAF) and oil
forced air forced (OFAF) cooling modes. The same finding was found for transformers with either
oil natural air natural (ONAN) or ONAF, and OFAF cooling modes in [6]. The TOT thermal models
in [5–7] utilize a number of input parameters which increase the accuracy of the simulated TOT.
A Levenberg–Marquardt model was proposed in [8] to determine the input parameters, such as TOT
rise at rated, oil time constant, and constant n, based on in-service TOT measured data. Based on the
case study of transformers with ONAF and OFAF cooling modes, it was found that the TOT thermal
model based on the Levenberg–Marquardt approach could represent the measured TOT better than the
Clause 7 model in IEEE C57.91-1995 [8]. This model requires the in-service measured TOT information,
which may not be available for some transformers in order to determine the input parameters. A TOT
thermal model was introduced in [9,10] based on the concept of outdoor and indoor heat transfer flows
of transformers. The indoor TOT thermal model equivalent circuit required a considerable number of
thermal resistances input parameters, such as windings, oil, core, tank, cooling air, ventilation holes,
walls, and ceiling. The model also considered the heat generated in the distribution cabinet, qcabin ,
such as heat from busbars, fuses, disconnectors, and switches. Based on the case study of a transformer
with an ONAN cooling mode, the TOT thermal model could represent quite well the measured TOT
for indoor and outdoor conditions. The TOT thermal models in [9,10] are beneficial for comprehensive
thermal analysis for individual outdoor or indoor transformers.
Nonlinear thermal resistance is one of the important parameters in the thermal-electrical analogy
thermal models. This parameter was commonly determined based on the Nusselt, Prandtle, and
Grashof numbers according to heat transfer theory [5,8–11]. Through this approach, the oil viscosity
was included in the final form of the nonlinear thermal resistance equation due to its sensitivity
toward temperature. The viscosity coefficient was also calibrated in [12] in order to determine the
nonlinear thermal resistance for the transformer with a rating over 6300 kVA. Another approach
to determine the nonlinear thermal resistance was based on measured values outside the tank and
average winding temperatures from the short-circuit heating experiment in [19]. The nonlinear
thermal resistance was represented by an exponential form in [20,21] where the element of the heat
conductance matrix and constants were determined based on a genetic algorithm (GA) [20,21]. Recently,
a moisture-dependent nonlinear thermal resistance was proposed in [22] where it considered the
cellulose thermal conductivity and area/thickness.
This paper presents a simplified approach to estimate the TOT based on a proposed concept of the
pathway of energy flow that is adopted from heat transfer theory to construct the nonlinear thermal
resistance and lumped capacitance circuit. An alternative approximation of the convection coefficient,
h, is proposed, which eliminates the requirement of the viscosity for application in the thermal-electrical
analogy. The main contribution of this study is the application of the pathway of energy based on heat
transfer principles and the redefinition of nonlinear thermal resistance. The derivation of the TOT
thermal model is presented and tested based on the measured TOT of seven transformers with either
Energies 2017, 10, 1843 3 of 15

ONAN or ONAF cooling modes. The TOT thermal model is also tested with the step-loading of a
transformer
Energies 2017,with an ONAF cooling mode from previous studies.
10, 1843
Energies 2017, 10, 1843
3 of 15
3 of 15

2. Heat
2. Transfer
2. Heat
Heat Theory
Transfer
Transfer TheoryApplication
Theory Applicationin
Application inTransformers
in Transformers
Transformers
2.1.2.1.
Concept of of Heat Transfer in Transformers
2.1. Concept
Concept of Heat
Heat Transfer
Transfer inin Transformers
Transformers
TheThe
Theconcept
concept
concept ofof
ofenergy
energytransfer
energy transferisis
transfer isone
oneof
one of the
of the simplest
the simplest approachesto
simplest approaches
approaches totoevaluate
evaluatethe
evaluate the
the heat
heat
heat transfer
transfer
transfer in in
in
transformers.
transformers. In general, heat transfer occurs there is a temperature difference
transformers. In general, heat transfer occurs when there is a temperature difference in a medium or or
In general, heat transfer occurs when there is a temperature difference in a
in medium
a medium or
between
between
between media
media
media [23].
[23].InIn
[23]. transformers,
Intransformers,
transformers,the the heat
the heat sources
sources originate
heat sources originatefrom
originate fromload
from loadand
load andno-load
and no-load
no-load losses
losses
losses which
which
which
distribute
distribute through
through oils
oils and
and transfer
transfer totothethe tank/radiator
tank/radiator of
of the
the transformer
transformer
distribute through oils and transfer to the tank/radiator of the transformer [24]. HST is the hottest [24].
[24]. HST
HST is the
is thehottest
hottest
temperature
temperature
temperature inside
inside
inside aa transformer
a transformer
transformerand and
andisis normally
isnormally located
locatedatat
normallylocated windings,
atwindings,
windings,and and the
and the coolest
the coolest temperature
coolest temperature
temperature is
is AT,
AT,iswhich which
AT, which is located
is located outside
outside
is located outside the
thethe transformer.
transformer.
transformer.Based Based
Basedonon the
onthe heat
the heat transfer
heat transfer theory, heat
theory,heat
transfer theory, heatororenergy
or energy
energy will
will
will
transfer
transfer
transfer from
from
from the
the hottest
hottest
the temperature
hottesttemperature
temperatureto to the
to the coolest
the coolest temperature;
coolest temperature;
temperature; in in transformers,
intransformers,
transformers,from from
from HST
HSTHST to the
to to
thethe
ATAT
ATviavia
via conduction
conduction
conduction and and convection
andconvection through
convectionthrough
through all all media,
media, including
all media, includingwindings,
including windings,insulation
windings, insulation
insulation papers,
papers,
papers, oils,
oils,
oils,
and
and tanks/radiators.
tanks/radiators.This This pathway
Thispathway of
pathwayofofenergyenergy transfer
energytransfer concept
transferconcept is applied
conceptisisapplied in this
appliedininthis study
thisstudy to determine
studytotodetermine
determinethe
and tanks/radiators.
the
the thermal
thermal resistance
resistance equivalent
equivalent circuit,
circuit, which
which can be
be seen
canseen seen in
in Figure
Figure 1.
thermal resistance equivalent circuit, which can be in Figure 1. 1.

Figure 1. Pathway of energy in transformers.

Figure
Figure 1.
1. Pathway
Pathway of
of energy
energy in
in transformers.
transformers.

Where
Where q1 𝑞𝑞is
1 is
thethe conductionthrough
conduction through copper
copper of the winding,𝑞q22 isisthe
the winding, theconduction
conductionthroughthrough thethe
Where 1 is the conduction through copper of the winding, 𝑞2 is the conduction through the
insulation
insulation paper, q 𝑞 is the convection from insulation paper to the oil, 𝑞 q is
insulation paper, 𝑞3 is the convection from insulation paper to the oil, 𝑞4 is the convection from thethe
paper, 3 is
3 the convection from insulation paper to the oil, 4 4 is the
theconvection
convection from
fromthe
oiloil
to to
oil thethe
to tank,
tank,
the tank, q5 𝑞𝑞is55 the
is
is the conduction
conduction
the conduction through
through
throughthe the tank/radiator
tank/radiator of
thetank/radiator of transformers,𝜃𝜃θℎ𝑠
of transformers,
transformers, is
is the
hs is
ℎ𝑠 the HST, 𝜃𝜃θ𝑜𝑖𝑙
theHST,
HST, oil is
𝑜𝑖𝑙
is the
theisTOT, TOT, and
andand
the TOT, 𝜃
θ a is𝜃𝑎the is the AT. Comprehensive analysis of the thermal conductivity
AT. Comprehensive analysis of the thermal conductivity through insulation through insulation
𝑎 is the AT. Comprehensive analysis of the thermal conductivity through insulation
paper
paper
papercancan
canbebe characterized
characterized
be characterizedbased based
basedonon consideration
onconsideration
consideration on on the
the porous
porouscharacteristics.
porous characteristics.Fractal
characteristics. Fractal
Fractal geometry
geometry
geometry is
is is
among
among
among the
thethe common
common
common approaches
approaches
approaches that
thatthat can
can can be
be used used to
to model
be used model the
the pore
to model pore
the structure structure
pore structure for
for modellingmodelling
for modelling the
the thermal
the
thermal
thermal conductivity
conductivity conductivity
through insulation through
through insulation
insulation paper
paperIn[25,26].
paper [25,26]. [25,26]. In
In this
this study, thestudy,
this study,
thermal the thermal
thermal conductivity
theconductivity conductivity of
of
of insulation
insulation
insulation paper was
paper was representedrepresented by a single
by a single thermal resistance
thermal for
resistance for the simplification
for the simplification of the thermal
of the resistance
thermal
paper was represented by a single thermal resistance the simplification of the thermal
resistance equivalent
equivalent circuit.
resistancecircuit. circuit.
equivalent

2.2.2.2.
2.2. Nonlinear
Nonlinear
Nonlinear Thermal Resistances
ThermalResistances
Thermal Resistances
All media in transformers have their own thermal
thermal resistances which act like dissipation elements.
AllAll media
media inin transformershave
transformers havetheir
theirown
own thermal resistances
resistanceswhich
whichact
actlike
likedissipation
dissipation elements.
elements.
The
The thermal
thermal resistance
resistance is
is nonlinear
nonlinear due
due to
to the
the presence
presence of
of the
the convection
convection coefficient,
coefficient, h.
h. Detailed
Detailed
The thermal resistance is nonlinear due to the presence of the convection coefficient, h. Detailed
descriptions
descriptions of the nonlinear thermal resistance are presented
presentedinin Section 3.1. Based on the pathway
descriptions of of
thethe nonlinear
nonlinear thermalresistance
thermal resistanceare arepresented inSection
Section3.1.
3.1.Based
Based onon the
the pathway
pathway of
of
of energy
energy transfer
transfer concept
concept shown
shown in
in Figure
Figure 1,
1, the
the heat
heat transfer
transfer thermal
thermal resistance
resistance equivalent
equivalent circuit
circuit is
energy
is
transfer
represented
concept
by Figure
shown
2.
in Figure 1, the heat transfer thermal resistance equivalent circuit
is represented by Figure 2.
represented by Figure 2.

Figure
Figure2. Thermal resistance equivalent
equivalent circuit.
Figure 2.
2. Thermal
Thermal resistance
resistance equivalent circuit.
circuit.

According
Accordingtoto
According Figure
Figure 2,
toFigure 2, there
2, thereare
there are
are three locations
three
three of
of temperatures,
locations
locations of temperatures,
temperatures, which are 𝜃
are
whichwhich 𝜃ℎ𝑠 , 𝜃𝑜𝑖𝑙θ,, and
are
ℎ𝑠 , 𝜃𝑜𝑖𝑙 hs
, θoil𝜃𝜃,𝑎𝑎..and
and The
The θ a .
heat
The sources
heat in
sources transformers
in transformersoriginate from
originate load
from losses,
load 𝑞
heat sources in transformers originate from load losses, 𝑞𝑐𝑢 , andcu ,
losses,
𝑐𝑢 andq no-load losses, 𝑞 , which are added
no-load losses, 𝑞𝑓𝑒 , whichf eare addedare
, and no-load losses,
𝑓𝑒 q , which
in
addedthe thermal
inthermal
in the resistance
the thermal equivalent
resistance
resistance circuit,
equivalent
equivalent as shown
circuit,circuit, in
as shown
as shown Figure
in Figure 3.
in 3.
Figure 3.
Energies 2017,
Energies
Energies 10,10,
2017,
2017, 1843
10, 1843
1843 of415
44 of of 15
15
Energies 2017, 10, 1843 4 of 15

Figure
Figure3. Thermal resistance equivalent
equivalent circuitwith
with heatsource.
source.
Figure 3.
3. Thermal
Thermal resistance
resistance equivalent circuit
circuit with heat
heat source.
Figure 3. Thermal resistance equivalent circuit with heat source.
Previous
Previous
Previous study
study
study showed
showed
showed that
that thethe
that thermal
thermal
the thermal resistance
resistance
resistance ofof tank/radiator
tank/radiator
of could
tank/radiatorcould
couldbe be discarded
bediscarded
discardeddue due
duetototothe
the Previous
perfect study characteristic
conductor showed that the thermal
where the resistance
thermal of tank/radiator
resistance was could be close
practically discarded
to due[3].
zero to In
the perfect
perfect conductor
conductor characteristic
characteristic where where the thermal
the thermal resistance
resistance waswas practically
practically closeclose to zero
to zero [3].[3].
InInthis
thestudy,
this perfecttheconductor
focus ischaracteristic
on the TOT where
thermalthe thermal
resistanceresistance
that was practically
relates with the close to zero [3].
convection In
thermal
this
study,thisstudy,
thestudy, the
focus the focus
is on is
the TOT on the TOT thermal resistance that relates with the convection thermal
resistance of the focus on thermal
is tank. the TOTresistance
thermal that relates
resistance thatwith the with
relates convection
thelaw thermal thermal
convection resistance of
theresistance
oil to the of the oil
tank. oil
Due
to the
to to
thethis
tank. Due
Due to
reason, to this
thethis reason,
reason, the
application
application
theofapplication
Fourier’s
of
of Fourier’s
law Fourier’s
of heat law of of heat
conductionheat conduction
conduction
and the dual
resistance of the oil to the tank. Due to this reason, the application of Fourier’s law of heat conduction
and
and the
the dual
dual phase
phase lag
lag method
method are
are not
not considered
considered in
in the
the computation.
computation. The
The final
final form
form of
of the
the heat
heat
phase lag method are not considered in the computation. The final form
and the dual phase lag method are not considered in the computation. The final form of the heat of the heat transfer thermal
transfer
transfer thermal
thermal resistance
resistance equivalent
equivalent circuit
circuit can
can be
be re-drawn,
re-drawn, as
as shown
shown in
in Figure
Figure 4.
4.
resistance
transferequivalent circuit can
thermal resistance be re-drawn,
equivalent as shown
circuit can in Figure
be re-drawn, 4.
as shown in Figure 4.

Figure 4. Thermal resistance equivalent circuit with a heat source and without the thermal resistance
Figure
Figure 4.
Figure Thermal
4. 4. resistance
Thermal
Thermal resistanceequivalent
resistance equivalentcircuit
equivalent circuit with
circuit heat
with aaheat source
heatsource and
sourceand without
andwithout
without the
the thermal
thermal
the resistance
resistance
thermal resistance
of the tank/radiator.
of the
of tank/radiator.
the tank/radiator.
of the tank/radiator.

2.3.2.3.
Lumped Capacitance
2.3.2.3. Lumped
Lumped
Lumped Capacitance
Capacitance
Capacitance
The Thecurrent
The current form of the
form of thethermal
thermalresistance
resistance equivalent circuit
equivalent circuit inin Figure
Figure 4 can
4 can onlyonly
be usedbe used
underunder
The current
current form form of of the
the thermal
thermal resistance
resistance equivalent
equivalent circuit
circuit in in Figure
Figure 44 can can only
only be be used
used under
under
steady steady
steady state state
state condition.
condition.
condition. InIn In order
order
order toto
to determine
determine
determine HST
HST and
and
and TOT
TOT
TOT under
under
under transient,
transient,
transient, the
the lumped
the lumped
lumped capacitance
capacitance
capacitance
steady state condition. In order to determine HST and TOT under transient, the lumped capacitance
approach
approach was was applied.The
applied. TheBiotBiotnumber
number of the the thermal resistance must bebebe
determined before thisthis
approach
approach was
was applied.The
applied. TheBiot Biotnumber
number of thermal
of the thermal resistance
thermal resistance
resistance must
must
must determined
be determined
determined before
before
before this
this
approach
approach
approach can can
canbe be be implemented.
implemented.
implemented. The
The
The Biot
Biot
Biot number
number
number is
is
is the
the
the ratio
ratio
ratio ofof
of the
the
the thermal
thermal
thermal resistance
resistance
resistance of conduction
of of conduction
conduction
approach can be implemented. The Biot number is the ratio of the thermal resistance of conduction
over the convection [3,23]. In transformers, the Biot number is less than 1 since the thermal resistance
over
over thethe
over the convection
convection
convection [3,23].
[3,23].
[3,23].In In
In transformers,
transformers,
transformers, the Biot
the Biot number
number
number is
isless
is lessthan
less than111since
than sincethe
since thethermal
thermal resistance
resistance
of conduction is normally smaller than convection [3]. Through application of the the
lumpedthermal resistance
capacitance
of of
of conduction
conduction
conduction is isnormally
normally smaller
normallysmaller thanconvection
than convection [3].
[3]. Through
Throughapplication
applicationof ofofthe lumped
the lumped capacitance
capacitance
method, the iscapacitance smaller
is addedthan to the convection
heat transfer[3].thermal
Through application
resistance equivalentthe lumped capacitance
circuit which can
method,
method,
method, the
the capacitance
capacitance
the capacitance isis added
added
is 5.
added to
to the
the heat
heat transfer
transfer thermal
thermal resistance
resistance equivalent
equivalent circuit
circuit which
which cancan
be represented by Figure Theretoare thetwoheat transfer thermal
capacitances that areresistance
included, equivalent
known as oil circuit which
and winding can
be
be be represented
represented
represented by
byEach
by Figure
Figure
Figure 5. There
5.5.There
Thereare are two
aretwotwo capacitances
capacitances
capacitances that are
that are
that included,
are included,
included, known
known
known as
asasoil and
oiloil
andandwinding
winding
winding
capacitances. of the capacitances represent the storage element of either the oil or winding in
capacitances.
capacitances.
capacitances. Each
EachEach ofof the
the
of the capacitances
capacitances
capacitances represent
represent
represent the
the
the storage
storage element
elementofof
storageelement either
ofeither
eitherthe the
theoiloil
oiloror winding
windinginin
orwinding inthe
the thermal resistance equivalent circuit.
the
the thermal
thermal thermal resistance
resistance
resistance equivalent
equivalent
equivalent circuit.
circuit.circuit.

Figure 5. Thermal resistance equivalent circuit with lumped capacitance.

Figure 5. Thermal resistance equivalent circuit with lumped capacitance.

Figure
Figure 5.
5. Thermal
Thermal resistance
resistance equivalent
equivalent circuit
circuit with
with lumped
lumped capacitance.
capacitance.
2.4. Thermal-Electrical Analogy
The thermal-electrical analogy parameters are through variable, across variable, dissipation
element, and storage element [3]. The through variable, heat source, q, in the thermal approach is
analogous to current, I in the electrical approach. Meanwhile, the across variable, temperature, θ,
in the thermal approach is analogous to voltage, v, in the electrical approach. The dissipation element
 Energies 2017, 10, 1843 5 of 15

2.4. Thermal-Electrical Analogy

The thermal-electrical analogy parameters are through variable, across variable, dissipation
element,
Energies 2017, 10,and
1843storage element [3]. The through variable, heat source, q, in the thermal approach is5 of 15
analogous to current, I in the electrical approach. Meanwhile, the across variable, temperature, 𝜃, in
the thermal approach is analogous to voltage, v, in the electrical approach. The dissipation element
variable, thermal
variable, thermal resistance,R𝑅
resistance, , ,ininthe
th𝑡ℎ thermalapproach
the thermal approach is analogous
is analogous to electrical
to electrical 𝑅𝑒𝑙 , in Rel ,
resistance,
resistance,
in thethe electricalapproach,
electrical approach,while
while the storage
storageelement
elementvariable,
variable, thermal capacitance,
thermal 𝐶𝑡ℎ , C
capacitance, inththe
, inthermal
the thermal
approach
approach is analogous
is analogous totoelectrical capacitance, 𝐶𝑒𝑙
electrical capacitance, C,elin thethe
, in electrical approach.
electrical Based on
approach. Figure
Based on5Figure
the 5
thermal-electrical circuit can be divided into two models, known as the HST
the thermal-electrical circuit can be divided into two models, known as the HST and TOT models, and TOT models, as
shown in Figure 6a,b, respectively. In this study, only the TOT model was developed and examined.
as shown in Figure 6a,b, respectively. In this study, only the TOT model was developed and examined.

(a) (b)

Figure 6. Thermal-electrical analogy circuit: (a) hot-spot temperature; and (b) top-oil
Figure 6. Thermal-electrical analogy circuit: (a) hot-spot temperature; and (b) top-oil temperature.
temperature.

3. Top-Oil Temperature
3. Top-Oil Thermal
Temperature Model
Thermal Model

3.1. Definition of Nonlinear

3.1. Definition OilOil
of Nonlinear Thermal
ThermalResistance
Resistance
According to the
According to heat transfer
the heat theory
transfer theoryin in
[5,23],
[5,23],the thenonlinear
nonlinearoiloil thermal resistanceisisdefined
thermal resistance definedasas the
theproduct
inverse inverse product of convection
of convection coefficient, h and area,
coefficient, h andAarea,
which A which
is given is by
given by Equation
Equation (1).nonlinear
(1). The The
nonlinear
oil thermal oil thermal
resistance, resistance,
Roil 𝑅𝑜𝑖𝑙 (1)
in Equation in Equation
refers to (1) refers to oil-to-tank
nonlinear nonlinear oil-to-tank thermal resistance,
thermal resistance, Roil −tank in
Figure𝑅𝑜𝑖𝑙−𝑡𝑎𝑛𝑘
6b: in Figure 6b:
1
R𝑅oil𝑜𝑖𝑙== (1) (1)
ℎ𝐴
hA
The convection
The convection coefficient,
coefficient,h hininEquation
Equation (1)
(1) may
may bebeaffected
affectedbyby multiple
multiple factors,
factors, including
including the the
medium thermal characteristics. According to the heat transfer theory, the convection
medium thermal characteristics. According to the heat transfer theory, the convection coefficient, coefficient, h
varies
h varies with
with thethe temperaturedifference
temperature difference between
between the
theobject
object(transformers
(transformers tank/radiator)
tank/radiator)and the
andfluid
the fluid
(transformers
(transformers oil) oil)
[23].[23]. In this
In this case,h hisisapproximated
case, approximated bybythe
theexpression
expression in in
Equation (2) [23]:
Equation (2) [23]:
ℎ = 𝐶(𝜃𝑠 − 𝜃∞ )𝑛 (2)
h = C (θs − θ∞ )2n (2)
where n and C are constants and the unit of C is W/m ·K(1+n). The constant, n, contributes to the
nonlinearity of the convection coefficient, h. In Equation (2), 𝜃 represents the TOT, 𝜃𝑜𝑖𝑙 , and 𝜃∞
where n and C are constants and the unit of C is W/m2 ·K(1+n)𝑠. The constant, n, contributes to the
represents the AT, 𝜃𝑎 . The TOT rise, ∆𝜃𝑜𝑖𝑙 , is determined by subtracting 𝜃𝑎 from the 𝜃𝑜𝑖𝑙 where it
nonlinearity of the convection coefficient,
can be simplified to Equation (3):
h. In Equation (2), θ s represents the TOT, θ oil , and θ∞
represents the AT, θ a . The TOT rise, ∆θoil , is determined by subtracting θ a from the θoil where it can be
simplified to Equation (3): ℎ = 𝐶(∆𝜃𝑜𝑖𝑙 )𝑛 (3)
n
h = C ( ∆θ
Substitution of Equation (3) into Equation (1) leadsoil to the final proposed equation of nonlinear (3)
)
oil thermal resistance
Substitution as shown
of Equation in Equation
(3) into Equation(4):
(1) leads to the final proposed equation of nonlinear oil
thermal resistance as shown in Equation (4):

1
Roil = (4)
CA(∆θoil )n

The thermal resistance is nonlinear due to the presence of the constant n, which represents the
heat transfer convection modes in transformers, as shown in Equation (4).
The redefinition of the nonlinear oil thermal resistance based on approximation of convection
coefficient, h, in Equation (4) is among the key contributions of this study. A comparison with
similar forms of nonlinear thermal resistances such as Susa and Tang models was also carried out.
Energies 2017, 10, 1843 6 of 15

The nonlinear thermal resistance for Susa Model can be obtained based on the following approach.
The relationship of Nusselt, Prandtle, and Grashof numbers is given by Equation (5) [5]:

Nu = C [ Gr · Pr ]n (5)

The Nusselt, Prandtle, and Grashof numbers can be determined based on Equations (6)–(8):

h· L
Nu = (6)
k
coil ·µ
Pr = (7)
k
L3 ·ρoil 2 · g · β·(∆θoil )
Gr = (8)
µ2
Substitutions of Equations (6)–(8) into Equation (5) lead to the Equation (9):
n
L ·ρoil 2 · g· β·(∆θoil ) coil ·µ
 3
h· L
= C· · (9)
k µ2 k

Since only viscosity is sensitive toward temperature, Equation (9) can be reduced to Equation (10)
known as the convection coefficient, h [5]:

∆θoil n
h = C1 ·( ) (10)
µ

where C1 is obtained from Equation (11):

h 1− n 3n−1
in
C1 = C · ρoil 2 · g· β·k( n ) · L( n ) ·coil (11)

Substitution of Equation (10) into (1) leads to Equation (12) which is the final form of the nonlinear
thermal resistance for Susa model [5]:

µn
Roil = (12)
CA(∆θoil )n

The nonlinear thermal resistance for Tang model can be derived based on the following approach.
According to [20,21], the thermal conductance can be represented by Equation (13):

G = a(∆θoil )b (13)

where G is the element of the thermal conductance matrix, ∆θoil is the TOT rise, a and b are the
constants. The reciprocal of Equation (13) leads to the final form of nonlinear thermal resistance for
Tang model, as shown in Equation (14) [20,21]:

1
Roil = (14)
a(∆θoil )b

The proposed nonlinear thermal resistance does not consider viscosity as in the Susa model.
The nonlinear thermal resistance in the Susa model considers multiple oil parameters, such as
characteristic length, L, oil thermal conductivity, k, specific heat of oil, coil , oil density, ρoil , gravitational
constant, g, and oil thermal expansion coefficient, β, while the proposed nonlinear thermal resistance
only considers the constant C that is derived from the heat transfer theory. The derivation of nonlinear
thermal resistance in the Susa model is different from the proposed nonlinear thermal resistance
where it is based on Nusselt, Prandtle, and Grashof numbers. Meanwhile, the nonlinear thermal
Energies 2017, 10, 1843 7 of 15

resistance in the Tang model is derived based on the exponential form of heat conductance matrix
element. The constants, a and b in the Tang model are obtained based on GA through the in-service
measured data.

3.2. Derivation of Top-Oil Temperature Thermal Model

Based on the TOT thermal-electrical analogy equivalent circuit in Figure 6b, the differential
equation representation can be determined based on Equation (15) [5]:

dθoil (θ − θ a )
q f e + qcu = Coil + oil (15)
dt Roil

Substitution of Equation (4) into Equation (15) leads to Equation (16):

dθoil (θ − θ a )
q f e + qcu = Coil + oil 1 (16)
dt n
CA(∆θoil )

The nonlinear oil thermal resistance at rated, Roil, rated , TOT rise at rated, ∆θoil,rated , and oil time
constant, τoil , can be determined according to Equations (17)–(19):

1
Roil, rated = (17)
CA(∆θoil,rated )n
 
∆θoil,rated = q f e + qcu · Roil,rated (18)
rated
τoil = Roil,rated ·Coil (19)

The ratio of load losses at the rated to the no load losses, R can be computed according to
Equation (20):
qcu
R= (20)
qfe
The ratio of the load current to the load current at rated, K can be calculated according to
Equation (21):
I
K= (21)
Irated
Substitutions of Equations (17)–(21) into Equation (16) lead to Equation (22), which is the final
form of the TOT thermal model:

1 + R·K2 dθ ( θ − θ a )1+ n
·∆θoil,rated = τoil oil + oil (22)
1+R dt ∆θoil,rated n

The TOT thermal model in Equation (22) is in the same form as in [8], however the approach in
terms of derivation of the TOT thermal model is different. The nonlinear oil thermal resistance, Roil ,
in [8] was obtained through Nusselt, Prandtle, and Grashof numbers. The TOT model in [8] also took
into consideration on the influence of oil viscosity in the TOT thermal model. In addition, the input
parameters such as the TOT rise at rated, ∆θoil,rated , oil time constant, τoil and constant, n, in [8] were
estimated based on the Levenberg-Marquardt method. The TOT thermal model in Equation (22) does
not require the information on the viscosity for determination of nonlinear oil thermal resistance,
Roil where simplification is carried out through approximation of convection coefficient, h, of the heat
transfer convection between the oil and the transformer tank.

3.3. Input Data

The TOT thermal model was tested on seven transformers with ONAN and ONAF cooling modes.
The seven transformers were named as TX1, TX2, TX3, TX4, TX5, TX6, and TX7. In addition, the TOT
Energies 2017, 10, 1843 8 of 15

thermal model was also tested on the step loading of a 250 MVA transformer with ONAF cooling
mode obtained from [1,5] and named as TX8. Both TX3 and TX4 have identical design except the
cooling mode for TX3 is ONAN and TX4 is ONAF. The input parameter for these transformers ratings
are shown in Table 1 for constant loading and Table 2 for step loading. The oil time constant, τoil ,
was calculated based on [2] using the input parameters obtained from the nameplate rating such as
the weight of core and coil in kg, the weight of the tank and fittings in kg, and transformer oil in L.
Based on the convection heat transfer, the value of constant, n is 0.25 for laminar oil circulation [27],
which indicates free convection (ONAN cooling modes). It is used to control the sensitivity of the TOT
model curve which depends on the cooling modes and oil circulation of transformers. The constant,
n, has different values for forced convection (ONAF and OFAF cooling modes) [5,6]. In this study,
the constant n for TE1 and TE2 were calibrated based on the measured TOT in order to increase the
accuracy of the simulated TOT. The same procedure was carried out for the TOT thermal model, which
can be seen in Table 3.

Table 1. Input parameters for the TOT thermal model.

Transformer/Winding
Parameters TX1 TX2 TX3 TX4 TX5 TX6 TX7
11/0.433 kV 33/11 kV 33/11 kV 33/11 kV 132/11 kV 132/33 kV 132/33 kV
Rating (kVA) 300 15,000 30,000 30,000 30,000 60,000 90,000
Cooling Modes ONAN ONAF ONAN ONAF ONAN ONAN ONAF
Load (p.u.) 1.00 1.00 1.00 1.00 1.00 1.00 1.00
R 4.52 7.56 10.47 11 5.79 4.66 10
∆θoil,rated (K) 30.3 51.23 59.2 58.87 38.9 44.07 47.87
τoil (min) 129 152 200 207 239 295 165
θoil,i (K) 31.9 44.9 29.7 29.9 31.8 32.3 41.7
(IEC) Oil exponent, x 0.8 0.8 0.8 0.8 0.8 0.8 0.8
(IEC) Constant, k11 0.5 0.5 0.5 0.5 0.5 0.5 0.5
(IEEE) Exponent, n 0.8 0.9 0.8 0.9 0.8 0.8 0.9

Table 2. Input parameter for TOT thermal model for step loading [1,5].

Transformer/Winding
Parameters TX8
250/118 kV
Rating (kVA) 250,000
Cooling Modes ONAF
1.00 pu/3 h + 0.60 pu/3 h + 1.50 pu/2 h + 0.30
Load (p.u.)
pu/3 h + 2.10 pu/0.33 h
R 10.17
∆θoil,rated (K) 38.3
τoil (min) 168
θoil,i (K) 38.3

Table 3. Constant n used in this study.

Constant n
Transformers
TE1 TE2 Thermal Model
TX1 0.1 1 0.1
TX2 0.75 0.9 1
TX3 0.75 0.9 1
TX4 0.75 0.9 1
TX5 0.4 0.95 0.65
TX6 0.4 0.95 0.65
TX7 0.25 0.95 0.5
TX8 0.1 0.95 0.25
Energies 2017, 10, 1843 9 of 15

3.4. Comparison with Previous Thermal-Electrical and Standard Models

In this study, two thermal-electrical models in [3,5] were used to evaluate the performance of the
TOT thermal model. Thermal-Electrical 1 (TE1) model was proposed by [5] as seen in Equation (23).
The model in Equation (24) was introduced by [3] and termed as Thermal-Electrical 2 (TE2) model:

1 + R·K2 dθ ( θ − θ a )1+ n
·µ pu n ·∆θoil,rated = µ pu n ·τoil oil + oil (23)
1+R dt ∆θoil,rated n

1 + R·K2 1 dθ 1
· (∆θoil,rated ) n = τoil oil + (θoil − θ a ) n (24)
1+R dt
The TOT thermal model was also compared with the existing models in standards known
as the Exponential model in IEC 60076-7 and the Clause 7 model in IEEE C57.91-1995, as seen in
Equations (25) and (26) [1,2]. The constants, x, k11 , and n used for both the Exponential and Clause 7
thermal models were chosen based on Table 1.
x !
1 + R·K2

θoil = θ a + ∆θoil,i + ∆θoil,rated · − ∆θoil,i · f 1 (25)
1+R
 
−t
θoil = θ a + (∆θ TO,U − ∆θ TO,i )· 1 − e τoil + ∆θ TO,i (26)

Maximum error and root mean square (RMS) error were used to analyze the simulated TOT,
as can be seen in Equations (27) and (28):
 
maximum error = θoil, calculated − θoil, measured  (27)
s
2
∑in=1 (θoil, calculated − θoil, measured )
RMS error = (28)
n
where θoil, measured is the measured TOT, θoil, simulated is the simulated TOT, and n is the number
of samples.

4. Results and Discussions

4.1. Constant Loading

The comparison between the TOT thermal model and the existing TOT Thermal-Electrical,
Exponential (IEC 60076-7) and Clause 7 (IEEE C57.91-1995) models can be seen in Figures 7–13. For TX1,
the simulated TOT for all thermal models overshoot during the first 300 min of the temperature rise
test as seen in Figure 7. The maximum errors for the TOT thermal, TE1, TE2, Exponential, and Clause
7 models are 5.65 ◦ C, 5.98 ◦ C, 5.03 ◦ C, 4.79 ◦ C, and 7.92 ◦ C. Meanwhile, the RMS errors are 2.72 ◦ C,
2.89 ◦ C, 2.41 ◦ C, 2.23 ◦ C, and 4.02 ◦ C. The measured TOT for TX2 is well represented by the TOT
thermal, TE1, TE2, and Exponential models, as shown in Figure 8. However, the simulated TOT based
on the Clause 7 model is lower than the measured TOT during the first 300 min of the temperature rise
test. The maximum errors of the TOT thermal, TE1, TE2, Exponential, and Clause 7 models are 3.72 ◦ C,
2.77 ◦ C, 2.09 ◦ C, 2.15 ◦ C, and 12.92 ◦ C, while the RMS errors are 1.34 ◦ C, 1.33 ◦ C, 0.97 ◦ C, 1.03 ◦ C, and
5.33 ◦ C. The simulated TOT based on TE2 and Exponential models are much closer to the measured
TOT than the TOT thermal, TE1, and Clause 7 models for TX3. The same pattern is observed for TX4
and it is expected since both transformers have the same design, except for its cooling modes. For TX3,
the maximum errors for TOT thermal, TE1, TE2, Exponential, and Clause 7 models are 6.53 ◦ C, 6.32 ◦ C,
0.91 ◦ C, 3.30 ◦ C, and 8.25 ◦ C, while for TX4, the values are 6.66 ◦ C, 5.71 ◦ C, 3.35 ◦ C, 8.37 ◦ C, and 8.24 ◦ C.
On the other hand, the RMS errors of TX3 for the TOT thermal, TE1, TE2, Exponential, and Clause 7
 Energies 2017, 10, 1843 10 of 15

models are 2.76 ◦ C, 2.49 ◦ C, 0.56 ◦ C, 2 ◦ C, and 5.17 ◦ C, while for TX4, the values are 2.36 ◦ C, 2.28 ◦ C,
1.38 ◦ C, 3.61 ◦ C, and 5.20 ◦ C, respectively.
The TOT thermal, TE1, and TE2 models can represent the measured TOT for TX5 quite well as
compared with the Exponential and Clause 7 models. The maximum and RMS errors for TOT thermal
Energies 2017, 10, 1843 10 of 15
model are10,1.43 ◦ ◦
Energies 2017, 1843C and 0.67 C. For TE1, TE2, Exponential and Clause 7 models, the maximum 10errors
of 15
◦ ◦ ◦ ◦ ◦ ◦ ◦ ◦
Clause 7 model for TX7 is lower than the measured TOT during the initial stage of the temperatureC,
are 1.42 C, 1.64 C, 7.44 C, and 5.30 C, while the RMS errors are 0.70 C, 1.04 C, 3.72 C, and 2.65
respectively.
Clause
rise test.7 On
modeltheThe
for
othersame
TX7 pattern
is
hand, lower is observed
than
the simulated for TX6
the measured
TOT based and
onTX7,
TOT duringexcept the simulated
the initial
the Exponential stage of
model is TOT
the based
thanon
temperature
higher thethe
Clause
rise test. 7On
measured model
TOT for TX7
theduring
other theisfirst
hand, lower
the500 than
min.the
simulated measured
TOT
The basedTOT
maximum during
onerrors
the for the
TX6initial
Exponential stage
for model
TOT isofhigher
the temperature
thermal, thanTE2,
TE1, the
rise
measuredtest. On
TOT the other
during hand,
the firstthe
500 simulated
min. The TOT based
maximum on the
errors Exponential
for TX6
Exponential, and Clause 7 models are 2.01 °C, 1.98 °C, 2.85 °C, 8.04 °C, and 9.04 °C, while for TX7, for model
TOT is higher
thermal, than
TE1, TE2, the
measured
Exponential,
the TOT
values areand during
2.14Clause
°C, 2.03the first
7 models 500
°C, 2 °C,are min.
2.01
6.67 The
°C,°C,
and maximum
1.98
6.84°C,
°C.2.85errors for
°C, 8.04 °C,
Meanwhile, TX6
theandfor
RMS TOT
9.04 thermal,
°C, while
errors TE1,
of TX6for TE2,
forTX7,
the
Exponential, and TE2,
Clause 7°C,
models are ◦ C, 1.98 ◦ C, 2.85 ◦ C, 8.04 ◦ C, and 9.04 ◦ C, while for TX7,
the
TOT values
thermal, areTE1,
2.14 °C, 2.03
Exponential, and2.01
2 °C, 6.67 °C, and
Clause 76.84 °C. Meanwhile,
models are 0.81 °C,the 0.72RMS errors
°C, 1.54 °C,of4.54
TX6°C,
forand
the
the values are 2.14 ◦ C, 2.03 ◦ C, 2 ◦ C, 6.67 ◦ C, and 6.84 ◦ C. Meanwhile, the RMS errors of TX6 for
TOT thermal, TE1, TE2, Exponential, and Clause 7 models
4.24 °C, while for TX7, the values are 0.95 °C, 0.88 °C, 1 °C, 3.39 °C, and 4.33 are 0.81 °C, 0.72 °C, 1.54 °C, 4.54
respectively. °C, and
the°C,TOT thermal, TE1, TE2, Exponential, and Clause ◦ C, 0.72 ◦ C, 1.54 ◦ C, 4.54 ◦ C,
4.24 while for TX7, the values are 0.95 °C, 0.88 °C, 1 7°C,
models
3.39 °C,areand
0.814.33 °C, respectively.
and 4.24 ◦ C, while for TX7, the values are 0.95 ◦ C, 0.88 ◦ C, 1 ◦ C, 3.39 ◦ C, and 4.33 ◦ C, respectively.
70 24
Measured
70 24
Thermal
MeasuredModel
60 20
TE1
Thermal Model
60 20
(°C)(°C)

(°C)(°C)
TE2
TE1
50 16
Clause 7 IEC
TE2Exponential
Temperature

Temperature
50 IEEE Clause 7 16
Clause
TE1 7 IEC Exponential
Temperature

Temperature
40 Error 12
IEEE Clause 7
40 TE1 Model
Thermal 12
Error
30 Thermal
TE2 Model 8
Exponential
Top-oil

30 TE2 8

Error
20 Exponential 4
Top-oil

Error
20 4
10 0
10 0
0 -4
0 100 200 300 400 500 600 700 800 900 1000
0 -4
0 100 200 300 400 Time500
(min) 600 700 800 900 1000
Time (min)
Figure 7. TOT based on different thermal models for TX1.
Figure
Figure 7. 7.
TOTTOT based
based onon different
different thermal
thermal modelsfor
models forTX1.
TX1.

90 48
90 48
80 40
80 Measured 40
70 MeasuredModel 32
(°C)(°C)

Thermal
(°C)(°C)

70 Exponential TE1
Thermal Model 24 32
Temperature

60
Temperature

Thermal Model TE2

TE1
Exponential
Temperature

60 TE2 24
Temperature

50 IEC
TE2Exponential 16
Thermal Model
50 TE2 TE1 Clause 7 IEEE Clause 7
IEC Exponential 16
40 TE1 Clause 7 IEEE Clause 7 8
Error
40 Error 8
Top-oil

Error

30 0
Top-oil

Error

30 0
20 -8
20 -8
10 -16
10 -16
0 -24
0 100 200 300 400 500 600 700 800 900 1000
0 -24
0 100 200 300 400 Time500
(min) 600 700 800 900 1000
Time (min)
Figure
Figure 8. 8.
TOTTOT based
based onon different
different thermal
thermal models
models forfor TX2.
TX2.
Figure 8. TOT based on different thermal models for TX2.
 Energies 2017, 10, 1843 11 of 15
Energies 2017, 10, 1843 11 of 15
Energies
Energies2017,
2017,10,
10,1843
1843 1111ofof1515

100 64
100 Measured 64
10090 Measured 64
Thermal Model 56
Measured
90 Thermal Model 56
9080 TE1 Model
Thermal 5648

(°C)
TE1

(°C)
80 TE2 48
(°C)
TE1

(°C)
8070
Top-oil Temperature (°C) TE2 4840
IEC Exponential

Error Temperature (°C)

Temperature
70 Exponential TE2 40

Temperature
IEC Exponential
Temperature
7060 Exponential IEEE Clause 7 4032

Temperature
IEC Exponential
IEEE Clause 7
60 Exponential
Clause 7 Error 32
6050 IEEE
ErrorClause 7 3224
Clause 7
50 Clause TE1 Error 24
5040 TE2 7TE1 2416
TE2 TE1 Thermal Model
40 16
Top-oil

TE2 Thermal Model

Error
4030 168
Top-oil

Thermal Model

Error
30 8
3020 80
20 0
2010 0 -8
10 -8
10 0 -8-16
00 100 200 300 400 500 600 700 800 900 1000
-16
0 0 100 200 300 400 Time
500(min) 600 700 800 900 1000
-16
0 100 200 300 400 Time
500(min) 600 700 800 900 1000
Time (min)
Figure 9. TOT based on different thermal models for TX3.
Figure 9. TOT based on different thermal models for TX3.
TOTbased
Figure9.9.TOT
Figure basedonondifferent
differentthermal
thermalmodels
modelsfor
forTX3.
TX3.
100 80
100 Measured 80
10090 Measured 8070
90 Thermal Model 70
Measured
9080 Thermal
TE1 Model 70
Thermal Model 60
(°C)

TE1

(°C)
80 TE2 60
(°C)

TE1

(°C)
8070 TE2 60
IEC Exponential 50
Top-oil Temperature (°C)

Error Temperature (°C)

TE2
Temperature

70 50

Temperature
IEC Exponential
Temperature

7060 IEEE Clause 7 5040

Temperature
IEC Exponential
IEEE Clause 7
60 Exponential Error 40
6050 Exponential Thermal Model IEEE
ErrorClause 7 4030
TE2
Exponential Thermal Model
50 TE2 Error 30
5040 TE2 TE1 Thermal Model 3020
Clause 7
40 TE1 Clause 7 20
Top-oil

TE1

Error
4030 Clause 7 2010
Top-oil

Error
30 10
3020 100
20 0
2010 0 -10
10 -10
10 0 -10
-20
00 100 200 300 400 500 600 700 800 900 1000
-20
0 0 100 200 300 400 Time500(min) 600 700 800 900 1000
-20
0 100 200 300 400 Time500(min) 600 700 800 900 1000
Time (min)
Figure 10. TOT based on different thermal models for TX4.
Figure 10.
Figure TOT based
10. TOT based on
on different
different thermal
thermal models
models for
for TX4.
TX4.
Figure 10. TOT based on different thermal models for TX4.
80 28
80 Measured 28
80 Measured 28
70 Thermal Model 24
Measured
70 Thermal
TE1 ModelModel 24
70 Thermal 24
(°C)

TE1
(°C)

60 TE2 20
(°C)

TE1
(°C)

60 TE2 20
Top-oil Temperature (°C)

IEC Exponential
Error Temperature (°C)

60 TE2 20
Temperature

Temperature

IEC Exponential
Temperature

50 IEEE Clause 7 16
Temperature

50 Clause 7 IEC Exponential

IEEE Clause 7 16
50 Clause 7 Error
IEEE 16
40 Exponential
Clause 7 ErrorClause 7 12
40 Exponential Error 12
Exponential Thermal Model
TE2 Thermal
40 TE2 Thermal Model 12
30 TE1Model 8
Top-oil

TE2
Error

30 TE1 8
Top-oil

Error

30 TE1 8
20 4
20 4
20 4
10 0
10 0
10 0
0 -4
00 100 200 300 400 500 600 700 800 900 1000
-4
0 0 100 200 300 400 Time500(min) 600 700 800 900 1000
-4
0 100 200 300 400 Time500(min) 600 700 800 900 1000
Time (min)
Figure 11.
Figure TOT based
11. TOT based on
on different
different thermal
thermal models
models for
for TX5.
TX5.
Figure 11. TOT based on different thermal models for TX5.
Figure 11. TOT based on different thermal models for TX5.
 Energies2017,
Energies 2017,10,
10,1843
1843 12of
12 of15
15
Energies 2017, 10, 1843 12 of 15

80 56
80 Measured 56
Measured
70 Thermal Model 48
70 Thermal Model 48
TE1
TE1
C)C)

C)C)
60 TE2 40
Temperature(°(°

Temperature(°(°
60 TE2 40
IEC Exponential
IEC Exponential 32
Top-oilTemperature

50 IEEE Clause 7

ErrorTemperature
50 Clause 7 IEEE Clause 7 32
Clause 7 Error
40 Exponential Error 24
40 Exponential 24
TE1 Thermal Model
30 TE1 Thermal Model 16
30 TE2 16
Top-oil

Error
TE2
20 8
20 8
10 0
10 0
0 -8
0 0 100 200 300 400 500 600 700 800 900 1000
-8
0 100 200 300 400 Time500(min) 600 700 800 900 1000
Time (min)

Figure12.
Figure 12. TOT
TOTbased
basedon
ondifferent
differentthermal
thermalmodels
modelsfor
forTX6.
TX6.
Figure 12. TOT based on different thermal models for TX6.

90 28
90 Measured 28
80 Measured
Thermal Model 24
80 Thermal Model 24
TE1
70 TE1 20
C)C)

C)C)
70 TE2 20
Temperature(°(°

Temperature(°(°
TE2
60 IEC Exponential 16
IEC Exponential 16
Top-oilTemperature

60 Exponential IEEE Clause 7

ErrorTemperature
Exponential IEEE Clause 7
50 TE1 Error 12
50 TE1 Thermal Model Error 12
40 Thermal Model 8
40 TE2 8
TE2 Clause 7
30 Clause 7 4
Top-oil

Error
30 4
20 0
20 0
10 -4
10 -4
0 -8
0 0 100 200 300 400 500 600 700 800 900 1000
-8
0 100 200 300 400 Time500(min) 600 700 800 900 1000
Time (min)

Figure 13. TOT

Figure13. TOTbased
basedon
ondifferent
differentthermal
thermalmodels
modelsfor
forTX7.
TX7.
Figure 13. TOT based on different thermal models for TX7.

4.2. Step
4.2. Step Loading
Loading
4.2. Step Loading
The performance
The performance of of the
the TOT
TOT thermal
thermal model
model isis also
also tested
tested on
on step
step loading
loading ofof TX8
TX8 as
as shown
shown inin
The performance of the TOT thermal model is also tested on step loading of TX8 as shown in
Figure14.
Figure 14.The
Thesimulated
simulatedTOT TOT based
based ononTOTTOT thermal
thermal andand
TE1TE1 models
models are quite
are quite closeclose to measured
to measured TOT.
Figure 14. The simulated TOT based on TOT thermal and TE1 models are quite close to measured
TOT.
The The simulated
simulated TOT based TOT on based
bothon TE2both
andTE2 and Exponential
Exponential models overshoot
models slightly slightly overshoot
once the once the
loading
TOT. The simulated TOT based on both TE2 and Exponential models slightly overshoot once the
loading increases
increases from 0 tofrom 0 to and
1.0 p.u. 1.0 p.u.
0.6 toand1.50.6
p.u.to On
1.5 p.u. On the
the other otherthe
hand, hand, the simulated
simulated TOT based TOTon based
the
loading increases from 0 to 1.0 p.u. and 0.6 to 1.5 p.u. On the other hand, the simulated TOT based
on the 7Clause
Clause model 7ismodel
slightlyislower
slightly
thanlower than measured
measured TOT during TOTtheduring the same
same loading loading
period. period. A
A significant
on the Clause 7 model is slightly lower than measured TOT during the same loading period. A
significantbetween
deviation deviation between
simulated TOTsimulated
based onTOT based onmodel
Exponential Exponential model TOT
and measured and measured TOT
is found once theis
significant deviation between simulated TOT based on Exponential model and measured TOT is
found once
loading the loading
decreases from 1.5 decreases
to 0.3 p.u.from
The1.5 to 0.3 p.u.
maximum The for
errors maximum
the TOTerrors
thermal,for TE1,
the TOT
TE2,thermal, TE1,
Exponential,
found once the loading decreases from 1.5 to 0.3 p.u. The maximum errors for the TOT thermal, TE1,
TE2,Clause
and Exponential,
7 models andare 5.77 ◦7C,
Clause 5.70 ◦ C,
models 5.77◦ °C,
are4.22 5.70◦°C,
C, 7.70 4.22 7.76
C, and ◦ C, while
°C, 7.70 °C, andthe7.76
RMS°C,errors
while are
the
TE2, ◦Exponential, and Clause 7 models are 5.77 °C, 5.70 °C, 4.22 °C, 7.70 °C, and 7.76 °C, while the
RMS C,
2.02 2 ◦ C,are
errors 1.91 ◦ C,°C,
2.02 ◦ C,1.91
2 °C,
4.78 and°C, ◦ C,°C,
2.844.78 and 2.84 °C, respectively.
respectively.
RMS errors are 2.02 °C, 2 °C, 1.91 °C, 4.78 °C, and 2.84 °C, respectively.
Energies 2017, 10,
Energies 2017, 10, 1843
1843 13 of
13 of 15
15

100 40
Measured
90 Thermal Model 35
80 TE1 30
Top-oil Temperature (°C)

Error Temperature (°C)

TE2
70 IEC Exponential 25
60 IEEE Clause 7 20
Error
50 15
Exponential TE1
40 TE2 Thermal Model Clause 7 10
30 5
20 0
10 -5
0 -10
0 100 200 300 400 500 600 700 800 900 1000
Time (min)

Figure 14. TOT based on different thermal models for TX8.

4.3. Discussions
Under constant loadings of seven transformers, the highest maximum maximum and RMS errors for the
TOT thermal model are 6.66 ◦°C C and 2.76 ◦°C,C, which are lower than Exponential and Clause 7 models
where the highest maximum errors are 8.37 ◦°C C and 12.92 ◦°C C while the highest RMS error are 4.54 ◦°C C
and 5.33 ◦
5.33 °C, respectively.
C, respectively. TheThe performance
performance of TE1
of TE1 and and TE2 models
TE2 models are slightly
are slightly better
better than TOTthan TOT
thermal
thermal modelthe
model where where themaximum
highest highest maximum 6.32 ◦are
errors areerrors C and6.325.03 ◦ C while
°C and 5.03 °C thewhile
highesttheRMS
highest
errorRMS
are
error◦
2.89 are 2.892.41
C and ◦
°C andC. 2.41 °C. Meanwhile,
Meanwhile, under
under step step loading,
loading, the TOTthethermal
TOT thermal
modelmodelperformsperforms
betterbetter
than
than Exponential
Exponential and Clause
and Clause 7 models,
7 models, and comparable
and comparable with with TE1TE2
TE1 and and models.
TE2 models. The model
The TE1 TE1 model is
is able
able to represent the TOT better than the TOT thermal model owing
to represent the TOT better than the TOT thermal model owing to its comprehensive consideration to its comprehensive
consideration
on the viscosity.onHowever,
the viscosity.
due However, due to
to the minimal the minimal
information information
requirement byrequirement
the TOT thermalby the TOT
model,
thermal model, the
the performance could performance
be considered could be considered
as reasonable as reasonable
as alternative approachas alternative
to determine approach
the TOT toof
determine
transformers.theFurthermore,
TOT of transformers. Furthermore,
the TOT thermal model is the TOT
able thermal transformers
to evaluate model is able to different
with evaluate
transformers
cooling modeswith different
since cooling
the oil time modesτoil
constant, since the oil time
is calculated constant,to𝜏the
according 𝑜𝑖𝑙 is calculated
respective according
cooling to
modes.
the respective cooling modes.
5. Conclusions
5. Conclusions
Based on the proposed pathway of the energy transfer concept and thermal-electrical analogy
through
Basedutilization of the nonlinear
on the proposed pathway thermal
of theresistance and lumped
energy transfer capacitance
concept method, an alternative
and thermal-electrical analogy
approachutilization
through to determine of the
theTOT is developed.
nonlinear thermalThe key component
resistance of the TOT
and lumped thermal model
capacitance method,is the
an
alternative approximation of the convection
approach to determine the TOTcoefficient,
is developed.h, proposed
The keybased on heatoftransfer
component the TOTtheory that
thermal
eliminates
model the alternative
is the requirementapproximation
of the viscosity.ofThe
theTOT thermal model
convection was tested
coefficient, on seven
h, proposed transformers
based on heat
with either
transfer ONAN
theory that or ONAF cooling
eliminates modes. Based
the requirement of theon the caseThe
viscosity. study,
TOTthethermal
TOT thermal model
model was can
tested
represent the measured TOT quite well where the cooling
highestmodes.
maximum and ◦C
on seven transformers with either ONAN or ONAF Based on RMS errors
the case arethe
study, 6.66
TOT
◦ C, respectively. Under step loading, the performance of TOT thermal model is comparable
and 2.76 model
thermal can represent the measured TOT quite well where the highest maximum and RMS
with
errorsother types°C
are 6.66 ofand
thermal
2.76 models where the Under
°C, respectively. simulated
stepTOT is quite
loading, theclose to measured
performance TOT thermal
of TOT and the
maximum and RMS errors are 5.77 ◦ C and 2.02 ◦ C. Overall, the performance of the TOT thermal model
model is comparable with other types of thermal models where the simulated TOT is quite close to
is comparable
measured TOTwithand the
the TE1 and TE2
maximum andmodels
RMS and
errorsbetter than°C
are 5.77 theand
Exponential and Clause
2.02 °C. Overall, 7 models.
the performance
of the TOT thermal model is comparable with the TE1 and TE2 models and better than the
Acknowledgments: The authors would like to thank Ministry of Education, Universiti Pertahanan Nasional
Exponential
Malaysia andand Clause Putra
Universiti 7 models.
Malaysia for funding under Putra IPB schemes, (GP-IPB/2014/9440801).
Special thanks to Malaysia Transformer Manufacturing Sdn. Bhd. for the technical support.
Acknowledgments: The authors would like to thank Ministry of Education, Universiti Pertahanan Nasional
Author Contributions: The research study was carried out successfully with contribution from all authors.
Malaysia
The main and Universiti
research Putra Malaysia
idea, simulation worksfor
andfunding under
manuscript Putra IPB were
preparation schemes, (GP-IPB/2014/9440801).
contributed Special
by Muhammad Hakirin
thanks
Roslan.to Malaysia
Norhafiz Transformer
Azis contributedManufacturing Sdn.preparation
on the manuscript Bhd. for the technical
and researchsupport.
idea. Mohd Zainal Abidin Ab Kadir,
Jasronita Jasni and Zulkifli Ibrahim assisted on finalizing the research work and manuscript. Azalan Ahmad
Author Contributions: The research study was carried out successfully with contribution from all authors. The
main research idea, simulation works and manuscript preparation were contributed by Muhammad Hakirin
Roslan. Norhafiz Azis contributed on the manuscript preparation and research idea. Mohd Zainal Abidin Ab
Energies 2017, 10, 1843 14 of 15

gave several suggestions from the industrial perspectives. All authors revised and approved the publication of
the paper.
Conflicts of Interest: The authors declare no conflict of interest.

Nomenclature
A Area
C Parameter with unit W/m2 ·K(1+n)
Coil Oil thermal capacitance
Cwdg Winding thermal capacitance
f1 Increase exponent function of Top-oil temperature
h Convection coefficient
I Load current
K Load factor
n A constant
qcu Heat generated by load losses
qfe Heat generated by no load losses
R Ratio of load losses at rated current to no load losses
Rwdg Winding thermal resistance
R paper Paper thermal resistance
R paper−oil Nonlinear paper to oil thermal resistance
Roil −tank Nonlinear oil to tank thermal resistance
Rtank/radiator Tank/radiator thermal resistance
Roil Nonlinear oil thermal resistance
Roil,rated Nonlinear oil thermal resistance at rated
Rth−hs Hot-spot thermal resistance
θa Ambient temperature
θoil Top-oil temperature
θhs Hot-spot temperature
θoil,i Initial value of Top-oil temperature
θoil, calculated Simulated value of Top-oil temperature
θoil, measured Measured value of Top-oil temperature
∆θoil,rated Top-oil temperature rise at rated
∆θ TO,U Ultimate Top-oil temperature rise
∆θ TO,i Initial Top-oil temperature rise
τoil Oil time constant
µ pu Per unit value of oil viscosity

References
1. Standard, I. 60076-7 Power Transformer—Part 7:‘Loading Guide for Oil-Immersed Power Transformers’;
IEC: Geneva, Switzerland, 2005.
2. IEEE. IEEE Guide for Loading Mineral-Oil-Immersed Transformers; IEEE Std C57.91-1995; IEEE: Piscataway
Township, NJ, USA, 1996.
3. Swift, G.; Molinski, T.S.; Lehn, W. A Fundamental Approach to Transformer Thermal Modeling. I. Theory
and Equivalent Circuit. IEEE Trans. Power Deliv. 2001, 16, 171–175. [CrossRef]
4. Swift, G.; Molinski, T.S.; Bray, R.; Menzies, R. A fundamental approach to transformer thermal modeling. II.
Field verification. IEEE Trans. Power Deliv. 2001, 16, 176–180. [CrossRef]
5. Susa, D.; Lehtonen, M.; Nordman, H. Dynamic thermal modelling of power transformers. IEEE Trans.
Power Deliv. 2005, 20, 197–204. [CrossRef]
6. Susa, D.; Lehtonen, M. Dynamic thermal modeling of power transformers: Further development-part I.
IEEE Trans. Power Deliv. 2006, 21, 1961–1970. [CrossRef]
7. Susa, D.; Lehtonen, M. Dynamic thermal modeling of power transformers: Further development-part II.
IEEE Trans. Power Deliv. 2006, 21, 1971–1980. [CrossRef]
Energies 2017, 10, 1843 15 of 15

8. Chen, W.G.; Pan, C.; Yun, Y.X. Power transformer top-oil temperature model based on thermal–electric
analogy theory. Eur. Trans. Electr. Power 2009, 19, 341–354.
9. Iskender, I.; Mamizadeh, A. An improved nonlinear thermal model for MV/LV prefabricated oil-immersed
power transformer substations. Electr. Eng. 2011, 93, 9–22. [CrossRef]
10. Mamizadeh, A.; Iskender, I. Analyzing and Comparing Thermal Models of Indoor and Outdoor
Oil-Immersed Power Transformers. In Proceedings of the IEEE Bucharest Power Tech, Bucharest, Romania,
28 June–2 July 2009; pp. 1–8.
11. Su, X.; Chen, W.; Pan, C.; Zhou, Q.; Teng, L. A simple thermal model of transformer hot spot temperature
based on thermal-electrical analogy. In Proceedings of the International Conference on High Voltage
Engineering and Application (ICHVE), Shanghai, China, 17–20 September 2012; pp. 492–495.
12. Ben-gang, W.; Chen-zhao, F.; Hong-lei, L.; Ke-jun, L.; Yong-liang, L. The improved thermal-circuit model
for hot-spot temperature calculation of oil-immersed power transformers. In Proceedings of the IEEE
Information Technology, Networking, Electronic and Automation Control Conference, Chongqing, China,
20–22 May 2016; pp. 809–813.
13. Zhou, L.; Wu, G.; Yu, J.; Zhang, X. Thermal overshoot analysis for hot-spot temperature rise of transformer.
IEEE Trans. Dielectr. Electr. Insul. 2007, 14, 1316–1322. [CrossRef]
14. Li, R.; Yi, Y.Y. Study on Prediction of Top Oil Temperature for Transformers Based on Bayesian Network
Model. In Proceedings of the MATEC Web of Conferences, Hong Kong, China, 11–12 June 2016.
15. Yi, Y.Y.; Li, R. Application of Gray Neural Network Combined Model in Transformer Top-oil Temperature
Forecasting. In Proceedings of the MATEC Web of Conferences, Hong Kong, China, 11–12 June 2016.
16. He, Q.; Si, J.; Tylavsky, D.J. Prediction of top-oil temperature for transformers using neural networks.
IEEE Trans. Power Deliv. 2000, 15, 1205–1211. [CrossRef]
17. Taghikhani, M.A. Power transformer top oil temperature estimation with GA and PSO methods.
Energy Power Eng. 2012, 4, 41–46. [CrossRef]
18. Assunção, T.; Silvino, J.; Resende, P. Transformer top-oil temperature modeling and simulation. Trans. Eng.
Comput. Technol. 2006, 15, 240–245.
19. Radakovic, Z.; Feser, K. A new method for the calculation of the hot-spot temperature in power transformers
with ONAN cooling. IEEE Trans. Power Deliv. 2003, 18, 1284–1292. [CrossRef]
20. Tang, W.; Wu, Q.; Richardson, Z. Equivalent heat circuit based power transformer thermal model. IEEE Proc.
Electr. Power Appl. 2002, 149, 87–92. [CrossRef]
21. Tang, W.; Wu, Q.; Richardson, Z. A simplified transformer thermal model based on thermal-electric analogy.
IEEE Trans. Power Deliv. 2004, 19, 1112–1119. [CrossRef]
22. Cui, Y.; Ma, H.; Saha, T.; Ekanayake, C.; Martin, D. Moisture-Dependent Thermal Modelling of Power
Transformer. IEEE Trans. Power Deliv. 2016, 31, 2140–2150. [CrossRef]
23. Bergman, T.L.; Incropera, F.P.; Lavine, A.S. Fundamentals of Heat and Mass Transfer; John Wiley & Sons:
New York, NY, USA, 2011.
24. Perez, J. Fundamental principles of transformer thermal loading and protection. In Proceedings of the
63rd Annual Conference for Protective Relay Engineers, College Station, TX, USA, 29 March–1 April 2010;
pp. 1–14.
25. Xiao, B.; Wang, W.; Fan, J.; Chen, H.; Hu, X.; Zhao, D.; Zhang, X.; Ren, W. Optimization of the fractal-like
architecture of porous fibrous materials related to permeability, diffusivity and thermal conductivity.
Fractals 2017. [CrossRef]
26. Xiao, B.; Chen, H.; Xiao, S.; Cai, J. Research on relative permeability of nanofibers with capillary pressure
effect by means of Fractal-Monte Carlo technique. J. Nanosci. Nanotechnol. 2017, 17, 6811–6817. [CrossRef]
27. Incropera, F.; DeWitt, D. Fundamentals of Heat and Mass Transfer; Wiley: New York, NY, USA, 1996.

© 2017 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access
article distributed under the terms and conditions of the Creative Commons Attribution
(CC BY) license (http://creativecommons.org/licenses/by/4.0/).