Towards Online Ageing Detection in TO - A Review
Towards Online Ageing Detection in TO - A Review
Towards Online Ageing Detection in TO - A Review
net/publication/364366006
CITATIONS READS
0 29
4 authors, including:
Arshad Arshad
Glasgow Caledonian University
51 PUBLICATIONS 434 CITATIONS
SEE PROFILE
Some of the authors of this publication are also working on these related projects:
Flashover Characteristics of Silicone Rubber Materials under Various Contaminated Conditions for Application in Outdoor Polymeric Insulators View project
All content following this page was uploaded by Ugochukwu Elele on 19 October 2022.
Abstract: Transformers play an essential role in power networks, ensuring that generated power gets
to consumers at the safest voltage level. However, they are prone to insulation failure from ageing,
which has fatal and economic consequences if left undetected or unattended. Traditional detection
methods are based on scheduled maintenance practices that often involve taking samples from in situ
transformers and analysing them in laboratories using several techniques. This conventional method
exposes the engineer performing the test to hazards, requires specialised training, and does not
guarantee reliable results because samples can be contaminated during collection and transportation.
This paper reviews the transformer oil types and some traditional ageing detection methods, including
breakdown voltage (BDV), spectroscopy, dissolved gas analysis, total acid number, interfacial tension,
and corresponding regulating standards. In addition, a review of sensors, technologies to improve
the reliability of online ageing detection, and related online transformer ageing systems is covered in
this work. A non-destructive online ageing detection method for in situ transformer oil is a better
alternative to the traditional offline detection method. Moreover, when combined with the Internet of
Things (IoT) and artificial intelligence, a prescriptive maintenance solution emerges, offering more
advantages and robustness than offline preventive maintenance approaches.
Figure1.1.Degradation
Figure Degradation Mechanism
Mechanism of Transformer
of Transformer Oil. Oil.
Transformer
2. Review oil insulation failure has rippling consequences on the primary purpose of
Methodology
power generation, the safety of personnel and the work environment, as well as enormous
economic consequences. Consequently, their volume, purity, and reliability cannot be
compromised. Ageing in transformer oil is primarily evidenced by interfacial tension
and acidity.
The transformer ageing detection method is classified as either intrusive or non-
intrusive, destructive or non-destructive, and offline or online. Intrusive detection tech-
niques make contact with the transformer oil as opposed to non-intrusive methods. De-
structive methods alter (in the short or long run) the transformer oil properties being
measured as opposed to non-destructive methods. The online detection method involves
live ageing detection of the transformer oil while operating as opposed to offline ageing
detection, which only involves sample collection for laboratory analysis and interpretation.
In-service transformer oil ageing cannot be prevented [11], but it can be maintained at
a lower rate through prescriptive maintenance steps using online sensors, thus avoiding
catastrophic effects/outages. The report by [18] points out that aged transformer liquids
can be treated by drying out, degassing, reclamation, refining and reconditioning. This
will ensure that the in-service transformer functions effectively up to its expected lifespan
(approx. 40 years) [11].
2. Review Methodology
This review was conducted with the research question—how can ageing detection
in transformer oil systems be improved? To effectively answer this question, this paper
Sensors 2022, 22, 7923 3 of 25
focused on nine (9) research themes summarised in Table 1. Figures 2 and 3 show the
statistical plots, and Tables 2 and 3 summarise the screening methodology and thematic
references, respectively.
Themes
1. Review of various transformer oil types
2. Review of transformer ageing characterisation techniques
3. Review of various ageing classification techniques
4. Accelerated thermal ageing technique
5. Fiber-Optic Sensor
6. Superhydrophobicity and online ageing detection
7. Machine learning models for online ageing detection
Sensors 2022, 22, x FOR PEER REVIEW 8. IoT and online ageing 4 of 26
Sensors 2022, 22, x FOR PEER REVIEW 4 of 26
9. Review of related systems
Statistical Plot
Statistical Plot of
of Articles
Articles
Conference Paper
Conference Paper
Generic
Generic
Journal
Journal
Patent
Patent Count
Count
Technical Paper
Technical Paper
00 10
10 20
20 30
30 40
40 50
50 60
60
Technical Paper
Technical Paper Patent
Patent Journal
Journal Generic
Generic Conference Paper
Conference Paper
Count
Count 3
3 00 56
56 26
26 19
19
Figure 2.
Figure 2. Statistical Plot
Plot of
of Articles.
Articles.
Figure 2.Statistical
Statistical Plot of Articles.
Year Plot
Year Plot
50
50
40
40
30
30
20
20
10 Count
Count
10
00
Before
Before Year
Year Year
Year Year
Year Year
Year Year
Year Year
Year Year
Year Year
Year
2015
2015 2015
2015 2016
2016 2017
2017 2018
2018 2019
2019 2020
2020 2021
2021 2022
2022
Count
Count 41
41 55 22 66 10
10 77 77 13
13 13
13
Figure 3.
Figure 3. Year
Year Distribution
Distribution of
of Articles.
Articles.
Figure 3. Year Distribution of Articles.
This statistic
This statistic did
did not
not include
include reference
reference sources
sources that
that did
did not
not uniquely
uniquely contribute
contribute toto
answering the research question. Since 2019, there has been significant research
answering the research question. Since 2019, there has been significant research interest interest
in this
in this subject
subject area
area (see
(see Figure
Figure 3);
3); however,
however, very
very few
few works
works have
have been
been reported
reported onon related
related
systems (see Table 3; review of related systems), emphasising the need for additional
systems (see Table 3; review of related systems), emphasising the need for additional re- re-
search on the online ageing detection of transformer
search on the online ageing detection of transformer oil. oil.
3. Types
3. Types of
of Transformer
Transformer OilOil
Mineral oil
Mineral oil and
and ester
ester oil
oil are
are two
two common
common examples
examples ofof transformer
transformer insulating
insulating liquids,
liquids,
with mineral oil having enjoyed over a century of use (compared to ester oil) and
with mineral oil having enjoyed over a century of use (compared to ester oil) and made made
Sensors 2022, 22, 7923 4 of 25
Themes References
1. Transformer oil types [4,18–27]
2. Review of transformer ageing characterisation techniques [7,16,17,21,28–49]
3. Review of various ageing classification techniques [29,46]
4. Accelerated thermal ageing technique [7,50–54]
5. Sample ageing detection sensors [32,38,55–61]
6. An overview of superhydrophobicity [62–78]
7. Machine Learning and Uncertainty Quantification [79–85]
8. IoT and online ageing [86–96]
9. Review of related systems [9,10,57]
This statistic did not include reference sources that did not uniquely contribute to
answering the research question. Since 2019, there has been significant research interest in
this subject area (see Figure 3); however, very few works have been reported on related
systems (see Table 3; review of related systems), emphasising the need for additional
research on the online ageing detection of transformer oil.
Other alternative oils include edible coconut oil, Karanji oil [26], silicon oil, castor oi
Figure 4. Reversible
and sesame esterification
oil, which showedreaction [25].
acceptable
limits in terms of pour point, acidity, DDF and
Figure 4. Reversible esterification reaction [25].
BDV, and which can be further improved by purification [27].
Other alternative oils include edible coconut oil, Karanji oil [26], silicon oil, castor oil
and Other
sesamealternative
oil, whichoils include
showed edible coconut
acceptable oil,
limits in Karanji
terms oil [26],
of pour silicon
point, oil, castor
acidity, DDF andoil
and
4. sesame oil,
Transformer which
Oil showed
Ageing acceptable limits
Characterisation in terms of
Techniques
BDV, and which can be further improved by purification [27]. pour point, acidity, DDF and
BDV, and which can be further improved by purification [27].
Insulation ageing is the gradual decrease in the dielectric strength of an insulator in
4. Transformer Oil Ageing Characterisation Techniques
operation to Oil
4. Transformer
complete
Ageing
the insulation breakdown
Characterisation Techniques
from electrical and environmenta
Insulation ageing is the gradual decrease in the dielectric strength of an insulator
stresses. Ageing affects transformer oil’s chemical and physical properties, and severa
Insulation
in operation toageing
complete is the
thegradual
insulationdecrease in the dielectric
breakdown strength
from electrical andof an insulator in
environmental
ageing techniques are used to understand this. This section details some of the curren
operation to complete the insulation breakdown from electrical
stresses. Ageing affects transformer oil’s chemical and physical properties, and several and environmental
ageing detection
stresses. Ageing affectsmethods and their
transformer oil’scorresponding
chemical and regulating
physical standards.
properties,
ageing techniques are used to understand this. This section details some ofand the several
current
ageing
ageing techniques are usedand
detection methods to their
understand this. This
corresponding section details
regulating some of the current
standards.
4.1. Breakdown
ageing detection Voltage
methods(BDV) and theirTestcorresponding regulating standards.
4.1. Breakdown Voltage testing
This electrical (BDV) Testmethod is used to ascertain the offline age of a transformer oi
4.1. Breakdown
sampleThis(see Voltage
electrical
Figure (BDV)
testing
5 [28]). Test
method is used to ascertain
The breakdown voltage theisoffline age of aattransformer
the voltage which the oil oil sampl
sampleThis(see Figure
electrical 5 [28]).
testing The
method breakdown
is used tovoltage is
ascertain the
the voltage
offline
becomes conductive (evidenced by spark) [17], usually reported in [kV/mm] units. It i at
age which
of a the oil
transformer sampleoil
becomes(see
sample conductive
Figure 5of(evidenced
[28]). by spark) [17], usually reported in [kV/mm] units.oil It is also
also the measure the The breakdown
ability voltage
of insulation toiswithstand
the voltage at which the
electrical stress sample
[29]. The trans
the measure
becomes of the
conductive ability of
(evidenced insulation to withstand electrical stress [29]. The transformer
former
oil isthe
oil
placed
is placed
in a test
a testby
incontaining
cellability
spark)
cell [17], usually
containing
hemispherical
reported inelectrodes.
hemispherical
electrodes. The BDV
[kV/mm] units. The ItBDVis test i
also measure of the of insulation to withstand electrical stresstest
[29].is The
guided by
trans-
guided
the IECoilby the
60156 IEC
and 60156standards.
IS in
6792 and IS 6792 standards.
former is placed a test cell containing hemispherical electrodes. The BDV test is
guided by the IEC 60156 and IS 6792 standards.
According to [17], in the initial phase of electric field application, thermal agitation
occurs in the oil, leading to the formation of microscopic cavities. Further application of
Sensors 2022, 22, 7923 6 of 25
the electric field leads to gas development and microbubble formation. Moreover, further
exposure results in a gas production rate exceeding bubble formation and, consequently,
the production of disruptive discharge. The voltage at which this occurs is called the
breakdown voltage or the disruptive discharge voltage value [30]. An example instrument
used in characterising oil samples with the BDV test is the BA75 analyser. Aged transformer
oil negatively correlates with the BDV value, but a low BDV value does not necessarily
imply ageing. BDV is a pointer to the severity levels of transformer oil impurity according
to BS EN 60422 [31].
The disruptive discharge indicates an insulation failure from a breakdown voltage
test instrument. The discharge completely bridges the insulation under the test, reducing
the voltage between the test electrodes to zero. According to IEC 60060-1 [30], disrup-
tive discharge may sometimes occur momentarily and is referred to as non-sustained
disruptive discharge.
The minimum acceptable BDV value is 30 kV/mm for transformers operating from
230 kV; 28 kV/mm for transformers rated between 69 kV and 230 kV; and 23 kV/mm for
those rated at most 69 kV [32].
Figure7.7.Acid
Figure Acid number
number titrator
titrator in GCU
in GCU Lab. Lab.
For the liquid–liquid interface which characterises the transformer oil, some of the
methods used to evaluate the IFT include the ring method (according to Du Noüy), plate
method, rod method, drop volume method, spinning drop method and pendant drop
method. Most IFT instruments mostly use the Du Noüy ring method. It involves slowly
lifting a platinum ring from a liquid interface, which relates directly to the interfacial
tension. Platinum surfaces are chemically inert, easy to clean, optimally wetted, and
generally form a zero contact angle. The platinum ring hangs parallel to the sample and
is allowed to sink into the liquid. As the ring is pulled upwards, the interfacial tension
generates a force on the ring, which is measured (the maximum force on the ring when
pulled out of the liquid) to evaluate the interfacial tension σ.
Mathematically:
Fmax
σ= ·C (3)
2πD
σ represents the interfacial tension, F is the force acting on the platinum ring, D is the
diameter of the ring and C is the correction factor (taking into account the additional
volume of liquid extracted together with the ring). Mathematically, C can be evaluated as:
r
γ
C = 0.725 + 4.014 × 10−4 × + 0.01287 (4)
∆ρ
γ represents the interfacial tension without correction in mN/m, and ∆ρ represents the
density difference between water and the insulating liquid at measuring temperature in
gcm−3 . The Du Noüy ring method is recommended by the ASTM D971 and BS EN 62961
standards. Automatic tensiometers calculate the σ and C values. According to BS IEC
62961 [47], automatic tensiometers can overcome the film destructive nature of the Du
Noüy ring method by automatically determining the maximum force value on the ring for
detachment from the film before detachment and reverse the platform movement promptly
prior to detachment without tearing the film. This allows for accurate, time-consistent
serial measurements. The ring method has some limitations, including the need for a
Sensors 2022, 22, 7923 10 of 25
Figure 8. Different
Figure 8. Different aged
aged samples
samples(L-R
(L-R3-Day,
3-Day,10-Day,
10-Day,15-Day,
15-Day,20-Day,
20-Day,and
and30-Day)
30-Day)[7].
[7].
AA better
better setup
setup that
that replicates
replicates aa power
power transformer
transformer more
more closely
closely was
was reported
reported by
by
Bouaicha, Fofana and Farzaneh [53]. Their setup included metallic catalysts (zinc,
Bouaicha, Fofana and Farzaneh [53]. Their setup included metallic catalysts (zinc, copper, copper,
aluminium
aluminium and and iron)
iron) to
to simulate
simulate the
the metallic
metallic components
components inin the
the transformer
transformer (windings
(windings
and
and core). Oil-impregnated pressboards were used to simulate the kraft
core). Oil-impregnated pressboards were used to simulate the kraft paper
paper insulation,
insulation,
and silica gel was rightly positioned to simulate the transformer breathing mechanism. The
and silica gel was rightly positioned to simulate the transformer breathing mechanism.
half-life rule could have been used to replicate the thermal degradation even more closely.
The half-life rule could have been used to replicate the thermal degradation even more
The ASTM D1934-95 standard provides the specification for oxidative ageing using two
closely. The ASTM D1934-95 standard provides the specification for oxidative ageing us-
procedures (without a metal catalyst and with a metal catalyst) [54].
ing two procedures (without a metal catalyst and with a metal catalyst) [54].
7. Cross-Capacitive and Fibre Optic Ageing Detection Sensors
7. Cross-Capacitive and Fibre Optic Ageing Detection Sensors
Machine learning IoT solutions will improve HV insulation condition monitoring [55]
Machine
and sensors learning
play IoT solutions
an essential will improve
role in generating data HV insulation
for machine condition
learning monitoring
diagnostics.
[55] and sensors play an essential role in generating data for machine learning diagnostics.
7.1. Cross-Capacitive Sensor
The authors of [56] reported the effectiveness of parallel plate capacitive sensors with
thin hydrophilic porous oxide metal film as moisture content detection sensors. However,
one of the setbacks highlighted with the parallel plate capacitive sensor is its instability due
to variations in the geometric parameters, such as a gap shift, bending of electrodes, or even
overlap of electrodes. Therefore, the cylindrical cross-capacitor proposed by Thompson
and Lampard is used to overcome these geometric issues. It comprises four (4) identical
cylindrical electrodes separated by gaps. The capacitance between the diagonal opposite
pair of electrodes is called the cross-capacitance.
Sensors 2022, 22, 7923 12 of 25
∈ × loge 2
Cc = ×l (8)
π
Cc represents the cross-capacitance; ∈ represents the dielectric permittivity; and l represents
the length of the electrodes.
In the simulation experiment (using ANSYS Maxwell Version 15.0) performed by [57],
the cross-capacitance value did not vary with changes in the structural design of the
cross-capacitor (circular or square electrode). Insignificant changes with different uniform
insulation gaps were also reported. The capacitance value depends on the length of
the electrode and the dielectric medium within the electrode. This is an improvement
over parallel plate capacitors that depend on the distance between the plates, the area of
plates and the permittivity of the dielectric. Furthermore, since the electrode lengths are
fixed, the capacitance depends solely on the oil dielectric. As a result, the cross-capacitive
sensor is more robust and suitable for transformer applications. It is also temperature-
independent [57].
Cross-capacitive sensors have been reported to be effective in measuring moisture
concentration, dielectric constant, resistivity, humidity and metal detection. However,
changes are usually made to the electrodes’ sensing film or geometric arrangement.
Rahman, Islam, Khera and Khan [56] designed a novel cross-capacitive sensor applica-
ble for contact measurements of moisture and 2-FAL. An AD 7150 capacitance-to-voltage
sensor was utilised to infer the capacitance of the dielectric medium. The results showed a
linear response to moisture and different 2-FAL concentration levels.
Figure 9.
Figure 9. Optical
Optical fibre
fibre classification.
classification.
An optical
Different fibre consists
sensitivity of theare
responses core, the cladding
expected based on andthe thechoice
coating. The corecomponent
of material, is the light
transmission
refractive indexareaandofmode
the fibre that can be ofFor
of propagation. either glass an
example, or plastic material.
experiment The cladding
conducted by [61]
provides
for ageinga monitoring
lower refractive of anindex at the liquid
industrial core interface
coolant to allow for
showed thata silicon
reflection within
fibres the
exhibit
core for
more light transmission
efficient behaviour than through
polymer thefibre
fibre(when
basedusedon the forprinciple of total internal
acidity monitoring) reflec-
in terms of
tion [59].and
stability Therefore, the cladding
repeatability. However, ensures
polymer a decrease in scattering
fibres have loss and offers
better sensitivity, additional
biocompatibility
protection
and bending to tolerance
the core. The [59].coating offers environmental and mechanical protection to the
Anfibre
optical optical
corefibre
and consists
is mainlyofmade the core,fromthe cladding
plastic [59]. and the coating. The core is the
light The
transmission area of of
general structure theanfibre
opticalthatfibre
can sensor
be of either glass orofplastic
is composed material.
an optical sourceThe(or
cladding provides
transmitter), a lower
the sensing refractive
element and index at thedetector.
the optical core interface to allow
The sensing forisadeveloped
area reflection
within the coreafor
by uncladding lightportion
choice transmission
of the fibrethrough
opticthe fibre
cable. The based on source
optical the principle of totala
can be either
internal reflection [59]. Therefore, the cladding ensures
light-emitting diode (LED) or a laser diode. LEDs are generally suited for multimode a decrease in scattering loss and
fi-
offers
bres andadditional protectionby
are characterised to low
the core.
power, The coating
less offers and
bandwidth environmental
a maximumand mechanical
throughput of
protection
1 Gbps. Laser to the optical
diode fibre core
sources offerand more is mainly
advantagesmadethan fromLED plastic [59]. at the expense of
sources
cost. The
Laser general
diodesstructure
are faster, of allow
an optical fibre sensor
single-mode is composed
or multimode of an optical
transmission source (or
(Fabry-Perot
transmitter), the sensing element and the optical detector. The
lasers) and can achieve a throughput of at least 10 Gbps. Distributed feedback (DFB) lasers sensing area is developed
by
anduncladding
vertical cavity a choice portion of the
surface-emitting fibre
lasers optic cable.
(VCSELs) do not The optical
enjoy the source can of
versatility beFabry–
either
aPerot
light-emitting diode (LED) or a laser diode. LEDs are generally
lasers as they are only suited for single-mode fibres and multimode fibres, respec- suited for multimode
fibres
tively.and are characterised by low power, less bandwidth and a maximum throughput of
1 Gbps. Laserfibre
Optical diode sources
sensors areoffer more advantages
classified as intrinsicthanandLED sources
extrinsic at the[59].
sensors expense of cost.
Intrinsic fi-
Laser diodes are faster, allow single-mode or multimode transmission
bre optic sensors have a sensing area within the fibre optic cable (the fibre is the sensor), (Fabry-Perot lasers)
and
whichcanisachieve
usually aachieved
throughput of at least 10
by uncladding Gbps. Distributed
a portion feedback
of the fibre optic (DFB)
cable. lasers and
The developed
vertical cavity surface-emitting lasers (VCSELs) do not enjoy
sensing area makes direct contact with the material to be measured, and a property of the versatility of Fabry–Perot
lasers
light isasmodulated,
they are only suited
which for single-mode
infers the input. The fibres
fibreand multimode
optic cable is afibres,
light respectively.
waveguide for
Optical fibre sensors are classified as intrinsic and
extrinsic fibre optic sensors to the external sensor. The intrinsic sensing mode extrinsic sensors [59]. Intrinsic
is more
fibre optic sensors have a sensing area within the fibre optic cable (the fibre is the sensor),
common with most optical fibre sensors. For the intrinsic sensor configuration, the meas-
which is usually achieved by uncladding a portion of the fibre optic cable. The developed
urand alters light wave parameters, such as wavelength, phase, intensity and polarisation,
sensing area makes direct contact with the material to be measured, and a property of light
which can be used to infer the measurand quantity. For this reason, fibre optic-based sen-
is modulated, which infers the input. The fibre optic cable is a light waveguide for extrinsic
sors (OFS) are either phase-modulated, intensity-modulated, wavelength-modulated or
fibre optic sensors to the external sensor. The intrinsic sensing mode is more common with
polarisation-modulated [59].
most optical fibre sensors. For the intrinsic sensor configuration, the measurand alters
Sensors 2022, 22, 7923 14 of 25
light wave parameters, such as wavelength, phase, intensity and polarisation, which can
be used to infer the measurand quantity. For this reason, fibre optic-based sensors (OFS)
are either phase-modulated, intensity-modulated, wavelength-modulated or polarisation-
modulated [59].
Intensity-modulated OFSs are common for intrinsic fibre optic configurations. The
sensing principle is based on the attenuation of the optical signal induced by the monitored
parameter [59]. Following the loss of signal power, which characterises intensity-modulated
OFSs, higher optical signals are recommended (multimode fibres). Intensity-modulated
OFSs, although simple to implement and adaptable for distributed instrumentation, require
standardisation and source referencing to impede false interpretations. Optical intensity
attenuation is mainly achieved through microbending and evanescent field absorption.
Microbends are bends too small to be detected by the human eye when external pressure is
applied to a defined portion of the fibre optic sensing area. Optical intensity attenuation
occurs during microbending when propagated guided signals radiate into non-guided
modes. According to [59], the microbend affects the critical angle that allows the propagated
signal to remain in the guided mode. Microbend OFSs have found applications in strain,
pressure, temperature, vibration, humidity and pH measurements [59]. Evanescent field
sensors rely on intensity loss due to the uncladding of a portion of the fibre optic cable.
This ensures that the total internal reflection does not lead to signal loss which depends on
the refractive index of the sensing medium. For transformer oil applications, the refractive
index of aged samples varies proportionally to the age of the oil samples. Polishing the
sensing area (superhydrophobicity) of evanescent field sensors can free the sensing area
from impurity clogging when in contact with test samples.
A single-mode fibre-sensing configuration comprises an optical fibre, a sensing compo-
nent, a photodetector and a light source, with the fibre acting as the transmission medium
and the sensing element. However, the single-mode–multimode–single-mode (SMS) config-
uration is reportedly more effective, and the multimode section of the configuration is used
as the sensing element. The single-mode step-index polymer optical fibre is reportedly
adequate for monitoring transformer oil breakdown [38]. The authors of [58] used a step-
index multimode polymer optical fibre to design a water-quantity sensor for transformer
oil application. The sensing area was determined by mechanically uncladding a portion
(2 cm) of the multimode fibre. The authors of [58] summarised the design process for a
multimode polymer step index sensor generally applicable for other related measurements.
The setup is summarised below.
µ is the modal fractional power, Pclad and Pcore represent the powers carried on the core and
cladding, respectively, and V is the fibre’s normalised frequency. The normalised frequency
is a unitless quantity expressed as:
2πa
q
V= n2 core − n2 clad (11)
λ
a is the fibre core radius, λ is the wavelength of the optical source, and ncore and nclad
represent the refractive indices of the core and cladding, respectively. With constant source
wavelength, core diameter, and core refractive index, Pclad becomes proportionally related
to the clad refractive index, nclad (combining Equations (9) and (10)). The uncladded region
(sensing area) makes direct contact with the ageing transformer oil sample, and so nclad
becomes replaced with noil , and Pclad is replaced by Poil , representing the power lost to the
transformer oil owing to changes in the refractive index of the oil.
Ageing increases the refractive index of the transformer oil and consequently increases
the power lost to the transformer oil. This power will not get to the fibre optic output, thus
defining a sensing area (uncladded) that responds to the transformer oil ageing behaviour.
Figure10.
Figure 10. Classification
Classificationof
ofwettability
wettabilityof
ofaasurface [66].
surface[66].
The
The contact
contact angle
angle is
is measured
measured statistically
statistically and
and dynamically
dynamically when
when the
the liquid
liquid resides
resides
on
on the
the surface
surface and
and relates
relates to
to the
the surface
surface tension
tension between
between the
the different
different interfaces.
interfaces. Young’s
Young’s
model
model relates
relates the
the contact
contact angle
angle with
with the
the surface
surface tension
tension of
of the
the different
different interfaces
interfaces forfor
smooth (or refined) surfaces:
smooth (or refined) surfaces:
γsv − γsl
cos θ = 𝛾 − 𝛾 (12)
cos 𝜃 = γlv (12)
𝛾
where θ represents the contact angle (or Young’s contact angle); γsv represents the surface
where 𝜃
tension represents
between the contact angle
the solid–vapour (or Young’s
interface; contactthe
γsl represents angle);
surface𝛾 tension
represents the sur-
between the
face tension interface;
solid–liquid between and the solid–vapour
γlv representsinterface;
the surface 𝛾 tension
represents the surface
between tension be-
the liquid–vapour
tween theThe
interface. solid–liquid interface;
Wenzel model and and 𝛾 represents model
the Cassie–Baxter the surface
relatetension
to roughbetween theThe
surfaces. liq-
uid–vapour
Wenzel modelinterface. The as:
is expressed Wenzel model and the Cassie–Baxter model relate to rough
surfaces. The Wenzel model is expressed as:
γsv − γ
cos θ = 𝛾 − sl𝛾 ·r (13)
cos 𝜃 = γlv ·𝑟 (13)
𝛾
where r represents the roughness ratio (r = 1 for a smooth surface; and r > 1 for a rough
surface).
Creating a superhydrophobic surface involves several approaches, such as chemical
Sensors 2022, 22, 7923 16 of 25
where r represents the roughness ratio (r = 1 for a smooth surface; and r > 1 for a rough
surface).
Creating a superhydrophobic surface involves several approaches, such as chemical
vapour deposition, wet chemical reactions, electrochemical deposition, the layer-by-layer
method, self-assembly, electrospinning, plasma treatments, sol-gel, etc. Studies [68–77]
have aimed at providing low surface energy (using fluorine or silicon-containing molecules;
for rough surfaces) and high micro-nanostructure surface roughness (by the introduction
of pores or microstructure; for non-rough surfaces) [63,78]. Surface energy is associated
with the intermolecular forces between two media interfaces.
Most works on superhydrophobic coating are applied to outdoor insulators, biomedi-
cal instruments [65] and non-intrusive measurement technology. However, more research is
needed to assess the potency of superhydrophobicity in guaranteeing the reliability of intru-
sive online ageing detection technology for transformer oil applications prone to ABPs. This
will involve applying a superhydrophobic coating to the sensing area of the uncladded op-
tical fibre configuration, ensuring that the optical properties are not significantly impacted.
The reliability can be tested by subjecting the superhydrophobic optical fibre sensor to
different aged samples and measuring the precision with historical data. While the presence
of ABPs is expected to impact the precision of a test conducted without superhydrophobic
coating, the precision result should be much better with superhydrophobic coating.
where ŷ is the estimated dependent variable, b̂0 is the constant term, b̂1 indicates the
sensitivity of y to x, and ∈ is the regression error [79].
The corresponding multiple regression model is specified as:
where ŷ is the estimated dependent variable, b̂0 is the model constant, b̂1 , b̂2 , . . . b̂k are the
estimated parameter values and model sensitivities, and ∈ is the regression noise [79].
Sensors 2022, 22, 7923 17 of 25
A more detailed overview of the linear regression model covering model specification,
model estimation, maximum likelihood, statistical inference, regularisation and linear
model diagnostics can be found in [80].
Regression and non-binary classification models can be constructed using decision
trees and neural networks. Decision trees are tree-like graphs in which each leaf node
denotes the result of cumulative decisions, and each branch node denotes an option among
numerous options [81]. Artificial neural networks (ANN) are a supervised machine learning
technique mimicking biological networks using connected layers. A detailed review of
decision trees and artificial neural networks can be found in [82,83].
IoT data collection can produce inconsistent, sparse and noisy data, reducing the data
analysis’ confidence and uncertainty. The dependability of data analysis is increased by
managing and quantifying model uncertainty. Additionally, it strengthens confidence in
the model’s judgments [84]. A detailed review of uncertainty quantification for machine
learning models and deep learning models highlights the use of Bayesian neural networks
and Bayesian physics informed networks for deep learning uncertainty quantification,
as well as gaussian process regression (GPR) and physics-informed neural networks for
traditional machine learning [85].
Security challenges abound and pose a significant threat, especially when connected
to cloud technologies [89]. A feasible solution will include measures to detect cyber attacks
in hacking or phishing and alert experts when necessary. Additional layers of security can
be added to make different cyberattack activities impossible. Edge computing helps to
solve security challenges by allowing computing and data storage to occur at the edge of
the internet instead of the cloud.
The IoT can be integrated to track ageing and make proactive decisions to ensure opti-
mal transformer performance for the online ageing detection of transformer oil. Although
there has not been significant research output in this area for transformer oil, the authors
suggest adopting Figure 11 as the framework for online ageing detection in transformer
oil. An intensity-modulated optical fibre sensor, cross-capacitive sensors, and ageing gas
sensors (examples include hydrogen gas sensors and carbon dioxide gas sensors) can be
utilised for the sensing layer. National Instrument NI USB 6008/6009, Arduino microcon-
Sensors 2022, 22, x FOR PEER REVIEWtroller, and Raspberry Pi are some examples of data acquisition devices for ageing data.
19 of 26
ThingSpeak, ThingWorx, Google Cloud, Oracle IoT and Microsoft Azure IoT are examples
of IoT platforms that can provide the network, services and interface layer.
Figure
Figure11.
11.IoT
IoTservice-oriented architecture[86].
service-orientedarchitecture [86].
Thingspeak has
Thingspeak has reportedly succeeded
succeeded in inseveral
severalIoT
IoTapplications
applicationsandandcan
canbebeapplied
applied to
transformer oil ageing systems. Thingspeak is an open-source IoT platform
to transformer oil ageing systems. Thingspeak is an open-source IoT platform [90] com- [90] compatible
with MATLAB
patible with MATLAB (for AI (for
capabilities), microcontrollers
AI capabilities), and DAQs.
microcontrollers It has been
and DAQs. It hasutilised for
been uti-
real-time
lised room condition
for real-time monitoring
room condition systems,systems,
monitoring water monitoring systems and
water monitoring analytics,
systems and an-air
qualityair
alytics, monitoring systems, traffic
quality monitoring control
systems, systems,
traffic and
control transformer
systems, subsystemssubsys-
and transformer [90–96].
When
tems used with
[90–96]. When MATLAB
used with (for example),
MATLAB (forthe channel the
example), ID and API ID
channel keyand
areAPI
inputted
key areto
enable sensor data to be read with the help of the ESP8266 WIFI module
inputted to enable sensor data to be read with the help of the ESP8266 WIFI module and and Arduino
microcontroller.
Arduino In addition,
microcontroller. the MATLAB
In addition, data analysis
the MATLAB data toolkit,
analysisSimulink and Simscape
toolkit, Simulink and
can be employed for data cleaning, analysis, predictive model development,
Simscape can be employed for data cleaning, analysis, predictive model development, and the design
of frontends
and the design and of user interfaces.
frontends and user interfaces.
IoT has applications
IoT has applications in inhealthcare,
healthcare,aviation,
aviation,intelligent
intelligenthomes,
homes,agriculture,
agriculture,food
foodpro-
pro-
cessing, security surveillance, reliability engineering, pharmaceutical, retail
cessing, security surveillance, reliability engineering, pharmaceutical, retail and logistics, and logistics,
recyclingand
recycling andentertainment
entertainment[87,89].
[87,89].This
Thiscan
canbe beintegrated
integratedforforthe
theonline
onlineageing
ageingdetection
detection
of transformer oil.
of transformer oil.
11. Example of Related Systems
11. Example of Related Systems
11.1. A Non-Destructive, Non-Intrusive Design Using an Antenna
11.1. A Non-Destructive, Non-Intrusive Design Using an Antenna
Rohit, Sisir and Marley [9] implemented a non-intrusive, non-destructive technique
Rohit,
for the Sisir and
condition Marley [9]ofimplemented
assessment a non-intrusive,
transformer oil non-destructive
as shown in Figure 12. A horn technique
antenna
for
wasthe condition
placed overassessment of transformer
an open vessel oil aschamber
in an anechoic shown in(aFigure
room 12. A hornto
designed antenna was
stop sound
placed over or
reflections anelectromagnetic
open vessel in anwaves)
anechoic
to chamber (a room designed
prevent interference, filledto stopoil
with sound reflec-
to study its
tions or electromagnetic
characteristics. waves)chamber
The anechoic to prevent interference,
was maintained filled
at awith oil to study
temperature of its ◦ C and
20 charac-
teristics. The anechoic chamber was maintained at a temperature of 20 °C and 50% hu-
midity using a temperature controller and a dehumidifier; this is because the frequency
response of the insulating liquid is sensitive to temperature and humidity variation. The
parameters monitored to detect ageing were moisture content (MC) and total acid number
Sensors 2022, 22, 7923 19 of 25
50% humidity using a temperature controller and a dehumidifier; this is because the
frequency response of the insulating liquid is sensitive to temperature and humidity
variation. The parameters monitored to detect ageing were moisture content (MC) and
total acid number (TAN). Laboratory samples were initially used before extensions were
made to actual in-service transformer oil samples for validation. Accelerated thermal
ageing and oxidation ageing approaches were used to study transformer insulation’s early,
midlife, and afterlife properties with several ageing samples. The frequency response plot
clearly showed the degree of ageing present in the transformer oil. A linear fit was derived
from the plot of the reflection coefficient, |S11 | against the moisture content and the TAN.
Guranteeing constant humidity and temperature for an in situ transformer will involve
additional costs to implement new controlled designs. This solution is also limited by
the need to significantly re-mechanise existing transformer designs to accommodate the
Sensors 2022, 22, x FOR PEER REVIEW anechoic chamber. An evanescent wave absorption intrusive fibre optic solution offers20 of 26
a better solution in terms of cost, online adaptability, simplistic design and immunity to
external interference.
Figure12.
Figure Antenna design
12.Antenna design setup
setup[9].
[9].
11.2. An Intrusive Ageing Detection Design Using a Cross-Capacitance Sensor
11.2. An Intrusive Ageing Detection Design Using a Cross-Capacitance Sensor
Rahman et al. [57] used a cross-capacitive sensor to predict the dielectric constant of a
Rahmaninsulating
transformer et al. [57]oil
used
withaacross-capacitive
known contamination sensor to of
level predict the the
2-Fal, and dielectric constant of
combination
a transformer
of moisture andinsulating oil with
2-FAL, mineral a synthetic
oil, known contamination
ester, soybean oillevel
andofrapeseed
2-Fal, andoil. the
Thecombina-
oil
cellofmeasured
tion moisturetheanddielectric
2-FAL, constant
mineral of oil,the various ester,
synthetic solutions. The plot
soybean of the
oil and dielectric
rapeseed oil. The
oilconstant against different
cell measured concentrations
the dielectric constant for the various
of the oil samples shows linear
solutions. characteristics.
The plot of the dielectric
The results
constant weredifferent
against reported for the simulatedfor
concentrations cross-capacitance
the oil samples ofshows
variouslinear
transformer oil
characteristics.
concentrations. The sensor was calibrated, and a linear calibration result was reported
The results were reported for the simulated cross-capacitance of various transformer oil
with a maximum error of 0.86%, 0.33%, 0.91% and 0.64% for mineral oil, synthetic ester
concentrations. The rapeseed
oil, soybean oil and sensor was calibrated, The
oil, respectively. andsensor
a linear
wascalibration
validated forresult
onlinewas reported
ageing
with a maximum
detection error of 0.86%,technology.
using cross-capacitance 0.33%, 0.91% and 0.64% for
Cross-capacitance mineral
intrusive oil, synthetic
ageing detection ester
oil, soybean oil and rapeseed oil, respectively. The sensor was validated
technology offers good repeatability but is prone to electromagnetic interference. for online
Theageing
detection
moisture using
contentcross-capacitance technology.
affects the permittivity, making it Cross-capacitance intrusive
difficult to discriminate betweenageing
ageing detec-
andtechnology
tion moisture content
offersdetection when usingbut
good repeatability thisissolution.
prone to electromagnetic interference. The
moisture content affects the permittivity, making it difficult to discriminate between age-
11.3. An Intrusive Ageing Detection Design Using Fibre Optic Technology
ing and moisture content detection when using this solution.
The authors of [10] as shown in Figure 13 used a single-mode–multimode–single-mode
(SMS) fibre optic configuration to track the ageing of transformer oil. Four samples of
11.3. An Intrusive Ageing Detection Design Using Fibre Optic Technology
palm-based transformer oil were used (S1, S2, S3, S4), with S1 representing the pure sample
andThe
S2 toauthors of [10] samples
S4 representing as shown in Figure
of varied age. 13
Theused a single-mode–multimode–single-
lightwave refractive index was used
mode (SMS)
for the fibre optic
calibration; configuration
changes in transformertooil’s
trackagethe ageing
affect of transformer
the lightwave oil.
refractive Four(RI).
index samples
ofTwo ageing characterisation
palm-based transformer oiltechniques
were usedwere(S1,
used to S3,
S2, determine the S1
S4), with agerepresenting
of the unknown the pure
samples:
sample andthe
S2breakdown voltage test
to S4 representing (BDV, with
samples 25 kV) age.
of varied and the
TheUV–Vis spectrum
lightwave using index
refractive
appropriate standards.
was used for the calibration; changes in transformer oil’s age affect the lightwave refrac-
tive index (RI). Two ageing characterisation techniques were used to determine the age of
the unknown samples: the breakdown voltage test (BDV, with 25 kV) and the UV–Vis
spectrum using appropriate standards.
mode (SMS) fibre optic configuration to track the ageing of transformer oil. Four samples
of palm-based transformer oil were used (S1, S2, S3, S4), with S1 representing the pure
sample and S2 to S4 representing samples of varied age. The lightwave refractive index
was used for the calibration; changes in transformer oil’s age affect the lightwave refrac-
tive index (RI). Two ageing characterisation techniques were used to determine the age of
Sensors 2022, 22, 7923 20 of 25
the unknown samples: the breakdown voltage test (BDV, with 25 kV) and the UV–Vis
spectrum using appropriate standards.
Figure 13.
Figure 13. Experimental
Experimental configuration [10].
configuration [10].
A light source (WSL-100) was injected through the SMS fibre optic configuration,
and the MS9740A optical spectral analyser was used to display the plots (wavelength,
absorption and refractive index). The resonant wavelength shift plot against the refractive
index showed linear behaviour.
This configuration can be improved for real-time ageing detection with a photodetector,
DAQ device and the IoT.
as hydrogen gas, cross-capacitive and fibre optic sensors. Fibre optic sensors have the
advantage of electromagnetic interference immunity, are lightweight, explosion-proof, and
have improved sensitivity. However, there is no significant research output on determining
the OQIN of transformer oil online using fibre optic sensors. In addition, superhydropho-
bicity can be applied to intrusive fibre optic sensors based on evanescent field absorption
to improve the detection reliability of these sensors subject to measurement error from
APBs clogging the fibre optic sensing area. Integrating fibre optic sensors for transformer
oil ageing detection with the IoT and artificial intelligence (for online OQIN determination
and classification) will result in a prescriptive maintenance solution which offers more
advantages and robustness than traditional preventive (offline) maintenance approaches.
Specifying the machine learning model’s uncertainty increases confidence in the model’s
judgement, and this can be accomplished using a gaussian process regression (GPR) and
physics-informed neural networks for traditional machine learning.
Author Contributions: Conceptualization, U.E., A.N., A.A. and I.F.; methodology U.E., resources
A.N., A.A. and I.F.; writing—original draft preparation, U.E.; writing—review and editing, U.E., A.N.,
A.A. and I.F.; supervision A.N., A.A. and I.F.; funding acquisition A.N., A.A. and I.F. All authors have
read and agreed to the published version of the manuscript.
Funding: The APC was funded by Glasgow Caledonian University, UK Repository Team.
Institutional Review Board Statement: Not Applicable.
Informed Consent Statement: Not Applicable.
Data Availability Statement: Not Applicable.
Conflicts of Interest: The authors declare no conflict of interest.
Abbreviations
ABP Ageing ByProducts
ATT Accelerated Thermal Temperature
BOT Base Operating Temperature
BDV Breakdown Voltage
DDF Dielectric Dissipation Factor
DFB Distributed Feedback
DSC Differential Scanning Calorimetry
DGA Dissolved Gas Analysis
EFA-FOS Evanescent Field Absorption-based Fibre Optic Sensor
FMECA Failure Mode Effect and Criticality Analysis
FTIR Fourier Transform Infrared Spectroscopy
HMWCA High Molecular Weight Carboxylic Acids
IoT Internet of Things
LED Light Emitting Diode
LIBS Laser Induced Breakdown Spectroscopy
LMWCA Low Molecular Weight Carboxylic Acids
MO Mineral Oil
MTTR Mean Time to Repair
NN Neutralisation Number
OFS Optical Fibre Sensor
OQIN Oil Quality Index
TAN Total Acid Number
TF Time Factor
TGA Thermogravimetric Analysis
UV-Vis Ultraviolet-Visible Spectroscopy
VCSELs Vertical Cavity Surface Emitting Lasers
WCA Water Contact Angle
Sensors 2022, 22, 7923 22 of 25
References
1. Khalil, M. Qualitative and Quantitative FMECA on 220 kV Power Transformers. In Proceedings of the 2018 IEEE International
Conference on Environment and Electrical Engineering and 2018 IEEE Industrial and Commercial Power Systems Europe
(EEEIC/I&CPS Europe), Palermo, Italy, 12–15 June 2018; pp. 1–8.
2. Balamurugan, S.; Ananthanarayanan, R. Condition Monitoring Techniques of Dielectrics in Liquid Immersed Power
Transformers—A Review. In Proceedings of the 2018 IEEE Industry Applications Society Annual Meeting (IAS), Portland, OR,
USA, 23–27 September 2018; pp. 1–7.
3. Paul, D.; Goswami, A.K. Determination of Health Index of Insulating Oil of In-Service Transformer & Reactors based on IFT
Predicted by Multi-Gene Symbolic Regression. IEEE Trans. Ind. Appl. 2022, 58, 5935–5943.
4. Martins, M.A.G.; Gomes, A.R. Comparative study of the thermal degradation of synthetic and natural esters and mineral oil:
Effect of oil type in the thermal degradation of insulating kraft paper. IEEE Electr. Insul. Mag. 2012, 2, 22–28. [CrossRef]
5. Hadjadj, Y.; Fofana, I.; Sabau, J.; Briosso, E. Assessing insulating oil degradation by means of turbidity and UV/VIS spectropho-
tometry measurements. IEEE Trans. Dielectr. Electr. Insul. 2015, 22, 2653–2660. [CrossRef]
6. Wang, Z.; Chen, W.; Zhou, W.; Zhang, R.; Song, R.; Yang, D. A Few-shot Learning Method for Aging Diagnosis of Oil-paper
Insulation by Raman Spectroscopy Based on Graph Theory. IEEE Trans. Dielectr. Electr. Insul. 2021, 28, 1892–1900. [CrossRef]
7. Alshehawy, A.; Mansour, D.-E.; Ghali, M.; Lehtonen, M.; Darwish, M. Photoluminescence Spectroscopy Measurements for
Effective Condition Assessment of Transformer Insulating Oil. Processes 2021, 9, 732. [CrossRef]
8. Qian, Y.; Wang, Y.; Zhao, Y.; Fan, S.; Fu, Q.; Wang, H. Application of Infrared Spectroscopy in Oil Quality Detection. In Proceedings
of the 2020 IEEE International Conference on High Voltage Engineering and Application (ICHVE), Beijing, China, 6–10 September
2020; pp. 1–4.
9. Sangineni, R.; Nayak, S.K.; Becerra, M. A Non-intrusive and Non-destructive Technique for Condition Assessment of Transformer
Liquid Insulation. IEEE Trans. Dielectr. Electr. Insul. 2022, 29, 693–700. [CrossRef]
10. Razzaq, A.; Zainuddin, H.; Hanaffi, F.; Chyad, R.M.; Abdul Razak, H.; Latiff, A.A. Measurement of ester-based transformer oil
aging using tapered single mode-multimode-single mode fiber structure. Microw. Opt. Technol. Lett. 2020, 62, 559–564. [CrossRef]
11. Rao, U.M.; Fofana, I.; Betie, A.; Senoussaoui, M.L.; Brahami, M.; Briosso, E. Condition monitoring of in-service oil-filled
transformers: Case studies and experience. IEEE Electr. Insul. Mag. 2019, 35, 33–42. [CrossRef]
12. Mihajlovic, D.; Ivancevic, V.; Vasovic, V.; Lukic, J. Cellulose Degradation and Transformer Fault Detection by the Application of
Integrated Analyses of Gases and Low Molecular Weight Alcohols Dissolved in Mineral Oil. Energies 2022, 15, 5669. [CrossRef]
13. Polužanski, V.; Kartalović, N.; Nikolić, B. Impact of Power Transformer Oil-Temperature on the Measurement Uncertainty of
All-Acoustic Non-Iterative Partial Discharge Location. Materials 2021, 14, 1385. [CrossRef]
14. Sarathi, R.; Koperundevi, G. Understanding the Discharge Activities in Transformer Oil under AC and DC Voltage Adopting
UHF Technique. Int. J. Chem. Mol. Eng. 2008, 2, 49–56.
15. Niasar, M.G.; Edin, H.; Wang, X.; Clemence, R. Partial Discharge Characteristics Due to Air and Water Vapor Bubbles in Oil. Int.
Symp. High Volt. Eng. 2011, 11–21. Available online: https://www.researchgate.net/publication/264561918_Partial_discharge_
characteristics_due_to_air_and_water_vapor_bubbles_in_oil (accessed on 13 June 2022).
16. Alshehawy, A.M.; Mansour, D.-E.A.; Ghali, M. Condition Assessment of Aged Transformer Oil Using Photoluminescence-Based
Features. In Proceedings of the 2021 IEEE 5th International Conference on Condition Assessment Techniques in Electrical Systems
(CATCON), Kozhikode, India, 3–5 December 2021; pp. 282–285.
17. Sai, R.S.; Rafi, J.; Farook, S.; Kumar, N.; Parthasarathy, M.; Bakkiyaraj, R.A. Degradation studies of electrical, physical and
chemical properties of aged transformer oil. J. Phys. Conf. Ser. 2020, 1706, 012056. [CrossRef]
18. Fofana, I. 50 years in the development of insulating liquids. IEEE Electr. Insul. Mag. 2013, 29, 13–25. [CrossRef]
19. Munajad, A.; Subroto, C. Fourier Transform Infrared (FTIR) Spectroscopy Analysis of Transformer Paper in Mineral Oil-Paper
Composite Insulation under Accelerated Thermal Aging. Energies 2018, 11, 364. [CrossRef]
20. Bandara, K.; Ekanayake, C.; Saha, T.K. Compare the performance of natural ester with synthetic ester as transformer insulating
oil. In Proceedings of the 2015 IEEE 11th International Conference on the Properties and Applications of Dielectric Materials
(ICPADM), Sydney, Australia, 19–22 July 2015; pp. 975–978.
21. Perrier, C.; Marugan, M.; Beroual, A. DGA comparison between ester and mineral oils. IEEE Trans. Dielectr. Electr. Insul. 2012, 19,
1609–1614. [CrossRef]
22. Mehta, D.M.; Kundu, P.; Chowdhury, A.; Lakhiani, V.K.; Jhala, A.S. A review of critical evaluation of natural ester vis-a-vis
mineral oil insulating liquid for use in transformers: Part II. IEEE Trans. Dielectr. Electr. Insul. 2016, 23, 1705–1712. [CrossRef]
23. Gockenbach, E.; Borsi, H. Natural and Synthetic Ester Liquids as alternative to mineral oil for power transformers. In Proceedings
of the 2008 Annual Report Conference on Electrical Insulation and Dielectric Phenomena, Quebec, QC, Canada, 26–29 October
2008; pp. 521–524.
24. Jing, Y.; Timoshkin, I.V.; Wilson, M.P.; Given, M.J.; MacGregor, S.J.; Wang, T.; Lehr, J.M. Dielectric properties of natural ester,
synthetic ester midel 7131 and mineral oil diala D. IEEE Trans. Dielectr. Electr. Insul. 2014, 21, 644–652. [CrossRef]
25. Housel, T. Synthetic Esters: Engineered to Perform. 2022. Available online: https://www.machinerylubrication.com/Read/2970
3/synthetic-esters-perform (accessed on 6 September 2022).
26. Maharana, M.; Nayak, S.K.; Sahoo, N. Karanji oil as a potential dielectrics liquid for transformer. IEEE Trans. Dielectr. Electr. Insul.
2018, 25, 1871–1879. [CrossRef]
Sensors 2022, 22, 7923 23 of 25
27. Naranpanawe, W.M.L.B.; Fernando, M.A.R.M.; Kumara, J.R.S.S.; Naramapanawa, E.M.S.N.; Kalpage, C.S. Performance analysis of
natural esters as transformer liquid insulation-Coconut, castor and sesame oils. In Proceedings of the 2013 IEEE 8th International
Conference on Industrial and Information Systems, Kandy, Sri Lanka, 17–20 December 2013; pp. 105–109.
28. Monzón-Verona, J.M.; González-Domínguez, P.I.; García-Alonso, S.; Vaswani Reboso, J. Characterization of Dielectric Oil with
a Low-Cost CMOS Imaging Sensor and a New Electric Permittivity Matrix Using the 3D Cell Method. Sensors 2021, 21, 7380.
[CrossRef]
29. Hadjadj, Y.; Fofana, I.; Jalbert, J. Insulating oil decaying assessment by FTIR and UV-Vis spectrophotometry measurements. In
Proceedings of the 2013 Annual Report Conference on Electrical Insulation and Dielectric Phenomena, Shenzhen, China, 20–23
October 2013; pp. 1310–1313.
30. BS EN 60060-1:2010; High-Voltage Test Techniques. General Definitions and Test Requirements. 2011. Available online:
https://bsol.bsigroup.com/en/Bsol-Item-Detail-Page/?pid=000000000030200685 (accessed on 13 June 2022).
31. BS EN 60422:2006; Mineral Insulating Oils in Electrical Equipment. Supervision and Maintenance Guidance. 2009. Available
online: https://bsol.bsigroup.com/en/Bsol-Item-Detail-Page/?pid=000000000030081812 (accessed on 13 June 2022).
32. Hayber, E.; Tabaru, T.E.; Güçyetmez, M. Evanescent Field Absorption-Based Fiber Optic Sensor for Detecting Power Transformer
Oil Degradation. Fiber Integr. Opt. 2021, 40, 229–248. [CrossRef]
33. Smith, B.C. Fundamentals of Fourier Transform Infrared Spectroscopy; CRC Press: Boca Raton, FL, USA, 2011.
34. Alshehawy, A.M.; Mansour, D.-E.A.; Ghali, M.; Rezk, A. Evaluating the impact of aging in field transformer oil using optical
spectroscopy techniques. In Proceedings of the 2017 IEEE 19th International Conference on Dielectric Liquids (ICDL), Manchester,
UK, 25–29 June 2017; pp. 1–4.
35. Liu, Y.; Bao, K.; Wang, Q.; Zio, E. Application of FTIR Method to Monitor the Service Condition of Used Diesel Engine Lubricant
Oil. In Proceedings of the 2019 4th International Conference on System Reliability and Safety (ICSRS), Rome, Italy, 20–22
November 2019; pp. 175–180.
36. BS EN 60567:1993; IEC 60567:1992: Guide for the Sampling of Gases and of Oil from Oil-Filled Electrical Equipment and for the
Analysis of Free and Dissolved Gases. 1993. Available online: https://bsol.bsigroup.com/en/Bsol-Item-Detail-Page/?pid=0000
00000000292653 (accessed on 13 June 2022).
37. IEC 60599:1999; Mineral Oil-Impregnated Electrical Equipment in Service-Guide to the Interpretation of Dissolved and Free Gases
Analysis. 2000. Available online: https://bsol.bsigroup.com/en/ (accessed on 13 June 2022).
38. N’cho, J.S.; Fofana, I. Review of Fiber Optic Diagnostic Techniques for Power Transformers. Energies 2020, 13, 1789. [CrossRef]
39. Adams, M.J.; Romeo, M.J.; Rawson, P. FTIR analysis and monitoring of synthetic aviation engine oils. Talanta 2007, 73, 629–634.
[CrossRef]
40. IEEE Guide for Acceptance and Maintenance of Insulating Oil in Equipment. 2006. Available online: https://ieeexplore.ieee.org/
document/7442048 (accessed on 13 June 2022).
41. BS EN 62021-1:2003; Insulating Liquids. Determination of Acidity. Automatic Potentiometric Titration. 2003. Available online:
https://bsol.bsigroup.com/en/Bsol-Item-Detail-Page/?pid=000000000030069710 (accessed on 13 June 2022).
42. BS EN 62021-2:2007; Insulating Liquids. Determination of Acidity. Colourimetric Titration. 2007. Available online: https:
//bsol.bsigroup.com/en/Bsol-Item-Detail-Page/?pid=000000000030156457 (accessed on 13 June 2022).
43. Danesh, A. PVT and Phase Behaviour of Petroleum Reservoir Fluids; Elsevier: Amsterdam, The Netherlands, 2001.
44. Interfacial Tension Measurement. 2022. Available online: https://www.biolinscientific.com/measurements/interfacial-tension#:
~{}:text=Interfacial%20tension%20measurement%20between%20two,is%20immersed%20in%20another%20liquid (accessed on 13
June 2022).
45. Fofana, I.; Bouaïcha, A.; Hadjadj, Y.; N’Cho, J.S.; Aka-Ngnui, T.; Béroual, A. Early stage detection of insulating oil decaying.
In Proceedings of the 2010 Annual Report Conference on Electrical Insulation and Dielectic Phenomena, Cancun, Mexico,
16–19 October 2011; pp. 1–4.
46. Gray, I. A Guide to Transformer Oil Analysis. Transformer Chemistry Services. IAR Gray 2009, 1–12.
47. BS EN IEC 62961:2018; Insulating Liquids. Test Methods for the Determination of Interfacial Tension of Insulating Liquids.
Determination with the Ring Method. 2018. Available online: https://bsol.bsigroup.com/en/Bsol-Item-Detail-Page/?pid=0000
00000030352874 (accessed on 13 June 2022).
48. Wilhelmy Plate Method. 2022. Available online: https://www.kruss-scientific.com/en/know-how/glossary/wilhelmy-plate-
method (accessed on 13 June 2022).
49. Bailey, A. Surface and Interfacial Tension. 2022. Available online: https://thermopedia.com/content/30/ (accessed on 13 June 2022).
50. Ghosh, D.; Khastgir, D. Degradation and Stability of Polymeric High-Voltage Insulators and Prediction of Their Service Life
through Environmental and Accelerated Aging Processes. ACS Omega 2018, 3, 11317–11330. [CrossRef] [PubMed]
51. Vasovic, V.; Lukic, J.; Mihajlovic, D.; Pejovic, B.; Radakovic, Z.; Radoman, U.; Orlovic, A. Aging of transformer
insulation—Experimental transformers and laboratory models with different moisture contents: Part I—DP and furans
aging profiles. IEEE Trans. Dielectr. Electr. Insul. 2019, 26, 1840–1846. [CrossRef]
52. Fofana, I.; Bouaïcha, A.; Farzaneh, M.; Volat, C.; Sabau, J. On the stability of mineral insulating oils under electrical stress. In
Proceedings of the 2009 IEEE Electrical Insulation Conference, Montreal, QC, Canada, 31 May–3 June 2009.
53. Bouaicha, A.; Fofana, I.; Farzaneh, M. Application of modern diagnostic techniques to assess the condition of oil and pressboard.
In Proceedings of the 2008 IEEE International Conference on Dielectric Liquids, Poitiers, France, 30 June–3 July 2008; pp. 1–4.
Sensors 2022, 22, 7923 24 of 25
54. D1934; Standard Test Method for Oxidative Aging of Electrical Insulating Petroleum Oils by Open-Beaker Method. 2012. Available
online: http://www.astm.org/Standards/D1934 (accessed on 13 June 2022).
55. El-Hag, A. Application of Machine Learning in Outdoor Insulators Condition Monitoring and Diagnostics. IEEE Instrum. Meas.
Mag. 2021, 24, 101–108. [CrossRef]
56. Rahman, O.; Islam, T.; Khera, N.; Khan, S.A. A Novel Application of the Cross-Capacitive Sensor in Real-Time Condition
Monitoring of Transformer Oil. IEEE Trans. Instrum. Meas. 2021, 70, 1–12. [CrossRef]
57. Rahman, O.; Islam, T.; Ahmad, A.; Parveen, S.; Kheraand, N.; Khan, S.A. Cross Capacitance Sensor for Insulation Oil Testing.
IEEE Sens. J. 2021, 21, 20980–20989. [CrossRef]
58. Mahanta, D.K.; Laskar, S. Water Quantity-Based Quality Measurement of Transformer Oil Using Polymer Optical Fiber as Sensor.
IEEE Sens. J. 2018, 18, 1506–1512. [CrossRef]
59. Maria de Fátima, F.D.; Radwan, A. Optical Fiber Sensors for IoT and Smart Devices; Springer: Berlin/Heidelberg, Germany, 2017.
60. Lee, B.H.; Kim, Y.H.; Park, K.S.; Eom, J.B.; Kim, M.J.; Rho, B.S.; Choi, H.Y. Interferometric Fiber Optic Sensors. Sensors 2012, 12,
2467–2486. [CrossRef] [PubMed]
61. Riziotis, C.; El Sachat, A.; Markos, C.; Velanas, P.; Meristoudi, A.; Papadopoulos, A. Assessment of fiber optic sensors for aging
monitoring of industrial liquid coolants. Opt. Compon. Mater. XII 2015, 9359, 308–315. [CrossRef]
62. Arshad; Momen, G.; Farzaneh, M.; Nekahi, A. Properties and applications of superhydrophobic coatings in high voltage outdoor
insulation: A review. IEEE Trans. Dielectr. Electr. Insul. 2017, 24, 3630–3646. [CrossRef]
63. Sharma, K.; Hooda, A.; Goyat, M.; Rai, R.; Mittal, A. A review on challenges, recent progress and applications of silica
nanoparticles based superhydrophobic coatings. Ceram. Int. 2022, 48, 5922–5938. [CrossRef]
64. Esmeryan, K.D.; I Radeva, E.; Avramov, I.D. Durable superhydrophobic carbon soot coatings for sensor applications. J. Phys. D
Appl. Phys. 2015, 49, 025309. [CrossRef]
65. Kumar, R.; Kumar Sahani, A. Role of superhydrophobic coatings in biomedical applications. Mater. Today Proc. 2021, 45, 5655–5659.
[CrossRef]
66. Song, J.; Fan, L. Temperature dependence of the contact angle of water: A review of research progress, theoretical understanding,
and implications for boiling heat transfer. Adv. Colloid Interface Sci. 2021, 288, 102339. [CrossRef] [PubMed]
67. Sethi, S.K.; Manik, G. Recent Progress in Super Hydrophobic/Hydrophilic Self-Cleaning Surfaces for Various Industrial Applica-
tions: A Review. Polym.-Plast. Technol. Eng. 2018, 57, 1932–1952. [CrossRef]
68. Shirtcliffe, N.J.; McHale, G.; Atherton, S.; Newton, M.I. An introduction to superhydrophobicity. Adv. Colloid Interface Sci. 2010,
161, 124–138. [CrossRef]
69. Li, X.-M.; Reinhoudt, D.; Crego-Calama, M. What do we need for a superhydrophobic surface? A review on the recent progress in
the preparation of superhydrophobic surfaces. Chem. Soc. Rev. 2007, 36, 1350–1368. [CrossRef]
70. Lakshmi, R.; Bharathidasan, T.; Basu, B.J. Superhydrophobic sol–gel nanocomposite coatings with enhanced hardness. Appl. Surf.
Sci. 2011, 257, 10421–10426. [CrossRef]
71. Fernández-Blázquez, J.P.; Fell, D.; Bonaccurso, E.; del Campo, A. Superhydrophilic and superhydrophobic nanostructured
surfaces via plasma treatment. J. Colloid Interface Sci. 2011, 357, 234–238. [CrossRef] [PubMed]
72. Jafari, R.; Asadollahi, S.; Farzaneh, M. Applications of Plasma Technology in Development of Superhydrophobic Surfaces. Plasma
Chem. Plasma Process. 2012, 33, 177–200. [CrossRef]
73. Song, X.; Zhai, J.; Wang, Y.; Jiang, L. Fabrication of Superhydrophobic Surfaces by Self-Assembly and Their Water-Adhesion
Properties. J. Phys. Chem. B 2005, 109, 4048–4052. [CrossRef] [PubMed]
74. Ji, J.; Fu, J.; Shen, J. Fabrication of a Superhydrophobic Surface from the Amplified Exponential Growth of a Multilayer. Adv.
Mater. 2007, 19, 2411. [CrossRef]
75. Zhang, L.; Chen, H.; Sun, J.; Shen, J. Layer-by-Layer Deposition of Poly(diallyldimethylammonium chloride) and Sodium Silicate
Multilayers on Silica-Sphere-Coated Substrate—Facile Method to Prepare a Superhydrophobic Surface. Chem. Mater. 2007, 19,
948–953. [CrossRef]
76. Momen, G.; Farzaneh, M.; Jafari, R. Wettability behaviour of RTV silicone rubber coated on nanostructured aluminium surface.
Appl. Surf. Sci. 2011, 257, 6489–6493. [CrossRef]
77. Ishizaki, T.; Hieda, J.; Saito, N.; Saito, N.; Takai, O. Corrosion resistance and chemical stability of super-hydrophobic film
deposited on magnesium alloy AZ31 by microwave plasma-enhanced chemical vapor deposition. Electrochim. Acta 2010, 55,
7094–7101. [CrossRef]
78. Wu, H.; Luo, J.; Huang, X.; Wang, L.; Guo, Z.; Liang, J.; Zhang, S.; Xue, H.; Gao, J. Superhydrophobic, Mechanically Durable
Coatings for Controllable Light and Magnetism Driven Actuators. J. Colloid Interface Sci. 2021, 603, 282–290. [CrossRef] [PubMed]
79. Kissell, R.; Poserina, J. Chapter 2—Regression Models. In Optimal Sports Math, Statistics, and Fantasy; Kissell, R., Poserina, J., Eds.;
Academic Press: Cambridge, MA, USA, 2017; pp. 39–67.
80. Su, X.; Yan, X.; Tsai, C. Linear regression. WIREs Comp. Stat. 2012, 4, 275–294. [CrossRef]
81. Tong, J.C.; Ranganathan, S. Computational T cell vaccine design. In Computer-Aided Vaccine Design; Tong, J.C., Ranganathan, S.,
Eds.; Woodhead Publishing: Cambridge, UK, 2013; pp. 59–86.
82. de Ville, B. Decision trees. WIREs Comp. Stat. 2013, 5, 448–455. [CrossRef]
83. Titterington, M. Neural networks. WIREs Comp. Stat. 2010, 2, 1–8. [CrossRef]
Sensors 2022, 22, 7923 25 of 25
84. Polužanski, V.; Kovacevic, U.; Bacanin, N.; Rashid, T.A.; Stojanovic, S.; Nikolic, B. Application of Machine Learning to Express
Measurement Uncertainty. Appl. Sci. 2022, 12, 8581. [CrossRef]
85. Siddique, T.; Mahmud, M.S.; Keesee, A.M.; Ngwira, C.M.; Connor, H. A Survey of Uncertainty Quantification in Machine
Learning for Space Weather Prediction. Geosciences 2022, 12, 27. [CrossRef]
86. Li, S.; Xu, L.D.; Zhao, S. The internet of things: A survey. Inf. Syst. Front. 2014, 17, 243–259. [CrossRef]
87. Smanchat, S. Internet of things: A review of applications and technologies. Suranaree J. Sci. Technol. 2014, 21, 359–374.
88. Lo, C.; Chen, C.; Zhong, R.Y. A review of digital twin in product design and development. Adv. Eng. Inform. 2021, 48, 101297.
[CrossRef]
89. Bandyopadhyay, D.; Sen, J. Internet of Things: Applications and Challenges in Technology and Standardization. Wirel. Pers
Commun. 2011, 58, 49–69. [CrossRef]
90. Razali, M.A.A.; Kassim, M.; Sulaiman, N.A.; Saaidin, S. A ThingSpeak IoT on Real Time Room Condition Monitoring System. In
Proceedings of the 2020 IEEE International Conference on Automatic Control and Intelligent Systems (I2CACIS), Shah Alam,
Malaysia, 20 June 2020; pp. 206–211.
91. Nasution, T.H.; Muchtar, M.A.; Seniman, S.; Siregar, I. Monitoring temperature and humidity of server room using Lattepanda
and ThingSpeak. J. Phys. Conf. Ser. 2019, 1235, 12068. [CrossRef]
92. Miry, A.H.; Aramice, G.A. Water monitoring and analytic based thingspeak. Int. J. Electr. Comput. Eng. (Malacca Malacca) 2020, 10,
3588. [CrossRef]
93. Kelechi, A.H.; Alsharif, M.H.; Agbaetuo, C.; Ubadike, O.; Aligbe, A.; Uthansakul, P.; Kannadasan, R.; Aly, A.A. Design of a
Low-Cost Air Quality Monitoring System Using Arduino and ThingSpeak. Comput. Mater. Contin. 2022, 70, 151–169. [CrossRef]
94. Sachenko, A.; Osolinskyi, O.; Bykovyy, P.; Dobrowolski, M.; Kochan, V. Development of the Flexible Traffic Control System Using
the LabView and ThingSpeak. In Proceedings of the 2020 IEEE 11th International Conference on Dependable Systems, Services
and Technologies (DESSERT), Kyiv, Ukraine, 14–18 May 2020; pp. 326–330.
95. Ravindran, V.; Ponraj, R.; Krishnakumar, C.; Ragunathan, S.; Ramkumar, V.; Swaminathan, K. IoT-Based Smart Transformer
Monitoring System with Raspberry Pi. In Proceedings of the 2021 Innovations in Power and Advanced Computing Technologies
(i-PACT), Malaya, Kuala Lumpur, 27–29 November 2021; pp. 1–7.
96. Mahanta, D.K.; Rahman, I. IoT Based Transformer Oil Temperature Monitoring System. In Proceedings of the 2022 International
Conference on Sustainable Computing and Data Communication Systems (ICSCDS), Erode, India, 7–9 April 2022; pp. 975–978.