Measurement 113 (2018) 38–45

Contents lists available at ScienceDirect

Measurement
journal homepage: www.elsevier.com/locate/measurement

Monitoring power transformers oils deterioration using structured laser MARK

illumination planar imaging
Guy-Oscar Regnimaa, Amidou Betiéa,b, Thomas Koffia, Olivier K. Baguia, Issouf Fofanab,
⁎
Abaka Kouacoua, Jérémie Zoueua,
a
Laboratoire d’Instrumentation, Image et de Spectroscopie (L2IS), INPHB, DFR-GEE, B.P. 1093 Yamoussoukro, Cote d’Ivoire
b
Research Chair on Aging Power Equipment Infrastructure (ViAHT), Université du Québec à Chicoutimi, Chicoutimi, Québec, Canada

A R T I C L E I N F O A B S T R A C T

Keywords: Reliable quality assessments of oils in power transformers are important as they provide valuable information
Extinction coefficient regarding the proper functioning of transformers. Thus, an early and accurate diagnostic of power transformers
Structured illumination oils can prevent potential failures of transformers. In this paper, an imaging technique known as Structured Laser
Transformers oil diagnosis Illumination Planar Imaging (SLIPI) was used to monitor the extinction coefficient μe in various oil samples. The
Dense turbid media
proposed technique offers the advantage of extracting the light intensity contribution from singly scattered
photons and rejecting most of the light intensity from photons that have been scattered many times. This leads to
more accurate and reliable measurement of the extinction coefficient μe , in optically dense oil samples. The
variation of the extinction coefficient was therefore determined as a function of oil aging. The results demon-
strate that SLIPI is reliable as a practical measurement method for the diagnosis of power transformer oils and
present an attractive solution, alternative to the conventional methods such as Dissolved Decay Products,
Interfacial Tension and Turbidity.

1. Introduction very important to monitor the condition of the transformer by com-

Power transformers are key components for electrical energy gen-
eration and delivery [1]. Degraded oil in transformers affects the power
delivery and can lead to functioning failures. It is therefore critical to
control and anticipate the proper functioning of power transformers by
monitoring their insulation. The diagnosis of transformers is thus vital
for their proper maintenance and to improve their operating conditions,
which ensure a reliable and efficient supply of electricity [2]. The oil
content and condition tell us about the degradation and the Aging of
power transformers; therefore, it serves as diagnostic target as it con-
tains approximately 70% of these diagnostic information [3–5]. Thus,
the diagnosis of oils has long been the focus of several research projects,
ultimately leading to the development of various diagnostic tools.
Among these diagnostic tools, electrical-based methods allow to de-
termining certain properties non-destructively such as the relative
permittivity and dissipation factor, the capacitance, polarization index,
resistivity, dielectric strength, partial discharges, electrostatic charging
tendency, etc [6]. The results of these measurements provide an
average value of the whole insulator volume, electrically and geome-
trically heterogeneous, and they do not characterize the status in a
specific point of the material. The role of these methods is nevertheless
parison with previous or suggested limit values. Physicochemical
methods, such as the water content in oil [7–9], and the analysis of
dissolved gas in the insulation oil are also widely used. Various stan-
dards/methods have been developed for the interpretation of these
dissolved gas analysis. The most commonly used interpretation
methods are [10]: the IEEE C57.104–1991, Doernenberg, Rogers, IEC
60599 and the Duval’s triangle. The main problem concerning this
technique lies in the determination of the limit values for simple gases.
However, gas chromatographs are very expensive. Other techniques
such as the total acid number in oil [11], the interfacial tension (IFT)
[12] are also widely used. IFT test require experienced user and also,
some precautions have to be taken as indicated in the American Society
for Testing and Materials (ASTM), ASTM D971 in order to do mea-
surements. Methods based on spectral analysis also exist, including
spectrophotometric absorption measurements of the oil samples. In this
case, light that passes through the contaminated oil is attenuated by the
absorbing elements [13]. The results obtained allow assessing the re-
lative amount of dissolved decay products (DDP) in oil. Furthermore, it
should be noted that these different diagnostic methods mentioned
above are for the most part time consuming and very expensive.
Moreover, regarding the methods based on spectral analysis, not all of

⁎
Corresponding author.
E-mail addresses: betie@yahoo.fr (A. Betié), Jeremie.zoueu@inphb.edu.ci (J. Zoueu).

http://dx.doi.org/10.1016/j.measurement.2017.08.019
Received 6 May 2017; Received in revised form 7 August 2017; Accepted 11 August 2017
Available online 23 August 2017
0263-2241/ © 2017 Elsevier Ltd. All rights reserved.
G.-O. Regnima et al. Measurement 113 (2018) 38–45

them provide knowledge about the degree of degradation of the oils

namely due to the high density of some of them. For spectrophotometry
using the Beer-Lambert-Bouguer within the limits of its application, it is
necessary to dilute the solution by a solvent thus modifying the initial
state of this solution. Otherwise, there are measurement errors due
mainly to multiple scattering of light [14].
In this article, a new assessment technique based on optical imaging
called SLIPI (Structured Laser Illumination Planar Imaging) is proposed
[14]. This technique could serve as a better alternative approach for the
diagnosis of oils in power transformers. The main advantage of the
technique is its capability in providing reliable results even within
optically dense liquids as recently demonstrated in [14]. Thus, one does
not need, for the dense samples, to perform dilution processes,. In ad-
dition to being fast and cost-effective, it has a simplified optical setup.
No need to shift the lines pattern to record images and it is suitable for
recording series of successive images. The most important advantage is
its ability to suppress multiple scattering effects on instantaneous
imaging in turbid media [14].
The SLIPI method consists in imaging, from the side, a modulated
light sheet crossing the turbid sample. Then, the intensity contribution
from multiple light is suppressed after post-processing of the recorded
image, and an exponential intensity decay is obtained as light crosses
the sample. By making a curve fit on this decay and using the Beer-
Lambert law, the extinction coefficient of the sampled liquid is obtained
[14]. Note that Bagui et al. [15] have used the SLIPI technique to es- Fig. 1. Overview of the transformer oil samples. The first range of transformer oils from
tablish ground coffee classification. S1 to S4 are pale yellow and are unaged (0 h). The second range of transformer oils from
S5 to S8 are brown and are those aged at 1000 h. Their appearance indicates bad con-
In this article, the extinction coefficient of transformer oils is mea-
dition. The last range from S9 to S12 are black and are those aged at 2000 h. Their
sured using the so-called single-phase SLIPI approach [14,16]. The re- appearance indicates extremely bad condition. (For interpretation of the references to
sults are compared with those obtained from the methods commonly color in this figure legend, the reader is referred to the web version of this article.)
used for the diagnosis of power transformer oils, such as Dissolved
Decay Products (DDP), Interfacial Tension (IFT) and Turbidity. A de-
2.2. Analysis of the oils using Turbidimetry
scription of the equipment and methodologies used is provided, fol-
lowed by the results and finally the discussion and conclusion are
Turbidity defines the disorder of a transformer’s oil caused by sus-
presented.
pended solids which are generally invisible to the naked eye. Usually,
an oil contains suspended solids, which consist of a large number of
particles of different sizes. The suspended particles are responsible of
2. Description of the investigated oils and standard analysis
the turbidity of the oil in the transformers. The instrument used to make
the measurements is the nephelometer 2100AN of the manufacturer
2.1. Description of the oils
HACH [17]. Measurement of turbidity has been performed according to
the ASTM D 6181 [18]. The turbidimeter uses the principle of the in-
The preparation of oil samples of transformer has been made in
teraction between an incident light and the suspended solids in oil
laboratory, at the University of Quebec in Chicoutimi, Canada. The si-
leading to several phenomena such as diffusion, reflection, absorption
mulation of the aging of the different oils is made by taking into ac-
and refraction. These suspended solids in oil, according to their size,
count the conditions of aging in the power transformers. Aging was
kind, shape, refractive index and its intensity cause a dispersion of the
carried out in a mechanical convection oven whose temperature was set
incident light in all directions [19]. The optical system consists of a
to 115 °C. To accelerate the aging, catalysts such as the powder of
tungsten filament lamp, lenses and apertures for concentrating light, a
copper (3 g per liter of oil) were used. Several samples were prepared.
detector placed at 90 degrees from the light detector, a forward scatter
The list of samples analysed is shown in the Table 1.
detector, a backscatter detector and a transmitted light detector. The
The paper behaves as a filter and decreases the dissolved decay
instrument measures the Turbidity using the complete set of detectors
products (see Fig. 1).
(ratio measurement). The microprocessor of the instrument uses a
mathematical calculation from the signals ratio of each detector. The
Table 1 advantages of using the ratio for measurements include excellent line-
Description of Transformer oil samples.
arity and stability of calibration and ability to measure turbidity in
Name of the samples Samples description presence of color.
The turbidity is expressed in NTU (Nephelometric Turbidity Unity).
Sample S1 0 h before test of stability As advantages, the Turbidimetry is very accurate and useful for mea-
Sample S2 0 h with 150 strains and strong electric discharges
suring very low turbidity (less than 5 NTU). However, there is a number
Sample S3 0 h with 0% of ratio paper-oil
Sample S4 0 h with 10% of ratio paper-oil of limitations: the high cost, high power needed and the fragility of the
Sample S5 1000 h before test of stability system [19].
Sample S6 1000 h with 150 strains and strong electric discharges Some guidelines for turbidity values of oils are given in Table 2.
Sample S7 1000 h with 0% of ratio paper-oil
The results of turbidity obtained for each sample of transformer oils
Sample S8 1000 h with 10% of ratio paper-oil
Sample S9 2000 h before test of stability as a function of aging duration are shown on Fig. 2.
Sample S10 2000 h with 150 strains and strong electric discharges In the analysis of the results obtained (Fig. 2), the increase of the
Sample S11 2000 h with 0% of ratio paper-oil values of turbidity reflects the aging of the power transformers oils
Sample S12 2000 h with 10% of ratio paper-oil [19,21]. According to the guidelines suggested for turbidity given in

39
G.-O. Regnima et al.
Table 2
Guidelines suggested for turbidity [20].
Oil condition Turbidity (NTU)
Good oils 0–1
Proposition an oils 1–4
Marginal oils 4–10
Bad oils 10–30
Very bad oils 30–150
Extremely bad oils > 150
Fig. 2. Turbidity of each sample of transformer oils as a function of aging duration. When
the aging duration is high, the turbidity for each sample of power transformer oil in-
creases.
table 2 [20], classification of the oils samples has been made. At 0 h, all
the oil samples are in the range of “Good oils”. the oil ssamples with
aging duration equals to 1000 h are “Marginal oils” and those aged at
2000 h correspond to “Marginal or bad oils”.
2.3. Analysis of the oils using interfacial tension
The interfacial tension (IFT) allows evaluating the oxidation degree
of the mineral oils, by detecting the presence of polar compounds [22].
It represents the force in dynes per centimeter required to rupture a
platinum ring at the interface between the oil and distilled water [23].
Measurement of the interfacial tension to assess the condition of the
liquid insulation is performed according to the ASTM D971-12 [12]
using a surface tensiometer. The instrument used is a DuNouy type from
Fisher Scientific brand tensiometer which uses a plane ring of Pla-
tinum/iridium alloy. The measurement shall be made within 60 s of
formation of the interface. When certain contaminants such as oxida-
tion products are present in the oil, the strength of the oil film is
weakened, requiring less tension to break the film. For oils in service, a
decreasing value of IFT points out the accumulation of contaminants,
oxidation products or both. It is a measurement indicating the presence
of oxidation products that can attack the solid insulation and influence
the cooling of the transformer. The more the oil degrades, the more this
value has a tendency to decrease.
The IFT test provides information on polar contaminants and oxi-
dation products. However, due to the high hydrophilic nature of the oil,
the interfacial tension test may not be very sensitive to changes in the
fluid quality [24]. Moreover, IFT test requires trained person and also,
40
Measurement 113 (2018) 38–45
Fig. 3. Interfacial tension of each sample of transformer oils as a function of aging
duration. When the aging duration is high, the IFT for each sample of power transformer
oil decreases.
some precautions have to be taken as indicated in the ASTM D971 in
order to do measurements.
Some guidelines for IFT values of oils exist. Good and clean oils have
IFT number between 40 to 50 dynes per centimeter. For IFT value
below 25 dynes per centimeter, the oil is recommended to be reclaimed
because the oil is contaminated by sludge, which begins at around 22
dynes per centimeter [25].
The results of IFT values obtained for each sample of transformer
oils are shown on Fig. 3.
Fig. 3 shows a decay of the values of IFT. Indeed, for in-service
transformer oil, the decrease of the values of IFT indicates the accu-
mulation of contaminants and/or oxidation products in the oil. One
may notice that with increasing aging duration, the IFT value decreases
[19,21]. This value is constant and is equal to 15 dynes per centimeter
from 1500 h. Classification of transformer oils samples has been es-
tablished by using data reported in [24]. At 0 h, all the oils samples
have IFT values higher than 25 dynes per centimeter. They are refereed
as “Good oils”. The oils with aging duration at 1000 h and 2000 h have
IFT values lower to 22 dynes per centimeter. They are classified as “bad
quality oils” and reclamation is recommended .
2.4. Analysis of the oils using dissolved decay products
The measurement of the relative quantity of Dissolved Decay
Products in oil or DDP is carried out by spectrophotometry UV/VIS
(ultraviolet/visible) according to the ASTM 6802. The instrument used
for measurements is a Thuramed T60 UV/Vis spectrophotometer. The
spectrophotometer sheds a monochromatic incident light I0 through a
length L (the length of the cuvette) of the oil sample to test and mea-
sures the absorbance A (size linked to the quantity of light absorbed by
the solution). From this study one determines the absorbance A and
transmittance T defined respectively by the Beer-Lambert-Bouguer law:
I
A = −logT and T = = e−μe L
I0 (1)
where I is the intensity transmitted from the light and µe is the ex-
tinction coefficient.
For each wavelength, the absorbance of each power transformer oil
is measured and the data collected are used to plot the absorbance
G.-O. Regnima et al.
Table 3
Guidelines for Dissolved Decay Products (DDP) [20].
Oil condition DDP (a.u)
Good oils 0–10
Proposition an oils ± 10–25
Bad oils 25–50
Very bad oils 50–300
Extremely bad oils > 550
curve A (on the ordinate) as a function of the Wavelength λ (on the
abscissa). The graph thus obtained constitutes the UV/visible spectrum.
The numerical integration of the surface below the absorbance curve
allows determining the relative amount of the Dissolved Decay Products
(DDP: peroxides, Aldehydes, ketones and organic acids) in liquid sam-
ples.
This technique is used routinely for the quantitative study of organic
molecules (molecules composed essentially of carbon and hydrogen)
and functional groups by measuring the absorbance of the oil samples.
While being very useful, these measurements are, unfortunately, no
longer valid when applied to optically dense scattering media such as
turbid liquids. This limitation is due to both the strong reduction of
desired unscattered light and the large increase of unwanted scattered
photons reaching the detector. Using more sensitive sensors or brighter
light sources can compensate the reduction of the desired signal.
However, the effects of multiple light scattering are more challenging to
account for [14].
Table 3 highlights the guidelines for DDP values, expressed in ar-
bitrary unit (a.u.).
The results of DDP obtained for each sample of transformer oils as a
function of aging duration are shown on Fig. 4.
Concerning the evolution of the DDP in the power transformer oils
seen on Fig. 4, there is a general increase in the values of the areas
calculated under the curves as the aging duration increases from 0 h to
2000 h. The increase of the values of DDP reflects the aging of the
power transformer oils [19,21].
From the analysis of the results of the physicochemical techniques
used, it emerges that power transformer oils undergo a rapid dete-
rioration as aging duration increases. Thus, oils aged at 2000 h are
more degraded than those at 1000 h which in their turn are degraded
Fig. 4. Dissolved Decay Products of each sample of transformer oils as a function of aging
duration. When the aging duration is high, the quantity of DDP for each sample of power
transformer oil increases.
41
Measurement 113 (2018) 38–45
with respect to those of 0 h (fresh oils).
3. Single-phase SLIPI experiment
3.1. Experimental setup
The SLIPI technique has been developed in 2008 by Edouard
Berrocal and Elias kristensson [26,27]. This technique, which has been
both demonstrated by simulation and by experimentation, possesses a
large capacity for the suppression of multiple light scattering in the
planar spray imaging, thereby strongly improving the visualization
[28–31]. With the conventional method, there is a multitude of scat-
tering in all directions which seriously affect the expected results. The
main motivation for developing this new technique is to eliminate
blurring effects due to multiple scattering of light in order to obtain
qualitative and quantitative information such as the extinction coeffi-
cient [32]. While the transmission measurement records the light in-
tensity of a single beam crossing the sample of interest (where the in-
itial and final intensities are recorded), SLIPI makes the imaging of the
spatially modulated light sheet. (from 90° angle).
In this contribution, the single-phase SLIPI approach [14,16] is used
for the measurements of the extinction coefficients. Fig. 2 shows a
schematic of the experimental setup. A sample of transformer oil into a
silica cuvette, is illuminated by a spatially modulated laser sheet. The
light sources consist of two continuous wave (CW) laser beams of
450 nm and 638 nm, which are recombined along the same optical path
using a dichroic mirror. One laser is activated at a time and a neutral
density wheel is used to adjust the incident irradiance in order to op-
timize the signal to noise ratio, while avoiding saturation. The laser
sheet is performed by means of a succession of spherical and cylindrical
lens. Subsequently, a reflecting mirror is placed at 45° which returns the
laser sheet on the sample after passing through a Ronchi grating (2 lines
pairs/mm) creating intensity modulation. The images are recorded
using a 14 bit Electron Multiplying CCD camera, Luca (r) from Andor
with 1002 × 1004 pixels and the final image is the results of the ac-
cumulation over 200 single-shot images. This camera located at 90°,
perpendicular to the direction of propagation of the laser sheet, cap-
tures images of the modulation created into the cuvette. The camera
objective is set to F# = 5.6 and the exposure time ranges between
0.002 and 0.015 s to optimize the dynamic range for each sample of oils
concentration. The camera linked to a computer records the different
images obtained. Fig. 5 is the schematic of the experimental set up.
3.2. Single-phase SLIPI post-processing
The single-phase SLIPI approach used in the current study and
shown in [14] is based on an analytical tool known as lock-in detection
[33]. Lock-in detection is most often associated with analysis of tem-
porally varying signals but it works equally well for spatially modulated
signals. To explain the principle of lock-in detection, consider a 1D
signal, I(x), with a superimposed periodic variation of an amplitude IS
in space:
I (x ) = IS sin(2πνx + ϕ) + IMS (x ) (2)
−1
where ν equals the spatial frequency (mm ) of the modulation and ϕ is
the spatial phase, which is unknown. The IMS term represents any un-
wanted non-modulated intensity contribution also being detected, such
as, multiple light scattering or any surrounding light background. The
purpose of the lock-in analysis is to extract IS and reject IMS. To achieve
this end, the signal I is multiplied with two reference signals R1 and R2,
created computationally, that have a relative phase shift of π/2 :
R1 (x ) = sin(2πνx ) and R2 (x ) = cos(2πνx ) (3)
Multiplying I with these reference signals yields:
G.-O. Regnima et al. Measurement 113 (2018) 38–45

Fig. 5. SLIPI optical set-up: After expanding the

incident laser beam, a light sheet is formed using
a cylindrical lens focusing the beam into the
cuvette. The light sheet is modulated just prior to
enter the cuvette containing the oil sample by a
Ronchi grating of 2 lp/mm frequency. The images
are recorded using an EM−CCD camera.

1
I1 (x ) = IS (cos(ϕ)−cos(4πνx + ϕ)) + IMS sin(2πνx )
2 (4)

1
I2 (x ) = IS (sin(ϕ) + sin(4πνx + ϕ)) + IMS cos(2πνx )
2 (5)

The frequency analysis of I1 and I2 reveals three components; (1) a

DC component, (2) one modulated with 2ν and (3) one modulated with
ν. The two latter components can be suppressed by means of a low-pass
filter in the Fourier domain, with a cut-off frequency less than ν, re-
sulting in the following expressions:
∼ 1∼ ∼ 1∼
I1 (x ) = IS cos(ϕ) and I2 (x ) = IS sin(ϕ)
2 2 (6)

where the tilde assignment indicates the applied frequency filtering.

∼
From these, IS can finally be extracted by calculating:
∼ ∼ ∼
IS = 2 (I1)2 + (I2)2 (7)

An illustration of the single-phase SLIPI process is given in Fig. 6.

From a modulated image, the amplitude of the modulation is extracted
using Eqs. (3–7). The modulated component, IS, corresponds to the
single light scattering which reduces exponentially with distance. By
applying an exponential fit to IS the extinction coefficient, µe, can be
directly extracted as stated in the Beer-Lambert-Bouguer law (see Eq.
(1)).

4. Results and discussion

The experimental results of SLIPI measurements are presented from

Figs. 7–10 for the different samples of power transformers oils. The
values of the extinction coefficients obtained at the different wave-
lengths (450 nm, 638 nm) are reported on Figs. 9 and 10.
The results of the extinction coefficients values obtained from the
curves on Fig. 9 are reported in the Fig. 11. These curves show how the
extinction coefficients of power transformers oils change as a function
of aging duration.
In regard to the results obtained with SLIPI technique (Fig. 9), for an
Fig. 6. Principle of single-phase SLIPI. The example shows the signal from a structured
in-service transformer oil (2000 h and 1000 h), the exponential decay
laser sheet, with cross-sections extracted from two different depths, marked as A and B.
of the light intensity as function of distance through the cuvette (a and Notice the decrease in amplitude from column A to B. The 1 D Fourier transforms of
b respectively) is more pronounced than the exponential decay of light curves A and B, shows the reduction in strength of the 1st order peak (modulation fre-
intensity as function of distance through the cuvette (c) obtained for the quency). This frequency is then isolated using frequency filtering (red area) after applying
fresh oils (0 h). These observations point out the effect of transformers the lock-in algorithm. Finally, the exponential decay is revealed, as shown in the SLIPI
oils aging. In fact, its express the fact that SLIPI technique is able to see image and related curve IS. (For interpretation of the references to color in this figure
legend, the reader is referred to the web version of this article.)
the effect of aging on power transformers oils.
From the presented results on Fig. 10, one can divide for each power

42
G.-O. Regnima et al.
transformer oil samples S9, S10, S11 and S12, the extinction coeffi-
cients obtained for 450 nm and 638 nm illumination wavelengths such
that R = μe (450 nm)/ μe (638 nm) . R corresponds to the ratio of the ex-
tinction cross-sections (σe(450 nm)/σe(638 nm)). This ratio calculated
for each sample is more or less equal to R = 26. It value is constant and
is, therefore, an optical parameter characterizing the probed liquid.
This implies that measurements are reliable. It means that power
transformer oil particles diffuse and absorb the blue light 26 times more
than the red light. Thus from this experimentation, behave power
transformer oil when it is in interaction with a radiation can be assess. It
can be observed that it would be better to use red laser at 638 nm il-
lumination wavelength to perform measurements in the case of power
transformer oils with aging duration superior or equal to 2000 h.
The curves in Fig. 11 show for 450 nm illumination wavelength
used, an increase in the values of the extinction coefficients of different
power transformers oil samples when the aging duration increases. By
comparing curves, the extinction coefficients of the in-service trans-
former oils (2000 h and 1000 h) are higher compared to the one of fresh
oils (0 h). The increase of this optical parameter indicates that power
transformers oils are subjected to an aging. In fact, in the case of fresh
43
Measurement 113 (2018) 38–45
Fig. 7. Images of the structured light sheet
crossing the cuvette containing the samples
S5, S6, S7 and S8 of power transformer oils
for 450 nm illumination wavelength. The
corresponding SLIPI images are compared in
the figures below. As these oils are mostly
absorbing, the blurring effects due to mul-
tiple scattering are less apparent. However
there is still some remaining undesired
scattered light which must be suppressed
using SLIPI. The resulting SLIPI images de-
pict, thus, the correct light extinction where
an exponential fitting of the intensity decay
provide the corresponding extinction coeffi-
cient.
oils (0 h), the extinction coefficients are very low which show that they
are not deteriorated. When the oils are in service, they change their
appearance with the time and become more and more optically dense.
Moreover, there is an increase in the concentration of chromophores in
the samples. This aging is reflected in some research works by an in-
crease of the absorbance as a function of the aging duration or wave-
length [34,35]. Furthermore, there is a relationship linking the absor-
bance to the extinction coefficient. This relationship is given by the Eq.
(8):
μe = A. ln(10)/ L (8)
where L is equal to 16 mm, corresponding to the length of the cuvette
used in the measurements.
A is the absorbance value.
From Eq. (8), it is clear that if the absorbance increases, the ex-
tinction coefficient will increase too. Thus, the values of the IFT, DDP
and turbidity measurements agree with the obtained results. The in-
service transformer oils (2000 h and 1000 h) have high extinction
coefficients than the new oil (0 h). So the SLIPI has the great ability to
analyse the oil condition to large scale.
Fig. 8. Images of the structured light sheet
crossing the cuvette containing the samples
S9, S10, S11 and S12 of power transformer
oils for 638 nm illumination wavelength.
The corresponding SLIPI images are com-
pared in the figures below. As these oils are
mostly absorbing, the blurring effects due to
multiple scattering are less apparent.
However there is still some remaining un-
desired scattered light which must be sup-
pressed using SLIPI. The resulting SLIPI
images depict, thus, the correct light ex-
tinction where an exponential fitting of the
intensity decay provide the corresponding
extinction coefficient.
G.-O. Regnima et al. Measurement 113 (2018) 38–45

1 Fig. 9. Curves obtained by vertically integrating the SLIPI

2000 hours S11 µe = 1.989 mm-1
image results of the power transformer oils for 2000 h (a),
2000 hours S12 µe = 2.011 mm-1 1000 h (b) and 0 h (c) for 450 nm illumination wavelength.
Normalized intensity [au]

0.8 2000 hours S10 µe = 1.998 mm-1 The exponential decay of the light intensity as function of
distance through the cuvette is obtained here and the ex-
2000 hours S9 µe = 1.966 mm-1
tinction coefficient has been deduced from those decays.
(c) 1000 hours S7 µe = 0.406 mm-1
0.6
1000 hours S8 µe = 0.387 mm-1
(a) 1000 hours S6 µe = 0.537 mm-1
0.4 1000 hours S5 µe = 0.513 mm-1

0 hour S1 µe = 0.053 mm-1

0.2 0 hour S4 µe = 0.064 mm-1

(b) 0 hour S3 µe = 0.061 mm-1

0 hour S2 µe = 0.059 mm-1

0
0 2 4 6 8 10 12 14
Distance [mm]

For 450 nm illumination wavelength (Fig. 9), the extinction coeffi-

cient of fresh oils is between 0.053 mm−1 and 0.064 mm−1. For those
at 1000 h, the extinction coefficient is between 0.387 mm−1 and
0.537 mm−1. Finally, at 2000 h, it is between 1.966 mm−1 and
2.011 mm−1.
The results of the accuracy measurements have been obtained by
performing for the same sample of transformer oil for 450 nm illumi-
nation wavelength, about ten measurements of the extinction coeffi-
cient. The extinction coefficient measurement for each sample in-
cluding the deviation of the measurements are given in Table 4. Table 4
summaries also the guidelines for extinction coefficients values at
450 nm wavelength, expressed in mm−1.
These different intervals found will constitute therefore, the inter-
vals of decision which gives us the state of good or bad quality of power
transformer oils.
The increase of the extinction coefficients values determined by
means of SLIPI reflecting the level of degradation of oils, justifies the
evolution of the different physicochemical parameters determined. By
comparing both the different results obtained (SLIPI technique and
physicochemical methods), a good agreement was observed.
Besides, note that SLIPI technique has been able to detect a change
in the value of the extinction coefficient around by 0.1 [14]. In the case
of aging oils without color large change, SLIPI has demonstrated its
Fig. 11. Evolution of the extinction coefficient of each sample of transformer oils as a
ability in detecting a change in the value of extinction coefficient which function of aging duration at λ=450 nm illumination wavelengths.
approach 0.01. This order of magnitude was obtained by making the
subtraction between the extinction coefficients for each aging duration.
5. Conclusion
It is therefore a technique which is able to detect the slightest aging
effect of power transformer oils.
In this article, a new technique of optical imaging called SLIPI has
been investigated. Since SLIPI technique gives excellent results, it can

1 Fig. 10. Curves obtained by vertically integrating the SLIPI

[ 450 nm ] image results of the power transformer oils for 2000 h for
2000 hours S11 µe = 1.989 mm-1 450 nm (a) and 638 nm (b) illumination wavelengths. The
exponential decay of the light intensity as function of dis-
Normalized intensity [au]

0.8
2000 hours S12 µe = 2.011 mm-1 tance through the cuvette is obtained here and the extinction
(b) coefficient has been deduced from those decays.
2000 hours S10 µe = 1.998 mm-1
0.6
2000 hours S9 µe = 1.966 mm-1

[ 638 nm ]
0.4 2000 hours S11 µe = 0.076 mm-1

(a) 2000 hours S12 µe = 0.079 mm-1

0.2
2000 hours S10 µe = 0.078 mm-1

2000 hours S9 µe = 0.075 mm-1

0
0 2 4 6 8 10 12 14
Distance [mm]

44
G.-O. Regnima et al.
Table 4
Guidelines for extinction coefficient values at 450 nm.
Oil condition Extinction coefficient (mm−1)
Good oils 0–0.0640.006
Proposition an oils 0.064 ± 0.006–0.387 ± 0.092
Marginal oils 0.387 ± 0.092–0.537 ± 0.014
Bad oils 0.537 ± 0.014–1.996 ± 0.024
Very bad oils 1.966 ± 0.024–2.011 ± 0.02
Extremely bad oils > 2.011 ± 0.02
be used to serve as a tool for diagnosis of the state of the power
transformers oils for their deterioration and aging. It is a non-destruc-
tive technique with no need to perform dilution like the UV–visible
spectroscopy technique. The results obtained have been compared to
existing physicochemical techniques and were found very satisfactory.
The technique of SLIPI offers a real opportunity and a better alternative
for diagnosis of oils compared to other techniques which are time
consuming and expensive.
Acknowledgement
The authors would like to thank the International Science Program
(ISP) of Uppsala University for the donation of equipment and financial
support as well as the TWAS. The authors would like to thank Assoc.
Prof. Edouard Berrocal for useful discussions and valuable guidance.
Conflicts of interest
The authors declare no conflict of interest.
Appendix A. Supplementary material
Supplementary data associated with this article can be found, in the
online version, at http://dx.doi.org/10.1016/j.measurement.2017.08.
019.
References
[1] H. Boyer, M. Norbert, R. Philippe, Cours de construction du matériel électrique-
Tome 3, La Capitelle ed., France, 1981.
[2] J. Zhang, X. Qian, Y. Zhou, Application of extension neural network in transformer
fault diagnosis, J. Comp. Eng. Appl. 47 (7) (2011) 8–11.
[3] S.D. Myers, Presentation on transformer oil testing, http://www.slideshare.net/
JoeloRoss/transformer-oil-testing-sd-myers, 2016 (accessed 11.10.16).
[4] L.O. Séraphin, Bien-fondé des analyses d’huiles pour le diagnostic d’état des
transformateurs immergés. http://www.laboratoireoksman.com/Analyse_
transformateur.pdf, 2016 (accessed 11.10.16).
[5] I. Fofana, Insulating oil - another vital part of transformer body, In the IEEE seminar
Section DEIS Chapter, Ottawa-Montreal, Canada, June 17th 2010.
[6] R. Fournié, Les isolants en électrotechnique: Essais, Mécanisme de dégradation,
Applications industrielles, Eyrolles ed., France, Paris, 1990.
[7] Y. Du, Moisture equilibrium in transformer paper-oil systems, IEEE Electr. Insul.
Mag. 15 (1) (1999) 11–20.
[8] T.V. Oommen, On-line moisture sensing in transformers, in: Proceedings of the
electrical electronics insulation conference, Boston, USA, 7-10 October 1991.
[9] B.K. Gupta, Direct determination of moisture in solid oil-paper insulation, in:
Conference record of IEEE international symposium on electrical insulation,
Ontario, Canada, 7–10 June 1998.
[10] T.K. Saha, Review of modern diagnostic techniques for assessing insulation condi-
tion in aged transformers, IEEE Trans. Dielectr. Electr. Insul. 10 (5) (2003)
903–917.
[11] American Society for Testing and Materials (ASTM), Annual Book of ASTM
Standards, in: Electrical insulation and electronics, 2005, p. 577.
45
Measurement 113 (2018) 38–45
[12] American Society for Testing and Materials (ASTM), D 971-12 Standard test method
for interfacial tension of oil against water by the ring method, vol. 10.03, 2012.
[13] N. Thakur, O.P. Rahi, Transformer diagnosis using UV-spectrophotometer, Int. J. of
Electron. Signals Syst., www.interscience.in, 2015 (accessed 18.05.15).
[14] G.O. Regnima, T. Koffi, O. Bagui, A. Kouacou, E. Kristensson, J. Zoueu, E. Berrocal,
Quantitative measurements of turbid liquids via structured laser illumination planar
imaging where absorption spectrophotometry fails, Appl. Opt. 56 (13) (2017)
3929–3938, http://dx.doi.org/10.1364/AO.56.003929.
[15] O.K. Bagui, K.A. Kaduki, E. Berrocal, J.T. Zoueu, Structured laser illumination
planar imaging based classification of ground coffee using multivariate chemo-
metric analysis, Appl. Physics Research 8 (3) (2016) 32–44, http://dx.doi.org/10.
5539/apr.v8n3p32.
[16] E. Berrocal J. Johnsson E. Kristensson M. Aldén Single scattering detection in turbid
media using single-phase structured illumination filtering. J. Eur. Opt. Soc-RAPID-7
2012.
[17] HACH. 2100AN Laboratory Turbidimeter Manuel d'Utilisation, 2nd ed., http://
www.hach.com/2100an-laboratory-turbidimeter-epa-115-vac/productdownloads?
id=7640450972, (accessed 06.06.15).
[18] American Society for Testing and Materials (ASTM), D 6181 – 03 Standard test
method for measurement of turbidity in mineral insulating oil of petroleum origin,
in: Electrical insulation and electronics, USA, ASTM, 2003, p. 5.
[19] J.S. N’cho, I. Fofana, Y. Hadjadj, A. Beroual, Review of physicochemical-based di-
agnostic techniques for assessing insulation condition in aged transformers,
Energies 367 (9) (2016), http://dx.doi.org/10.3390/en9050367 10.
[20] I. Fofana, A. Bouaicha, M. Farzaneh, J. Sabau, D. Bussières, Decay Products in the
Liquid Insulation of Power Transformers, IET Electr. Power Appl. 4 (2010)
177–184.
[21] K.S. Kassi, 1. Fofana, F. Megnefi, A. Betie, Z. Yeo, Decay products in mineral oil: a
comparative study between hybrid and conventional insulation systems, IEEE
International Conference on Liquid Dielectrics, Bled, Slovenia, June 30 - July 3,
2014.
[22] American Society for Testing and Materials (ASTM), D 971 – 99a standard test
method for interfacial tension of oil against water by the ring method1, in:
Electrical insulating liquids and gases, electrical protective equipment, USA, ASTM,
1999, p. 3.
[23] American Society for Testing and Materials (ASTM), Standard Test Method for
Interfacial Tension of Oil against Water by the Ring Method. D 971, http://www.
biolinscientific.com/zafepress.php?url=%2Fpdf%2FAttension%2FApplication
%20Notes%2FAT_AN_10_astm.pdf (accessed 12.02.16).
[24] J.S. N'Cho, I. Fofana, A. Beroual, T. Aka-Ngnui, J. Sabau, The gassing tendency of
various insulating fluids under electrical discharge, IEEE Trans. Dielectr. Electr.
Insul. 18 (5) (2011) 1616–1625.
[25] FIST Volume 3–5, Maintenance of Liquid Insulation: Mineral Oils and Askarels,
http://www.worldcat.org/title/maintenance-of-liquid-insulation-mineral-oils-and-
askarels/oclc/45032750, 1992, (accessed 10-01-16).
[26] E. Berrocal, Multiple scattering of light in optical diagnostics of dense sprays and
other complex turbid media, PhD. Dissertation, School of Engineering, Cranfield
University, England, 2006.
[27] E. Kristensson, Structured laser illumination planar imaging SLIPI. Applications for
spray Diagnostics, PhD. Dissertation, Division of combustion physics, Lund
University, Sweden, 2012.
[28] E. Kristensson, E. Berrocal, M. Richter, S.G. Pettersson, M. Aldén, High-speed
structured planar laser illumination for contrast improvement of two-phase flow
images, Opt. Lett. 33 (2008) 2752–2754.
[29] E. Berrocal, E. Kristensson, M. Richter, M. Linne, M. Aldén, Application of struc-
tured illumination for multiple scattering suppression in planar laser imaging of
dense sprays, Opt. Express 16 (22) (2008) 17870–17881, http://dx.doi.org/10.
1364/OE.16.017870.
[30] E. Kristensson, E. Berrocal, M. Richter, M. Aldén, Nanosecond structured laser il-
lumination planar imaging for single-shot imaging of dense sprays, Atomization
Spray 20 (2010) 337–343.
[31] M. Storch, Y. Nath Mishra, M-. Koegl, E. Kristensson, S. Will, L. Zigan, E. Berrocal,
Two-phase SLIPI for instantaneous LIF and Mie imaging of transient fuel sprays,
Opt. Lett. 41 (2016) 5422–5425.
[32] E. Kristensson, E. Berrocal, M. Aldén, Extinction coefficient imaging of turbid media
using dual structured laser illumination planar imaging, Opt. Lett. 36 (2011)
1656–1658.
[33] M.L. Meade, Advances in lock-in amplifiers, J. Phys. E. Sci. Instrum. 15 (1982)
395–403.
[34] J.S. N’cho, A. Beroual, I. Fofana, T. Aka-Ngnui, J. Sabau, Verification of Oil
Reclamation Process by Turbidity and Spectrophotometry Measurements, J. Energy
Power Eng. 6 (2012) 703–712.
[35] R. Shoureshi, V. Permana, R. Wood, R. Swartzendruber, M. Simoes, Intelligent
transformer monitoring system utilizing neuro-fuzzy technique approach, power
systems engineering research center, school of engineering and computer science,
University of Denver, USA, 04-27 July 2004.