TRANSFORMER OIL DETECTION FEASIBILITY

WITH A LOVE WAVE SENSOR

Arthur S. Souza 1, Raimundo C. S. Freire 1 , Luiz A. M. M. Nobrega 1* ,
Marlo A. Santos 1,3,4, Ollivier Tamarin 2,4, Dominique Rebière 4, Corinne Dejous 4,
João V. J. de Melo1
1
Federal University of Campina Grande, Aprígio Veloso, 882, Campina Grande, Brazil
2
Université de Guyane, 97337 Cayenne, French Guiana
3
Federal Institute of Pernambuco, Av. Professor Luiz Freire, 500, Recife, Brazil
4
Univ. Bordeaux, IMS, CNRS UMR 5218, Bordeaux INP, 351 Cours de la Libération, 33405 Talence, France
*[email protected]

Keywords: Love Wave Sensor, Oil, Power Transformer.

Abstract
In this paper, the use of a Love wave sensor is presented to verify the feasibility of detecting transformer oil. It was demonstrated
that the Love wave sensor response exhibits significant variation with oil on its acoustic path, when compared to the bare sensor.
We noted a difference of approximately -17.249  0.552 dB in insertion loss magnitude and a frequency shift of approximately
125 kHz, when observing the transmission coefficient (S21). Moreover, the phase of the S21 parameter exhibits a good linearity
over the working region of the sensor as well as a phase shift in the sensitive region when the oil is introduced, which can also
be considered as a frequency shift. This result is a good indication of the feasibility of using a Love wave sensor as chemical
sensor for transformer oil.

1 Introduction parameters of liquids [11-13] and among them, the viscosity of

automotive oil [14].
Electrical transformers are one of the most significant and
expensive parts of an Electric Power System. The permanent This paper presents the use of a LW sensor in the VHF range
operation of transformers can generate failures, which can lead to verify the feasibility of detecting transformer oil. For this,
to safety issues and unexpected interruptions of the system. measurements of the LW sensor were carried out with real
Such failures can also entail significant financial costs [1]. In transformer oil, and compared with air. This work is organized
this way, the preventive diagnosis can prevent potential as follows: the first part is dedicated to the materials and
damages of the power transformers, thus increasing the methods; the second presents the experimental setup used for
reliability of the system, reducing costs and eventual problems measurement; later the results obtained and the discussions;
[2]. finally, the conclusions and future perspectives are presented.

Most transformers use mineral oil inside with the function of 2 Materials and Methods
insulating and cooling. Consequently, the service lifetime of
transformers directly depends on the physico-chemical state of 2.1 Transformer Oils Characteristics
the oil, whose parameters deteriorate during operation due to
Transformer oil is produced by refining a given fraction of
humidity and oxidative thermal aging [3-4]. Therefore,
hydrocarbons collected during distillation of a quantity of
monitoring the oil quality in a predictive way is of great
crude oil. The process is carried out in such a way that the
importance, since the state of conservation of the transformer
resulting oil has the characteristics within the permitted limits
is directly linked to that of the oil. Several works address this
[15]. It can be paraffinic or naphthenic in origin [16]. Among
problem and bring solutions for oil quality monitoring, as can
the most important physical characteristics to observe are
be seen in [5-7].
permittivity, resistivity, power factor and viscosity. Typically,
oil with optimum quality, have a relative permittivity εr = 2.2,
Surface acoustic wave (SAW) sensors, based on the Love wave
resistivity of 109 Ω.cm or greater, and a dissipative power
(LW), with shear horizontal polarization (SH), have been used
factor of approximately 0.001 at 60 Hz [17]. The kinematic
as an alternative with high sensitivity to detect chemical agents
viscosity according to the ASTM D3487 standard must not
in liquids or gases. Among the applications, the use as a
exceed the value of 76.0 cSt for 0 °C, 12.0 cSt for 40 °C and
biosensor [8-9] stands out. This type of sensor can typically be
3.0 cSt for 100 °C [18]. Such characteristics allow the oil to
used in a delay line or resonator configuration. Researchers
function mainly as an insulator, preventing the electric arc, and
also explore the use of wireless solutions for liquid detection
cooling, as the core generates heat due to eddy current losses
[10]. Finally, we also noted works that deal with the use of
depending on the type of material [19]. Therefore, monitoring
SAW sensors for applications that aim to quantify physical
these parameters in a predictive way can increase the

1
transformer's useful lifetime, which is directly correlated to the
state of conservation of the oil.

2.2 Love wave Sensor

The principle of the LW SAW sensor used for the experiment
is represented on Fig. 1 a). A typical device is shown in
Fig. 1 b). This device is designed by IMS Laboratory in
Bordeaux – France, and fabricated by LAAS CNRS
b c
Laboratory in Toulouse - France. The sensor consists of two
interdigitated titanium gold electrodes (IDTs) with 88 double Fig. 1 (a) Typical Love wave device principle adapted from
fingers and 150 nanometers deposited on a piezoelectric quartz [23], (b) Love wave dual delay-line when a liquid is added on
substrate. An additional silicon dioxide thin film called the surface acoustic path of the LW sensor, (c) test cell with
“guiding layer” of 4 micrometer thick is then deposited on the PDMS microfluidic chip when a liquid is added.
device (quartz and IDTs) with open contacts for electrical
connection. It is based on electroacoustic transduction in the 2.3 Sensor Test Protocol
delay line configuration. When a liquid is deposited on its
surface including IDTs and acoustic path, the acoustic and Initially, the sensor was cleaned using distilled water, isopropyl
electromagnetic waves propagation are modified in magnitude, alcohol and acetone. Then, the measurement of the sensor
frequency, phase velocity and input impedance. These changes exposed to air was made. Subsequently, using a micropipette
occur from the physical characteristics of the liquid used, such with 50 microliters of transformer oil, the liquid was inserted
as permittivity, conductivity and viscosity [13, 20]. In Fig. 1 b) onto the surface of the sensor, being restricted to the acoustic
and c) it is shown the liquid localization for our measurements path, using a short chamber PDMS chip [22]. After each
is restricted onto the acoustic path between the IDTs of the measurement (in air and with fluid), the test cell designed by
sensor with a PDMS (polydimethylsiloxane) microfluidic chip IMS Laboratory (cf. Fig. 2 d)) was disconnected and
(called short chamber PDMS chip as excluding the IDT disassembled. Subsequently, the process is carried out again in
regions) also designed and fabricated by IMS laboratory. The the cell for a new measurement. In order to minimize errors,
measurement principle of the LW sensor is carried out from the this process was repeated several times.
incident and reflected energies in both ports (input and output
IDTs). As presented in Fig. 1 a), when electrically interrogated,
3 Experimental Setup
the input IDT transforms the RF energy (Vin) into an acoustic
wave based on the piezoelectric effect. The acoustic wave
propagates along the acoustic path towards the output IDT,
where there is a new conversion to an RF wave (Vout). The
relationship between the input and output energy (performed
from the S parameters, for example) allows the
characterization of wave propagation in term of resonance
frequency, insertions losses, and phase. Thus, from S11 and S21
parameters, it is possible to estimate the impedance change of
the sensor in the real and imaginary part and the acoustic LW
propagation properties in order to extract information about the
liquid. Thus, in a first approach, the acoustic characteristics, a
such as the propagation of acoustic wave properties, are
modified when the liquid is inserted. Several works have been
carried out in this sense using glycerol, distilled water, salt
water and isopropyl alcohol. They all go in the direction of
determining the conductivity but also the viscosity from the
characteristics of the liquid deposited onto its surface [10, 14,
21, 22]. b c

Fig. 2 (a) Experimental setup with LW sensor in wired

configuration, (b) test cell with sensor, (c) transformer oil and
a (d) all parts of the test cell with the PDMS chip.

2
The experimental setup is shown in Fig. 2 a). The computer (1) transmission coefficient S21. With air (blue curve) and oil (dash
is connected via USB cable to the nanoVNA (with an output orange curve).
power of 0 dBm and a frequency range from 10 kHz to
1.5 GHz) (2), this connecting via coaxial cable (20 cm of Table 1 Result of measurements of the reflection and
model RG 405) (3) to the sensor in the test cell, mainly transmission coefficients.
composed of a printed circuit board with electrical connexions
(5), a metal basis which contains the sensor (6) and the short
S11 Freq. S21 Freq.
chamber PDMS chip (7). A full 2-port calibration SOLT
(dB) (MHz) (dB) (MHz)
(Short-Open-Load-Through) was performed at the end of the
cables over the 100 to 130 MHz range. After cleaning and -1.260 -32.761
Air 118.771 Air 119.065
calibrating with appropriate solutions and protocol (4), the test  0.015  0.242
cell containing the LW sensor has been connected and the S -1.264 -50.010
Oil 118.803 Oil 118.940
parameters were measured. This process was repeated ten  0.025  0.310
times in the same way to minimize errors and assess the -0.004 -17.249
∆ 0.032 ∆ -0.125
repeatability.  0.04  0.552

4 Results
From the analysis of Fig.3 and Table 1 it was possible to verify
The magnitudes of the reflection and transmission coefficients that when exposed to air the sensor presented an average for
and phase of the sensor in the experimental setup, with and the maximum value of the magnitude of the transmission
without liquid deposited onto its surface on the acoustic path, coefficient (S21) of -32.761  0.242 dB at a frequency of
as previously described, are presented in Fig. 3 and Table 1. 119.065 MHz, for the magnitude of the reflection coefficient
(S11) the minimum value was -1.260  0.015 dB at 118.771
MHz. When the liquid was inserted, the maximum of S21 varied
to -50.010  0.310 dB with a frequency shift of 118.940 MHz,
while S11 had a minimum value of -1.264  0.025 dB for a
frequency of 118.803 MHz. It was possible to verify that there
is a variation of approximately -17.249  0.552 dB for the
magnitude of the transmission coefficient and a shift of
125 kHz when the liquid is added. The reflection coefficient
varied from -0.004  0.04 dB with a frequency shift of 32 kHz
which remains non-significant. These results are due to the fact
a that when oil is added to the acoustic path, there is an
attenuation of the S21 signal due the variation in the acoustic
properties imposed by the liquid, that is, a reduction in the
magnitude of the recovered output signal when compared to
the input signal. Observing the phase of S21 parameter, it can
be observed that in the working region of the sensor, it is linear
and that there is a phase shift, also corresponding to a
frequency shift, when the oil is inserted in the sensitive region.
This leads to the conclusion that the sensor is sensitive to the
detection of transformer oil, that is, when the liquid is
deposited on the sensitive acoustic path, there is a significant
b variation in the S21 sensor response when compared to air.
Moreover, the non-significant variation of S11 coefficient
validates the use of the test cell with PDMS chip without oil
leakage from the acoustic path toward the IDTs.

5 Discussions
The results showed that the sensor is sensitive to oil insertion
in its acoustic path, with a good repeatability. In addition, other
works of the literature focused on the influence of liquids on
the response of such sensor [11-14], based on their different
parameters (such as viscosity and molecular weight, for
c
example), using different protocols and applications. This
would open future possibility of applying such a sensor as a
Fig. 3 (a) Magnitude of the transmission coefficient S21,
chemical sensor for oil transformer that would allow remote
(b) magnitude of the reflection coefficient S11, (c) phase of the
monitoring. This aspect is of great value, since the useful life

3
of transformers is directly linked to the quality of the oil [5] Bryakin, I. V., Bochkarev, I. V., Khramshin, V. R., et al.:
contained in it. As a suggestion for future works, it is the ‘Power Transformer Condition Monitoring Based on
measurement of viscosity, permittivity, conductivity and Evaluating Oil Properties’. Machines, 2022, 10, (8), pp. 630.
breakdown voltage of transformer oil using LW sensor with
[6] Degeratu, S., Rotaru, P., Rizescu, S., et al.: ‘Condition
long chamber PDMS Chip [22]. For this, it is necessary to
validate that the sensor is capable of distinguishing different monitoring of transformer oil using thermal analysis and other
qualities of oil. Therefore, it is necessary to measure these techniques’. Journal of thermal analysis and calorimetry, 2015,
parameters so that they can support the distinction of the 119, pp. 1679-1692.
qualities of the oils. [7] Pahlavanpour, B.; Wilson, A.: ‘Analysis of transformer oil
for transformer condition monitoring’. In: IEE Colloquium on
6 Conclusion An Engineering Review of Liquid Insulation (Digest No.
1997/003). IET, 1997. p. 1/1-1/5.
In this paper, a Love wave sensor was presented in order to
[8] Tamarin, O., Comeau, S., Dejous, C., et al.: ‘Real time
verify the feasibility of detecting transformer oil. For this, tests
were made with transformer oil, using the acoustic path of the device for biosensing: design of a bacteriophage model using
sensor. The results were compared with air. It was found that Love acoustic waves’. Biosensors and Bioelectronics, 2003,
when oil is inserted on the acoustic path, there is a variation in 18, (5-6), pp. 755-763.
the sensor response when compared to air. The difference from [9] Länge, K., Rapp, B. E., Rapp, M.: ‘Surface acoustic wave
the peak of S21 was approximately -17.249  0.552 dB with a biosensors: a review. Analytical and bioanalytical chemistry’,
frequency shift of approximately 125 kHz. Together with 2008, 391, pp. 1509-1519.
results in the literature, this leads to an indication of the
[10] Santos, M. A., Freire, R. C. S., Hallil, H., et al.: ‘Remote
feasibility of using the Love wave sensor as a chemical sensor
for transformer oil in future works. For this, it is suggested to Measurement of a VHF Love Wave Sensor for Liquid
test oils with different stages of degradation, that is, from new Detection’. In: 2022 Wireless Power Week (WPW). IEEE,
oil to unusable oil, and evaluate the physicochemical 2022. pp. 536-540.
parameters (permittivity, conductivity, breakdown voltage and [11] Raimbault, V., Rebière, D., Dejous, C.: ‘A microfluidic
viscosity) of the oil. surface acoustic wave sensor platform: Application to high
viscosity measurements’. Materials Science and Engineering:
7 Acknowledgments C, 2008, 28, 5-6, pp. 759-764.
The authors want to acknowledge the CAPES (Coordination [12] Raimbault, V., Rebière, D., Dejous, C., et al.: Acoustic
for the Improvement of HigherLevel Education Personnel), the Love wave platform with PDMS microfluidic chip.’ Sensors
National Council for Technological and Scientific and Actuators A: Physical, 2008, 142, pp. 160–165.
Development (CNPq), the Support Program for Centers of [13] Raimbault, V., Rebière, D., Dejous, C., et al.: ‘High
Excellence (PRONEX). The authors also want to warmly thank frequency microrheological measurements of PDMS fluids
Jean-Luc LACHAUD and Serge DESTOR, from IMS lab using saw microfluidic system.’ Sensors and Actuators B:
(Univ. Bordeaux, CNRS, Bordeaux INP), for their technical
Chemical, 2010, 144, 2, pp. 467-471.
commitment in the test cell fabrication and the associated
electronic setup. [14] Jakoby, B., Scherer, M., Buskies, M., et al.: ‘An
automotive engine oil viscosity sensor’, IEEE Sensors Journal,
8 References 2003, 3, (5), pp. 562-568.
[15] Rouse, T. O.: ‘Mineral insulating oil in transformers’,
[1] Bossi, A., Dind, J. E., Frisson, J. M., et al.: ‘An international IEEE Electrical Insulation Magazine, 1998, 14, (3), pp. 6-16.
survey on failures in large power transformers in service.’ [16] Lipshteĭn, R. A., Shakhnovich, M. I.: ‘Transformer oil’.
Cigré Electra, 1983, 88, pp. 21-48. Israel Program for Scientific Translations; [available from the
[2] Islam, M. M., Lee, G., & Hettiwatte, S. N.: ‘A review of US Department of Commerce, Clearinghouse for Federal
condition monitoring techniques and diagnostic tests for Scientific and Technical Information, Springfield, Va.], 1970.
lifetime estimation of power transformers. Electrical [17] Hwang, J. W. G.: ‘Elucidating the mechanisms behind
Engineering,’ 2018, 100, pp. 581-605. pre-breakdown phenomena in transformer oil systems’, PhD
[3] Wilson, A.C.M.: ‘Insulating liquids’. Their uses, thesis, Massachusetts Institute of Technology, 2010.
manufacture and properties, 1980. [18] ASTM D3487-00, ‘Standard Specification for Mineral
[4] Abdi, S., Safiddine, L., Boubakeur, A., et al. ‘The effect of Insulating Oil Used in Apparatus’, the United States: the
water content on the electrical properties of transformer oil’. American Society for Testing and Material, 2000.
In: Proceedings of the 21st International Symposium on High [19] Rengaraj, R., Venkatakrishnan, G. R., Moorthy, P., et al.:
Voltage Engineering: Volume 2. Springer International ‘Transformer Oil Health Monitoring Techniques—An
Publishing, 2020. pp. 518-527. Overview’. Advances in Smart System Technologies: Select
Proceedings of ICFSST 2019, 2021, pp. 135-154.

4
[20] Tamarin, O., Rube, M., Lachaud, J. L., et al.: ‘Mobile
acoustic wave platform deployment in the Amazon river:
Impact of the water sample on the Love wave sensor response’,
2020, Sensors, 20, 1, pp.72.
[21] Rube, M., Tamarin, O., Sebeloue, M., et al.:
‘Unconventional protocol for SAW sensor: multi- physic
response enrichment in liquid medium’, IEEE Sensors Journal,
2021, 22, (12), pp. 11345-11354.
[22] Choudhari, A., Rube, M., Sadli, I., et al.: ‘Love wave
acoustic sensor response in high turbidity liquid environment’.
In: 2022 IEEE Sensors, Dallas, TX, USA., 2022. pp. 1-4.
[23] Zimmermann, C.: ‘Conception, réalisation et étude de
micro-capteurs à ondes de love pour applications en milieu
gazeux: Cas de la détection de composés organophosphorés,’,
PhD thesis, Bordeaux 1, 2002.