21, rue d’Artois, F-75008 PARIS D1_203_2010 CIGRE 2010

http : //www.cigre.org

A novel optical fiber sensor system for temperature monitoring of power

transformers

J. H. LEE*1, M. H. SONG2, J. K. LEE3, S. H. KANG4, D. K. KIM5 AND J. S. IM6

1
Hoseo Univ., 2 Chunbuk Nat’l Univ., 3 Andong Nat’l Univ., 4 Myongji Univ. , 5 KETEP
and 6 Mokpo Maritime Nat’l Univ.
KOREA

SUMMARY

A precise information on spatial distributions of temperature are important for diagnosing

power transformers and evaluating its life because a power system failure will result in an
enormous loss in tangible as well as social. The optical fiber Bragg grating (FBG) sensors,
which have been studied intensively for last decade, can be very efficient tools for
applications to above mentioned purpose because these are immune to EMI and can be highly
multiplexed, which enables efficient quasi-distributed temperature sensing along tens of km
range.
We proposed a fiber-optic temperature monitoring system with an array of over 10 sensor
gratings. The temperature-induced Bragg wavelength variations are accurately monitored by a
scanned tunable wavelength filter. Differential measurement with temperature-stabilized
reference gratings and a Gaussian curve-fitting algorithm has been used to enhance
measurement accuracy.
In laboratory, we monitored the temperature of 20kVA oil-filled pole transformer using the
FBG sensors as well as the conventional thermocouple sensors. The results showed that the
FBG sensor could measure the temperature with less random noise and less influence from
electrical noises by switching or semiconductor operations.
The developed FBG sensor system has been applied to 3000kVA mold transformer which
was in operation on the real distribution network for 2 months. The performance of FBG
sensor has been proven to be stable enough for real scale application

KEYWORDS

Optical Fiber Sensor - Bragg grating - Power Transformer - Temperature

[email protected]
INTRODUCTION

One of the major factors which degrade the performance of power equipments such as
tranformers, generators, polymeric insulated cables, etc. is thermal stress exceeding the rating
temperature. In order to protect the equipment from unexpected failure or accident, the
continuous monitoring of temperature at desired spots with reliable accuracy is considerd as
very necessary method because the failure of power system usually results in an immense loss
of profit in consumers.
Most of conventional temperature monitoring systems are not suitable to measure the
temperature at the location to which the high electric fields are applied during the operation
because some parts of the systems are metalic or conductive materials. In addition, compact
physical dimensions and installation environment make it more difficult to implement bulky
conventional sensors. In this work, the optical FBG sensors has been proposed as a new
monitoring system to overcome the drawback of conventional sensors for power equipments,
especially for power transformers.
Since the first observation of photosensitivity in fiber [1], the use of FBGs in both
communications and sensing applications has been intensively studied in the last two decades
[2-7], because it has many important advantages. First, a FBG only reflects Bragg wavelength
and transforms the detected physical quantities to shifts in reflected Bragg wavelengths.
Because of this wavelength-encoding characteristic, the sensed information is independent of
source power fluctuations, total light level, losses in the connecting fibers and couplers and
the other environmental noise sources. Secondly, FBGs can be easily multiplexed in a serial
fashion, allowing many of them to be used on a single fiber, which enables efficient quasi-
distributed sensing [8-10]. In addition, the reflected Bragg wavelength shifts show linear
response to the change of fiber grating’s properties, which means any external physical
quantities applied to the gratings such as strain, pressure, temperature or vibration can be
recovered from the measured Bragg wavelength shifts.

EXPERIMENTS

50:50 coupler
BBS
Index
Sensor FBG
matching
Ref. FBG gel

Sync
F-P Function out
filter Generator

LPF
Photo Detector
Fig. 1 Schematic diagram of FBG temperature sensor system

Figure 1 is the experimental setup of proposed FBG temperature sensor system in this work.
An ASE broadband source with spectral range of 1530 nm ~ 1610 nm was used to illuminate
FBG sensor array which consists of 10 grating sensors with nominal center wavelength from

1
1542.09 ~ 1558.96 nm. A MEMS Fabry-Perot (FP) filter with 4 GHz bandwidth was used as a
wavelength scanning filter.
The reflected lights from the FBGs pass through the FP filter of which passband is modulated
by a ramp signal. Then the light is detected after passing through the FP filter, transforming
the wavelength-domain profiles into time-domain ones. Now the Bragg wavelength shifts are
measured by locating the peaks in the PD signal.
Figure 2 shows the PD signal with 12 temporal peaks that correspond to the Bragg
wavelength distribution. The peaks at both far ends are reference grating (RG) signals and
others are sensor grating (SG) signals. We used two RGs to suppress errors by the possible
non-linear action of FP filter. Wavelength variations of SGs are calculated relatively from
wavelength interval of the 2 RGs measured in each scan cycle.

(a)

(b)

 ,t

(c )

t
Fig. 2 FBG reflection spectra in optical(a) and time(b) domain. (b): FP filter scanning.

As described above, the accuracy of temperature measurement strongly depends on the

accuracy of peak locations. By several reasons, such as strain gradients and irregular stress
distribution applied to sensor gratings, the reflection spectra of grating sensors can be
distorted, resulting in inaccurate peak locations. To alleviate this problem, we applied
Gaussian line-fitting algorithm to the temporal profiles as in figure 3.

5.0

4.0
Amplitude [V]

3.0

2.0

1.0

0.0

1547.0 1547.2 1547.4 1547.6 1547.8 1548.0

Wavelegth [nm]
Fig. 3 Gaussian line-fitting(redline) for detecting more accurate peak location

2
FBG reflection peaks are assumed to be in Gaussian shapes, and the optimal fitting curve is
determined when the mean square error (MSE) between Gaussian function and temporal peak
is minimal. This also enhances the measurement resolution, because the center position of
Gaussian peak can be determined in the much smaller scale than the quantization error.
The SGs were put into a water vessel to test the feasibility of the FBG temperature sensor
system. Figure 4 shows the output variation when the SG was put into cold water. The upper
trace is the case with raw peak detection and the lower case is the case when the Gaussian
line-fitting algorithm was used. It is apparent that the line-fitting algorithm reduces the
random walks of the measurement results, enhancing resolution of the measurement.
30
Raw data Temperature [℃]

25

20

15

10

0 10 20 30 40 50 60 70
30 T im e [m in ]
Gaussian fitted data Temperature [℃]

25

20

15

10

0
0 10 20 30 40 50 60 70
T im e [m in ]

Fig. 4 Temperature measurements with raw peak detection (upper) and Gaussian line-fitting
(lower).

FBG reflection peaks are assumed to be in Gaussian shapes, and the optimal fitting curve is
determined when the mean square error (MSE) between Gaussian function and temporal peak
is minimal. This also enhances the measurement resolution, because the center position of
Gaussian peak can be determined in the much smaller scale than the quantization error.
The SGs were put into a water vessel to test the feasibility of the FBG temperature sensor
system. Figure 4 shows the output variation when the SG was put into cold water. The upper
trace is the case with raw peak detection and the lower case is the case when the Gaussian
line-fitting algorithm was used. It is apparent that the line-fitting algorithm reduces the
random walks of the measurement results, enhancing resolution of the measurement.

RESULTS AND DISCUSSION

The developed FBG sensor system applied to 20 kVA oil filled pole transformer in the
laboratory to simulate the power transformer. The temperature of the pole transformer can be
controlled by the overload test power supply. The conventional thermocouples of the same
number of FBG sensors were installed at the points adjacent to corresponding FBG sensors.

Fig. 5 (a) and (b) are the examples of measured temperatures from pole transformer by FBG
sensor and thermocouple sensor, respectively. It is showed that the data from FBG sensor has

3
less random noise than that from thermocouple and FBG sensor is less influenced by noises
such as switching or power semiconductor operations.

(a) (b)
Fig. 5 Measured temperatures at the same point: (a) from FBG sensor and (b) from
thermocouple sensor.

Fig. 6 Tested 3000kVA mold transformer with the 10 FBG sensors to monitor temperature
distributions for 2 months in operation.

Fig. 6 shows that the 10 FBG sensors were adhered on the 3000kVA mold transformer and its
temperature distribution continuously monitored for 2 months in operation.
All the sensors successfully measured temperatures during the period of 2 months and Fig.
7(a) is one example of the 10 monitored and recorded results. The measurement was
performed from May to July. The curve of monitored temperature increases as the function of
day. During the period of monitoring, the daily power consumption has been stable so that the
increase of measured temperature might be due to rise in ambient temperature. Fig 7(b) is the
average and average mean of ambient temperature during the monitoring period acquired
from Korea Meteorological Administration and it shows the coincident trend with the
monitored data curve.

4
 50

45
FBG8 Temperature[? ]

40

35

30

25

20
0 10 20 30 40 50
T im e [d a y ]

(a) One exapmle of monitored temperature (b) The average and average mean of ambient
temperature during the monitoring period
acquired from Korea Meteorological
Administration
Fig. 7 Monitored temperature from 3000kVA mold transformer in operation and daily
ambient temperature data from KMA.

CONCLUSION

A FBG array temperature sensor system has been proposed and developed for the reliable
temperature monitoring on power transformers. The temperature-induced Bragg wavelength
shifts are detected by a wavelength-scanned Fabry-Perot filter, and two reference gratings
were used to compensate any nonlinear action of the wavelength filter
In the laboratory, we monitored the temperature of 20kVA oil-filled pole transformer using
the FBG sensors as well as the conventional thermocouple sensors. The results showed that
the FBG sensor could measure the temperature with less random noise and less influence
from electrical noises by switching or semiconductor operations.
The developed FBG sensor system has been applied to 3000kVA mold transformer which
was in operation on the real distribution network for 2 months. The performance of FBG
sensor has been proven to be stable enough for real scale application.

ACKNOWLEDGEMENT

This research was supported by Korea Ministry of Knowledge Economy Grant # R-2008-25.

BIBLIOGRAPHY

[1] K. O. Hill, Y. Fujii, D. C. Johnson, and B. S. Kawasaki, “Photosensitivity in optical

fiber waveguides: Application to reflection filter fabrication,” Applied Physics Lett., 32,
647-649, 1978.
[2] J. A. R. Williams, I. Bennion, K. Sugden, and N. J. Doran, “Fibre dispersion
compensation using chirped in-fibre Bragg grating,” Electron. Lett., 30, (12) 985-987,
1994.
[3] A. D. Kersey, M. A. Davis, H. J. Patrick, M. LeBlanc, K. P. Koo, C. G. Askins, M. A.
Putnam, and E. J. Friebele, “Fiber Grating Sensors,” J. Lightwave Technol., 15, 1442-
1462, 1997.

5
[4] A. D. Kersey, T. A. Berkoff, and W. W. Morey, “Multiplexed fiber Bragg grating strain-
sensor system with a fiber Fabry-Perot wavelength filter,” Opt. Lett., 18, 1370-1372,
1993.
[5] S. H. Yun, D. J. Richardson, and B. Y. Kim, “Interrogation of fiber grating sensor arrays
with a wavelength-swept fiber laser,” Opt. Lett., 23, 843-845, 1998.
[6] M. Song, S. Yin, and P. B. Ruffin, “Fiber Bragg grating strain sensor demodulation with
quadrature sampling of a Mach-Zehnder interferometer,” Appl. Opt., 39, 1106-1111,
2000.
[7] C.C. Chan, W. Jin, H.L. Ho, D.N. Wang, and Y. Wang, “Improvement of measurement
accuracy of fibre Bragg grating sensor systems by use of gas absorption lines as multi-
wavelength reference,” Electron. Lett., 37, 742-743, 2001.
[8] W. W. Morey, G. Meltz, and W. H. Glenn, “Fiber Bragg grating sensors,” in Proc. SPIE,
1169, 98, 1989.
[9] W. W. Morey, G. Meltz, W. H. Glenn, “Bragg-grating temperature and strain sensors,”
in Proc. OFS, 526, 1989.
[10] W. W. Morey, J. R. Dunphy, and G. Meltz, “Multiplexing fiber Bragg grating sensors,”
in Proc. SPIE, 1586, 216, 1991.