Documente Academic
Documente Profesional
Documente Cultură
69 85 1.0 20.0
138 56 7.0 50
where Tf is the failure rate in the period under consideration [%], Rate (%) 1.71 2.33 1.46 0.40
Nf is the number of failures in the period under consideration,
Ne,i is the number of devices in each year i considered, and t is
the number of years of the period considered. fault, making it difficult to locate it, especially if its intensity is
Analyzed quantitatively, the slightly higher rates of the low. Particularly interesting is the use of a noninvasive method
138- and 230-kV transformers are mainly due to the fact that such as the acoustic PD detection method, which allows for
there are fewer of them. Table 3 lists the failure rates of the four monitoring of the evolution of PDs even while the device is in
voltage classes of transformers, without taking into consideration operation.
the reserves. Throughout its operation, a power transformer has to withstand
The transformer failure rates in CELG’s system are relatively numerous stresses that generally result in the degradation of the
low, which is attributable to the use of predictive techniques. oil–paper insulation system by decomposition of the paper and/
To further reduce the failure rates, the company’s maintenance or oxidation of the oil. Degradation reduces the quality of this
engineering department is seeking new predictive techniques, insulation. Partial discharges can lead to winding breakdowns
with emphasis on the detection of partial discharges (PDs) using and may cause accelerated aging. Partial discharges must
the acoustic method. be inferred to build an early warning system. In this context,
PDs serve as an important measuring parameter for on-line
Predictive Techniques monitoring [6].
The dissolved gas analysis (DGA) of insulating oil is To illustrate the above, the photograph in Figure 4 shows the
sensitive to some types of incipient faults (defects). To quantify failure of a 138-kV/13.8-kV, 33-MVA power transformer due to a
the efficiency of this technique in detecting defects in CELG’s lightning surge. Systematic DGA monitoring of this transformer
equipment, a comparison was made of the total number of showed a very slight increase in gas but failed to warn about the
transformer disruptions that could have been detected by DGA need to remove this device from service, ultimately leading to
and the disruptions effectively detected by this technique. damage of its bushing (Figure 5).
This comparison revealed an efficiency of approximately This fault could have been prevented by PD detection,
75%. However, sampling of transformer oil for DGA is done preventing the defect from developing into a short circuit
periodically, according to the chromatography software program between the winding spirals and, hence, failure. This indicates
CELG uses and to the specificity of each device. Thus, between the need for integrating predictive maintenance techniques
consecutive samplings, the device may undergo adverse to improve diagnostic efficiency and ascertain the state of
operating conditions, which may trigger or accelerate incipient transformer insulation.
faults and cause the device to fail before the next sampling, Therefore, this article highlights the combination of DGA
masking the efficiency of DGA. It is therefore understood that with the detection of PDs by acoustic emission [7], which allows
the efficiency of the DGA technique, per se, is higher than PD activity to be pinpointed in the transformer while in service.
75%. In addition to DGA, the detection of PDs may increase
the overall efficiency of the predictive technique as DGA has Conclusions
low sensitivity for detecting PDs [5] and this could possibly Although the failures rates and the number of disruptions that
lead to inaccuracy in the analytical method. Furthermore, DGA occurred during the 33-year period under study were relatively
does not allow for the identification of the site of an incipient low, it is important to implement predictive techniques other than
DGA that are sensitive to incipient faults in power transformers.
This is needed for the detection of defects involving windings,
bushings, and tap changers, which when taken together, account
for 69% of the disruptions caused by these components. These
techniques include the measurement of PDs by the acoustic
emission method, which in conjunction with DGA, could
increase the maintenance efficiency, reducing disruptions, and
therefore the quality of electric power supply.
Acknowledgments
This work was carried out in collaboration with the
Maintenance Engineering Division of CELG Distribuição,
CELG D, and the Federal University of Goiás School of
Electrical and Computer Engineering (EEEC/UFG) through a
Figure 5. Damage to bushing due to internal fault. partnership in an R&D Project - ANEEL.