This article describes the actions that were taken to check the condition of the insulation of a 230/115/45 kV transmission transformer in operation since 1967. The continuous monitoring of selected dielectric indicators was undertaken and different available methods were applied. The state of bushings insulation was checked by monitoring the absolute values of its capacitance (C) and dissipation factor (DF). Additionally, partial discharge (PD) measurements were performed with sensors at the bushing measuring taps and with ultra-high frequency (UHF) senor inserted in the transformer tank. Different challenges were faced during design and installation of the system and during the evaluation of the monitoring data. These are described in the paper.
Monitoring of Bushing C and DF
The utility decided to continuously monitor absolute values of DF and C of its resin bonded paper (RBP) bushings of the 230/115/45 kV transmission transformer. The reference signal is taken from a group of 0.2 accuracy class VTs located in the substation. This pure resistive signal is compared with the mainly capacitive leakage current measured at the bushing tap in order to derive the dissipation angle and then the DF. The temperature compensation is applied on DF values taking the insulation system of the bushing - resin bonded paper - into account.
The diagnosis of the bushing is performed by analyzing the DF trend, its magnitude and rate of change. In general, a continuous operation is not recommended when a predefined level of 1.5% is reached. On the other hand, bushings with DF values above this level but with a quite stable trend may stay in operation. Additional off-line diagnosis is recommended in cases when the DF doubles its value in a short period of time, e.g. in six months.
The trend diagrams of DF and the ambient humidity at the monitored bushings are charted. Figure 1 shows the comparison of the highest DF monitored value with the pre-set threshold limit and with the values from the off-line measurements performed after taking the transformer out of service for maintenance. The good state of the bushings insulation is confirmed.
The trend of the bushings capacitance together with the trend of ambient humidity are charted. The diagnosis is performed by comparing the capacitance values from continuous monitoring with the values from off-line measurements. The difference gives the capacitance variation C that has to remain within certain limits. The change of bushing capacitance when one capacitive grading layer is short-circuited for bushings of different rated voltage is charted. Taking the 230 kV rated voltage of the bushings of the monitored transformer into account, the presence of about 30 capacitive grading layers can be assumed. The short circuit of a single layer would lead to an increase of the bushing capacitance of about 3.3 %. Such a value can be set up as a threshold limit of the capacitance variation. Figure 2 shows the comparison of the bushing capacitance values coming from continuous monitoring and off-line measurements with the preset threshold limits. It can be noticed, that the values coming from monitoring are very close to those from the off-line measurements. The short circuit of one of the bushing grading layers does not necessarily lead to the immediate failure of the bushing, in case of resin bushings, but it creates a higher electric field on the healthy layers.
The accuracy of the applied monitoring system was < ±0.8 pF for capacitance measurement and ±0.01 % for dissipation factor measurement. The errors introduced by the VT class - calculated according to  - have to be considered.
Monitoring of PD
The PD activity was measured by a conventional detection method at the bushing measuring taps and an unconventional UHF method by placing a sensor inside the transformer tank.
Additionally, an acoustic method was applied for a precise localization of the PD sources. All these methods have a complementary character. The conventional type measurements indicate the defects in both, bushings and winding insulation. The UHF measurements are only sensitive to detect winding problems.
Unconventional UHF method
The UHF PD signal was measured in the frequency range from 0.1 to 2 GHz by an antenna type sensor installed inside the transformer tank. In order to gain information about the UHF frequency content of the signal, off-line and on-line frequency sweeps were performed. An off-line frequency sweep was performed during the installation of the monitoring system, while the transformer was de-energized. The off-line spectra gives information about the sources of interferences produced by other equipment in the substation. These sources are discarded when the analysis of the on-line detected PD signal is performed. The on-line sweep indicates the PD activity in the frequency range from 450 to 650 MHz. The trend line of the UHF signal and PRPD pattern of the signal (at the central frequency 560 MHz and bandwidth of 70 MHz) has been increasing since May 2014 and is shown in Figure 9.
The pattern, synchronized with the voltage in phase U, is generated by the PD source located around phase V. The PRPD patterns at frequencies above
1 GHz were also investigated and no PD activity was identified.
Three independent observers (sources) of the PD signal are necessary to build a 3PARD diagram, one signal for each axis of the diagram. In Figure 3, such a diagram was built with the UHF signal and with the two signals measured at the bushings of phase V and W. After the cluster separation, the PD activity was confirmed at the phases V and W. The PRPD patterns from the back transformation of clusters surrounded by red rectangles in Figure 3 show the same shape and phase position with those have to be mapped. Detecting the PD activity in the UHF range it is an important indication that the PD sources are located inside the transformer tank and not inside the bushings (C and DF results are confirmed).
Acoustic Localization of PD Sources
In order to obtain a more precise localization of the PD sources, acoustic PD measurements were performed in July 2014.
Figure 4 shows the shape of the acoustic PD signals and final position of the sensors that allowed the calculation of the PD source coordinates. According to the results of the acoustic method, the PD activity takes place at the exit leads of the HV winding of phase V. No PD source could be localized at phase W using the acoustic method.
In the next step, the transformer was de-energized and the bushing of phase V was dismantled for an internal investigation. PD activity traces around the phases V and W were found using an endoscope. Having the proof of the on-going PD activity, the utility decided to lower the oil level and dismantle all other bushings for a detailed investigation. Off-line C and DF measurements were also performed, confirming the results of the monitoring system. Several carbonization traces on the surface of the HV exit leads were found at phase V (Figure 5) and phase W (Figure 6). Carbonization traces were identified on the inner layers of the insulation, as well.
The following conclusions can be formulated:
- A catastrophic failure of the transmission transformer was prevented through continuous monitoring of bushing C, DF and PD;
- The high accuracy of bushing insulation C and DF measurements was obtained with absolute measurements, taking the reference signal from a group of voltage transformers located in the same substation;
- With the combined application of conventional and UHF PD measurements, the presence of the PD activity was confirmed inside the transformer tank;
- The advanced technique of synchronous multi-channel PD measurements was applied in order to recognize the PD source type as well as to identify and separate noise;
- With acoustic PD measurements, the location of the PD defect was precisely identified leading to a shorter time required for the internal investigation.
Author: L.V. Badicu, OMICRON Energy Solutions GmbH, Berlin, Germany