Practical investigation of partial discharge using ultra high frequency (UHF) method in high voltage system

  • Febri Arwan Nugraha

Student thesis: Master's Thesis


In the last two decades, PD detection gained in popularity as a reliable condition monitoring technique that could detect incipient faults and provide real time diagnostic information on insulation condition of the asset. The urgency of this PD detection is not only required in monitoring the insulation of any HV equipment in running system, but also in any HV equipment in manufacturing or assembling stage. Recently, the ultra-high frequency (UHF) technique has been used as a nonintrusive method for the detection and localization of PD, particularly in GIS and in transformers [1], [2]. It has shown promising potential to be used as a complement to the IEC 60270 standard method [3], but some aspects are still not fully understood and research in this area is ongoing. Aspects needing clarification include detectable range and identification of different PD sources, types and properties of UHF sensors, attenuation of emitted waves propagating in different media, and signal processing. This thesis is intended to fill the gap in bringing more understanding to the detection of partial discharge using the UHF method. It reports the results of an experimental investigation on the characteristics of the two types of PD in air gap and void within solid insulation using the UHF method combined with the high-frequency current transformer (HFCT) method. Different types of antenna having different frequency ranges were used to detect the electromagnetic (EM) radiation signals emitted from the PD source: three monopole rod antennas and a horn antenna. The UHF signals are correlated with the measured apparent charge displacement using the HFCT method. The effect of gap distance, voltage type, voltage magnitude and polarity and locations of the UHF sensor relative to the PD source and to each other were examined. PD activity is inherently a non-stationary signal and hence analysis of such signals using the FFT technique may yield limited information about the discharge characteristics. Therefore, the B - time frequency distribution (B-TFD), a member of the time-frequency analysis family, is implemented to the measured signals in order to identify the frequency content, as well as the duration of the discharge. The results indicate that, although PD inception voltage for negative DC discharge is higher than for positive DC discharge, the PD emission from positive DC typically has a higher amplitude than PD emission from negative DC. Typical emission from the positive DC source has a dominant frequency band between 50 - 500MHz, whereas the negative DC source produces PD emission between 50 – 350MHz. The results show that the emitted EM signals from AC discharge typically has its energy focused in two dominant frequency bands. These frequency range located below 100MHz and between 200-390MHz, regardless the voltage magnitude and the sensors location. In addition, these dominant frequency components have around 200 – 400ns duration on average. It is found that the voltage magnitude and sensors location have affected the amplitude and duration of the dominant frequency component from both DC and AC corona signals. Furthermore, Analysis of detected signals using horn antenna pointed out that the typical electromagnetic radiation signals emitted from voids discharge within XLPE sample have only one dominant frequency band which spanned between 1.7 – 3.3GHz. The duration of this band was relatively constant around 80ns. Moreover, prior and upon the electrical treeing development, the amplitude of time-frequency analysis plot was the only parameter that changed. It tended to decrease as electrical treeing developed. In conclusion, it has been demonstrated that the UHF method along with time – frequency analysis method have potential to be implemented as an effective tool to detect and recognize different types of PD activity. Further work is in progress to investigate other types of PD sources, and explore the feasibility of the UHF method for use as an effective tool for PD detection and classification, with potential application in power cable diagnosis and transformer diagnosis.
Date of Award2015
Original languageAmerican English
SupervisorNoureddine Harid (Supervisor)


  • Applied sciences
  • Discharge
  • Practical
  • UHF
  • Voltage
  • Electrical engineering
  • 0544:Electrical engineering

Cite this