Condition monitoring of high voltage equipment is mandatory to prevent failures, reduce maintenance and operating costs and also to extend plant life. Partial discharge measurement is an important condition monitoring technique to assess the insulation condition of high voltage equipment like power cables, generators, motors transformer’s, ct’s pt’s etc. The partial discharge measurement is a non-destructive diagnostic test to locate incipient fault in an insulation system. Compared to all other electrical product, extruded power cables system is most vulnerable to partial discharge failures. This paper summarises the various partial discharge measuring techniques, issues and signal processing techniques available to eliminate the interferences associated with PD signal extraction.
Definition of Partial Discharge
Partial Discharge is the localized electrical discharge that only partially bridges the insulation between conductors and which can or cannot occur adjacent to a conductor . Partial discharges (PD) are in general a consequence of local electrical stress concentrations in the insulation or on the surface of the insulation. Corona is a form of partial discharge that occurs in gaseous media around conductors which are remote from solid or liquid insulation. Partial discharges are often accompanied by emission of sound, light, heat, and chemical changes. It is measured in units of charge known as pico-Coulombs or millivolts.
Occurrence of Partial Discharges
It has been well recognised that one of the most common causes of insulation system failure occurs due to presence of void inclusions. For example, in case, extruded cables during manufacture or oil impregnated paper insulating systems such as power transformers and capacitors, voids are generally introduced due to maintenance of poor vacuum during the impregnation cycle. When the insulation system is subjected to a voltage stress such weak points occluded results in local over-stressing and undergo partial discharges activity. Although, the magnitude of such discharges is usually small they are known to cause progressive deterioration and eventually breakdown (failure) of the insulation due to their chemical effects.
Sequence of Discharges
The dielectric constant of the void inclusion usually a gas or a mixture of gases is often lower than that of the surrounding dielectric so that the electric stress in the void is higher than in the dielectric. Hence, void breakdown at a lower stress. When the void is subjected to a stress greater than its breakdown value (ignition voltage) discharge will occur regularly at each half cycle of the applied voltage wave. The number of the discharges that occur in each half cycle is mainly determined by the integer times that the applied voltage peak exceeds the ignition voltage of the void.
The insulation of all power equipment’s may contain several cavities at different locations and these in turn may contain different discharges sites. Consequently, if discharges are made visible on an oscilloscope a complex picture will appear. Constant, intermittent and irregular sequence of discharges may appear separately or superimposed. Surface discharges may also occur if there exists a stress component parallel to the dielectric surface.
Failures due to Partial Discharge
Partial Discharge leads to chemical and mechanical destruction of adjoining materials. Insulation breakdown is the number one cause of electrical failures. In medium and high voltage equipment, partial discharges are the first indication of insulation breakdown. The failure rate of Electrical equipment due to Partial discharge based on IEEE report is as given below. .
PD Detection in Power Cable System
Partial discharges in power cable systems are caused by various defects such as voids, shield protrusions, contaminants, improper installation of accessories, improper stress control at accessories and defects developed due to ageing. Off line and on line measurement techniques are available for PD Detection and measurement.
Off-Line Partial Discharge Testing Techniques
Offline Detection requires de-energization of the cable under test and in some cases, completely removed from the distribution system. It has the advantage of close control of the test voltage, and if necessary raising the voltages above the normal operation voltage.
Partial discharge measurement on long length extruded cable necessitates the use of high KVA rating, PD free source transformers in order to drive higher capacitive currents. To overcome this difficulty, alternate methods like VLF (Very Low Frequency), OWTS (Oscillatory Wave Test Set), DAC etc., are used. Very-low frequency (VLF) source allows the cable to be energized at 0.1Hz therefore preventing any large capacitive current from being drawn and minimizing the size of the test set. While off line testing is a valuable resource, this method is costly due to the requirement of disconnection the cables from the substation.
On-line PD Measurements
Online PD Detection is based on detecting the discharges at operating voltage using proper sensors, without the use of any additional source. In this case, the testing is conducted during real operating conditions, under typical temperature, voltage stresses, and vibration levels. It is a non-destructive test and does not use over voltages that could adversely affect the equipment. Online Partial Discharge Testing is relatively inexpensive compared to offline testing which requires interruption of service and production.
Issues with Online PD Measurements
Compared to Off line methods, the sensitivity of PD measurements will be much poor due to higher interferences and noises. Another issue with on-line PD measurements appears when more than one source of pulse-shaped signals is present in the electric HV system under testing. In these cases, an adequate selection of the non-conventional measuring technique with the most suitable sensors and the implementation of effective signal processing tools are essential to achieve a correct evaluation of the insulation condition.
Capacitive or inductive sensors are used for Partial discharge detection to measure the high frequency pulses generated due to Partial discharges. Generally, capacitive sensors are used for offline PD detection and Inductive sensors for online detection. Typical inductive sensors are ferrite based high frequency current sensors (HFCT). The sensors provide wideband measurement s from 100 kHz to 30 MHz. These split core sensors are inductively coupled to the earth link of each phase cable.
Figure 1: Typical form of disturbances in time domain
a) Broad band noise (b) Narrow band noise (c) Pulse noise
Figure 2: Typical form of disturbances in Frequency domain
(a) Broad band noise (b) Narrow band noise (c) Pulse noise
Noises and Interferences
On-line partial discharge measurements are hampered with external and internal disturbances, which decrease the sensitivity of the measurements. There are three common types of disturbances namely (a) narrowband interference, (b) broadband noise and (c) pulse-shaped interference. Figure 1 & 2 shows the typical form of disturbances in time and frequency domain.
It can be seen from Figure 2 that a broadband signal is random in nature and its energy is equally spread along the whole frequency band. A narrowband signal is concentrated on a narrow frequency band and in time domain it is continuous and strongly oscillating. A pulse-shaped signal is discontinuous in time domain and in frequency domain its power is widespread and usually concentrated on lower frequencies.
Narrowband interference is usually the most dominant source of disturbance in on-line partial discharge measurements and thus increases the noise level. Its amplitude can be a lot higher than the amplitude of the partial discharge pulses. The main sources of narrowband interference are radio transmitters. Broadcasting stations, mobile phones, etc. The level of these kinds of disturbances varies strongly as a function of time of day and place. Narrowband interference is picked up by the cable shield, overhead lines, badly shielded RMU components or the sensors themselves. Narrowband interference can be compressed by conventional notch filters or modern signal processing tools and through that the sensitivity of the measurement can be increased.
Stochastic broadband noise signals result from numerous natural and artificial sources like thermal noise, lightning noise, measurement instrument noise or other man-made noise. Broadband noise is present within the entire frequency spectrum of the partial discharge signals and thus sets a fundamental limit for partial discharge detection. Most of the broadband noise is concentrated on frequencies below 30 MHz.
Pulse Shaped Interference
Pulse-shaped interference frequently occurs during partial discharge measurements. Examples of pulse-shaped interference are thyristor pulses, switching transients and partial discharges from adjacent systems. The pulse shape of this kind of interference can be identical to the partial discharge pulses from the tested cable. Thus this type of interference can lead into false results if proper discrimination between the partial discharge pulses from the measured cable part.
Noise Elimination through Signal Processing Techniques
The discrete Wavelet transform (DWT) is found to be a powerful signal processing tool to denoise the PD signals. Wavelets are mathematical functions that divide the data into different frequency components, and then study each component with a resolution matched its scale. They have advantages over traditional Fourier transform methods where the time-based signal is converted to frequency based, where all information regarding time will be lost. Since Partial discharges are non-stationary signals with discontinuities and sharp spikes, the time information is also vital in analysing the PD signal.
The general wavelet-based denoising method proceeds in three steps: decomposing the noisy signal, thresholding the wavelet coefficients, and reconstructing the signal as shown in Figure 3. In the first step, the signal is transformed into the wavelet domain and represented by weighted coefficients. This stage involves defining, the type and order of mother wavelet, and the decomposition levels. The first step is to decompose the signal to a maximum level using the selected mother wavelet. The levels for decomposition need to be selected with careful analysis so that the PD signal will not get attenuated and at the same time maximum reduction of noise level is to be ensured. In each level, the signal is down sampled by 2. The tree structure of wavelet method of dividing signals to detail and approximation coefficients are shown in Fig 4.
The second step is crucial for efficient noise cancellation. In this step, it is necessary to set a proper threshold in such a way that all interferences and noises are suppressed. Wavelet denonising methods can be carried out using either hard or soft thresholding. Hard thresholding processes data in such a way that those wavelet coefficients whose absolute values (x) are greater than the threshold (λ) are kept and those less than the threshold are set to zero. Soft thresholding sets the wavelet coefficients below the threshold to zero. The coefficients greater than threshold are kept and then shrunk towards zero. The reconstruction is defined by the inverse discrete wavelet transform which involves up-sampling and filtering steps. At the completion of this stage, the denoised signal is represented in time domain.
Wavelet packet transform (WPT) is a direct expansion from the DWT pyramid tree algorithm where each branch of the tree has two sub-branches. It is the generalization of DWT in that both the low-pass and the high-pass output undergo splitting at the subsequent level. Therefore, WPT is seen to have the capability of partitioning the high-frequency bands to yield better frequency resolution.
The complete binary tree resulted from WPT contains many nodes. It follows that the terminal nodes (leaves) of every connected binary subtree of the complete tree form an orthogonal basis of the signal space. Therefore, to achieve the best denoizing performance, there is a need of choosing the best nodes subset (best tree) for representing a signal in wavelet packet domain.
The tree structure of wavelet packet transform is as given in Fig.5.
In wavelet packet analysis, the details coefficient as well as the approximation coefficients are split.
Partial discharge (PD) measurements provide valuable information for assessing the condition of high voltage (HV) and EHV Power Cable systems, contributing to their quality assurance. The availability of significant Online PD measuring techniques makes it possible to acquire the PD signals from cable system, without interruption. However in Online PD measurements apart from the signal acquisition the signal processing also plays an important role in ascertaining the actual magnitude, its significance and location. The digital signal processing technique based on wavelet technique is highly imperative in proper extraction of PD signals.
If you want to share thoughts or feedback then please leave a comment below.