The development of UV cameras that permit corona detection even during daytime is quite a positive development for electric utilities. This is because these cameras offer yet another practical tool for routine inspection of lines and equipment. At the same time, however, it is often not possible to reach any definitive conclusions concerning the condition of insulators based only on visual data – especially if they are of the composite type. This INMR article from 2004, by Prof. Ravi S. Gorur, discussesd research of the day regarding means to assess the severity of corona observed on composite insulators – whether with silicone or EPDM housings. Such information has special importance when it comes to identifying which specific types of discharge patterns are most likely to pose problems to the integrity of these insulators.
Corona discharge can be an especially significant threat to the integrity of composite, non-ceramic insulators (NCIs) due to the organic nature of their housing materials. For transmission class insulators (69 kV and above), corona can occur not only in contaminated but also even in clean environments. This is important to emphasize since, based on service experience worldwide, it seems that the majority of NCI applications are actually under relatively clean or only light-contamination conditions.
On NCIs, corona can be present locally for long periods of time due to such factors as inadequate hardware design, damaged hardware or deficient interfaces due to improper design and/or manufacturing. In practice, it is difficult to avoid corona on transmission networks, especially under wet and contaminated conditions. Hence, a knowledge of the corona discharge magnitude as well as the damage threshold of the housing material is essential. Fig. 1 shows the damage caused by corona during service, with this particular insulator obviously in an advanced stage of degradation. Damage of this type leading to cracks in the housing can expose the fiberglass core to moisture and premature failure can occur by tracking, erosion or brittle fracture. Hence, there is a necessity to periodically inspect NCIs and replace degraded insulators in a timely manner.
Generally-speaking, the methods currently used by utilities to inspect NCIs in-service include:
• Acoustic emission
• Radio interference voltage (RIV) measurement
• Infra-red (IR) thermography
• Electric-field measurement
• Visual observation
Acoustic emission and radio interference voltage measurements both have a sensitivity to background noise. These methods therefore cannot be used to reliably determine the precise location of discharge on an insulator. This presents a significant drawback since discharges on insulator hardware cannot always be avoided and may not even be a problem; however such discharges on the housing can be a harbinger of serious problems to come.
Determination of surface temperature by IR thermography has yielded some success. However, this method cannot be used to detect small defects which are incapable of causing any significant rise in surface temperature. The reliability of IR thermography is also reduced when performed under hot or sunny conditions. Electric field measurement along an insulator has been employed more successfully for ceramic than for non-ceramic insulators due to inherent differences in the construction of these insulators. At the same time, it has been shown that this method can generate erroneous results in the presence of moisture. This method also requires deployment from the tower or a bucket truck and hence is not always convenient. For all these reasons, visual observation by knowledgeable personnel either from the ground or from helicopters has remained up to now perhaps the single most effective method for inspecting NCIs.
Principle of Operation of Corona Camera
Visual observation of corona with the unaided eye is difficult because corona emits weak radiation mostly in the ultra-violet (UV) spectral range and is therefore invisible. Solar radiation in this region is much stronger, thus masking the corona signals. In the so-called “solar-blind” region (240-280 nm), corona emission is much weaker. However, the solar background is now virtually negligible. It is these signals generated by corona in the solar-blind region which are used by specialized cameras to detect corona.
Image Analysis & Processing
The image captured by a corona camera is generally stored in a computer as a matrix of pixels in a JPG file. This image then requires a number of processing and noise removal steps to facilitate subsequent mathematical operations and quantitative analysis. Basically, the JPG file represents the light intensity levels at different locations of the image matrix.
By using a MATLAB Program, the pixel intensities (represented by the elements of the image matrix) are summed up to obtain the cumulative intensity. The maximum pixel intensity level (Im) of the background image is determined. In the output corona image, pixels which have intensities higher than Im are considered illuminated (pixels which have intensities higher than Im are considered ‘1’ or high – otherwise ‘0’ or low). The number of illuminated pixels is a measure of the illuminated pixel area. The cumulative intensity and the illuminated pixel area are referred to as transformed image parameters in the subsequent sections of this article.
A 115 kV-rated silicone rubber insulator removed from the field was chosen for the initial experiments. This insulator was selected because its line end fitting was found to be unusually rough. The hardware also displayed burn marks caused by lightning events. The voltage applied was increased gradually to the rated voltage in steps of 5 kV. The corona images and their corresponding PD magnitudes were recorded simultaneously. For convenience, the distance between the point of interest and the camera was selected to be 5 . This experimental set-up is illustrated in Fig. 2.
Any increase in the voltage from the inception value resulted in higher discharge magnitudes (as measured by the PD detector). The corona at 50 kV (0.75 p.u) of magnitude 30 pC from the weathered hardware of a silicone insulator under dry conditions and its corresponding intensity plot is shown in Fig. 3.
Findings & Discussion
By changing the voltage in steps of 5 kV, the corona discharge magnitude was varied and the corresponding transformed parameter values were determined. Figs. 5 and 6 show the results of the transformed parameters as a function of the PD discharge magnitudes. The discharge magnitude associated with corona can be seen to be proportional to the intensity of the luminous discharge and the discharge area.
There are several regions along a non-ceramic insulator which typically serve as preferential locations for corona discharges. These include:
• The hardware of the insulator, even under dry conditions, in the absence of a grading ring, (as shown in Fig. 6a).
• The hardware and the hydrophobic housing material, especially in the presence of water droplets (Figs. 6b and 6c).
• Interfaces created by joining individual sheds, especially in the presence of water droplets (Fig. 6d).
• The shank portion between the HV end fitting and the first shed, especially under wet conditions (as shown in Fig. 6e).
Knowledge of the discharge location is quite important in assessing the threat to an insulator due to corona. Discharges which occur on the hardware and/or far from the housing material are less damaging than are those discharges occurring on or in close proximity to the housing. The ratio of the cumulative intensity to the area of the illuminated pixels is a measure of the corona discharge density and could therefore be a good parameter to predict degradation. It might appear that the corona corresponding to images Figs. 6d and 6e has a less damaging effect than the corona corresponding to the image in Fig. 6a. However, Fig. 7 indicates otherwise.
Knowing the corona discharge magnitude is therefore only part of the problem. The other part is to determine the corona degradation characteristics of different housing materials. Research explored this aspect and the time for manifestation of damage from corona was determined for several different NCI housing materials. Since this is dependent on formulation, it is important to have such information in order to determine if the observed corona during field inspection is indeed harmful to the insulator. Users of NCIs therefore should ideally create a database listing the discharge magnitude (or density) for every inspection. This will help establish if there are particular designs and field locations where corona can be more problematic than for others. Appropriate maintenance can therefore be performed in a timely manner.
The images provided by a corona camera can be translated into quantifiable parameters and there is good correlation between the visual image and the corona discharge magnitude. The results obtained
have been applied to field inspection of NCIs confirming that it is indeed possible to use corona cameras for troubleshooting.