805 VOLTAGE DISTRIBUTION OF A STRING ILUSULATOR UNDER DC VOLTAGES K.N.Ravi, Dr. M. Ramammrty, Dr. M.S. Naidu Central Power Research Indian Institute of Institue, Bangalore Science, Bangalore INTRODUCTION The e & of pollution is more severe in the case of DC sys-tems than in the case of AC sysems since the build up of pollution on an HVDC insulator is non uniform and more severe and the duration of the current pulses is longer [ 1,2 ]. Both these factors are due to unipolar nature of the voltage. Although the insulator failure under HVDC does not cause the line drop, it shortens the creepage distance and thus paves way for the pollution flashover along the insulator string [3]. In the case of an insulator string under DC voltages, the electrostatic force on any pollutant particle depends on its relative position with respect to the string and this in turn depends on the electric field intensity at the particle and its charge. This implies that the voltage distribution across the string is the important factor which is responsible for the non uniform distribution of the pollutant on the surface of the insulators [41. The variation of the voltage distribution of an insulator string with ambient conditions has been studied so far only for one hour or less [5,6]. The results of such a study for longer durations would help to determine the maximum voltage appearing across any given insulator at a particular ambient condition, and the failure mechanism due to surface erosion and or ion migration can be better understood. In this study, the variation of voltage distribution across an insulator string consisting of four porcelain disc insulators has been carried out at Pollution Laboratory of CPRI under ambient conditions, for 12 hours on any given working day, for relative humidity conditions varying fyom 19% to 70%, streb2hing over a period of about one year. EXPERIMENTAL SET UP AND TEST PROCEDURE Measurements have been carried out using the sphere
gap method, with spheres of 2 cm diameter. The experimental arrangement is shown in Fig.1. A 420kV, 20kVA, 3 Stage DC unit was used as the voltage source. A 25 mm aluminium tube was used to simulate the condumr. The two ends of the aluminium tube were rounded OB by inserting two tight fitting aluminium hemispheres. The sphere-sphere gap arrangement used was supported by two perspex rods fixed on either side of it and connected to metal plates at the top and b t b m end 2 57F 2 5yF f] 504z - L20kV,75 MA,3SlAGE DC SE1 FIG 1 Experimental setup The voltage distribution was debzmined using the following two procedures. Procedure A was used to de*rmine the individual voltages across the 4 insulators for an almost constant humidity, while Procedure B was used to determine the voltage distribution across any given insulator over a longer duration. The sphere-sphere gap spacing was varied from 1.7 to 5 mm and the voltage applied to the string from 10 lm 100 kv. -- Procedure A: In this, the spacing of the sphere -sphere gap was pre-set at a particular value and then it is connected across the specified insulator. The voltage to the string was raised till sparkover of the sphere gap occurs. This procedure was repeaed about 5 times and the average of these values is considered br evaluation of the percentage voltage shared by the insulator, which is expressed as: Sparkover voltage of the sphere-sphere gap V= XlOO % Voltage applied to the string to cause sparkover Procedure B: In this, the sphere-sphere gap ele-
7- ----1--- 60 CJ g 50 c 1 I
808 Insulator surface current depends on its surface resistance while the volume current depends on its temperature which generally remained constant. However,the air current depends on the variation in the ambient wind velocity,which was not significant in the present study. Then the voltage distribution would depend primarily on the insula~r surface resistance which is extremely sensitive the variations in the ambient relative humidity. Therefore, extensive investigations were carried out in the ambient and the results are shown in Fiqs.3,4 & 5.,, Rh ---.... I' Figure 3 shows the variation of the voltage across the line end insula-r for two difeerent patterns of relative humidity variations when the initial relative was in the range of 60 to 65%. Both the voltage characteristics shown by the solid and dotted fines show patterns in which they vary inversely with corresponding relative humidity values. Contrary to the above, volt- Fig 4 Variation of Ihe voltage across ground end Insulator (VI and relative humldlty (Rh) 0s a tunction ot time
809 age variations in the case of ground end insulator appears to have a direct relationship with relative humidity as can be seen *om Fig.4. When the relative humidity reduces, the surface current also reduces and thus the net charge reaching the pin of the qround end insulator decreases, causing a corresponding reduction in the voltage 1 ~ ~ --i~- - appearing across it. j i j i Fig s variation of vollage across an insulator as o tunctlon 01 time when!ne Initial j relative humidity>+% % When the initial relative humidity exceeds 60%, the voltage shared by the insulators exhibit distinct peaks, especially when the relative humidity reduces sharply from its initial value, as can be seen from Fig.5. These peaks were seen over the initial 3 hour period only. During the rising portion of any given peak, the relative humidity is high and therefore, there exists increased draining of the charges from the corresponding insulator cap resulting in a fall of the potential at the cap, which in turn causes an increase in the voltage appearing across it. The fall in the voltage across the insulator afkr the peak (Rh (50%) is due to reduced draining of charges from the insulator cap. CONCLUSIONS The following conclusions may be arrived at from the above results: i) The voltage distribution across the string measured over a period of about half an hour clearly shows a non linearity in the distribution. ii) Peaks in the voltage across the insulators were observed in the initial 3 hour period when the initial relative humidity was >60 %.
810 =) The variation in the voltage across a given insulator over a period of time seem to depend on the processes of charge accumulation and draining at the cap and pin of the middle two insulators, while it seems to depend on the accumulatbn and draining of charges at the cap of Line end insulator and at the pin of the ground end insulator. iv) The variatbn in the voltage showed by the insulators has a large spread. Therefore, it follows that under dry conditions the voltage across any given hsulatmr can rise to higher values. This suggests that the insulator can fail at these increased voltage levels if other conditions such as surface erosion and increase in the sodium ion concentration in the insulator volume due to pollution, are favourable. ACKNOWLEDGEMENTS We wish to thank Mr.S.Parameswaran, Additbnal Direcbr, CPRI and Dr.Channakeshava, Joint Dhx?ctmr, CPRI for their co-operatbn in carrying out the experiments. We also thank Mr.N.Vasudev and Mr.A.K.Mu]umdar for their help in conducting the experiments. REFERENCES [I] WANG ZUN et al, CIGRE 1986, Vol.11, pp 33-02. [2] MERKHALEV and VLADIMIRSKY, Vth I.S.H., Braunshweig, 1987, F.R.G. pp 63-18 [3] PEIXOTO et al, IEEE Transactions on Power Delivery, VOl.3, N0.2, 1988, pp 777-782 [4] KIMOTO et al, IEEE, Trans. Power Appar. and Systems, Vol. PAS-92, 1973, pp 943-949. [5] TRAIN and DUBE, IEEE Trans. Power Appar. and Systems, Vol. PAS-102, N0.8, August 1983, pp 2461-2475 [6] JANISCHEWSKYJ et al, Vth I.S.H., Braunshweig, 1987, F.R.G. pp 51-05.