Insulator Selection Tool IST 2018

Reference Manual Insulator Selection Tool IST 2018 Version 5.0.1

Contents Contents... 1 1 Statistical method for insulator pollution performance... 2 1.1 Calculation of the probability density function of site severity f()... 3 1.2 Calculation of the cumulative distribution function P()... 4 1.3 Calculation of the risk for flashover... 6 2 Economic evaluation... 7 3 References... 7

1 Statistical method for insulator pollution performance A statistical method suitable for computerized calculations of insulator pollution performance were developed by STRI and implemented in the Insulator Selection Tool (IST) and the Line Performance Estimator (LPE) [1-3]. Some practical examples of the implementation of this method are presented in [4-12]. In this statistical method, the main parameters characterizing the insulation are considered as statistical variables, defined by mean values and dispersions, as opposed to the deterministic method, where the parameters are assumed to be constant. The method can be used for both pollution and ice environments, although according to above the quantities describing the environmental stress on the insulators are different: the (Equivalent Salt Deposit Density, ESDD) level that is exceeded 2 % of the time is typically used for characterizing the statistical pollution stress, while the 2 % conductivity of the water film on the melting ice is used to characterize the icing stress (left curve of Figure 1-1). The Insulator Selection Tool (IST) covers only pollution, while the Line Performance Estimator (LPE) covers both pollution and ice. The statistical dimensioning of insulators entails the selection of the dielectric strength of an insulator with respect to voltage and environmental stresses, to fulfil a specific performance requirement. This is done by evaluating the risk for flashover of the candidate insulators and selecting those having an acceptable performance. With reference to Figure 1-1, the risk for flashover can be calculated as follows: - It is assumed that the insulators are energized at a voltage with constant amplitude, corresponding to the maximum continuous operating voltage - The variation of the pollution environmental stress (ESDD) at the site of interest is represented by the probability density function f(), which is expressed in terms of the site severity. These data normally come from service measurements and are approximated typically as lognormal distribution functions. - A cumulative distribution function P() describes the strength of the insulation, i.e., the probability for flashover as a function of the same measure of site severity, as was used to describe the environmental stress (ESDD). This data is normally obtained from laboratory tests and is approximated as a truncated Weibull function. - The P() function is then corrected from a single insulator tested in the lab to n insulators installed on the whole line or line section, and exposed to the same pollution events. - The multiplication of the f() and P() functions gives the probability density for flashover of the insulator at the given site, and the area under this curve expresses the risk for flashover during one pollution event. - Knowing the number of pollution events per year (e.g. salt storms in coastal areas, or light rain or dew in inland areas), the risk for flashover per year can be calculated. The statistical method is based on using pollution laboratory test results in terms of the flashover stress along the insulation length (in kv/m) and its standard deviation. One example of a standardized pollution tests is the Solid Layer method according to IEC 60507 (corresponding to an NSDD level of about 0,1 mg/cm 2 ).

Stress: density of occurrence Strength: probability of flashover Stress: f() Strength: P() f ( ) P ( ) Risk for flashover Pollution severity () Figure 1-1. Illustration of insulation design based on a probabilistic approach. The flashover data is corrected for the altitude of the line according to Cigré Technical Brochure No. 158 [18]: FO A = FO (1-0,05 A) for AC voltage FO A = FO (1-0,035 A) for DC voltage where FO A is the corrected flashover voltage, FO is the flashover voltage in standard atmospheric conditions, and A is the altitude (km). 1.1 Calculation of the probability density function of site severity f() The pollution severity is characterised by the type of pollution, the amount of pollution present and the frequency of wetting events that may lead to flashover. The pollution deposit on the insulators builds up during periods without extensive wetting. It may be washed or leached from the insulator during wetting events. The extent of this removal is related to the intensity and duration of the wetting. As a result, the pollution deposit on the insulators varies over time with maxima occurring usually at the start of wetting events. The wetting events form the basis for the dimensioning process as it represents the occasions with a significant risk for flashover. With sufficient pollution severity measurements available, a suitable distribution function can be fitted to obtain a statistical description of the pollution stress at the site. It is assumed that Equivalent Salt Deposit Density (ESDD) measurements were made and that a lognormal distribution provides the best fit to site severity data. The lognormal distribution is characterised by two parameters, normally the median value (i.e. the value that will be exceeded 50% of the time) and the standard deviation of LN(ESDD). The former can also be another parameter such as: - The mode, i.e. the most likely value in a sample - The mean value, i.e. the average value of the sample - The 2% value, i.e. the value that will be exceeded 2% of the time These values are illustrated in the Figure 1-2. If sufficient ESDD measurements are available, it is recommended that the standard deviation of LN(ESDD) be determined through a regression or maximum likelihood analysis of the measurement data. If sufficient data are not available, the standard deviation of LN(ESDD) can be estimated to be in the range 0,4 to 0,8 as was found from a

study of ESDD measurements performed at different locations around the world [2]. This study showed that the standard deviation of LN(ESDD) varies in a narrow range irrespective of the actual level of the ESDD measurements. Figure 1-2. Statistical distribution of ESDD (mg/cm²) as represented by a lognormal distribution density function. From a dimensioning point of view, it is preferred that the level of site severity be characterised by the 2% value, also known as the "statistical site severity", in order to determine the required insulator length with a high level of confidence. However, this requires that sufficient amount of ESDD measurements be available to allow estimation of the 2% value by statistical regression methods. 1.2 Calculation of the cumulative distribution function P() The probability for flashover is normally determined as function of the voltage at a constant pollution severity. This is because it is much easier and accurate to vary the voltage during the pollution test than it is to vary the pollution severity. Flashover tests are then performed at different levels of severity to obtain the insulator flashover stress as function of the pollution severity. For the tested insulator with its specific shed profile and material, this relationship can be described as follows: U 50( ) A l where U 50 is the voltage level having a 50% flashover probability; l is the insulation length; is the pollution severity; A and are the experimental constants derived from the laboratory tests. The standard deviation (or c if defining it in % of U 50) is also derived from the tests. Figure 1-3 presents a typical example of such a relationship determined from flashover laboratory tests at two test pollution severities (SDD levels). For the statistical dimensioning process, this information needs to be converted to express the probability for flashover as a function of the pollution severity at the service stress of the insulator, i.e. P(). Knowing A,, c and the maximum operating voltage U m, the flashover probabilities at various pollution levels are determined as shown in Figure 1-3 for the 10, 30 and 50% probability levels. (1)

Flashover stress (kv/m) 140 120 100 Fitted curve: Equation 1 Estimated U 50 from laboratory tests 85% confidence interval 80 10 30 50 60 40 Service stress at 420 kv 20 0 0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 0.5 Pollution severity (SDD; mg/cm 2 ) A = 68,5 = 0,136 Figure 1-3. Flashover stress of a typical 420 kv insulator as a function of pollution severity. Curves are shown for a flashover probability of 50%, 30% and 10%. The obtained P() can be approximated using different types of statistical distributions. The flashover value of an insulator for a particular pollution severity is usually described in terms of its 50% flashover voltage U 50 and standard deviation while assuming a normal distribution. There are however experimental data that indicate that the distribution function is truncated at n=2,5. That is, at a particular pollution stress, a voltage U 0 exists below which flashover is not possible i.e. a zero flashover probability. This is true from the physical point of view, because to initiate a pre-flashover over the dry band on the polluted insulator, a rather high voltage is obviously needed. Thus, P() can be described by a three-parameter Weibull distribution: P( ) 1 e k 1 U 1 U 0 The three parameters U 0, k and can be expressed in terms of U 50, and the truncation parameter n as follows: (2) U 0 U 50 n (3) 1,38 k ln n /( n 1) 1 1 ncln 2 k (4) nc (5) c Using Eq. 1 and 2, the probability of flashover as function of pollution severity may now be expressed as: U 50 (6)

k 1 1 0 P( ) 1 e (7) where 0 is the truncation value of pollution, which is derived from: 0 50 1 nc 1 (8) and 1 50 ( Al / U m) (9) where U m is the continuous operating voltage. Standard laboratory pollution tests aim at producing a uniform pollution deposit on the insulators. Under natural pollution conditions, however, the deposit is more non-uniform. The effect of the greater non-uniformity is that it leads to a greater spread in the flashover voltage under natural conditions. Like the non-uniformity in pollution deposit, the variations in wetting severity also contribute to the large spread of the flashover voltage under natural conditions. This can be accounted for by adjusting the insulator flashover characteristics obtained from laboratory tests to estimate the flashover characteristic under natural conditions. To account for differences in the normalized standard deviation of flashover voltage in artificial tests vs. natural conditions, 50 is replaced by 50n defined as: 50n Al U m 1 nc 1 nc where c a and c n are the normalized standard deviations in artificial tests and natural conditions, having typical values of 0,05-0,15 and 0,20, respectively. a n 1 (10) 1.3 Calculation of the risk for flashover The risk for flashover, R, of one insulator exposed to N p pollution events per year, is calculated according to Figure 1-2 as follows: R N p f P d (11) For the usual case where multiple (or parallel) insulators of the transmission line are exposed to the same conditions, the risk for flashover becomes higher. This can be quantified by calculating the flashover probability P n of a system comprising n insulators from the flashover probability of one insulator P 1. When all the insulators are exposed to the same conditions, P n is given by: Pn 1 1 P1 n (12) A more general approach is the case of a line that spans different levels of pollution and which has different types of insulators installed. Assuming that all the insulators on the line are affected by the

same pollution events, but to a different degree of severity, the probability of flashover of the system becomes: Pn n 1 1 P S i1 where S i expresses the pollution severity at insulator i in p.u. of the maximum severity along the line. Finally, the risk for a flashover on the line is calculated as: 1i i (13) R N p f P n d Additional references on statistical methods for insulation dimensioning are found in [19-23]. (14) 2 Economic evaluation IST can be used for comparing different insulator options with respect to purchase, installation, disposal, scheduled maintenance and any additional costs for the insulators, as well as corrective maintenance and outage costs associated with insulation failures. The present value of the total cost for each insulator option is calculated in the following way for a specified period using a specified interest rate: Firstly, the MTBF is calculated. Using the calculated MTBF for each insulator option, the present values of corrective maintenance costs and outage costs are calculated for each outage (provided that at least one outage takes place during the period). The present values of scheduled maintenance costs are calculated at each year during the period for the total number of insulators. The present value of the disposal costs is calculated at the end of the period for the total number of insulators. The present values of the purchase, installation, and additional costs are calculated at the beginning of the period for the total number of insulators. Finally, the present values described above are added to get the present value of the total cost for the total number of insulators. 3 References [1] C.S. Engelbrecht, R. Hartings, J. Lundquist: "On the statistical dimensioning of insulators with respect to pollution conditions", IEE Proceedings - Generation, Transmission and Distribution, Volume 151, Issue 03, May 2004, p.321-326. [2] C.S. Engelbrecht, I. Gutman, R. Hartings: "A practical implementation of statistical principles to select insulators with respect to polluted conditions on overhead a.c. lines", PowerTech 2005, St. Petersburg, Russia, June 27-30, paper 129. [3] M. Farzaneh, T. Baker, A. Bernstorf, J.T. Burnham, T. Carreira, E. Cherney, W.A. Chisholm, R. Christman, R. Cole, J. Cortinas, C. de Tourreil, J.F. Drapeau, J. Farzaneh-Dehkordi, S. Fikke, R. Gorur, T. Grisham, I. Gutman, J. Kuffel, A. Phillips, G. Powell, L. Rolfseng, M. Roy, T.Rozek, D.L. Ruff, A. Schwalm, V. Sklenicka, G. Stewart, R. Sundararajan, M. Szeto, R.Tay, J. Zhang: "Selection of station insulators with respect to ice and snow. Part II: Methods of Selection and

Options for Mitigation", IEEE Transactions on Power Delivery, Vol. 20, No.1, January 2005, p.p. 271-277 [4] I. Gutman, J. Lundquist: "New software programs for estimation of availability of overhead lines with respect to pollution, ice lightning&switching overvoltages", Proceedings of the International Conference "Suspension and post composite insulators: manufacturing, technical requirements, test methods, service experience, diagnostics", St. Petersburg, Russia, 4-9 October 2004, p.p. 37-41. [5] I. Gutman, K. Halsan, D. Hübinette, E.A. Solomonik, L.L. Vladimirsky: "New developed Insulator Selection Tool (IST) software: results of application using known Russian service experience", 12th Asian Conference on Electrical Discharge, Shenzhen, China, 19-22 November 2004, p.p. 10-15. [6] K. Halsan, D. Loudon, C. Engelbrecht, I. Gutman: "Ice and pollution testing of compact insulator strings in Statnett", 2003 World Conference on Insulators, Arresters & Bushings, Marbela, Spain, 16-19 November, 2003, p.p. 387-398. [7] K. Halsan, D. Loudon, C. Engelbrecht, I. Gutman: "Norwegian Utility Evaluates Insulation Alternatives to Upgrade 300 kv Transmission Network", Insulator News and Marker Report Quarterly Review, Issue 64, Quarter Two-2004, Volume 12, Number 2, 2004, p.p. 16-25. [8] I. Gutman, W.L. Vosloo: "Application of statistical principles of insulator dimensioning with respect to polluted conditions to select line insulators based on test station results", PowerTech 2005, St. Petersburg, Russia, June 27-30, paper 355. [9] I. Gutman, R. Hartings: "New methods for the pollution and ice testing of all types of outdoor insulation", Proceedings of the International Conference "Suspension and post composite insulators: manufacturing, technical requirements, test methods, service experience, diagnostics", St. Petersburg, Russia, 4-9 October 2004, p.p. 32-36. [10] W.L. Vosloo, I. Gutman, J.P. Holtzhausen: "Statistical determination of insulator performance at Koeberg insulator pollution test station", The 5th CIGRE Southern Africa Regional Conference, Cape Town, South Africa, 24-27 October, 2005, p.p. 180-187. [11] I. Gutman, J. Lundquist, K. Halsan, L.Wallin, E. Solomonik, W..L. Vosloo, "Line Performance Estimator Software: Calculations of Lightning, Pollution and Ice Failure Rates Compared with Service Records", CIGRE Session, Paper B2-205, Paris 2006. [12] I. Gutman, J. Lundquist, T Ohnstad, D. Hübinette, "Requirements on the Insulation for Different Applications of 400 kv Circuit Breakers in Pollution, Ice and Snow Environments", CIGRE Session, Paper A3-305, Paris 2006. [13] I. Gutman: "Selection Criteria for Line Insulators Based on Availability Requirements", INMR Worlds Congress on Insulators, Arresters & Bushings, Hong Kong, 27-30 November 2005. [14] C.S. Engelbrecht; A. Eklund, R. Hartings; R. Znaidi: "Field and laboratory testing for the choice of optimum composite insulator design for a marine-desert environment", CIGRÉ Session 2000, 33-202. [15] C.S. Engelbrecht; S.M. Berlijn; B. Engström; R. Hartings; K.Å. Halsan: "Dimensioning of insulators for salt pollution: A novel procedure and a laboratory test method", CIGRÈ Session 2002, 33-401. [16] C.S. Engelbrecht; R. Hartings; H. Tunell; B. Engström; H. Janssen; R. Hennings: "Pollution Tests for Coastal Conditions on an 800-kV Composite Bushing", IEEE Transactions on Power Delivery, Vol. 18, No. 3, July 2003. [17] C.S. Engelbrecht; R. Hartings; B. Engström; D. Hübinette; K. Halsan: "The Dry-Salt-Layer method, a laboratory pollution test-method for marine pollution: Its repeatability and a comparison of field and laboratory results", 13th ISH-2003, Netherlands 2003, Smit (ed.), 2003 Millpress, Rotterdam, ISBN 90-77017-79-8, p.p. 201. [18] CIGRE TF 33.04.01, "Polluted Insulators: A review of current knowledge", Technical Brochure, No. 158, June 2000. [19] Karady G, Dallaire D, Mukhedkar D, "Statistical method for Transmission line insulation design for polluted areas", IEEE/PES Winter Meeting, Paper No. WM A76 220-4, 1976.

[20] Sforzini M, Cortina R, Marrone G, "Statistical approach for insulator design in polluted areas, IEEE Trans. on Power Apparatus and Systems", Vol. PAS-102, 1983, pp. 3157-3166. [21] Naito K, Mizuno Y, Naganawa W, "A study on probabilistic assessment of contamination flashover of high voltage insulator", IEEE/PES Summer Meeting, Paper No. 94SM 445-7, PWRD, 1994. [22] Yamada K, Mizuno Y, Naito K, "Simulation of flashover risk of polluted insulator", IEEJ Tokai Regional Conference, Paper No. 80, 1994 and Second report, No. 1639, IEEJ all Japan Conference, March, 1995, (Both in Japanese). [23] Rizk F, El-Arabaty A, El-Sarky A, "Laboratory and field experiences with EHV transmission line insulators in the desert", IEEE Trans. on Power Apparatus and Systems, Vol. PAS-94, No. 5, Sept./Oct., 1975.