Whatever the application of electrical insulators on a power network, there are always two important forces that have to be considered: the environmental factors and overvoltage situations that will stress the insulation and the insulation strength in terms of dielectric withstand (i.e. dimensions, shed geometry and material). All future line performance will then derive from how well these opposing forces are kept in perfect balance.
For a long time now, various international standards (most notably IEC 815) provided guidelines to design engineers on how to correctly specify insulators made of porcelain and glass, depending on environment. These were based on decades of accumulated experience but did not always apply well in certain unique environments, as found in many countries of the developing world. More recently, IEC 60815-1,2,3 have been improved and expanded to reflect broader service experience and also to account for composite insulators, which in the past have typically been dimensioned the same as ceramic alternatives – even though this was not always required nor even desirable.
One of the most prominent test laboratories studying this topic has been STRI in Sweden, where a variety of novel test methodologies have been developed over the years, including the dust cycle. STRI has also been among the most active in studying the behavior of composite insulators, both in the laboratory and at its long-term outdoor test site that was among the first in the world to monitor natural ageing.
INMR meets with STRI insulation specialist, Igor Gutman, to discuss some of the current issues that apply to statistical insulator selection and dimensioning.
While statistical dimensioning of insulators seems to be a hot new topic in the power industry, according to Gutman it is certainly not new – with roots that actually go back as far as the early 1970s. At that time, he explains, Russian engineers were building a 500 kV line to the huge Aswan dam in Egypt and employed insulator strings composed of 23 cap & pin discs. However, flashovers soon began to appear every morning along the line. To deal with these, large numbers of insulators were removed and sent to a newly built nearby HV laboratory. In addition, test stations were constructed at various points along the line to provide additional input.
“Back then,” recalls Gutman, “all the data coming from these had to be analyzed by hand and therefore a statistical approach was created whereby the insulation would be redesigned to ensure some predicted acceptable outage rate per 100 km of line. This eventually resulted in a 27 to 36-unit string along with a maintenance program of frequent washing. The line was monitored for the following three years and indeed service experience was found to be quite close to what had been predicted. This was therefore the birth of the concept of statistical dimensioning.”
Gutman points to a number of recently published documents that confirm that similar statistical approaches are now being widely applied to insulator selection. These include CIGRE Brochure 361 on insulators under polluted conditions and IEC Technical Specifications IEC/TS 60815 “Selection and dimensioning of highvoltage insulators intended for use in polluted conditions”. Part 1 of these IEC Technical Specifications covers general principles while Part 2 focuses on AC glass and porcelain insulator dimensioning. Part 3 deals with AC polymeric insulators and Parts 4 and 5 will apply to DC ceramic and composite insulators. Similarly, IEEE has started to develop P1820 as a guide for selection of insulators with respect to various icing situations, including pure icing, cold fog, and icing combined with pollution. “This document,” notes Gutman, “also looks at insulation from a statistical dimensioning point of view.”
In order to demonstrate how statistical insulator dimensioning works in practice, Gutman reviews the basic principles of insulation coordination, which aims to balance environmental stress with dielectric withstand. “The optimum dimensions of insulators” he explains, “should reflect an accurate knowledge of their pollution situation, an estimate of their actual flashover strength and a knowledge of required line performance. All will determine proper insulation design – which can subsequently be verified by testing.” For example, Gutman explains that typical graphs used in IEC 60815-1 (2008) illustrate the strength and pollution stress curves plotted against pollution severity, which are multiplied to give the probability density for flashover.
The start of this process, says Gutman, begins with knowledge of the line’s topography, climatic conditions and especially the pollution severity that will affect it. Parameters such as maximum ESDD as well as its distribution along the insulator, number of pollution events (e.g. wetting by dew, drizzle and fog), and type of contaminant all play a role in this respect.
Apart from direct measurements of ESDD at insulators and measurements of leakage currents associated with ESDD and NSDD, pollution severity data is also collected by devices such as directional dust deposit gauges (DDDGs) which have been used for years by utilities such as Eskom in South Africa and which have now also become part of the IEC standard. Gutman mentions, as example, data gathered from Eskom tests stations. This consisted of statistical pollution data (Input 1) as well as leakage current data converted into pollution performance by plotting flashover voltage against ESDD (Input 2). He also points to a recent STRI project involving NamPower in Namibia, where test towers from a line were equipped with different types of insulators, all connected to leakage current measuring equipment. This provided the utility with data similar to that which could be obtained from a test station while DDDGs on the ground offered additional information after conversion into ESDD.
The other important part of the statistical dimensioning process focuses on calculating flashover strength and here Gutman points to the need to establish curves that show the 50% flashover voltage (U50 ) versus ESDD either by research in the laboratory (or in a test station) or directly through service experience (e.g. from historical outage rates). These can then be converted into pollution flashover strength of the insulators.
Gutman says that when it comes to estimating the flashover strength of composite insulators, the solid layer method is the most representative way to achieve this. “Eventually,” he observes, “it’s quite possible to have 100s of milliamps of leakage current on such an insulator whenever there is a temporary loss of hydrophobicity.” He contrasts this with the salt fog test (e.g. at salinities of 80 kg/m3 ), which typically shows maximum currents of less than 1 mA and which he claims is therefore not at all representative for composite insulators. Says Gutman, “standard salt fog tests just do not correlate with data from test stations.
For example, a modified solid layer test was just recently performed at STRI. In this case, the Swedish grid operator Svenska KraftNät was re-conductoring a 400 kV line and wanted to replace ceramic tension insulators with very high mechanical strength composite units that needed to be tested for flashover performance.
Says Gutman, “in this instance we started to decide to add a short recovery phase after application of the pollution layer. That means that, after washing and drying, there is preconditioning followed by application of the pollution. This is then followed by recovery before testing under voltage.” Gutman observes that past hydrophobicity recovery tests on a large composite bushing confirmed that six days was sufficient for almost full recovery. “In most service situations, there is always time for recovery,” he emphasizes.
Other methodologies for estimating flashover strength for the purposes of statistical dimensioning of insulators also exist, including rapid flashover tests, obtained in clean fog after one hour, as well as various ice and snow tests.