Reliable operation of transmission and distribution systems depends on correct selection of insulators. But proper dimensioning can be a challenge since it requires knowledge of both site contamination and insulator withstand to specific types of contaminants. Moreover, contamination level depends on a number of site-specific parameters. Wind, rain, humidity and the physical-chemical characteristics of pollutants all play a role while secondary parameters such as wind direction and rain intensity also have an influence. A priori characterization of any site is further complicated given the uncertainty when estimating each relevant parameter. Specific statistical methods are then necessary to reduce uncertainty and allow more robust estimation. Contamination measurements represent another approach to classify a site and can be performed directly on energized or non-energized insulators or using dust collectors (e.g. DDDGs). Both priori characterization of a site based on meteorological parameters as well as measurements using dust collection devices require a further step, namely evaluating the contamination that actually deposits on insulator surfaces. All such techniques then need feedback from past operating experience that can help further refine knowledge of the contamination level of a site.
Withstand to specific contamination is related to insulator characteristics and tests have been performed over the years at different laboratories involving a range of insulator types and profiles. Such data have been assembled to obtain mostly 2-parameter models that provide unified specific creepage distance (USCD) for any given contamination level. The final goal when selecting the proper insulator for a site is evaluating risk of failure so as to ascertain whether or not this value is lower or equal to acceptable risk. Acceptable risk usually varies with voltage level and type of installation (e.g. line or substation). Again, confidence in final selection comes from past field experience. This edited contribution to INMR by insulation specialist, Massimo Marzinotto, of Italian TSO, Terna, performs separate such analyses for line and station insulators since acceptable risk is generally different for the two.
Insulator Dimensioning & Risk
Dimensioning insulation can be performed through statistical methods or in a deterministic manner. The former allows estimating risk of failure while the latter is based on experience using safety factors (but where risk of failure remains unknown). Evaluation of risk of failure is performed using the following equation, which yields probability of flashover for an adverse contamination event:
where f(s) is probability density of the contamination (s) on the insulator surface depending on site characteristics as well as on insulator parameters (i.e. profile, diameter, inclination, etc.). F(s) is cumulative flashover probability function of the insulator/string for any contamination (s) given applied voltage and insulator characteristics (i.e. profile, diameter, creepage factor, type of contamination (whether A or B), hydrophobicity class, number of insulators in parallel, altitude and, in the case of Type A contamination, coefficient of uniformity ratio (CUR) and NSDD level. From risk of failure, it becomes possible to define number of expected flashover events each year, Ne:
where np is annual presumed number of unfavorable events (i.e. that can lead to flashover). Ne is compared with number of allowable flashover events each year (Na). Thus, insulator dimensioning is considered acceptable if Ne ≤ Na. As mentioned, F(s) can be estimated based on laboratory tests while f(s) can be assessed either from a measurement campaign or simulated using specific models. np is closely related to number of days per year that give rise to 1) wetting of the insulator and 2) conditions able to lead to flashover. This number has been evaluated at somewhere between 20 and 40 days per year but is general and needs to be refined on a case-by-case basis. Na depends on voltage level, line or substation and on specific decisions by the transmission system operator. Considering that flashovers due to insulator contamination can sometimes cause permanent outages, a possible value could be somewhere between 2 and 5 times lower than allowable number of back-flashover events from lightning striking a line or station. For example, for a line having an annual back-flashover rate of 0.1/100 km (i.e. 1 event every 10 years/100 km of line), Na should be 0.02 to 0.05/100 km (i.e. 1 event every 20 – 50 years/100 km of line). On the other hand, for a station where allowable yearly flashover from lightning events is on the order of 0.01 (i.e. 1 event every 100 years), Na could be 0.002 to 0.005 (i.e. 1 event every 200 to 500 years). Based on this case, it can be concluded that yearly allowable events, Na, for a substation is an order of magnitude lower than for 100 km of line. It is important to note that parameters can be time dependent, in particular np, as can be other parameters that contribute to estimating f(s). Updating parameters can be based on feedback from the field. For instance, if a new steel mill is built in the service area, np can change and f(s) would shift to the right. Usually, contamination maps are used even if they cannot give sufficient information for assessment of risk, R. In fact, these are contour maps that report only maximum value (or sometimes mean value) of contamination for each level but give no information on statistical distribution and standard deviation. Re-mapping is usually performed after a number of years to evaluate possible variations in site contamination level. But such activities are costly and usually undertaken only if the returns from the operation highlight the need.
Recently, for example, this activity has been started in Italy with 200 measurement sites uniformly distributed across the country. Such measurements involve sampling every 3 months over a period of two years, i.e. a total of 8 samplings distributed uniformly over the year (e.g. March, June, September and December). To simplify sampling, non-energized strings of 10 standard profile cap & pin glass insulators have been installed on selected towers at a height just below the lowest conductor. For each sampling, one insulator is selected as reference and a swab technique is employed in line with what is used for measuring ESDD and NSDD. The uppermost and lowest insulators in the string are not used in such measurements to avoid ‘end effects’. In order to give greater confidence in the validity of the data collected, this information is matched with data from units returned from service in the field. This helps re-define color (i.e. contour level) of the map and better understand the efficiency of these measurements. Moreover, if theoretical models of contamination can be improved, this will also help better estimate risk of failure, R, and expected number of flashover events yearly, Ne. To monitor specific sites across Italy that are known as heavy contamination areas, Terna (formerly ENEL) decided 20 years ago to adopt a monitoring system able to survey the severity of contamination deposited on insulators and give warnings whenever this exceeds specific thresholds. This monitoring system, called AMICO 2, allows severity of the contamination deposit to be evaluated through surface conductance of a post insulator exposed to natural contamination. The surface conductance measurement is performed without altering natural conditions and applying a voltage equal to 12.5 kV (rms) for 4-cycles only. Measurements are performed periodically taking into account the value of other recorded meteorological parameters such as wind speed and direction, temperature and humidity. The system is thus able to convert the surface conductivity to contamination level, as such giving a warning and also an alarm when specific thresholds are exceeded. Empirical contamination models based on meteorological parameters are another area where researchers are developing knowledge to yield robust models for predicting contamination on insulator surfaces. These models can only be validated from field returns.
Another interesting recent field of investigation is resilience of power systems in general and in particular resilience of overhead lines and substations against contamination. Insulators must be able to withstand even extreme contamination conditions and therefore number of expected flashover events each year, Ne, must be lower than normal. This imposes a more stringent approach to insulator/string dimensioning. Finally, insulators must also guarantee safety of the public, of maintenance crews, of other power system equipment as well as other infrastructure beneath a line or near a substation. The reliability of the insulator in this sense is difficult to quantify. Obviously, risk of mechanical failure can have great impact on public safety and also on collateral damage to other infrastructure. Consequently, these risks should be much lower than that accepted for a flashover event due to contamination. Risk evaluation in this sense is difficult to achieve due to the complexity of arriving at a statistical distribution of all stresses that can influence mechanical failure of the insulator. For this reason, a purely deterministic approach to the problem is preferable, allowing adequate margin between the maximum stress that an insulator will encounter during service and its mechanical withstand value. Such margins have to be sufficient to guarantee extremely low risk of mechanical failure.
Another relevant aspect is ageing since this affects insulator performance. Insulator ageing is the product of the combined electrical, mechanical and environmental stresses. Moreover, the greater the synergy among these stresses, the greater will be the rate of ageing. Synergy of these stresses is relatively high in the case of composite insulators but comparatively low for porcelain and glass insulators. This conclusion has been confirmed from experience over many years of service at different voltage levels and under different environmental conditions. In this regard, immediately after the first installations of composite and RTV-coated insulators, Terna decided to set up specific test protocols to be applied to insulators removed from the field after a number of years of service. Such test protocols aim to: 1) identify changes in the ageing process, if any, and 2) yield information to better diagnose insulator end-of-life. This way, final design of an insulator is refined based on field returns. Such fine-tuning allows for the type of information necessary to achieve confidence that insulator dimensioning has been performed properly.
Overhead Line Insulators
The Italian transmission grid originally employed both porcelain and glass insulators but glass has become the standard since the 1960s. Today, it is the only type allowed, except for heavily polluted service areas where either composite insulators or factory RTV-coated cap & pin insulators are mandatory. Experience with glass insulators has been broad and shown high reliability. For example, some insulators having more than 50 years of service reveal no signs of ageing. On the other hand, corrosion of the cap and of the pin has been observed in harsh service environments due to high leakage current. In such environments, it has even been necessary to periodically replace towers due to the corrosion. Also, in some cases, there has been discoloring of glass insulators after 30 years of operation but with no impact on electrical or mechanical performance. In Italy, special rings are used on the voltage side of strings with the aim of equalizing electric field, giving more uniformity along the insulator string and limiting radio interferences. Mention must be made of the risk of internal defects in the glass shell. Even though rare, these can lead to self-shattering under electric field stress. However, should this occur, mechanical performance of the string does not decrease below 90% of its rated value, even should all discs in the string self-shatter. This test is required to ensure a large safety margin should several units shatter during service. Self-shattering is easily detectable by naked eye from afar, without need for special instruments or meters. Consequently, even for high towers, maintenance crews can easily detect shattered units without need for climbing. Fig. 3 shows a ‘naked eye’ view of a string at about 100 m.
Unfortunately, glass insulators cannot be applied without special precautions in unusually harsh environments. This is because increasing creepage distance beyond a certain level does not yield reliable results – especially in AC where electric field distribution along the string is strongly capacitive. In such a cases, Terna has adopted two solutions: periodic insulator washing and application of grease. Both solutions are costly, especially for lines exposed to marine salt due to vicinity to the sea. In such cases, there have been situations where insulator washing was needed twice a year. Similarly, grease needed replacement every 2 to 3 years, with longer outages needed for this than for washing.
Starting in the late 1980s, the first pilot installations began with composite insulators in heavily polluted areas. However initial field experience was not satisfactory. Then, by the end of the 1990s and early 2000s composite insulators became the standard with broader application in polluted areas. Those years also saw the first pilot installations of factory produced RTV-coated glass insulators. In fact, positive field experience encouraged Terna to apply this technology to extra high voltage lines, becoming the first TSO to install such insulators on a 380 kV grid. Composite and RTV insulators immediately highlighted advantages compared to washing and grease, however uncertainty in regard to end-of-life made it clear that this issue needed to be investigated. Different laboratory tests on both field-aged composite as well as RTV-coated glass insulators have since begun and involved annual sampling to learn more about ageing processes.
Ageing is different for composite and RTV coated insulators. In general, ageing is usually driven either by one main process that prevails over minor processes or by a combination of different ageing processes. Insulators will age from mechanical, electrical and environmental stresses and their synergistic impact on ageing rate become stronger as time in service increases. In the case of porcelain and glass insulators, experience has demonstrated that electrical and mechanical stresses are not or are only weakly related. That means their synergistic result is negligible. On the hard hand, in the case of composite insulators, the synergy between mechanical and electrical stresses is high. In fact, most failures of composite insulators are usually mechanical but a consequence of previous electrical stress. Electric field under specific environmental conditions gives rise to tracking and erosion, thereby damaging the insulator sheath. The combined impact of electrical and mechanical stresses becomes ever stronger and, should mechanical stress prevail, there will be a mechanical failure (as shown in Fig. 4). Should electrical stress prevail, there will be permanent electrical failure. Mechanical failures give rise to line drops but electrical failures can be difficult to detect. This is due to high and sometimes varying fault resistance, misleading line protection to show an incorrect fault distance after line re-closing.
RTV-coated insulators demonstrate behavior that is similar to uncoated porcelain and glass insulators. Mechanical, electrical and environmental stresses all influence ageing rate but there is no or only weak synergy between mechanical and electrical stresses. As such, given the same environmental conditions, ageing of RTV-coated insulators will be less than for composite insulators. This conclusion is confirmed by service experience which shows that failure rate of RTV-coated insulators is in line with that of uncoated glass insulators. For this reason, Terna has decided not to install composite insulators on tension towers, on suspension towers with spans that cross roads, railways, rivers, etc. nor on strategic lines in the grid. In regard to end-of-life, it is also important to note that failed composite insulators completely lose electrical performance, resulting in permanent electrical breakdown. On the other hand, if RTV-coated glass insulators lose the enhanced electrical performance offered by the coating, the string is only subject to de-rating, i.e. its electrical performance becomes similar to that of the same uncoated string. This difference in behavior is significant. With RTV-coated insulators, a line can be re-energized if contamination level is not too high or, alternatively, operation can continue but with an increased number of expected annual flashover events, Ne. Maintenance is another aspect in regard to composite and RTV insulators. RTV insulators can be maintained similarly to uncoated cap & pin insulators. In the case of RTV-coated insulators, maintenance actions are low in cost since these are limited to naked eye visual checks for self-shattering without need for climbing towers, namely the same as for glass insulators. Low frequency visual checks from the tower can be performed to assess the condition of the coating and such visual checks can be performed sampling selected towers along the line.
Based on experience in Italy, significant damage to a coating (as for example shown in Fig. 5) does not alter behavior of the string under contamination if such damage is limited to only several units. Moreover, there have never yet been cases of serious damage found along entire strings, even after more than 10 years of service. In this respect, it is important to note that in-factory coating yields a higher level of reliability than field coating in terms of long lasting adhesion. As such, RTV coated line insulators can be installed on the Italian high voltage grid only if they have been pre-coated in the factory. Composite insulators are much more costly to maintain than RTV-coated glass insulators. Their inherent characteristic to abruptly lose mechanical or electrical performance requires a different approach to maintenance. This will involve annual visual inspection from the tower together with specialized equipment either able to assess infrared and ultraviolet emissions or to record field distribution along the insulator in order to detect critical hidden defects (see Fig. 6). DC lines are usually long and reliable methods to easily detect faulted insulators are a big advantage. For this reason, Terna has decided to use only uncoated or factory pre-coated glass insulators only.
Recently, a limited pilot installation was started on a 132 kV line equipped with half-coated RTV glass cap & pin insulators (see Fig. 7). If found to be reliable, this solution could prove an economical alternative to fully coated insulators for application in heavily contaminated areas, providing a solution somewhere between uncoated glass and full coating. Finally, a new natural contamination laboratory called LANPRIS went into operation in late spring of 2017 and is located in a contaminated area near the sea in southern Sardinia. Here, there is a combination of heavy marine pollution due to year round strong wind and industrial pollution from local iron and steel industries. This facility will be used to test insulators stressed at different specific creepage distances in order to gather more information on ageing processes.
Terna has always used brown glaze porcelain for solid core post insulators (e.g. for busbar, bay and switchgear supports) as well as hollow core insulators (e.g. for current and voltage transformers, breakers, arresters, bushings, cable terminations, etc.). The first trials with composite hollow core insulators started in the 1990s. Today, due to safety and security against risk of explosion due to internal arcing, composite hollow core insulators are mandatory for all new installations and also in all cases of refurbishment/replacement of old porcelain insulators. In this respect, all hollow core composite insulators must pass specific Terna tests, e.g. a 2000-hour test (also applicable for composite and RTV-coated line insulators) and a recommended DC inclined plane test. The latter is mandatory, even for hollow core insulators intended for AC applications. Moreover, in the case of composite insulators, Terna requires a specific maximum tangential electric field on the insulator surface. Also, for high temperature vulcanized (HTV) materials, at least 45% of the weight must be alumina trihydrate (ATH) filler. In addition, artificial pollution testing is not accepted on hollow core insulators alone.
To cope with the need for frequent insulator washing at substations located in extremely polluted environments, Terna has resorted both to greasing and to field RTV coating of porcelain. The former, although cheaper, has a more frequent replacement cycle. The interval between consecutive replacements of the grease relates to level of contamination but typically ranges from 3 to 6 years. Fig. 8 shows a post insulator that has been greased. By contrast, on site RTV coating, can reduce station section outage time since it is assumed that re-coating is necessary only after about 10 years (assumed end-of-life). Fig. 9 shows a busbar porcelain post insulator that was coated in the field and laboratory tested after 7 years of operation. Fig. 10 shows a field coated porcelain post insulator supporting the smoothing reactor of the Italy-Greece HVDC Intertie to cope with harsh environmental conditions at the Galatina Converter Station.
Outdoor insulation dimensioning is not an easy process. Many variables must be taken into account and the uncertainty in regard to the values of such variables often requires caution. For this reason, statistical methods can better manage uncertainties and lead to a more robust estimation of expected flashover events per year. Although behavior of alternative types of insulators/strings under different laboratory contamination levels can be reasonably inferred from the wide number of tests performed, estimation of insulator contamination over the course of a year is a difficult task because:
• measurements performed in the field are usually small in number and not necessarily extendable to a wide territory;
• theoretical models perform well in certain areas but poorly in other areas;
• even if environmental contamination level is well estimated, specific models need to be used to convert such data to insulator surface contamination.
Terna has gathered a lot of experience directly from field operation of its grid made up of almost 73,000 km of HV/EHV lines and 855 HV/EHV stations. Experience is vital in insulation dimensioning process since it adds a further confidence limit to results obtained.
Overhead line insulators and substation insulators have much in common but need to be treated separately since the peculiarities of each will create differences in selecting creepage. Italian high voltage lines are widely insulated with glass cap & pin insulators with highly satisfactory results in terms of both service life and maintenance needs. They offer a highly robust solution. Nevertheless, composite and RTV pre-coated glass insulators are used in areas with harsh contamination. Although composite insulators can give good results, it is complicated to predict their end-of-life since, in some case, abrupt failures have been recorded. Consequently, specific countermeasures need to be adopted in order not to incur critical situations. Pre-coated RTV glass insulators have the advantage to be considered an improvement over non-coated glass for application in heavily polluted areas. These offer high electrical and mechanical reliability, residual electrical performance at the end-of-life (e.g. in the worst case, electrical performance is similar to bare glass insulators) and are therefore a robust solution for the grid.
Italian substations are equipped mostly with brown glaze porcelain post and hollow insulators. However, since some years now all new hollow core station insulators and all the apparatus equipped with hollow core insulators must be composite type. This policy has been taken to increase safety and security against internal arcing faults. In harshly polluted areas, grease and in-field RTV coating on porcelain insulators are used to reduce risk of failure. Continuous research and field returns are of paramount importance to promote better selection of creepage and high grid reliability.