Upgrading Network from 38 kV to 110 kV (Video)
by Muhammad Shariq Hassan

The design enabled conversion of existing 38 kV overhead lines to 110 kV, utilizing similar structures in the same locations as those they replace. This approach not only increases capacity along selected line corridors but minimizes environmental and community impact.

The post Upgrading Network from 38 kV to 110 kV (Video) appeared first on .

Critical factors for asset management of bushings include consistent monitoring of insulation integrity, prompt detection of partial discharge activity, and regular assessment of mechanical and thermal stresses. Implementing these measures can optimize bushing performance and extend their operational lifespan.

The post Best Practice for Reliable Bushings Asset Management (Video) appeared first on .

FortisBC serves customers across British Columbia with some 7300 km of 63 kV, 138 kV, 161 kV and 230 kV overhead lines, most utilizing wood pole construction. In the past, the normal service life of these poles was about the same or less than that of the porcelain and glass insulators installed on the network. Replacement was therefore a relatively simple matter of exchanging insulators whenever there was external damage to them or when poles reached their end-of-life, as determined during routine inspections.

However, with recent application of polymeric insulators to this network combined with adoption of more durable structures made of steel, what once seemed a relatively straightforward decision has become complex. Now, structures and insulators can have significantly different service lives.

This edited contribution to INMR by Aram Khalil-Pour, Director of Engineering Services & Asset Management at FortisBC, in cooperation with Staff Engineer, Kevin Kendal, reviews one such case and the testing and other considerations that went into deciding on insulator end-of-life.

Hivolt Power System

China

EC Insulator Jiangxi Co., Ltd.

China

See more suppliers of Support Insulators for Substations

Monitoring the condition of in-service insulators is a constant requirement to obtain metrics on when replacement may be needed. Deciding on optimal frequency of such testing is often made according to a schedule based on number of years in service and then some pre-established repeat period thereafter. A different approach involves waiting either until visual defects are detected or until failures start to occur. Selecting the timing for testing depends on relative costs and degrees of acceptable risk and both need to be evaluated based on a line’s criticality.

Testing generally requires removing sample insulators from service and completing some form of destructive testing. While there do exist live-line test techniques, these do not provide the same level of certainty as do destructive tests and are better suited to deciding if there may be a need to remove samples for in-depth tests at a laboratory.

Results from such testing help asset managers determine whether insulators are truly at end-of-life and require urgent or scheduled replacement or if replacement might be premature.

Relying on test results to drive large capital investments comes with the possible downside that it becomes difficult to forecast the cost to replace insulators unless multiple testing cycles indicate clear trends. Also, more cycles of testing could mean a greater possibility that insulators are replaced prematurely. It can therefore be a balancing act to determine when it is best to start testing and the frequency with which these tests should be repeated thereafter.

In the case of FortisBC, the first round of insulator testing began in 2016 and involved a 230 kV line equipped with polymeric insulators that were suspected to be nearing their end-of-life after 25-30 years’ service. The line in question runs in a corridor containing two 230 kV lines and a single 138 kV line. Two of these lines are built on double circuit steel structures while the third runs in parallel but has a traditional wooden H-frame construction. The terrain consists mainly of lightly forested hillsides with low insulator contamination.

The 230 kV line in question was equipped with two different makes of polymeric insulators installed on tubular steel structures. During routine condition assessment in 2016, these composite insulators were showing signs of degradation although there had not yet been any insulator failures reported. Still, given the criticality of this line and based on the recommendation of a consultant, a small population of these (1 post and 7 suspension insulators from Manufacturer A and 3 suspension insulators from Manufacturer B) were removed and sent to a local high voltage laboratory for testing. These insulators were mid 1980s vintage with about +30 years in-service.

The following tests were conducted:
• Megger test;
• Dry power-frequency voltage flashover;
• ESDD measurement;
• Hydrophobicity;
• Dye penetration;
• Moisture penetration.

These were completed based on relevant standards and guidelines:
• CSA C411.1 – AC Suspension Insulators;
• CSA C411.4 – Composite suspension insulators for overhead lines > 75 kV;
• IEC TS 62073 – Guidance on the measurement of hydrophobicity of insulator surfaces;
• IEC 60060-1 – High-voltage test techniques – Part 1: General terminology and test requirements;
• STRI Guide 92/1 – Hydrophobicity Classification Guide.

Visual inspection was also completed on all sampled insulators. For Manufacturer A, there were signs of electrical activity, build-up of gray residue and small areas of erosion. These were considered to have failed visually. Insulators from Manufacturer B showed no evidence of degradation and only had small build-ups of dirt. These passed visual inspection.

First was the Megger test, which saw 5000V DC applied across the end fittings to yield an overall resistance value. All 11 insulator samples passed with high resistance values, as seen in Table 1.

A second electrical test was the dry power-frequency voltage flashover test, which sees a voltage of 75% of flashover rating applied to the insulators and gradually increased until flashover. This process was repeated 5 times for each insulator. All 3 insulators tested this way passed and exceeded 95% of manufacturer ratings (see Table 2).

Next came measurement of equivalent salt deposit density (ESDD). Three insulator samples were washed in a 2-litre bucket of de-mineralized water. Due to their length, each insulator was first cut into 3 sections. Resulting water temperature and conductivity were measured and overall ESDD was calculated based on total insulator surface area. According to IEC 60815, ESDD levels were considered light or very light.

The hydrophobicity test involved spraying water onto insulators for 20 to 30 seconds from 25 cm (±10 cm) away. Hydrophobicity classification (HC) of the tested area was then evaluated based on the STRI Guide 92/1 10 seconds after spraying. The sheath, top and bottom of three insulator sheds on three different insulators were tested. Two insulators had a hydrophobicity rating of HC1 for each shed, while the third had reduced hydrophobicity at the bottom of each shed. Based on their age in service, these levels were deemed acceptable, and all insulators were considered to have passed.

Next came the dye penetration test, which is designed to evaluate the condition of the core material based on penetration of dye through cut samples of each insulator. If the dye penetrates within 15 min of exposure, this is considered a failure. Ten samples were taken from each insulator. All samples of the first insulator tested had no dye penetration over a 15-min time frame. The second insulator, however, had dye penetration within 1 to 3 seconds after exposure for 8 of the 10 cut samples (the remaining 2 had no dye penetration). The final insulator had dye penetration within 8 to 10 seconds after exposure for 1 sample and after 1 minute for a second sample. There was no dye penetration for the remaining 8 cut samples.

Both insulators that had dye penetration were considered to have failed.

Finally, came the moisture penetration test to verify integrity of end fitting seals. This test submerges each of the end fittings in dye for 15 min and then cuts them to determine if any dye has penetrated. Three insulators were tested and all passed.

Based on findings from the first round of testing (2016), it was decided that the polymeric insulators on this 230 kV line were in an acceptable condition to remain in service. Despite a secondary recommendation to repeat testing within a 3-year time window and with an expanded sample of insulators, the second round of tests was slightly delayed.

In 2021, 27 insulators from the same 230 kV line as well as from the adjacent 138 kV line of the same vintage, were removed for laboratory testing. The test program was as in 2016 but this time with addition of a mechanical load test.

Results from the second round of tests showed:

• Further visible damage to insulators, including cracks in the sheath, sheds and metal fittings. Signs of galvanization flacking and corrosion were also identified on metal fittings;
• All insulators passed the Megger test;
• All passed the dry power-frequency voltage flashover test;
• Mechanical load testing was passed, with all insulator samples reaching their routine test load and only 1 of 8 samples not reaching their rated ultimate strength;
• Testing showed decreased hydrophobicity for multiple insulators, with 1 insulator having HC7 on its bottom sheds;
• Only 2 out of 10 insulators passed the dye penetration test.

Conclusions

One of the key questions arising from these types of findings relates to what utilities should do next. From an electrical and mechanical strength perspective, all insulators are still in an acceptable condition. However, due to the failed dye penetration tests in combination with reduced hydrophobicity, it seems clear that these insulators are at risk of accelerated further degradation. Based on the 2 test cycles (2016 & 2021) involving 38 insulators sampled from the total population of 774 on the line, it is still difficult to predict when these may start to fail.

Due to the criticality of the line as well as the age of the insulators, a risked-based decision was made to replace all insulators over the next 3 rehabilitation cycles, as follows:

1. During first cycle in 2025
All dead-end and heavy angle insulators on this critical line will be replaced due to their higher mechanical loads.

2. During second cycle in 2032
All remaining suspension insulators will be replaced.

3. During final cycle in 2040
All polymeric post insulators will be replaced, even though these had passed original testing, since they will already be past their originally expected service-life.

At FortisBC, replacement of in-service transmission insulators follows a condition-based approach which relies mainly on visual inspection, age and records of failure to determine when insulators are nearing end-of-life. At the same time, further testing of batches of these polymeric insulators will be undertaken prior to full line replacement to ensure that they are indeed at end-of-life.

The post Testing & Replacement Strategy for Polymeric Line Insulators: Case Study from Canada appeared first on .

This edited past contribution to INMR by Andrew J. Phillips of EPRI in the United States discussed such failures and proposed strategies to address risk of premature ageing of polymeric insulators due to high E-field.

DALEKOVOD OSO

Croatia

PFISTERER

Germany

See more suppliers of Laboratory & Field Testing Equipment

EPRI was among the first organizations to determine that high E-field and resulting discharge activity was a key contributor to premature ageing of polymeric insulators. This was identified as a primary ageing mechanism on 230 kV and 500 kV insulators based on results from the multi-stress ageing chamber and other testing in Lenox, Massachusetts and appropriate E-field limits were established. However, subsequent insulator failures at 115 and 138 kV later suggested that this phenomenon could also be a risk factor at lower system voltages. Ageing of insulator sections subject to localized high electric field is usually the result of stresses associated with one or more types of discharge activity:

• Continual corona activity from metallic end-fittings or grading rings under dry conditions;

• Discharges due to non-uniform wetting of the rubber material;

• Internal discharges, e.g. along the interface between the core and rubber housing or within the core itself.

Continuous corona activity from the metal end fittings contains sufficient energy to cause erosion on rubber as well as loss of galvanization on metal end fittings. Moreover, individual drops or relatively limited water patches on hydrophobic insulators can enhance localized E-field by a factor of up to 12 due to the high permittivity of water (εr = 80). In the high E-field regions of the insulator, such enhancement could result in corona activity from the edge of the water. Research has suggested that it is unlikely that water drop corona alone will result in significant degradation of the housing since temperature increase from this type of corona is minimal. However, there is ample evidence that the chemical by-products of corona – together with moisture – can cause serious material degradation. Formation of nitric acid is considered important in this respect. For example, it has been found that the pH on the surface of an insulator drops from an initial value of about 7 to 3.4 after only about 15 minutes of corona activity on a wet insulator surface. Moreover, it has been found that some formulations of silicone rubber can be especially vulnerable to deterioration when exposed to nitric acid. Evidence suggests that water drop corona can be just the initial phase of a more severe, degradation mechanism that can affect long-term insulator performance. This process is thought to be as follows:

1. Water drop corona in the high E-field regions results in localized loss of hydrophobicity. Regions affected have E-field magnitudes above the onset threshold for water drop corona (see Fig. 2);

2. Under wetting conditions, patches of surface water form in regions of lower hydrophobicity and are separated by dry regions or ‘bands’;

3. Localized arcs form, bridging gaps between water patches;

4. The energy and temperature of these localized arcs are significantly higher than that of water drop corona, further stressing the rubber;

5. With time, as affected regions lose hydrophobicity and completely wet out, E-field in the adjacent regions is enhanced above the water drop corona onset threshold under wetting conditions;

6. The ageing mechanism is then initiated in previously unaffected regions. In this manner, affected regions grow in size;

7. By-products formed by corona in combination with water, notably nitric acid, can be aggressive on the housing, resulting in cracks or corrosion of end fittings.

During wetting conditions, the surface of hydrophilic composite insulators, such as EPDM, is more likely to become covered with patches of water rather than distinct droplets. Dry regions separate these patches and, due to E-field enhancement, sparking can occur between patches. These discharges contain more energy than corona and could degrade rubber. Although such activity can also occur away from the high E-field region, observation during ageing tests suggests that it is more prevalent in regions with high E-field.

Sufficiently high E-field magnitudes can lead to discharge activity within any internal defects, e.g. voids, inclusions from poor bonding between sheath and core. This, in turn, could eventually lead to insulator failure either by destruction of the rod from discharge activity or by ‘flashunder’ (see Fig. 5).

Research has shown that not all insulators are equally affected by high electric fields. Important factors that influence the rate and level of degradation include:
• Type of rubber and design of weathershed system;

• Design of end fitting seal;

• Level, location and type of discharge activity, which is determined to a large extent by E-field along insulator, type and intensity of wetting, presence of contaminants and level of hydrophobicity of material.

Service Experience

Reports of insulator failures on 115 kV and 138 kV lines a decade ago prompted some utilities in the U.S. to initiate a study to better understand ageing mechanisms on insulators of this voltage class. EPRI’s failure database at the time showed that these were not isolated incidents but rather part of a trend of increasing failures on such insulators. In fact, on average, 7 failures/year had been reported to EPRI leading up to 2012. Breakdown of the different failure modes of the incidents recorded in the EPRI database for 115 kV to 138 kV insulators indicated that the dominant failure modes were stress corrosion cracking (brittle fracture) and flashunder. A high proportion of these were on the same insulator design and specifically on units manufactured between 1993 and 1999.

During the investigation, it was shown that all such failures could be attributed to continuous discharge activity from the end fitting under dry conditions. This continuing exposure to corona resulted in cracks in the rubber sheath and degradation of the end fitting seal. Once a seal is compromised, moisture can come into contact with the fiberglass rod, leading to brittle fracture. Brittle fracture is a mechanical failure of the rod due to acid attack and where the fracture exhibits one or more smooth planar surfaces – mainly perpendicular to the axis of the rod and giving the appearance of the rod being cut. As a consequence of these failures, utilities were forced to re-examine use of corona rings (or lack thereof) on 115/138 kV polymeric insulators. In cooperation with EPRI, utilities have since initiated a number of specific activities to assess risk to 115/138 kV polymeric insulators from premature ageing due to high electric fields. These included:

• Daylight discharge inspections;

• Detailed examination of insulators taken from service, failure investigations;

• E-field calculations.

These activities focused mainly on one particular design that suffered most failures.

Daylight Discharge Inspections

EPRI and 5 utility members together performed daytime discharge inspections on twelve 115 and 138 kV transmission lines to determine whether continuous discharge activity was occurring from end fittings under dry conditions. While these inspections were primarily directed towards one design, there were also opportunities to inspect other designs.

Conclusions from these inspections were:

• Corona discharge activity under dry conditions was observed on the end fittings of composite insulators installed on all twelve 115 kV and 138 kV transmission lines inspected, though not all insulators had corona activity;
• Corona discharges were more likely on dead-end strings and least on brace post configurations. This is not surprising since it is known from previous calculations that E-field is generally higher for dead-end insulators than for suspension units or brace post configurations;
• Corona activity was observed on 3 of the 4 insulator designs from different manufacturers (see Fig. 8);
• In one case, daylight corona observations were made before and after installation of a corona ring. This confirmed that addition of the ring eliminated corona from the end fitting.

Detailed Inspection

EPRI worked with several utilities to evaluate degradation on over 200 units of 115 and 138 kV insulators removed from service. All had been installed between 1994 and 2006, without corona rings, and were of the same design. 74 insulators removed from service were subjected to detailed examination comprising (1) visual inspection, (2) hydrophobicity measurement, (2) dye penetration test, (3) dissection and, in some cases, (4) mechanical testing. The remaining units were evaluated only by visual inspection. In all cases, the most severe degradation was observed in the same regions, i.e. where dry corona activity was seen during daylight discharge inspections. On some units it was found that degradation of the sheath and end fitting seal had progressed to the point that the rod was exposed to the environment. These were considered as high-risk units where failure was inevitable.

E-Field Calculations

3-D E-field calculations were performed at both 115 and 138 kV to obtain a better understanding of expected E-field distribution on these insulators as well as the parameters that influence it. Another goal was to evaluate possible remedial measures. Significantly, E-field calculations were made to account for the presence of all three phases (and in some cases even adjacent circuits) on structures. Calculations focused on structures where failures had already occurred or where corona has been observed. These calculations considered both E-field on end-fittings (to indicate likelihood of dry corona) and along the sheath (to indicate the likelihood for water induced corona). The following conclusions were drawn:
• Dead-end insulators have higher E-field magnitudes than suspension insulators;
• Single dead-end insulators have higher E-field magnitudes than double dead-end insulators;
• Addition of a hot line link results in slightly higher E-field magnitude on the insulator;
• There is a significant difference in E-field levels between different insulator designs (see Fig. 8). Small and slender end fittings tend to have higher E-fields in the region of the end fitting seal. Shape of end fitting dictates where the highest field occurs and accordingly whether or not any dry corona present will be in contact with the housing material;
• E-field magnitudes exceeded EPRI recommended limits on all designs of 115 and 138 kV polymeric insulators installed without corona rings;
• Addition of 8” corona rings at the live end of the insulator was in most cases sufficient to reduce E-field magnitudes to an acceptable level;
• Results of E-field modeling combined with corona camera inspection confirmed that failures that occurred on 115 and 138 kV insulators as well as degradation observed could be associated with high E-fields on the insulators;
• E-field limits need to be adjusted downward for insulators installed at high altitudes i.e. above 3300 ft (1000 m).

Population Assessment

Service experience suggests the need for corona rings on 115 kV and 138 kV polymeric insulators to protect against risk of premature ageing due to corona activity. Although this conclusion may seem simple, the implications could be quite profound, especially if large numbers of such insulators are already in-service. Utilities would then be faced with the challenge of identifying high-risk units and deciding the most appropriate and cost effective remedial actions. Fortunately, deterioration due to corona discharge activity develops slowly and this gives utilities time to conduct proper condition assessment. EPRI helped develop a population assessment strategy to address premature ageing of polymeric insulators on 115 and 138 kV lines due to high E-field. Key in this strategy is a set of tools that includes field guides, failure databases, E-field modeling techniques, corona inspection technologies and relevant accelerated ageing test results.

Conclusions

Starting about 20 years ago there were increasing reports of failures of polymeric insulators on 115 and 138 kV lines. These were particularly serious since they mostly involved critical dead-end insulators that pose a threat to system integrity as well as risk a downed conductor. Investigation suggested that these failures were due to high electric fields occurring close to or on the high voltage end fittings of affected insulators. Consequently, corona or grading rings may also be necessary for polymeric insulators installed at these voltage levels. Levels of dry corona activity from end fittings that occurred in-service were higher than expected based on laboratory testing. E-field modeling offered two explanations:

1. At 115 and 138 kV, proximity of nearby phases significantly increases surface E-field magnitude;

2. Most laboratory testing is done on suspension configurations but E-field magnitudes on dead-end and hard angle insulators are higher.

Laboratory Center of Koncar Electrical Engineering Institute

Croatia

CATU Test Laboratory

FRANCE

See more Laboratories

The post Mitigating Corona on Composite Line Insulators appeared first on .

At high voltages, electrically sufficient hydrophobicity is assumed given a receding contact angle of approximately 40° or a classification of HC 2. But at lower dynamic contact receding angles, significant current can flow through the pollution layer and occasional initial discharges (so-called dry band discharges) can ignite. If such discharges occur over a prolonged period, thermal and chemical damage can occur until the insulator fails.

This edited contribution to INMR by Professor Stefan Kornhuber of the University of Applied Science Zittau/Görlitz, Germany reviewed test methods for assessing hydrophobicity.

EB Rebosio SRL, A Gruppo Bonomi Company

Italy

Guangzhou MPC Power International Co. Ltd.

China

See more suppliers of Insulators for Substation Equipment

While the pollution flashover exponent for non-hydrophobic insulators is 0.25, in the case of hydrophobic insulators it falls between 0.01 and 0.1. This means less reduction in pollution flashover voltage with increasing conductivity of the pollution layer. Due to different stress and multi-stress situations, the hydrophobicity property of a material can diminish over time, which leads to lower flashover performance from the hydrophobic curve to the non-hydrophobic curve and significantly reduced flashover performance.

Silicone materials also exhibit a certain ability to recover hydrophobicity and insulators in service can revert back to their original hydrophobic pollution flashover performance. Given this, creepage distance reduction factors have been introduced for so-called hydrophobicity transfer materials (HTM). Requirements to be termed HTM include:

• retention of hydrophobicity against certain stresses such as partial discharges under wet conditions (water droplet corona);

• recovery of hydrophobicity after a rest period; and

• transfer of hydrophobicity into polluted surfaces.

Understanding hydrophobic behaviour and the processes that lead to its reduction and recovery is therefore crucial in order to select the proper insulator material and design. Hydrophobicity transfer is already included in the new version of IEC TR 62039 and results of multi-stress testing to investigate hydrophobicity retention and recovery (e.g. the Dynamic Drop Test) have been summarized by CIGRE WG D1.58.

Test Methods

1. Hydrophobicity Transfer Test

In the case of silicone-housed materials in particular, the polymeric structure can be covered by a foreign layer representing pollution. It is known that short and medium-length polymeric chains, i.e. the so-called low molecular weight (LMW) and medium molecular weight (MMW) components, have the ability to penetrate this layer and form a new hydrophobic layer (so-called ‘hydrophobization’). This is particularly necessary to prevent wetting of the layer and thus occurrence of dry band discharges, as described earlier (see Fig. 3).

Testing for this hydrophobicity transfer mechanism has been specified in the new revision of IEC 62039, published in 2021. In principle two different test methods are possible:

• Method A, with quartz powder according to CIGRE;

• Method B. with Kieselguhr according to DL/T 810.

Method A

Test specimens are covered with adhesive foil to obtain a window of 30 mm × 30 mm (L x W). Thickness of these foils defines thickness of the pollution layer and 0.36 mm thickness is to used. Inside the area marked by the foil window, the specimens are coated by applying a slurry made of 7.5 g untreated silica powder (i.e., not silanized, medium grain size of approximately 3 µm) as well as approximately 3.5 ml of a mixture of water and isopropanol (65% water and 35% isopropanol by volume) that has been homogenized by stirring. A clean medical blade is used to wipe off any excess slurry. Since isopropanol tends to evaporate, the slurry is to be used shortly after its preparation. The result is a smooth and even surface.

After application of the slurry, samples are stored for 96h in desiccators under controlled 53% ± 10% relative humidity and temperature of 23°C ± 2°C (e.g. using a saturated solution with magnesium nitrate or a climate chamber).

Method B

The area of the test specimens should be around 30 cm² to 50 cm² and their thickness between 3 mm and 6 mm. Polishing test specimens is not permitted before the test. The pollutants include Kieselguhr, in accordance with IEC 60507:2013 Table 2, and NaCl. The Kieselguhr is weighed and put on the surface of each specimen and the NaCl solution is dropped on the Kieselguhr using a pipette or syringe. The Kieselguhr and NaCl solution are mixed and then evenly applied on the specimen using a small paintbrush.

Desired amounts of Kieselguhr and NaCl are 0.5 mg/cm² and 0.1 mg/cm² respectively. Mass of Kieselguhr is calculated according to the area of the test samples. The 5 polluted specimens are then placed in a dust-proof container for 96h under standard laboratory ambient conditions (i.e. 40% to 70% relative humidity and 20°C to 25°C).

Evaluation Criteria

Both methods provide results which lead to the same acceptance criteria in regard to the static contact angle from 5 measurements:

• Average value: ≥90°;

• Minimum value: ≥80°.

The criterion is related to static contact angle, which is measurable anywhere across the globe. In addition, several publications as well as IEC 62073 show that the dynamic receding contact angle provides better information in regard to related leakage current behaviour. Dynamic receding contact angle is used in the evaluation and the static contact angle should be measured for conformity with the standard.

Comparison of Test Results

Several measurements were done at different laboratories to compare test methods during the course of development of the standard and related discussions. Fig. 3 for example compares results of Methods A and B, whereby the values shown are the average mean value of measurements according to the specification.

2. Dynamic Drop Test

Test Method 

The Dynamic Drop test (DDT) permits accelerated evaluation of hydrophobicity retention when the surface of a material is subjected to electrical micro-discharges caused by water droplets. This is achieved by supplying an electrolyte with defined volume conductivity and flow rate under simultaneous electric field stress.

Samples are arranged at an inclination angle to the horizontal axis (see Fig. 4). The evaluation criterion is time to loss of hydrophobicity. This stage is achieved when a conductive electrolytic path between high voltage and ground electrodes is formed, which can be detected by measurement of leakage current.

At the beginning of the test, the electrolyte forms discrete droplets that roll down the specimen surface with more or less constant frequency. Presence of the droplet can result in field intensification and hence in ignition of electrical micro-discharges which could lead to localized reduction in hydrophobicity. This process continues until the electrolytic path becomes completely hydrophilic and a wetted path bridges the electrode distance. The failure criterion is reached with increase in leakage current to the mA range.

The upper-grounded electrode is supplied with an electrolyte that continuously provides droplets onto its lower side. As such, droplets flow on the surface of the insulation material to the energized electrode at the bottom. As a result of continuous electrolytic and electrical field stress on the drop-off sliding track, its hydrophobicity property is lost.

At the start of the test, individual electrolyte residues are formed and increase progressively in number and size. At the end, the number of residues is so high that, in conjunction with a draining drop, a continuous electrolyte connection of both electrodes occurs, which can carry a resistive current. The time taken to reach this state is defined as the retention time (tA) and can be detected by a current measurement system. Table 1 summarizes test parameters during this investigation.

Reference Measurement

For reference measurement and validation of the test set-up, a special material was found, i.e. the medical silicone, Replisil 22 NO. Not used in high voltage engineering, it is two component curing material at room temperature. Figs. 5 and 6 show initial results with this medical silicone material.

In late 2021 again investigations were done for evaluating the test setups with the same procedure and the same medical silicone, which is shown in Fig. 7.

In summary, the medical silicone Replisil 22 NO seems to show suitable properties for evaluating the test set-up and for comparing measurements. However, once produced, this type of silicone exhibits changing behaviour over time and the same timing must therefore be satisfied.

Summary & Conclusions

The following key properties are necessary for reliable, long-term functionality of composite insulators manufactured with hydrophobicity transfer materials:

• retention of hydrophobicity against stresses such as partial discharges under wet conditions (water droplet corona);

• recovery of hydrophobicity after a rest period; and

• transfer of hydrophobicity into polluted surfaces.

The new version of IEC 62039 includes testing for this hydrophobicity transfer property using two different test methods (A and B), which nonetheless show similar results. However, having two tests for only one property seems not an ideal solution, Moreover, the static contact angle by itself is still not fully representative of behaviour for electrical purposes.

The Working Group behind the new document will now get the first response and experience and, based on this, adjust the next version.

For the Dynamic Drop Test (DDT), CIGRE WG D1.58 has been defining test set-up and the possible evaluation criteria since 2014. After several Round Robin tests, test description and procedure can be recognized and minimum retention time to validate a material have been reached. In terms of subsequent steps, a final brochure will be written and all remaining open questions will need to be solved in a new WG. Moreover, IEC WG 5 decided at the TC 112 meeting in 2019 that this test will soon be proposed as a new work item.

References
[1]    Bär, Christiane; Schmuck, Frank; Kornhuber, Stefan; Bärsch, Roland; Brade, Volker: Influence of the Material Composition on the Dynamic Hydrophobicity of Silicone Elastomers for high-voltage Outdoor Application. In: CIGRE Session 2018. Paris, 2018
[2]    IEC, TR 62039:2021 Selection guide for polymeric materials for outdoor use under HV stress. 2021.
[3]    J. Kindersberger and R. Bärsch, “Schlussbericht zu dem IGF-Vorhaben Prüfverfahren für die Bewertung wasserabweisender Eigenschaften polymerer Isolierwerkstoffe für Hochspannungsanwendungen (IGF-Vorhaben 17001 BG),” Nov. 2015.
[4]    C. Bär, R. Bärsch, A. Hergert, and J. Kindersberger, “Evaluation of the retention and recovery of hydrophobicity of insulating materials in high voltage outdoor applications under AC and DC stresses with the Dynamic Drop Test,” IEEE Trans. Dielect. Electr. Insul., vol. 23, no. 1, pp. 294–303, Sep. 2016.
[5]    A. Hergert, J. Kindersberger, C. Bär, and R. Bärsch, “Transfer of hydrophobicity of polymeric insulating materials for high voltage outdoor application,” IEEE Trans. Dielect. Electr. Insul., vol. 24, no. 2, pp. 1057–1067, Apr. 2017.
[6]    F. Schmuck, “Zur zeitraffenden Alterungsprüfung von Silikongummi-Oberflächen unter Fremdschichtbelastung und simultaner 50-Hz-Spannungsbeanspruchung,” Zittau, Techn. Hochsch., Diss., 1992., 1992.
[7]    IEC, TS 60815-4:2016 Selection and dimensioning of high-voltage insulators intended for use in polluted conditions – Part 4: Insulators for d.c. systems, Oct. 2016.
[8]    C. Bär, “Bewertung dynamischer Hydrophobieeigenschaften polymerer Isolierstoffe mit dem Dynamischen Tropfen-Prüfverfahren unter Wechsel- und Gleichspannungsbeanspruchung,” München/Zittau, Diss., 2016.
[9]    IEC, TS 60815-3:2008 Selection and dimensioning of high-voltage insulators intended for use in polluted conditions – Part 3: Polymer insulators for a.c. systems, Oct. 2008.
[10]  Christiane Bär, Roland Bärsch, Alexander Hergert, Josef Kindersberger: Evaluation of dynamic hydrophobicity properties with the dynamic drop test under ac and dc stress and the hydrophobicity transfer test. 19th International Symposium on High Voltage Engineering (ISH 2015_335), Pilsen, 2015
[11]  CIGRE WG D1.14: Evaluation of dynamic hydrophobicity properties of polymeric materials for non-ceramic outdoor insulation: retention and transfer of hydrophobicity. Technical Brochure 442, Paris, 2010
[12]  Roland Bärsch, Josef Kindersberger, Christiane Bär, Alexander Hergert: Prüfverfahren für die Bewertung wasserabweisender Eigenschaften polymerer Isolierwerkstoffe für Hochspannungsanwendungen, Schlussbericht zu dem IGF-Vorhaben 17001 BG der Forschungsvereinigung Elektrotechnik, München und Zittau, 2015
[13]  Alexander Hergert, Josef Kindersberger, Christiane Bär, Roland Bärsch: Stand und Entwicklung von Prüfverfahren zur Bewertung dynamischer Hydrophobieeigenschaften polymerer Isolierstoffe für den Hochspannungseinsatz. Beitrag der 4. ETG-Fachtagung: Grenzflächen in elektrischen Isoliersystemen, Dresden, 2013
[14]  Alexander Hergert: Test methods for evaluating the dynamic properties of hydrophobicity of polymeric insulating materials. Dissertation, Technische Universität München, 2017
[15]  Hergert, Josef Kindersberger, Christiane Bär, Roland Bärsch: Transfer of hydrophobicity of polymeric insulating materials for high voltage outdoor application. IEEE Transactions on Dielectrics and Electrical Insulation 24 (2017), 1057–1067
[16]  Heike Herzig, Stefan Kornhuber: Defined silicone rubber surface structures in a long-term test. IEEE 2nd International Conference on Dielectrics, Budapest, 2018
[17]  Cervinka, Rüdiger, Stefan Kornhuber, und Christiane Bär. „Investigation for a reference material for the dynamic drop test (DDT) with a reproduceable and repeatable retention time“. In 2019 IEEE 4th International Conference on Condition Assessment Techniques in Electrical Systems (CATCON), 2019.

The post Hydrophobicity Transfer & Retention appeared first on .

 

Among the technologies to assess the condition of power network assets are infrared and ultraviolet imaging. Both have been used for decades and both rely on specialized cameras that have evolved towards greater portability, improved data storage and sharing and greater ease in assessing the severity of whatever problems are identified. The continued existence of these two imaging technologies has made some wonder whether they are complementary and must be used together or whether they are alternatives that each can detect the same incipient problem.

The best answer is that each technology offers unique specific benefits when it comes to locating certain types of defect. In general, it can be said that infrared imaging records presence of internal heat due to elevated leakage current while ultraviolet imaging detects presence of corona due to high electric field near surfaces. The first phenomenon depends on an internally-generated heat source and the second on surface condition. Both situations represent a possible threat to continued safe operation of a line component or equipment at a substation and are reasons for closer scrutiny or immediate remedial maintenance.

DALEKOVOD OSO

Croatia

Phenix Technologies (A Division of Doble)

United States

See more suppliers of Laboratory & Field Testing Equipment

Not that long ago, infrared equipment used for inspection of power system assets was heavy and cumbersome. These days, however, an IR camera with several times the pixels of the bulkier versions it replaced is light and portable.

In IR inspection, the definitive factor in condition assessment is not absolute temperature but rather how temperature of the item being inspected compares to the reference temperature of similar nearby items. This is because what is crucial is not the temperature of any particular component being inspected but rather whether or not this temperature differs significantly from that of similar items located next to it.

IR imaging also requires the inspector to be aware of key meteorological parameters since it can be affected by humidity as well as wind conditions. For example, higher relative humidity makes any ‘hot spot’ appear cooler than it actually is. While there is no specific camera adjustment in the case of high wind, a trained operator has to be aware and take this into account when preparing an inspection report. One of the keys to all infrared imaging is emissivity since each item being inspected will emit at the same radiation as it absorbs. That means infrared imaging of each object has to be adjusted based on its known emissivity.

For example, during thermal inspection of a 138 kV Substation near Fort Myers, Florida, a hot spot was observed on the blade of a transmission switch. This is something that could have been caused by corrosion between the fingers and blade and probably only requires cleaning. A typical report by the inspector in such a case will pinpoint the location of such hot spots, indicating the device numbers and also the phase where these were detected. Moreover, findings of any inspection can have different consequences depending on system operator. Every utility has their own strategy on how to prioritize maintenance problems, whether urgent, serious or moderate. Sometimes it is a question of availability and what is happening elsewhere in the grid. Still, if a temperature differential of 150°F (66°C) is observed between the phases of the same load and at the same location, this is almost always regarded as ‘critical’. By contrast, localized temperature rises of up to about 20°F (6.7°C) can be due solely to increasing load or even just to changes in ambient conditions.

It is not unusual for inspectors to identify from 10 to 15 hot spots per inspection, with larger utilities often having 3 to 4 times this number. Typical locations of hot spots include distribution switches, which typically can account for about half of all registered during an inspection. Switches tend to have the most hot spots mainly because they are a moving part. Another common area for hot spots are bushings, generally due to loose or dirty connections.

Thermal imaging can be particularly useful when it comes to assessing condition of surge arresters. This is because they normally dissipate only little energy during steady state operation and therefore seldom exhibit a temperature much above ambient. Even some of the largest MOV arresters (e.g. up to 4 or 5 m in height) dissipate less than 50 watts and therefore do not create any discernable rise in temperature. This makes the effort to measure temperature gradients that much more challenging.

Experience has shown that better assessment of arrester temperature can be achieved in an environment without sunlight, allowing even less than a degree difference from ambient to be more easily observed. It is a step-by-step process such that, if an arrester displays even a slight difference from similar nearby units, it is placed on a watch list. If the arrester displays a difference of from 4 to 6°C, then it is marked to be re-checked in a few weeks. If the temperature gradient is greater than 7-10°C, the arrester is placed on an emergency change-out list.

This approach is common among many utilities. One point that is certain is that seldom does any surge arrester become 15°C hotter in comparison to nearby units. Once temperature starts to rise that much, leakage currents will also increase which in turn causes even faster heating. That means it will likely not survive much longer. Partial discharge does not itself generate enough of a heat signature to be detectable with thermal imaging.

There are several benefits to applying thermal imaging to arresters, starting with speed of data collection. Indeed, experts claim that there is no faster way to tell if an arrester is near end-of-life than a scan of its temperature. Accuracy from a distance is also excellent, especially with a long-range lens. The risk that an arrester is in the process of failing without also generating heat is low. At the same time, if an arrester has been damaged by lightning strike or switching surge only days after its last thermal scan, it may well fail before the next scheduled scan.

The potential for failure between successive scans is among the only drawbacks to thermal imaging of arresters. Another is that most IR devices currently used in the marketplace cannot transmit data remotely. As such, obtaining a thermal profile for an arrester typically requires someone to go on-site, often at night, and collect data. In spite of this, thermal imaging remains one of best options to monitor the health of an energized surge arrester.

Apart from substations, thermal imaging can also be applied efficiently to overhead lines. While much of the power industry is looking to increased application of fly-by techniques to monitor lines, driving a line route at around 20 mph (30 km/h) is sometimes preferable where possible – especially in the case of distribution lines. One of the advantages is that accuracy of measurements tends to be greater from the ground. The sky is the perfect cool background compared to the alternative of looking down from a helicopter to what could be numerous other heat sources on the ground.

One example of a situation where thermal imaging has proven valuable in assessing maintenance requirements involved a 500 kV line in southern China that had experienced brittle fracture of one of its composite I-string insulators. There was concern that this failure might be a precursor of a broader problem relating to unsuitable corona ring design for these insulators. A program was therefore instituted to inspect the entire population of line insulators to determine if the problem was widespread or just isolated to one unit which may have been defective or improperly installed. A total of more than 600 strings were examined over the course of the following months as part of a systematic inspection using a hand held infrared camera. The process revealed that 85 strings were ‘running hot’, i.e. had abnormally high temperatures near their live ends versus ambient.

To gather more data on the extent of the problem, ‘hot’ insulators were removed and separated into five categories, based on their elevated temperature gradient: 1-3°C; 3-5°C; 5-7°C; 7-10°C; and greater than 10°C. Each insulator was tagged with detailed information, including specific tower, phase sequence, location either to right or left side, etc. A total of 20 insulators, i.e. 4 units from each of these 5 categories were then sent for testing to study possible causes of overheating.

On the advice of experts, a maintenance program was established to monitor temperature profiles of all the line’s insulators. According to these experts as well as the Technical Guide for infrared diagnostics of live high voltage equipment (DL/T 664-2008), any string whose maximum temperature was found to exceed ambient by at least 5°C would be replaced, in most but not all cases by a double string. In the end, several dozen strings were replaced with insulators from a different supplier having the same basic shed geometry and creepage but featuring a different design of corona ring. While the original insulators were equipped with rings having a concave lower surface, rings on replacement insulators had convex upper and lower surfaces.

The post Thermal Inspections Help Prioritize Maintenance appeared first on .

This edited contribution to INMR by Jordan Edwards of CRWN.ai in Canada proposes a data-driven framework for continuous condition monitoring using IoT devices and machine learning. Case studies from controlled laboratory environments and real-world deployments have provided empirical validation of this approach.

Hitachi Energy Transformer Components and Service

Switzerland

DALEKOVOD OSO

Croatia

See more suppliers of Miscellaneous Equipment & Software

Limitations of Current Methods

Commercial inspection technologies (e.g. ultraviolet imagers, handheld ultrasonic diagnostic imagers, visual inspection) provide valuable yet incomplete insights on transmission assets. Inspections typically occur every 2 to 5 years, leaving significant gaps in asset condition awareness. Failures often manifest between inspections, discovered only after catastrophic flashovers or wildfires. Geographic remoteness further compounds these limitations. In North America, for example, there are over 566,000 kms of high voltage transmission lines, with large sections not readily accessible for unplanned repairs.

Importance of Timeliness

A failing insulator can present immediate risks. Once flashover occurs, damage is not confined to the insulator itself but may extend to structures and surrounding vegetation, increasing likelihood of pole fires. These risks necessitate expedient replacement of identified tracked insulators before they flashover. Because utilities generally do not know if their insulators are tracking, they lack actionable information to address this risk preemptively, without inspection.

While the typical mean restoration time in North America, as cited by NERC, is as little as 3 hours, average outage duration can climb dramatically when the failure is in a remote area, or if the infrastructure is damaged, such as when the failure causes scorching or a fire. Early detection creates the opportunity to plan and stage insulator replacements, preventing prolonged outages and pole replacements. Timely detection of possible degradation of insulation is therefore an essential requirement in monitoring capability.

Understanding Partial Discharge Factors

Partial discharge (PD) is a fundamental phenomenon in high voltage engineering, representing a localized electrical discharge that partially bridges the insulation between conductors. It occurs in regions where the localized electric field stress surpasses the dielectric strength of a small portion of the insulating medium, while the bulk of the material remains intact.

Occurrence of PD is both a primary symptom of existing insulation defects and a potent mechanism for progressive degradation, ultimately leading to catastrophic failure. For high voltage insulators, PD phenomena are primarily classified into three types: corona, internal (or void) discharge, and surface discharge.

Corona occurs in a gaseous dielectric, such as air, in a region of highly non-uniform electric field, typically around sharp points or edges of a high voltage conductor. Corona is fundamentally a discharge from the conductor into the surrounding gas. While it can occur near an insulator surface, particularly at the junction of metal hardware (like a cap or pin) and the insulating material, it does not require the solid dielectric surface for propagation. Its characteristics are therefore dominated by the properties of the gas (e.g. pressure and humidity) as well as geometry of the electrode.

Void discharge occurs within a gas-filled cavity or void completely enclosed by a solid or liquid dielectric material. These voids are typically manufacturing defects or the result of material ageing. The electric field inside the void is intensified relative to the surrounding dielectric due to the lower permittivity of the gas. When this intensified field exceeds breakdown strength of the gas, a discharge occurs within the void. Characteristics of internal discharges are governed by gas properties within the void and the electrical properties of the void’s surfaces.

Surface discharge occurs along the interface between a gaseous and a dielectric. Unlike corona, its path is determined by the insulator surface. The initiation and propagation of surface discharges is critically dependent on the condition of that surface. Factors such as contamination, moisture, and material degradation cause tracking that facilitates discharge, often at voltages well below the flashover voltage of a clean, dry insulator.

Tracked insulators are at significantly higher risk to catastrophically fail by way of flashover. Due to these risks, significantly tracked insulators should generally be replaced. Identifying tracked insulators is traditionally difficult due to the vast number of insulators deployed; inspection is infrequent, on the order of years. Often, tracked insulators are found once a flashover has caused a fire, either on the pole (commonly on the crossarm holding the insulator) or on nearby vegetation. Development of tracking is a process that can largely be attributed to PD activity, most notably, surface discharge.

Partial Discharge & Insulator Ageing

Environmental exposure contributes to material breakdown of polymeric insulators over time. Aged polymer insulators show rougher surfaces and higher porosity. As a result, they become more likely to emit corona because of E-field stress from rougher surfaces. Moreover, loss of surface hydrophobicity causes reduced corona inception voltages. Additionally, such insulators are more likely to express surface discharge as they age due to contaminants and moisture, which accelerates formation of carbonized tracks.

PD activity itself accelerates ageing of these types of insulators. Each PD event dissipates energy in a localized area, generating intense, but highly localized heat. Over time, the cumulative effect of repeated discharges leads to thermal decomposition of insulation materials. This process can result in formation of carbonized paths within the insulation, which are significantly more conductive than the parent material. Carbonized tracks reduce a polymeric insulator’s dielectric strength and create pathways for more severe discharge activity.

During PD events, high-energy electrons and ions collide with ambient gas molecules (i.e. oxygen, nitrogen and water vapor), generating highly reactive species such as ozone (O3) and various nitrous oxides (NOX). These species can directly attack the molecular chains of polymeric insulators. In the presence of moisture, a common condition for outdoor insulators, nitrous oxides can react to form nitric acid (HNO3). This acid is highly corrosive to organic materials, leading to chemical etching of the insulator surface. This degradation is particularly damaging for modern polymer insulators, since it can destroy their crucial property of hydrophobicity and make them more susceptible to formation of conductive water films, persistent surface contamination and subsequent flashover.

Partial Discharge & Environmental Conditions

PD is dynamically influenced by the surrounding environment. Among the various environmental factors, temperature and humidity are the most significant, exhibiting complex, often non-linear, and sometimes contradictory effects on PD activity. Understanding these influences is paramount for accurate insulation condition assessment, particularly for interpreting data from online monitoring systems where environmental conditions can fluctuate continuously.

Role of Temperature

A substantial body of research indicates that an increase in ambient temperature generally promotes PD activity. The most commonly-observed trend is a decrease in the partial discharge inception voltage (PDIV) – the voltage at which PD begins, and there is an increase in number and repetition rate of PD pulses at a given operating voltage.

Physically, this relates to the properties of the gas in which the discharge occurs. Reduction in gas density lowers the dielectric strength of the gas, making electrical breakdown and PD more probable at a lower electric field stress. Increased thermal energy enhances the mobility of charge carriers within and on the surface of dielectrics, facilitating processes such as charge injection from electrodes and charge extraction from trapping sites within the material. Depending on insulator geometry and material composition, this redistribution can lead to a higher concentration of electrical stress in certain regions such as within a gas-filled void, thereby increasing likelihood of PD.

Role of Humidity

In many experiments, an increase in humidity leads to a decrease in PD activity and a corresponding increase in the PDIV. This suppressive effect is rooted in the molecular properties of water. Water vapor is an electronegative gas, meaning its molecules have a high affinity for attaching to free electrons. When water molecules are present, they readily capture electrons, forming heavy, slow-moving negative ions. This process of electron attachment effectively removes the primary charge carriers from the system, arresting the avalanche mechanism and making it more difficult to initiate and sustain a discharge. This effect is most pronounced for corona, which is highly dependent on properties of the surrounding gas. But it also applies to surface discharges on clean, hydrophobic surfaces where a continuous water film does not form.

On contaminated or aged hydrophilic surfaces, moisture does not remain as discrete droplets but instead combines with surface pollutants (such as salts, dust, and industrial chemicals) to form a continuous, conductive electrolytic layer. This conductive film provides a path for leakage current to flow across the insulator surface. As this current flows it generates heat, causing some parts of the wet layer to evaporate faster than others. This non-uniform drying leads to formation of small, dry, and highly resistive gaps known as dry bands.

The entire voltage drop across a large section of the insulator becomes concentrated across these narrow dry bands, resulting in extremely high localized electric fields that cause intense arcing. Dry-band arcing is a form of surface discharge that is far more energetic and damaging than corona, and it can elongate and eventually bridge the entire insulator, causing complete flashover. In this regime, PD is dominated by surface conductivity and resistive heating, not electronegativity of the surrounding gas.

Role of Barometric Pressure

Numerous studies have confirmed that PDIV decreases substantially with decreasing air pressure. This is critical for design and operation of electrical equipment in high altitude regions, where the rated PDIV of an insulation system can be lower compared to the appropriate value at sea-level. Lower pressure not only reduces the PDIV but also alters the characteristics of the discharge itself.

As pressure decreases, number and magnitude of PD pulses tend to increase for a given voltage above inception. Moreover, the physical area affected by the discharge tends to expand, as the lower gas density allows streamers to propagate more easily. It is important to note that this trend does not continue indefinitely. At extremely low pressures approaching vacuum, dielectric strength starts to increase because there are too few gas molecules available to sustain an ionization avalanche.

Limitations of Partial Discharge Understating

A significant limitation of many foundational studies is their reliance on controlled laboratory experiments. These investigations often utilize new, clean material samples and employ standardized electrode configurations to create artificial PD sources. Most laboratory tests are designed to investigate the effect of one or two environmental stressors in isolation.

While this approach is invaluable for isolating and understanding specific physical mechanisms, direct extrapolation to the behavior of complex, geometrically intricate and field-aged insulators must be done with caution. In-service insulators are subjected continuously to multiple stressors, including electrical stress, thermal cycling, mechanical loads, UV radiation, as well as fluctuating pollution and precipitation.

The relationship between temperature and PD is not universally simple, and some studies reveal more complex or even contradictory behaviors. For example, research on silicone rubber nanocomposites has shown that while higher temperatures increase space charge dynamics (promoting PD), they can also induce further vulcanization or cross-linking within the polymer matrix. This structural change can create more trapping sites for charge carriers, which can immobilize them and impede growth of electrical trees.

Influence of humidity on PD is one of the most complex and debated topics in the field, with literature reporting seemingly contradictory results. Some studies conclude that humidity suppresses PD, while others find that it dramatically enhances it. This apparent contradiction is not a failure of the research but rather a reflection of the fact that humidity can trigger two distinct and competing physical mechanisms. In many controlled experiments, particularly those focused on clean, new insulator surfaces, increase in humidity leads to a decrease in PD activity. In stark contrast, for insulators in service, which are invariably exposed to contamination and ageing, increased humidity is a primary driver of severe surface discharge activity and a major factor in flashover events.

The individual effects of temperature and humidity are complex, but their combined influence is even more so, often resulting in highly non-linear interactions that cannot be predicted by simple superposition. Research conducted in climate-controlled chambers, where both variables can be precisely controlled, has been instrumental in revealing the nature of these synergistic effects.

Findings consistently show that the magnitude of effect on PD of one variable is strongly dependent on the other variable. At an ambient temperature of 30°C, increasing relative humidity was found to lower the PDIV. However, at the high temperature of 90°C, a surprising result was observed: at 75% RH, a high-conductivity area formed on the surface, leading to complex PD behavior that was difficult to explain with simple models. Other studies on solid insulation have also noted this strong non-linear relationship, concluding that there is no general rule to explain the combined effect of humidity and temperature on the PD phenomenon.

These strong, non-linear interactions have profound implications for the practical task of insulation condition monitoring. Seasonal variations in ambient temperature and humidity can cause large fluctuations in measured PD that are purely environmental in origin and not indicative of any change in the insulation’s physical condition. For example, one case study reported that the PD readings on a machine in winter (typically cold and dry) were over 100 times higher than the readings on the same machine in summer (hot and humid).

Accretion of ice and snow on insulators introduces a set of conditions under which corona and surface discharge behaviors diverge significantly. This scenario transforms the insulator’s dielectric properties, creating a new and complex environment for discharge activity. Initially, ice and snow can physically bridge the insulating gaps between the sheds of an insulator, effectively shortening the creepage distance that is designed to prevent flashover. This accretion is rarely uniform, especially when it occurs on an energized insulator.

Electrical forces and localized heating from minor discharge activity lead to formation of icicles, air gaps within the ice structure, and voids between the ice and the insulator surface. The electric field, which would normally be distributed along the full length of the insulator, becomes highly concentrated across these newly formed air gaps. Since the dielectric strength of air is much lower than that of ice or the insulator material, these gaps become the weakest points in the insulation system. As a result, intense PD activity initiates within these air gaps. This activity is a form of void or surface discharge, physically distinct from the corona that might occur at insulator end-fittings under clean conditions.

When ambient temperature rises to between -2°C and 0°C, a thin film of highly conductive water forms on the surface of the ice and icicles. The conductivity of this film is often elevated because impurities from pre-existing surface contamination or atmospheric pollutants captured by the precipitation that is expelled from the ice lattice as it melts become concentrated in the liquid water layer. This contaminated and conductive water film facilitates current flow across iced sections, causing full system voltage to be dropped across the remaining air gaps. This leads to initiation of powerful local arcs within these gaps. These arcs can then propagate along the conductive, melting ice surface, potentially elongating to bridge the entire insulator and cause complete flashover, often at a fraction of an insulator’s nominal withstand voltage.

These findings have significant practical implications for the power industry. The strong, non-linear dependence of PD activity on environmental conditions means that data from online monitoring systems cannot be used reliably for asset management purposes without first being rigorously correlated with comprehensive, time-synchronized weather data, including temperature and, critically, absolute humidity.

A novel remote sensing device and sophisticated AI technology has been developed which can accurately model the complexity of PD activity. The case study below describes this technology and explains the influence of these complex environmental factors on surface discharge activity.

Machine Learning Framework

Machine learning is a way for computers to “learn” patterns from data rather than being programmed with fixed rules. Instead of telling a computer exactly how to solve a problem, it is provided with many examples among which it discovers patterns and relationships.

The process starts with a collected dataset of inputs (e.g. images, numbers, text, or audio files) paired with category labels. A model structure is selected based on what type of input is being used, and how complex the patterns are likely to be. During training, the model tries to make predictions, compares them to the correct labels, and then adjusts its internal parameters to reduce the error percentage.

After many iterations through the data, the model improves its ability to recognize hidden patterns. After a suitable threshold of accuracy has been reached, the model can take in new, unseen data and make predictions or decisions, such as distinguishing between different types of PD phenomena. The accuracy of the model is evaluated not only in how often it guesses correctly overall but also where and when it guesses incorrectly. Data scientists determine what is acceptable for accuracy, precision, and confusion. Depending on the application, for example, there may be different tolerance for false positives versus false negatives.

In short, machine learning intends to teach computers to generalize from experience. It is widely used today in recommendation systems, speech recognition, medical diagnoses, and countless other applications where rules are too complex to hand-code. It is therefore also the perfect tool to use for detecting and classifying complex signals such as electrical partial discharge.

Data Science Lifecycle

Machine learning algorithms have been developed that are trained on labeled acoustic and radio frequency recording datasets of PD activity, learning statistical patterns that differentiate between healthy and degraded insulators. Once deployed, these models infer insulator condition from real-time sensor data. To create high-performing models that give reliable predictions, this process is used:

1. Controlled Experimentation

Experimentation is conducted in high voltage laboratories, such as Powertech Labs, in Vancouver. Different types and ranges of partial discharge are created, across the full spectrum of patent, aged, and degraded insulators. Equipment used includes:
1. Flat frequency response, high fidelity ultrasonic microphones;
2. High fidelity acoustic recording devices;
3. Radio Spectrum Analyzer;
4. Wide band radio antenna;
5. Solar blind ultraviolet corona camera;
6. Omicron MPD device to capture Phase Resolved Partial Discharge plots (PRPDs).

With this equipment, the following data can be detected and captured:
1. Acoustic data extending into ultrasonic frequencies;
2. Broadband radio frequency (RF);
3. Temperature;
4. Humidity;
5. Barometric Pressure;
6. Phase-Resolved Partial Discharge (PRPD) plots;
7. Ultraviolet-C lumen counts and overlaid images showing location of PD.

2. Data Cleaning & Exploratory Data Analysis

Data cleaning and exploratory data analysis (EDA) are essential steps before building a machine learning model because they ensure data quality and guide model design. Raw data often contains errors, duplicates or missing values that can mislead algorithms and reduce accuracy. Data cleaning fixes these issues, making the dataset more reliable.

EDA helps uncover patterns, distributions, correlations, and anomalies, providing insights into feature importance and potential transformations. It also identifies biases, outliers and data imbalances that could affect predictions. Without these steps, models risk being inaccurate, overfitted, or misinterpreted, leading to poor performance and unreliable real-world results.

3. Field Data Collection

To create a model that will accurately represent the real world, data collection must focus on recordings from real world conditions that are representative of existing transmission infrastructure. A data collection crew surveys each tower along target transmission lines, recording data from a wide variety of different insulator types, ages, tower construction methods, and voltage ranges. As above, recordings are taken in ultrasound, radio and ultraviolet. Measurements are taken for temperature, altitude, humidity, atmospheric pressure.

Data labelling is performed by correlating data samples to the patterns of known defects determined in the lab. These labels are compared to the readings taken with industry standard diagnostic devices (US diagnostic imager, UV Solar blind corona camera) to provide another layer of validation.

4. Model Training

CRWN.ai’s machine learning models, for example, typically use Convolutional Neural Networks (CNNs). In the training and validation of a CNN for this purpose, the audio is first preprocessed by converting WAV files into numerical representations such as spectrograms or Mel-frequency cepstral coefficients (MFCCs). The dataset is split into training, test, and validation sets. The CNN architecture is created with convolutional, pooling, and fully connected layers suited for 2D input (spectrogram images). Next, the model is optimized by iteratively minimizing an appropriate loss function through backpropagation.

The model is then trained by feeding it pre-processed audio data. Performance is monitored on the validation set, and hyperparameters adjusted. Finally, accuracy is measured against the test set.

5. Deployment

Once the models have been sufficiently tested and validated, they are uploaded to the CRWN Inferencing Engine in the cloud. Transmissions received from the devices deployed in the field will be run through the models and predictions delivered to the end-user.

6. Ground Truth Validation

Validation must be done by field inspection. To ensure that models continue to be accurate, and guard against model drift, regular ground truth validations must be performed. Model drift in machine learning refers to the decline in a model’s performance because the data it sees when deployed over time differs from the data on which it was trained. This happens when relationships between input data and dependent variables change due to evolving real-world conditions. In the case of transmission lines, this could mean that construction methods or materials change, or variables show up that were not discovered in the original training process.

Using industry standard inspection tools, measurements are taken and compared to the predictions made by the installed device in real time.

Technical Application & System Design

IoT Device (Cricket)

The Cricket device integrates ultrasonic, RF and environmental sensors into a self-contained solar powered unit. Safe and durable lithium iron phosphate battery packs provide a steady power supply for periods of darkness or low light levels. These devices are weatherproof, robust and designed for long deployments (~5 years) with the ability to survive harsh environmental conditions (-30°C to +60°C, high wind, dust).

The Cricket also includes a dedicated always-on sensor for detecting sudden flashover events. Should a sudden burst of energy centered around 500 kHz occur, the device will report immediately to the cloud and a message will be forwarded to the utility. A calculation of the distance from the discharge event is made to differentiate flashover events (located close to the device) from lightning or electromagnetic pulses caused by other sources.

The Cricket device is installed directly onto the tower by means of a pair of bolts and a ground clamp directly below the insulators. Devices are mounted on transmission structures without requiring the line to be de-energized or the use of heavy equipment.

Collection & Feature Processing

On an hourly recurring cycle, the device takes recordings of ultrasound and radio emissions in targeted frequency bands. These recordings are then processed through a feature extraction algorithm. Feature extraction reduces the recordings to only 2% of the size of the original sample. These features (a list of integers) are all that is needed by the machine learning model to detect and classify partial discharge.

Transmission

The acoustic, radio sample features and a set of environmental readings are packaged and sent from the Cricket devices over long range, low power radio to gateways (LoRa), or over LTE where available to a cloud system (e.g. CRWN Cortex). On the cloud servers, the features are run through a series of models to produce predictions and report these readings back to the utility.

Generating Inferences

By using a patent (pending) ultrasound and radio frequency sensor fusion system, it becomes possible to filter for only the electromagnetic events corresponding to partial discharge. If either ultrasound or radio frequency were used on their own, signals that falsely resemble partial discharge could then be confused for partial discharge. True partial discharge creates corresponding, but non-linearly related signals in radio and ultrasound.

A two-channel Convolutional Neural Network resolves the mapping between these two signals that indicates classifications and magnitudes of partial discharge. Further data analysis and rich trend modelling take place on the combined dataset. Patterns in temporal or environmental conditions can be modelled and actionable intelligence created. This arrangement allows offloading the highest resource-demanding process (machine learning model processing) to the high power computing resources available in the cloud.

Case Study: Alberta Field Deployment

Between Dec. 2024 and Aug. 2025, 100 structures were monitored at locations across Alberta. Over 52,000 readings were collected under diverse environmental conditions. Recordings of temperature, humidity, and barometric pressure were taken in conjunction with the ultrasound and radio samples.

Key findings included:

● Surface discharge peaked near freezing temperatures, consistent with condensation effects;
● Corona was more prevalent at high temperatures and low humidity;
● Predictive models achieved high calibration accuracy, validating physics-informed hypotheses;
● Individual variability between towers was significant, underscoring the necessity of tower-specific monitoring.

Table 1 provides summary statistics of continuous variables.

At each reading, proprietary ultrasonic and RF sensors were used to predict if the transmission structure was producing corona, surface discharge, or neither.

ABB and PEHLA Laboratories

Germany

Laboratory Center of Koncar Electrical Engineering Institute

Croatia

See more Laboratories

Conclusions & Future Work

Evidence demonstrates that environmental and temporal variability render periodic inspections inadequate. Continuous monitoring enables utilities to proactively identify and replace degraded insulators before catastrophic failure. Moreover, network-level analytics provide system-wide insights, allowing utilities to manage risk across entire corridors.

The work highlights the efficacy of combining IoT devices with machine learning to monitor the condition of electrical insulation. Compared to periodic inspections, continuous monitoring yields orders of magnitude more data, enabling predictive insights unattainable by conventional methods. Future work includes refining multi-factor ageing models, integrating weather data more deeply into predictive frameworks and expanding deployments to larger networks.

References
[1] F. Hassan, M. Lehtonen, M. Niemel¨a, and F. Aday, “Effects of environmental factors on partial discharge activity and estimation of insulation lifetime in electrical machines,” IEEE Access, vol. 8, pp. 108 491–108 502, 2020.
[2] Y. Ji, P. Giangrande, and W. Zhao, “Effect of environmental and operating conditions on partial discharge activity in electrical machine insulation: A comprehensive review,” Energies, vol. 17, no. 16, p. 3980, Aug. 2024.
[3] Z. Li et al., “Effect of temperature on partial discharges activity and electrical trees prop agation in xlpe,” IET Science, Measurement & Technology, vol. 18, no. 1, p. 300–309, Sep. 2024.
[4] A. Rumi, A. Cavallini, and L. Lusuardi, “Combined effects of temperature and humidity on the pdiv of twisted pairs,” in 2020 IEEE 3rd International Conference on Dielectrics (ICD), 2020, pp. 906–909.
[5] High-voltage test techniques – Part 1: General definitions and test requirements, Inter national Electrotechnical Commission Std. 60 060-1, 2010.
[6] High-voltage test techniques – Partial discharge measurements, International Elec trotechnical Commission Std. 60 270, 2000.
[7] D. Mello and S. Souza, “A comparison between the atmospheric correction factors used in IEC and ANSI applied to tests performed on post insulators,” in [1991] Proceedings of the 3rd International Conference on Properties and Applications of Dielectric Materials. IEEE, 1991, pp. 359–362.
[8] M. R. Rahimi, R. Javadinezhad, and M. Vakilian, “Dc partial discharge characteristics for corona, surface and void discharges,” in 2015 IEEE 11th International Conference on the Properties and Applications of Dielectric Materials (ICPADM). IEEE, Jul. 2015, pp. 260–263.
[9] CIGRE Working Group B2.58, “Pollution test of naturally and artificially contaminated insulators,” International Council on Large Electric Systems, Tech. Rep. 691, 2016. [Online]. Available: https://cigre.cz/dokumenty komise/d1/D1 44 TB Natural Pollution Testing CIGRE.pdf
[10] A. M. Nor and Q. Liu, “Effects of temperature on the partial discharge characteris tics of silicone oil,” in 2024 IEEE Conference on Electrical Insulation and Dielectric Phenomena (CEIDP). IEEE, Oct. 2024, pp. 1–4.
[11] INMR, “Practical guidelines for visual inspection & condi tion assessment of transmission insulation,” https://www.inmr.com/ practical-guidelines-for-condition-assessment-of-transmission-insulators/, Aug 2025, accessed: 2025-08-12. [Online]. Available: https://www.inmr.com/ practical-guidelines-for-condition-assessment-of-transmission-insulators/9
[12] Electrical Power Research Institute, “Plant support engineering: Switchyard component end-of-expected-life considerations and the need for planning,” EPRI, Tech. Rep., Dec 2008, accessed: 2025-08-12. [Online]. Available: https://restservice.epri.com/ publicdownload/000000000001016684/0/Product
[13] The Florida Public Service Commission Office of Auditing and Performance Analysis, “Review of electrical transmission structure inspection and maintenance procedures and processes,” Florida Public Service Com mission, Tech. Rep., Mar 2020, accessed: 2025-08-12. [Online]. Avail able: https://www.floridapsc.com/pscfiles/website-files/PDF/Publications/Reports/ General/Electricgas/2020 Trans Facil Maint Final Report.pdf
[14] Pacific Gas and Electric Company, Electric Transmission Preventive Maintenance Manual (05), Pacific Gas and Electric Company, Aug 2020, accessed: 2025-08- 12. [Online]. Available: https://www.pge.com/assets/pge/docs/outages-and-safety/ outage-preparedness-and-support/td-1001m-etpmm.pdf
[15] Southern California Edison Transmission and Distribution, “Transmission inspection and maintenance program (timp) 2022—second quarter issue,” Southern California Edison, Tech. Rep., Jun 2022, accessed: 2025-08-12. [Online]. Available: https://www.sce.com/sites/default/files/AEM/Supporting%20Documents/2023-2025/ Transmission%20Inspection%20and%20Maintenance%20Program%20%28TIMP%29.pdf
[16] M. Amin, M. Akbar, and M. Salman, “Composite insulators and their aging: An overview,” Science in China Series E: Technological Sciences, vol. 50, no. 6, pp. 697–713, Dec. 2007.
[17] D. Ghosh and D. Khastgir, “Degradation and stability of polymeric high-voltage insulators and prediction of their service life through environmental and accelerated aging processes,” ACS Omega, vol. 3, no. 9, pp. 11 317–11 330, Sep 2018.
[18] L. Mu, B. Wang, J. Hao, Z. Fang, and Y. Wang, “Study on material and mechanical characteristics of silicone rubber shed of field-aged 110 kV composite insulators,” Scientific Reports, vol. 13, no. 1, p. 16828, Oct. 2023.
[19] X. Li, Y. Zhang, L. Chen, X. Fu, J. Geng, Y. Liu, Y. Gong, and S. Zhang, “Study on the ageing characteristics of silicone rubber for composite insulators under multi-factor coupling effects,” Coatings, vol. 13, no. 10, p. 1668, Sep. 2023.
[20] F. L¨u, L. Niu, S. Wang, T. Jiang, W. Li, J. Guo, and H. Li, “Influencing of water droplets on corona inception characteristics of composite insulator recorded by UV imager,” IET Science, Measurement & Technology, vol. 14, no. 10, pp. 1049–1056, Dec. 2021.
[21] R. Ullah, M. Akbar, N. Ullah, S. Al Otaibi, and A. Althobaiti, “Understanding variations in the tracking and erosion performance of HTV-SR-Based composites due to AC-Stressed aging,” Polymers, vol. 13, no. 21, p. 3634, Oct. 2021.
[22] X. Qiao, Y. Ming, K. Xu, N. Yi, and R. Sundararajan, “Aging of polymeric insulators under various conditions and environments: Another look,” Energies, vol. 15, no. 23, p. 8809, Nov. 2022. [Online]. Available: https://www.mdpi.com/1996-1073/15/23/8809
[23] CIGRE Working Group D1.62, “Surface degradation of polymeric insulating materials for outdoor applications,” International Council on Large Electric Systems, Tech. Rep. 740, 2018.
[24] M. H. Ahmad, N. Bashir, Z. Buntat, Y. Z. Arief, A. A. Abd Jamil, M. A. Mohamed Piah, A. A. Suleiman, S. Dodd, and N. Chalashkanov, “Temperature effect on electrical treeing and partial discharge characteristics of silicone rubber-based nanocomposites,” Journal of Nanomaterials, vol. 2015, no. 1, p. 962767, Jan. 2015. [Online]. Available: https://onlinelibrary.wiley.com/doi/abs/10.1155/2015/962767
[25] L. Jiang et al., “Impact of air pressure variations on electrical vehicle motor insulation,” IET Generation, Transmission & Distribution, vol. 17, no. 16, pp. 3834–3844, 2023.
[26] M. Farzaneh et al., “Effects of ice and snow on the electrical performance of power network insulators,” IEEE Transactions on Dielectrics and Electrical Insulation, vol. 16, no. 4, pp. 1141–1153, Aug. 2009.
[27] D. D. Hinde, “Corona discharges on the surfaces of high voltage composite insulators,” Ph.D. dissertation, Queensland University of Technology, 2009. [Online]. Available: https://eprints.qut.edu.au/29320/
[28] V. M. Moreno and R. S. Gorur, “Impact of corona on the long-term performance of nonceramic insulators,” IEEE Transactions on Dielectrics and Electrical Insulation, vol. 8, no. 1, pp. 117–124, 2001.
[29] Z. Zhang et al., “A comparative study of abnormal heating composite insulators,” Sensors, vol. 23, no. 12, p. 5673, Jun. 2023.
[30] Y. Ji, P. Giangrande, W. Zhao, V. Madonna, X. Zhang, H. Zhang, and M. Galea, “Partial discharge investigation under humidity conditions via dissipation factor and insulation capacitance tip-up test,” IEEE Transactions on Dielectrics and Electrical Insulation, vol. 29, no. 4, pp. 1483–1490, 2022.
[31] M. Fenger and G. Stone, “Investigations into the effect of humidity on stator winding partial discharges,” IEEE Transactions on Dielectrics and Electrical Insulation, vol. 12, no. 2, pp. 341–346, Apr. 2005.
[32] B. Pinnangudi, R. S. Gorur, and A. J. Kroese, “Quantification of corona discharges on nonceramic insulators,” in 2005 Annual Report Conference on Electrical Insulation and Dielectric Phenomena. IEEE, 2005, pp. 743–746.
[33] M. Michelarakis, “Investigation of surface discharges in insulation gases through synchronisation of improved electrical and fast enhanced-imaging techniques,” Ph.D. dissertation, Cardiff University, 2022. [Online]. Available: https://orca.cardiff.ac.uk/ id/eprint/153964/1/Thesis – MichailMichelarakis – FinalSubmission.pdf
[34] M. Fenger and G. Stone, “How humidity affects partial discharge activity in stator windings,” in Proceedings: Electrical Insulation Conference and Electrical Manufacturing and Coil Winding Technology Conference (Cat. No.03CH37480), 2003, pp. 47–50.
[35] M. Levesque, E. David, C. Hudon, and M. Belec, “Contribution of humidity to the evolution of slot partial discharges,” IEEE Transactions on Dielectrics and Electrical Insulation, vol. 19, no. 1, pp. 61–75, 2012.
[36] Y. Lou, R. Caruana, J. Gehrke, and G. Hooker, “Accurate intelligible models with pairwise interactions,” in Proceedings of the 19th ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, ser. KDD ’13. New York, NY, USA: Association for Computing Machinery, Aug. 2013, p. 623–631. [Online]. Available: https://doi.org/10.1145/2487575.2487579
[37] ICF Consulting, “The Economic Cost of the Blackout: An Issue Paper on the Northeastern Blackout, August 14, 2003.”
[38] “Transforming the Grid to Revolutionize Electric Power in North America,” Bill Parks, U.S. Department of Energy, Edison Electric Institute’s Fall 2003 Transmission, Distribution and Metering Conference, October 13, 2003
[39] https://investor.pgecorp.com/news-events/press-releases/press-release-details/2019/PGE-Responds-to-Camp-Fire-Announcement-from-CAL-FIRE/default.aspx

The post Applying IoT Devices, AI & Machine Learning to Predict Transmission Line Failures appeared first on .

Insulation coordination has become a well-developed engineering practice and one where the characteristics of the system in terms of insulation and arresters cross paths. The task of coordinating insulation withstand with the desired performance levels of the system can be significantly different if arresters are applied versus not applied. This coordination task as well as the task of selecting and locating arresters can be simple while at other times complex, requiring computer simulation. However, as a good first approximation, system performance can be modeled using the formulas presented in IEC 60071-1, 60071-2 and 60071-4. These standards cover 99% of what needs to be known to perform a lightning or switching surge insulation coordination study. IEEE 1313.1 and 1313.2 are another excellent source for better understanding this engineering practice.

This edited past contribution to INMR by respected arrester expert, Jonathan Woodworth, reviewed the fundamentals to help power system engineers decide whether a comprehensive study is needed and what might be the benefits.

Guangzhou MPC Power International Co. Ltd.

China

PFISTERER

Germany

See more suppliers of Arresters

Insulation Characteristics

All insulation has limits to its withstand capability. Because it is not possible to insulate sufficiently to withstand all lightning surges, insulators are designed and tested to determine the levels at which they will flash over.

Insulation has two fundamental characteristics: lightning impulse withstand and switching impulse withstand, shown graphically in Fig. 1. The LIWV characteristics of external self-restoring insulation (see Fig. 2) are universally tested and verified under dry conditions. The actual direct length between the insulator terminals is the most significant factor in determining these fast impulse characteristics. The SIWV of external self-restoring insulation is universally tested under wet conditions because this withstand characteristic depends on an insulator’s creepage or leakage distance when wet (this being the total distance between the two terminals along the surface of all sheds).

Substation Insulation Coordination for Lightning Surges

Substations are subject to two types of overvoltages that can stress insulation. Even if a station is well shielded, lightning surges can enter indirectly. All stations connect to the rest of the system via incoming and outgoing overhead conductors. In the case of air-insulated substations, if lightning strikes these lines within the span of one or two towers from the substation, a surge is likely to enter the station along the conductors.

Even well shielded transmission lines can allow a fast rising surge to enter a nearby station if there is a backflash to the conductor during a switching or lightning surge (see Fig. 3). However, owing to the high insulation withstand on systems above 245 kV, such back-flashovers are much less probable then on systems below this voltage level. Moreover, they are rare on systems of 500 kV or higher.

Fast rising surges on an incoming conductor have a high probability of flashing over insulation in the station if there are no arresters. The amplitude of these incoming surges will be equal to the flashover level of the backflashed insulator. If the only mitigation tool is an arrester at the transformer, it will protect the transformer if properly coordinated with the transformer insulation. The arrester may even protect equipment on the surge side to some extent. In this type of coordination scenario, the probability of occurrence is quite low over the expected life of the transformer, which is probably 30-40 years. However, if a properly selected arrester is not located near the transformer, it only takes one such event to cause failure of a very expensive asset in the circuit.

Another important part of this coordination scenario is the state of the circuit breaker. If the breaker is in the open position, it will become an end point on the circuit. Because endpoints represent a significant change in impedance, voltage will be reflected and cause a doubling effect at the breaker. This voltage doubling effect (i.e. traveling wave theory) will likely cause the breaker insulator to flashover resulting in yet another path for current to flow to ground.

The voltage doubling effect can also occur if the breaker is open during operation to break the power frequency fault back at the tower. Since lightning seldom occurs as only a single stroke, another stroke along the original path can send a second fast rising surge down the same line. Due to these two potential open breaker scenarios, it is advisable to apply arresters at the line entrance of the station to eliminate the voltage doubling at the breaker and an almost certain flashover of its insulation.

Yet another variable to consider in substation coordination for lightning is the number of incoming lines to the station. Fortunately, more lines make it harder to flashover insulators at the substation but, at the same time, increase the likelihood of an incoming surge. Both factors are therefore used in the formulas used to determine proper coordination.

Separation Distance

Separation distance is another important consideration when it comes to protection of substations and their insulation coordination. Arresters will limit or clamp a fast rising surge according to their own UI characteristics immediately in their vicinity. However, as protected insulation is located further from the arrester, it is increasingly less protected from fast rising surges as described above. (Note: There are no separation distance issues for slow rising surges from switching sources.)

This reduced protection is again due to the effects of traveling waves and reflections. For this reason, the location of and distance between critical insulation points in the substation need to be known before a proper insulation coordination study can be completed. Of course, the non-self restoring insulation of the transformer is generally of highest consideration when it comes to separation distance issues. The formula for determining the farthest possible distance between an arrester and the transformer it protects is found in the above references as well as in IEC 60099-5. The higher the system voltage, the shorter the separation distance becomes because the ratio of transformer withstand voltage to system voltage is reduced.

Substation Insulation Coordination for Switching Surges

Switching surges are of concern only on systems of 245 kV and above since their magnitudes for systems below that level generally do not exceed 1.5 pu of the system phase-to-ground voltage. This is due to the fact that line capacitance, length and voltage are not high enough to result in challenging surges.

There are numerous sources of slow front switching surges at substations and circuit breakers or switching devices are involved in all forms of such surges. Fault and fault clearing overvoltages are generated in the unfaulted phase when the fault is first initiated and when the voltage is re-established.

Switching surge statistical level, known as the 2% voltage (see Fig. 5), range from 1-2 per unit of the crest phase-to-ground voltage if they are mitigated with pre-insertion resistors or arresters. However, if they are not mitigated, their levels can easily exceed 2.0 pu. When energization and re-energization surges are mitigated, load rejection surges need attention. Switching inductive and capacitive currents need particular attention when the associated breakers experience pre-strike or restrike. In this case the range of 2% voltages is 2 – 2.5 pu.

There are two methods used in the practice of insulation coordination for this scenario. The deterministic method is used exclusively when applied to non self-restoring insulation. When coordinating self-restoring insulation, statistical (also known as probabilistic) methods are almost universally used. The basic difference between these methods is that in the deterministic method absolute maximum and minimum values are coordinated. For example, the maximum residual voltage of an arrester for a slow front surge is coordinated and compared to the minimum withstand level of transformer switching impulse. When using the statistical method in determining the flashover rate of the 25 self-restoring post insulators in the substation, the probability of flashover occurrence and magnitude of the surge are used in the calculation. The results are a probability distribution representing the overall switching surge flashover rate.

Arrester Characteristics & Substation Insulation Coordination

Arresters are a fundamental part of insulation coordination in the substation. They are used universally to protect the non self-restoring insulation of power transformers. As stated above, the coordination of non-self restoring insulation is accomplished using the deterministic method. This is because there are no acceptable test methods that can determine the probability of disruptive discharge in oil/paper insulation systems. Therefore the only option is to accept a deterministic approach.

Arresters applied in substations characterized by three voltages relative to insulation coordination; the arrester operating voltage (U_c), the lightning impulse protective level (LIPL) and the switching impulse protective level (SIPL). These are shown in Fig. 6. For non self-restoring insulation, a deterministic comparison is undertaken. After the insulation and arrester characteristics are determined, they are coordinated to ensure there is ample safety margin between them. The comparative graph is shown in Fig. 7 and referred to as the Margin of Protection.

Environmental Effects on Insulation Coordination

Flashover voltages for air gaps depend on the moisture content and density of air. Insulation strength increases with absolute humidity up to the point where condensation forms on the insulator surfaces. Because insulation strength decreases with decreasing air density, longer strike distance is required to attain the same flashover voltage at 2000m elevation then at 100m above sea level. A detailed description of the effects of air density and absolute humidity are given in IEC 60-1 for different types of voltage stresses. When determining the co-ordination withstand voltage, it should be kept in mind that most adverse conditions from the strength point of view (i.e. low absolute humidity, low air pressure and high temperature) do not usually occur simultaneously. In addition, at any given site the corrections applicable for humidity and ambient temperature variations basically cancel each other. Therefore, the estimation of strength can usually be based on the average ambient conditions at the location.

When contamination from salt or industrial pollution is present, the response of external insulation to power-frequency voltages becomes important and may dictate longer creepage or leakage distances. This type of contamination does not adversely affect lightning and fast front withstand levels. Flashover of insulation generally occurs when the surface is contaminated and becomes wet due to light rain, snow, dew or fog without any significant accompanying washing effect.

Transmission Line Insulation Coordination

Transmission line insulation coordination is also separated into two categories: lightning and switching. The performance assessment methods are based on expected lightning and switching overvoltages and their corresponding insulation levels. Since line insulation is self-recovering, their performances are usually determined by the statistical method. The basic practices outlined in substation insulation coordination also apply to line coordination.

The sum of the back flashover rate (BFR) and shielding failure rate (SFR) determine the flashover rate (FOR), expressed in flashovers/100km/year. The back flashover rate is the most significant cause of outages on transmission lines. While the fast-rising surge associated with a backflash seldom makes it to the substation due to corona effects, the resulting fault current and breaker operation is felt over the entire length of the system. Oftentimes, a switching surge is experienced immediately following any lightning induced flashover.

Another significant variable usually involved in lightning flashover coordination is the system elevation. The CFO of a line insulator can be reduced by as much as 20% at higher elevations and, since transmission lines often traverse high elevations, this factor must be considered. For higher elevations, insulators might need to be lengthened or arresters applied. Both are excellent means of mitigation.

Switching impulse studies need only be considered for lines exceeding 245 kV. For lines below this level, switching surge magnitudes generally do not overstress normal insulation configurations but, for levels above 245 kV, the stresses can be significant.

The switching surge flashover rate (SSFOR) is determined by numerical integration of the Stress-Strength relationship. The stress in this case is the switching impulse voltage or switching overvoltage (SOV) quantified by a probability distribution. Strength is the switching impulse withstand voltage (CFO). IEC 60071-2 defines this process in detail. If arresters are used to mitigate the SSFOR, the evaluation method is modified to accommodate the change in the SOV since it will no longer be a normal distribution but instead a truncated distribution.

Distribution Systems Insulation Coordination

The practice of distribution system insulation coordination is limited. Still, there are a number of specific deterministic practices that are quite important. The margin of protection calculations for some system configurations can determine when and when not to use arresters. For example, on underground circuits where voltage doubling is common, a margin of protection calculation can reveal that applying an arrester only at the riser pole for systems above 25 kV can be essential. When this is the case, then open point arresters are recommended to provide a lower risk of cable failure.

On delta distribution systems, a close check of the margin of protection can often show that there is little margin compared to well-grounded systems. This is due to the fact that higher rated arresters are applied to these circuits to give them ample overvoltage withstand capability. By raising the operating voltage of the arrester, the clamping voltage is also increased and the margin between the transformer’s withstand curve and the arrester’s clamping curve is decreased.

Another factor that can have a major impact on insulation coordination on distribution systems is the lead lengths on arresters. Long leads can effectively render an arrester unable to protect non self-restoring insulation on distribution equipment.

Conclusions

While a number of variables can be involved in the engineering of insulation coordination and which can make this task quite complex, optimization of application of arresters can result in significant savings on insulation cost for all types of systems.

The post Fundamentals of Insulation Coordination appeared first on .