The first part is devoted to analyzing their role in case of indirect lightning, since this condition is more severe compared to the direct lightning on this type of network. The second part is devoted to assessment, for a realistic line configuration equipped with surge arresters, of the improvement which can be obtained by installation of shield wire.

Wish Power (Thailand) Co. Ltd

Thailand

Protektel sp. z o.o.

Poland

See more suppliers of Arresters

Lightning transients can cause significant disturbances in overhead distribution lines, leading to both temporary and permanent faults. Several mitigation strategies have been proposed to limit such effects, including: (i) increasing the critical flashover (CFO) of the line, (ii) installing surge arresters (SA), and (iii) implementing shield wires (SW).

The first part focuses specifically on the approach (iii), namely addressing the advantages that can be obtained in the use of SWs in overhead distribution lines when stressed by indirect lightning. Indeed, the role of the SWs in direct lightning events is well known and well addressed, whereas for the indirect lightning events, despite extensive investigations in the literature, the effectiveness of SWs remains controversial. Some studies report a substantial reduction in overvoltages when SW are used, while others observe negligible improvements. Moreover, even in the studies which agrees in the benefits which can be gained using SWs, conflicting conclusions often arise due to the specific parameters, such as the relative height of the SW compared to phase conductors. Recent efforts by the authors have clarified this ambiguity by identifying a key indicator which can be used to quantify the mitigation effect: the Shielding Factor (SF). The concept of SF is not new: this is defined as the ratio of the voltage induced on a phase conductor when the SW is present, to the voltage induced in the absence of SW. However, the literature has presented divergent conclusions regarding the significance of this ratio. While some studies suggest that lines equipped with SW exhibit low SF (i.e., effective mitigation), others report high SF values, indicating limited effectiveness. Additionally, no clear consensus exists on how line configuration parameters – such as SW height – affect the SF. For example, some researchers assert that variations in SW height significantly influence the mitigation effect, whereas others report minimal or no impact. In the traditional literature assessment, a further complication arises from the difficulty in assigning a precise SF value to any given line configuration: due to the stochastic nature of lightning events and the variety of influencing factors (e.g., distance between lightning channel and line, ground conductivity, grounding spacing), it was considered impossible in the past to predict the exact mitigation performance of a configuration a priori. This represented a strong limitation both in the assessment of the mitigation in existing lines and the design of new ones. For example, given a desired SF value (e.g. SF = 0.7), according to the consolidated literature there was no straightforward method to determine the required line geometry to achieve that value reliably.

In the second part, an assessment is devoted to a realistic line configuration equipped with SAs to show how the addition of a SW can significantly contribute to the mitigation of the overvoltages produced under both direct and indirect lightning events.

Role of Shield Wires in Mitigating Effects Due to Lighting

The effectiveness of SWs in mitigating both direct and indirect lightning effects has been extensively investigated by many researchers.

During direct lightning events, SWs protect the overhead power line by intercepting the lightning stroke and providing a low-impedance path to the ground. SWs are traditionally used in high-voltage transmission lines to protect them from direct lightning; while their use in distribution lines is strongly limited for this type of lightning due to the frequent inception of back flashover (BFO), caused by the high potential which usually takes place between the pole/tower structure and phase conductors. Insufficient CFO levels and poor grounding resistance can facilitate BFO. To effectively reduce the impact of direct strikes, the SW should be grounded at every pole, the line should guarantee an adequate CFO, and the grounding resistance should be kept low. This highlights the importance of proper design practices and grounding when implementing SWs in distribution networks for protection from direct lightning.

In indirect lightning events SWs play a different role: SWs can help lower the overvoltage levels caused by nearby lightning strokes. This mitigation is due to the electromagnetic coupling between the SW and phase conductors, which reduces induced voltages regardless of the SW physical position relative to the phases. Stronger coupling leads to greater voltage attenuation. Recent studies developed by the authors have proposed a revised theoretical framework for understanding the mitigation provided by SWs in overhead distribution lines. This new approach introduces two fundamental insights that aim to clarify longstanding ambiguities and inconsistencies in literature.

A. Aspect 1: Parameter Classification
Historically, the various parameters influencing the effectiveness of SWs have been grouped indiscriminately, which has contributed to conflicting conclusions in earlier assessments. The revised approach argues that it is essential to distinguish between two categories of parameters:

Internal parameters: These refer to characteristics of the line that are within the control of the line designer. They include the absolute height of the SW, its height relative to the phase conductors, grounding resistance, and grounding spacing.

External parameters: These are uncontrollable variables that are not related to the line design. They include the front time of the lightning current, the distance between the line and the lightning channel, offset of the lightning strike relative to the grounding point, and the ground conductivity.
Table I summarizes the classification of parameters as internal or external.

This clear distinction helps eliminate ambiguities and promotes a more structured understanding of the factors affecting lightning mitigation.

B. Aspect 2: Point of Mitigation Assessment
The second recent innovation concerns location along the line at which mitigation effectiveness should be evaluated. While existing approaches commonly assess the mitigation effect at the point of closest proximity between the line and the lightning channel, recent findings demonstrate that a more accurate evaluation should be conducted at the SW grounding point.

By jointly applying this revised parameter classification and refocusing the assessment point, it becomes possible to define the SF in a unique and controlled way. This represents a major shift in understanding, as it implies that the mitigation effect produced by the SWs can be precisely quantified and controlled since the parameter which quantifies it – the SF – assumes a unique and predictable value.

Methodology & Implications

The main implication of these recent insights is that the SF can now be treated as a line design specification in indirect lightning assessment. Designers can target specific SF values through the adjustment of internal parameters, which are all controllable: this forms the foundation for a more deterministic and rigorous approach to SW implementation in lightning protection strategies. The following briefly retraces the revised theoretical approach showing the essential steps needed to evaluate the role of SW in mitigating lightning-induced overvoltages in indirect lightning events. To enable a clear understanding of each influencing factor, the methodology begins with a simplified but insightful configuration: all conductors, including the SW, are assumed to be lossless and infinitely long; the ground is considered perfectly conducting, and the SW is grounded at only one point. Although such an idealized setup may not reflect real-world scenarios, it provides a valuable basis for analytical exploration. This allows development of generalized insights and facilitates the identification of key parameters affecting mitigation effectiveness. In this configuration, the mitigation effect, quantified by the SF at the grounding point, is expressed as:

where: v’_a (t) and v_b (t) are the voltages induced on the phase conductor and the SW, respectively; v_a (t) is the voltage that would be induced on the phase conductor in the absence of the SW; Z_bb is the self-surge impedance of the line; Z_ba is the mutual surge impedance between the SW and the phase conductor; R_b is the grounding resistance. The impedance terms Z_bb and Z_ba depend on the conductor heights and spacing, and are given by:

where: ζ₀ = 377 Ω is the impedance of free space; h_b and h_a are the heights above ground of the SW and phase conductor, respectively; r_b is the radius of the SW; s is their horizontal separation. As can be seen, the SF is influenced by both controllable line design parameters and external (uncontrollable) variables (see Table I). To deepen understanding, the methodology also examines the step-function case for lightning current, allowing the derivation of an analytical expression for the induced voltage v_k (t,h_k ) at height h_k on the k-th conductor:

where the sub-components are defined as:

Auxiliary definitions used include:

where: I is the step magnitude of the lightning current; β is the ratio of the stroke propagation speed to the speed of light c. This formulation provides a solid analytical framework for assessing the influence of key parameters and serves as the basis for subsequent evaluations and comparisons. It has been demonstrated that the exact solution of the induced voltage at the considered conductor location can be effectively approximated by a first-order expression: this simplified expression corresponds to the Rusck approximation, which is given by:

Leveraging this approximation, the induced voltage at the grounding point becomes directly proportional to the height of the conductor being considered: this leads to a new form of the SF, now expressed in terms of the conductor height:

Under this condition, which is completely justified both on a theoretical and a practical basis, the SF becomes dependent exclusively on internal parameters: the conductor arrangement (e.g., Z_ba, Z_bb), the respective conductor heights (h_a, h_b), and the grounding resistance R_b. As consequence, the influence of external parameters becomes negligible, and the protection effectiveness can be accurately quantified.

Evaluation of Protection Schemes With & Without Shield Wire

This section investigates the lightning performance of a distribution network equipped with different protection schemes, with particular focus on the role of a SW when added to a line equipped with surge arresters. The effectiveness of these configurations is influenced by several key parameters, the spacing between surge arresters (D_SA), the insulation strength of the line, quantified via the CFO, and, of course, the presence or absence of the SW.

The analysis is carried out on a test distribution line composed of ten towers. Each tower is 10 m tall and supports a three-phase conductor arrangement. The SW, when present, is installed 0.8 m above the topmost conductor ad with the approach presented in the previous section is assessed to be SF=0.65. The network configuration used for this assessment is illustrated in Fig. 1. The performance is evaluated under both direct and indirect lightning strokes, following the simulation procedure using the code EMTP detailed in Fig. 2. For direct strokes, a first-stroke lightning current is modeled as per the Heidler waveform, with parameters listed in Table II.

where η is a peak correction factor ensuring that the waveform peaks at the correct time and magnitude, I_peak is the peak current, τ₁ and τ₂ are time constants that shape the rise and the decay of the waveform, respectively, and n is a dimensionless shape factor that adjusts the sharpness of the waveform’s rise.

Each stroke is injected at the top of each of the ten towers, ranging from (T₁ ) to (T₁₀ ). For every simulated event, the occurrence of flashover is recorded, and the protection rate is computed accordingly: the protection rate is defined as the ratio of lightning events that do not cause flashovers to the total number of events, serving as a quantitative measure of the protection scheme’s effectiveness. The tower and pole impedances are calculated according to their height and base radius; the surge impedance (Z_t) of the tower results to be 300 Ω; the surge impedance (Z_P) of the pole results to be 210 Ω. The grounding resistance of both towers and poles is 25 Ω. In the present modeling framework, this constant Rg does not strictly represent DC grounding resistance but is rather treated as an effective equivalent resistance under transient conditions that lies within the typical range (between 18.2 Ω and 45.5 Ω, based on local soil conditions and grounding geometries provided by the industry) as reported for poles in distribution networks. A soil resistivity value of ρ= 33 Ωm was selected based on the average ground conductivity data provided for the test site region.

In the case of indirect strokes, using the same waveform for the lightning current, the lightning channel is located 50 m in front of and between each pair of towers along the span from (T₁ ) to (T₁₀ ). This setup allows for assessment of the line’s vulnerability under realistic exposure to nearby strokes. Results are presented as average protection rates for each CFO level, allowing an evaluation of how insulation strength correlates with overall lightning performance under different protective configurations.

In the simulations of Fig. 3 no SW is installed, and the influence of SA placement intervals on protection performance under direct lightning events is shown; the results indicate that decreasing the SA spacing significantly enhances the protection rate, with shorter intervals (e.g., 50 m) achieving over 90 % flashover prevention at higher CFO values. In contrast, larger intervals (e.g., 200 m) exhibit a significantly lower protection rate, demonstrating the importance of optimized arrester placement in mitigating lightning failures.

Fig. 4 illustrates the results for the same line configuration of Fig. 3, but this time under indirect lightning events. In this case, the lightning stroke is located 50 m in front and between each tower, spanning from T₁ to T₁₀ of Fig. 1. The results indicate that closer SA spacing significantly enhances protection performance, particularly at lower CFO levels. For instance, at a CFO of 100 kV, the protection rate exceeds 80% for the 50 m spacing but remains lower for wider spacing intervals. As the CFO increases beyond 150 kV, the variation in protection among different spacing configurations reduces, indicating diminishing returns from closer arrester placement at higher insulation levels.

Similar analysis is repeated this time introducing the SW with a SF=0.65: Fig. 5 shows how the combined use of SW and SAs can lead to significantly improved results.

Fig. 6 demonstrates the performance of the same configuration of Fig. 5 under indirect lightning conditions. In this case too, the SW plays a significant role in improving the protection rate, and the effectiveness of SWs is particularly pronounced at higher CFO values.

The study has demonstrated that combining SAs with SWs results in a consistent performance improvement of approximately 20–30%, as an average, in flashover resistance under both direct and indirect lightning conditions compared to using SAs alone.

This synergistic approach enhances the overall effectiveness of the lightning protection system. It is highlighted that strategic placement of SAs, when coordinated with appropriate SW coverage, offers a technically viable solution for improving lightning resilience in distribution networks.

Conclusions

The above discussion has been structured into two parts: in the first, it was explained how the role of the SW in overhead distribution lines in mitigating the overvoltages under indirect lightning events can be precisely assessed, in contrast with consolidated literature. This is a fundamental result, both on assessing the performance of existing lines, but especially in the design of new ones.

In the second part it has been shown that integrating SAs and SW offers superior performance, after exploring various protection combinations. An important aspect is strategic deployment of SAs, which involves optimizing their placement to achieve maximum protection with minimal investment, when combined with use of SW can be fruitful in improving line lightning performance.

GCC Electrical Testing Laboratory

Kingdom of Saudi Arabia

STRI

Sweden

See more Laboratories

References
[1] A. Piantini, and J. M. Janiszewski, “Lightning induced voltages on overhead lines: The effect of ground wires,” in Proc. 22nd International Conference on Lightning Protection (ICLP), Budapest, Hungary, Sep. 19-23, 1994, pp. R3b/1–R3b/5.
[2] P. Chowdhuri, and E. T. B. Gross, “Voltages induced on overhead multiconductor lines by lightning strokes,” Proc. IEE, vol. 116, no. 4, pp. 561-565, Apr. 1969.
[3] A. Andreotti, A. Pierno, V. A. Rakov, and L. Verolino, “Analytical formulations for lightning-induced voltage calculations,” IEEE Trans. Electromagnetic. Compat., vol. 55, no. 1, pp. 109-123, Feb. 2013.
[4] M. Brignone, D. Mestriner, R. Procopio, A. Piantini, and F. Rachidi, “Evaluation of the mitigation effect of the shield wires on lightning induced overvoltages in MV distribution systems using statistical analysis,” IEEE Trans. Electromagn. Compat., vol. 60, no. 5, pp. 1400-1408, Oct. 2018.
[5] A. Andreotti et al. “On the role of shield wires in mitigating lightning-induced overvoltages in overhead lines-Part I: A critical review and a new analysis,” IEEE Trans. on Power Del., vol. 38, n. 1, pp.335-344, Feb. 2023
[6] A. Andreotti et al. “On the role of shield wires in mitigating lightning-induced overvoltages in overhead lines-Part II: Simulation results for practical configurations.” IEEE Trans. on Power Del., vol. 38, n. 1, pp.345-352, Feb. 2023
[7] N. Ravichandran, A. Andreotti, A. Di Pasquale, M. Pagano, D. Proto, E. Stracqualursi, R. Araneo and L. D’Orazio, “Selection of viable distribution line surge arrester for prospective optimal protection, in 2023 AEIT International Annual Conference (AEIT), 2023, pp. 1-6
[8] N. Ravichandran, D. Proto, A. Andreotti, “Surge arrester optimal placement in distribution networks: A decision theory-based approach,” Electric Power Systems Research, vol. 234, p.110744, 2024
[9] N. Ravichandran, A. Andreotti, R. Araneo, J. Cao, J., L. D’Orazio, Y. Du, D. Proto, and E. Stracqualursi, “Improvement of the lightning performance of overhead distribution lines: Possible solutions,” in 2023 International Symposium on Lightning Protection (XVII SIPDA), 2023, pp.1-6
[10] N. Ravichandran, D. Proto, D., F. Mottola, and A. Andreotti, “Multi-Objective Optimization for Lightning Protection in Distribution Networks: A Novel Approach Based on Design of Experiments,” IEEE Access, vol. 13, pp. 45215-45226, 2025
[11] CIGRÉ, “Guidelines for representation of network elements when calculating transients,” CIGRÉ Technical Brochure 39, Tech. Rep., 1990.
[12] IEC 60071-2:2018 – Insulation Co-ordination – Part 2: Application Guidelines. Geneva, Switzerland: International Electrotechnical Commission (IEC), 2018.
[13] E-Distribuzione, “Internal report, Codice Contratto JA10112065,” E-distribuzione, Tech. Rep., 2021, internal document.
[14] O. ITU-R, “World atlas of ground conductivities,” 1999.

The post Role & Advantage of Shield Wire in Overhead Distribution appeared first on .

The hydrophobic nature of composite insulators, when new, creates difficulty in developing an appropriate artificial pollution test since any good method should deposit contaminants uniformly along the entire length of the insulator under test.

The post Preconditioning Methods & Options for Pollution Testing of Hydrophobic Insulators (Video) appeared first on .

A simple qualitative approach to evaluate condition of polymeric surge arresters analyzes the plot of instantaneous applied voltage versus total leakage current. The method focuses on identifying changes in the graphical pattern, which correlate with an increase in the resistive current component—a key indicator of degradation due to moisture ingress.

The post Qualitative Laboratory Evaluation of Polymeric Surge Arresters Exposed to Internal Humidity (Video) appeared first on .

This edited past contribution to INMR by Prof. JIA Zhidong at the Graduate School at Shenzhen, Tsinghua University in China and LI Yaozhong, Sr. Engineer with the Xinjiang Electric Power Research Institute, described findings from research on how best to deal with ageing damage to two types of silicone materials commonly used for electrical insulation: liquid silicone rubber (LSR) and RTV coatings. LSR is typically a two-part, addition-curable material with low viscosity and good fluidity and used as housings of substation apparatus such as transformer and wall bushings, surge arresters and circuit breakers. RTV coatings are usually one part condensation-curable materials that are applied to porcelain and glass insulators to improve pollution flashover performance.

Hunan Yangdong Porcelain Insulators

China

PPC Insulators

Austria

See more suppliers of Insulators for Substation Equipment

Reduction in mechanical properties is less important in the case of RTV coatings than for HTV silicone and LSR materials used in insulator housings. However critical ageing phenomena such as chalking, fading and peeling have also been observed on some RTV coated insulators after several years of service (see Fig. 2).

When such cases of ageing occur, common practice is to replace affected insulators with new ones. However, this course of action carries high cost, especially when it comes to wall bushings and other high voltage apparatus. At the same time, replacement can involve lengthy interruption times. For these reasons, repair and/or recoating are often better options.

Repairing Hollow Core Composite Insulators

Assessing Degree of Ageing

The first step when it comes to repairing a degraded hollow core composite insulator is to evaluate the extent of ageing and expected impact on performance. In this regard, in service insulators can be divided into three basic categories:

1. those not showing evidence of ageing;

2. those where ageing is deemed reparable; and

3. those where ageing has progressed to a point that they can no longer be repaired.

An optical microscope is used to study degree of ageing of a sample of the LSR housing. Fig. 3, for example, shows a typical magnified image of aged LSR sheds. A layered structure can be observed when dissecting the shed of this insulator such that the aged shed can be divided into 3 distinct layers. The surface of the shed is white and opaque and non-recoverable cracks are formed when bending the shed. This portion, with a thickness of from 0.3 to 0.5 mm, can be referred to as the ‘chalking layer’. Inside the chalking layer, the shed is relatively transparent with much lighter color than on the surface. This part is called the ‘transition layer’. Inside this layer is the core of the shed, referred to as the ‘non-chalking layer’, where all the normal properties of LSR are maintained.

The cutaway view shown in Fig. 4 demonstrates that the ageing process develops ‘outside-in’ and therefore degree of ageing can be evaluated by measuring the thickness of the chalking layer. Further study has shown that degree of ageing of LSR sheds can also be determined by the following parameters:

1. Mirror Glossiness

The 60 degree mirror glossiness of unaged LSR is above 60 but gradually decreases with ageing.

2. Shore Hardness

Shore hardness of LSR increases with ageing such that the normal value of 35 to 40 becomes 45 or higher.

3. Hydrophobicity

The hydrophobicity of normal LSR sheds is HC1 or HC2 but for aged shed can drop to HC6 or HC7. Table 1 summarizes the parameters that can help identify whether or not an aged LSR housing can be repaired or is beyond repair.

Proposed Repair Technique

The top priority when repairing a hollow core composite insulator is recovery of its initial hydrophobicity while causing no damage to those portions where there is no evidence of ageing. Fig. 5 shows the steps in this procedure.

Several years ago, successful repair of aged hollow core composite insulators was carried out at three substations in Guangdong Province using the procedure described in Fig. 5.

Repairing Degraded RTV Coated Insulators

Recoating Weather

Adhesion and hydrophobicity are the two most important criteria to evaluate the condition of RTV coatings. Hydrophobicity is mainly determined by the silicone material itself while adhesion depends largely on application conditions and techniques. For example, when applied during bad weather or where there is dust or moisture on the insulator surface, even high quality RTV coatings may suffer poor adhesion during service. Fig. 7 proposes a recoating procedure of RTV coated insulators that show obvious signs of ageing such as loss of hydrophobicity or poor adhesion.

Adhesion

Circle scratch tests were conducted as part of this research in order to evaluate the adhesion of RTV coatings. Influence of temperature, humidity, surface condition (dust or moisture) were taken into account.

Temperature

Temperature was set both at 5°C and 45°C, i.e. the lowest and highest temperatures that engineers might come across during recoating. Relative humidity (RH) was set as 60%. Circle scratch tests were conducted on samples, as shown in Fig. 8.

Test results showed that adhesion of both samples were ISO 1, deemed good for RTV coatings.

Humidity

Four tests were carried out under different humidity conditions. Temperature was set at 25°C and relative humidity was 10%, 33%, 75% and 100%. Fig. 9 shows the results.

Adhesion of samples at RHs of 33%, 75% and 100% was ISO1 whereas for the sample applied at RH of 10% did not couple, which indicates that proper minimum humidity is important during vulcanization.

Surface Condition

Surface condition of insulators has great impact on adhesion of RTV coatings. In re-coating practice, adhesion between the existing layer and the insulator surface is usually strong but it is of great importance to polish the surface from dust and moisture which, if present, will lower adhesion between the original coating and the new RTV coating (see Figs. 10 and 11).

For example, coating adhesion for both samples was lowered in the presence of dust. Testing was also conducted to assess the influence of moisture. In the four samples shown in Fig. 11, RTV coatings were sprayed and vulcanized with water droplets on the insulator surface. As with tests where there was dust on the surface, coating adhesion on all samples worsened.

Recoating Procedure

Spray Details

Spraying is recommended as the superior coating application method both in the IEEE standard and in the national standard of China. Yet details such as air pressure, distance between sprayer and surface and spraying time remain to be investigated. As part of this research, 6 different spraying conditions were simulated and spraying details and resulting appearance are shown in Table 2 and Fig. 12.

Those conditions linked to the spraying parameters for sample 4 proved most suitable to achieving optimum and even coating thickness. Successful re-coating of RTV coated insulators has already been carried out using these spraying parameters on 220 kV tension strings in Shanxi Province (see Fig. 13).

Summary & Conclusions

Ageing phenomena such as chalking and cracking of hollow core composite insulator housings made from LSR material can greatly impact outdoor performance. In the case of those housings deemed only ‘slightly aged’, repair is both time efficient and economical. Normal outdoor insulation performance can be regained by removing the chalking layer in the silicone rubber housing and spraying RTV coatings onto the aged surface. Mirror glossiness, Shore hardness and hydrophobicity are all parameters that can help evaluate extent of ageing of LSR housings and determine which insulators can be repaired and which have aged beyond repair.

To repair an aged insulator, the first step is to use a grinder to remove the chalking layer and then clean the surface with an air spray gun. The second is to treat the surface with an RTV silicone coating to improve the degraded hydrophobicity.

As far as RTV coated insulators are concerned, removing the original coating is inefficient and hence the better choice is to recoat new RTV coatings directly over the original layer. Weather, surface condition and application methods all influence adhesion between the new and the original coating. Tests have shown that adhesion of RTV coatings is best when applied at ambient temperatures from 5°C to 45°C. High relative humidity will not affect adhesion whereas extremely low humidity can lead to coupling failure. Dust and moisture have a great negative impact on adhesion and therefore must be thoroughly eliminated. Spraying is believed to be the best method to apply RTV coatings and ideal air pressure, spraying distance and time are 0.6 MPa, 20 cm and 5 s respectively.

ABB and PEHLA Laboratories

Germany

Laboratory Center of Koncar Electrical Engineering Institute

Croatia

See more Laboratories

The post Repairing Ageing Damage on Hollow Composite Insulators & RTV Coatings appeared first on .

To perform the test, the cable system was energized with a resonant test set to 330 kV for a period of 60 minutes. Three teams of technicians simultaneously measured PD at different cable joints using high-frequency current transformers (HFCTs) and mobile PD test equipment in a process referred to as ‘joint-hopping’.

Selection of appropriate measurement frequencies and verification of sufficient sensitivity are discussed and results analyzed.

PFISTERER

Germany

Phenix Technologies (A Division of Doble)

United States

See more suppliers of Laboratory & Field Testing Equipment

High voltage (HV) cable systems are the backbone of energy networks around the globe. While their purchase is costly, the installation and required civil works can be even more expensive. Full turnkey projects for long lengths of HV or EHV cable, installation, civil works, substations and associated needed infrastructure can exceed $1 billion. Once installed and energized, HV and EHV transmission class cables are typically critical assets for the energy grid and must not fail in service. To assure proper production, installation and life cycle management of such assets, the IEC, IEEE and CIGRE have all issued multiple Guides, suggestions and standards – all requiring proper testing of HV cables.

Before HV & EHV cables and cable components leave the factory, they must undergo a series of tests to ensure that they meet contractual specifications, which are often based on the aforementioned standards and guides. Some tests are designed to check the mechanical integrity of the cable conductor, the dimensions and multiple parts of the extrusion process during manufacturing.

To test the insulation of the cable, HV testing with voltages exceeding rated operating voltage (U0) combined with Partial Discharge (PD) measurement are widely used. These tests shall assure that no life limiting production-related damage or impurities are present in the insulation of the cable system.

Successful testing in the factory requires trained personnel and adherence to cable factory testing standards. Ideally, these tests are carried out in a controlled environment such as shielded rooms to reduce external background noise. However, even the most rigorous factory testing cannot guarantee that the cables will be installed correctly or that they will not be damaged during transport or laying.

Once cables are delivered to site and installed, most future owners require commission tests before accepting ownership. Such tests are designed to detect possible damage that might have been inflicted during transport or laying. Special focus lies on the accessories, which are often installed in the field under conditions that are far from the clean, shielded room conditions of factory setups. In fact, the primary purpose of on-site testing is to detect possible issues in the cable system resulting in limiting life.

According to CIGRE studies based on a population of 26,494 km of HV and EHV cable tested with AC (20 Hz – 300 Hz), only 0.01% of the cables systems showed non-passing issues in the cable sections themselves. However, the non-pass rates for terminations and joints were 2.8% and 0.75% respectively.

For on-site testing, overvoltage tests above nominal voltage phase-to-ground (U₀) and in combination with PD testing is now state-of-the-art. IEC and CIGRE both state requirements in regard to voltage levels and setup of PD measurement equipment. These requirements are rigorous to assure proper testing of HV cable systems to avoid both false positive and false negative test outcomes. Often, the parties carrying out the on-site tests follow such requirements, particularly in regard voltage levels and test durations.

To energize HV and EHV cables, especially with elevated voltages, huge and expensive generator-transformer test sets would be needed. Their size and weight would put many challenges in the project management involved in the testing. Therefore, to reduce the size and weight of the voltage sources for such a commissioning test, mobile, modular Resonant Test Systems (RTS) are often used to energize the cables. Many HV RTS usually have a fixed inductance. Together with the capacitance of the cable system, a resonance circuit is established and by means of tuning the frequency, the resonance frequency between 20Hz and 300Hz is being established as per Clause 16.3 of IEC 60840 and IEC 62067.

The longer the cable system at a certain voltage class, the greater the capacitance and resistive losses in the cable system under test. For long cable systems, such as the 22 km cable system described here, multiple RTS need to be used in series and parallel to drive the needed current and therefore voltage for the overvoltage tests.

IEC and CIGRE state clear requirements concerning the PD measurement, required sensitivity, maximum allowable PD from a cable system and thus the maximum background noise. These requirements are justified and are important to achieve. Unfortunately, for long cable systems, these requirements are seldom met, especially the demands for background noise. Even with state-of-the-art PD test systems, obtaining a background noise of less than 5pC can be challenging to impossible, based on the surroundings of the test area. For the time being, the only way to conduct PD measurements is by trying to reduce background noise and varying the PD measurement frequencies (center frequency) by assuring proper sensitivity at the same time.

Challenges of PD Measurement for Long HV Cable Systems

The primary challenge of PD measurement for long HV cable systems is the high attenuation of the PD signals and, as well, further magnitude deterioration at accessory/cable interfaces. This attenuation is caused by the resistance, inductance and especially capacitance of the cable and increases with cable length. As a result, the PD signals that reach the PD measurement sensors are weaker than for a shorter cable. This can make it difficult to detect and measure PD signals, especially at low levels. Even low levels of PD can indicate issues within the cable insulation and can lead to severe and even catastrophic failures with danger to assets and safety.

Another challenge is the potential for interference from external sources. In general, the cable acts like an antenna for airborne noise. The longer the cable, the better the antenna function. Interference can come from a variety of sources including power lines, other electrical equipment regulated by power electronics such as rotating machines, wireless networks, and even the grounding network. It can be especially important to filter out interference when possible PD signals are weak.

Test Setup

Cable System
The cable under test is specified as a 230 kV / 400 kV cable with U₀ = 220 kV and U_max = 420 kV with a length of 22 km and a copper conductor with a minimum diameter of 2500 mm². Each phase of the 3-phase cable is sectionalized by 33 joints thus creating 34 cable sections. Open-air end terminations are installed at one end of the cable. On the other GIS end terminations are present. The end terminations at both sides are grounded directly.

What Cable End to Energize From?
CConnecting an external voltage source such as an RTS to a GIS is possible by means of special adapters that require trained personnel to be connected correctly and must also be handled with care. Moreover, they have an economic impact on testing costs. While these costs are small in relation to the total project cost of the cable installation, they can become a significant factor in relation to the cost of commissioning testing by itself.

Testing through the GIS would result in the energization of more parts of the GIS system than the cable terminations only. These additional parts often have current transformers (CTs) or voltage transformers (VTs) installed that have stricter limitations concerning possible overvoltage and testing voltage frequencies. Depending on manufacturer of the GIS and CTs and VTs, the testing voltage frequencies are usually limited to 50 Hz to 80 Hz. This is a significant difference to the aforementioned 20 Hz and 300 Hz (IEC 60840 and IEC 62067) for cable testing.

These limitations would increase number of RTS systems needed since more power would be necessary to bring the resonant frequency up to 50 Hz – 80 Hz rather than slightly above 20 Hz. This increases complexity of the HV voltage source setup and therefore testing costs. Thus, it was decided to energize the cable from the open air end termination side.

Joints
In each phase, 33 joints connect the 34 sections of the cable to each other.

The individual cable sections between the joints vary from 610 m to 700 m. At every third joint bay, the cable sheaths are grounded. For the other joint bays, cross bonding is installed with over-voltage arresters towards ground.

Measurement Setup
Long AC high-voltage (HV) cables are installed with cross-bonding joints to reduce circulating currents in the cable screens. During AC commissioning testing of each phase individually, the cross-bonding configuration of the cable screens must be inactive and a straight-through cable screen configuration must be established. These configurations are often done in link boxes, which can be accessed from above. Sometimes these link boxes are above ground, but are often located in manholes or just below the surface in joint bays, referred to as ‘coffin boxes’.

HFCT
Link boxes provide easy access to the screens of the joints and often allow for the easy installation of High Frequency Current Transformers (HFCTs). HFCTs are non-invasive sensors that can be clamped around the cable screens to detect PD signals. The main advantage of placing HFCTs in link boxes is their proximity to the joints. This allows for the detection of PD signals with high sensitivity and accuracy.

In a factory environment, where individual cable reels and accessory components are tested, coupling capacitors are typically used to measure PD signals, as stated in the IEC 60270. This is so because the test objects in a factory environment, largely, behave like lumped capacitances. Coupling capacitors are connected in parallel with the cable conductor and provide a path for the PD signals to flow to the measuring equipment. However, coupling capacitors are not feasible for long cable systems since (1) these constitute distributed impedances and (2) are subject to attenuation of high frequency signals such as PD.

Cigre B1.28 states: “For shorter lengths of cable systems, terminal PD measurements can be performed, whereas for longer lengths of jointed cable systems a distributed PD measurement should be performed.” [1] This means that for long cable systems, it is important to measure PD signals at multiple locations (joints) along the cable length.

HFCTs are often the best sensors for distributed PD measurement because they are non-invasive and can be installed at any location along the cable route without the need to include PD sensors in the joints themselves.

Center Frequencies Setup
Many PD measurement systems allow the operator to determine the center frequency for each sensor. The main goal of such setting is to ensure proper PD signal reception and the proof thereof. To reduce the background noise, changing the center frequency can be helpful too, as long as it can be assured that PD signals are not missed.

The ultimate goal is to a good signal to noise ratio (SNR), with a high decoupling of possible PD signals and low background noise. Unfortunately, the background noise can be different at every joint position.

The center frequencies for the PD measurement should be chosen with care based on the setup and used sensors. Some users have the tendency to view the manufacturers transfer function of a given sensor and assume this will be the frequency response of the total measurement system. Sometimes they even perform what many people call “calibration” before arriving on site.

Using a calibrator to inject a signal into the HFCT is a sensitivity check, but should not be mistaken with a calibration according to the IEC standards for factory measurements on cable components or short cable lengths. For this case it must be noted, that for long lengths of cables, the cable system does no longer act like a lumped capacitance and the assumptions in IEC 60270 is no longer meaningful.

A sensitivity check of the HFCT will give the scale factor that can be used to convert the output of the HFCT to the actual PD magnitude. This scale factor is described as k in the IEC 60270. This scale factor is valid for the selected testing center frequency and bandwidth. It will most likely stay within +/- 5% for another HFCT of the same model.

In Fig. 4, the blue line represents an almost ideal transfer function of an HFCT. Operating the HFCT in the frequency range where the blue line is horizontal would be optimal, as the scale factor of the HFCT would the constant – at least in theory.

With the green line representing the FFT of the noise, the difference between the blue and the green line would be the Signal-to-Noise Ratio (SNR). Clearly, the higher the frequency, the higher the SNR. This can be illustrated in the following real background noise measurement at site.

Fig. 5 shows the background noise at 2 different center frequencies using the same scale factor (around k =5). Sometimes this leads to the misconception that the center frequency should be very high to obtain a good SNR or even that the center frequency can be chosen at will as a low background noise is supposedly the most admirable goal during test. Sometimes there is the attempt to change the center frequency at every joint to obtain the lowest background noise. Such procedure neglects the fact, that the HFCTs, the cable leads from the joint to the link boxes and the cable system itself, just to name the most important ones, create a complex network of impedances.

Such complex network will result in a frequency response that does not resemble the blue line in Fig. 4. The effective frequency response can show multiple resonance frequencies that affect sensitivity of the PD measurement at different center frequencies. When viewed in the frequency spectrum, there will be frequencies with high attenuation and frequencies with high amplification.

Fig. 6 shows an exaggerated plotted example of multiple resonance frequencies. It shows multiple areas of amplification and attenuation combined with attenuation towards higher frequencies. The attenuation towards higher frequencies is symbolic of the attenuation of signals along the cable.

If the measuring center frequency is set to a position where there is a lot of attenuation/damping visible in the spectrum, the background noise may appear to be low. Inexperienced persons might conclude that this frequency is ideal for getting close to the desired background noise level of less than 5 pC, as suggested in IEC and CIGRE. This is sometimes emphasized by the fact that the scale factor k of the HFCT at this frequency is similar to other center frequencies that are +/- 1MHz. Unfortunately, at a minimum of a resonance frequency, not only is the background noise low but the ability to detect PD from the cable system can also be close to zero. Conversely, there will be center frequencies where the PD setup will show high responses to high-frequency signals. In these areas of the frequency spectrum, the PD return is high but the background noise is high as well.

Fig. 7 shows an example of two selected measurement center frequencies with their respective bandwidths in a spectrum with resonances. The red center frequency will return only a small amount of energy from the cable system. Thus, the background noise will appear to be low but possible PD signals will also be very low. It is important to understand that the possible resonances in the frequency spectrum shown in Figs. 6 and 7 are the result of the aforementioned complex network of the chosen sensor, the physical location of the sensor, and the cable system under test. Therefore, these frequencies cannot be determined before arriving on site. They can only be determined once the HFCTs are placed at the cable system in multiple joints.

Often, the available time planned for the commissioning test is limited and cable installers and the owner push for rapid completion of the test. Moreover, it is in the economic interest of many parties to finish testing quickly. This pushes testing engineers to limit the number of center frequencies used at each measurement point, such as a joint bay.

If the PD testing setup in combination with the cable under test has resonance frequencies that are powerful enough to limit the measurement of PD at certain test locations, it can be difficult for the test engineer to determine this. The best possibility is to inject powerful non-classic calibration signals at one end of the cable system, having a variety of rise-times, pulse widths and fall times, measure this calibration signal at all testing points, such as the joints.

However, with cables above a couple of kilometers in length, this will be difficult since long lengths of jointed cable circuits constitute a distributed network of impedances. Therefore, conventional PD measurement, including conventional calibration, do not apply to measurement of PD on such circuits.

High Center vs Low Center Frequencies
The ability to detect calibration signals injected at one end of the cable is limited by attenuation of the signal, presence of noise and resonance frequencies of the cable system. Lower center frequencies (e.g. much below 3 MHz) allow for longer detection distances but this also increases amount of background noise. This is because both the PD signal and the background noise are attenuated by the cable. But the PD signal is attenuated more quickly – at least it seems that way. As a result, a PD signal becomes more difficult to distinguish from the background noise at lower center frequencies.

For center frequencies above 3 MHz or 4 MHz, the calibration signal can only be detected a few joints down the cable and in some cases even less depending on the design of the cable system. As a rule of thumb, the higher the center frequencies, the more localized the measurement will be around the test position (e.g. joint), thus reducing the “ability” of the HFCT to detect signals from larger distances.

Choosing higher center frequencies can be beneficial in the presence of high noise sources. Partial discharge from the HV source setup can be the origin of noise, especially if it cannot be mitigated properly. On long cables, these high center frequencies will make it difficult to detect the calibration signal over multiple joints and to ensure that any resonance frequencies in the setup are identified.

Selecting Proper Center Frequency
One way to ensure proper center frequency selection is to test at multiple joints simultaneously. This can be done by injecting a calibration signal through the HFCT at one joint bay and verifying that the signal can be detected at neighboring joint bays. The HFCT is positioned around the sheath cable at the link box in the same way that it will be positioned to test for PD. A wire loop from the calibrator is either threaded through the HFCT, as in the sensitivity check, or the HFCT is directly connected to the calibrator using a short 50 ohm coaxial cable.

If this procedure is repeated for different joints, the attenuation of the cable under test can be recorded for all center frequencies being used. This information can then be used to select the best center frequencies for PD testing. Of course, for such procedure multiple test teams need to work in parallel on neighboring joints.

It is important to note that PD testing on HV or EHV cable systems typically involves scanning for PD over a range of frequencies. This makes is easier to distinguish possible PD from noise source and localize these PD.

Voltage Source Setup
According to CIGRE TB 841 and TB 728, the cable system shall be without PD at 1.5 * U₀ = 1.5 * 220kV = 330kV.

The used Resonance Test System (RTS) can energize a cable up to a testing voltage of 260 kV with a current of 70 A – 90 A. The longer the cable, the higher the capacitance. For longer cables, often 2 or more RTS need to be used in parallel to feed enough current to provide sufficient recharging current.

For cable tests above 260 kV, RTS need to be used in series, with the second reactor on insulators to reach voltage of around 520 kV. For this commissioning test, extra reactors are needed to elevate the voltage up to 330 kV, as required by CIGRE.

The RTS uses an adjustable frequency to tune to the resonance frequency in an LC network. The inductance (L) is part of the test system and the capacitance (C) is represented by the cable. Due to the cable’s length, the capacitance totals about 5000 nF, according to the specification sheet.

To drive enough current to energize the 22 km cable, four RTS units must be connected in parallel, with aforementioned extra reactors on insulating stands.

With 4 RTS and additional 4 reactors, the voltage source setup alone requires multiple technicians for minimum a day. The 4 RTS need to be synchronized with all voltage taps on the exciters being correctly set. This system needs to be connected to multiple diesel generators working synchronously as well to supply enough feeding power.

The HV connection from the voltage source to the test object is very challenging and takes multiple days to set up properly. A major challenge is the wind, which picks up sand and proximity to water. The combination of humidity, especially in the morning, and fine sand is highly challenging as dust settles on the conductors of the setup.

Due to the space required for the voltage systems, as well as the space needed to manoeuvre the trailers and diesel generators, in this case it was not possible to set up close enough to the cable under test. Instead, the setup was placed about 120 meters away, as shown in Fig. 10.

Under these conditions, it is economically difficult to create a setup that does not show any external discharges since this would take many days to set up and large investments in proper insulating stands. The current setup (Fig. 10) will result in PD during the measurement at the near end of the cable and thus properly interpreting the data requires expertise and skill. For the above testing project, the end client understood the issue and agreed to a commissioning test under these conditions.

Measurement Procedure

Testing a 33-joint-bay HV cable system with 2 end terminations can be a challenging task. One of the key challenges is that it is not feasible to test at all locations simultaneously. This is because fiber optic cables are not available to connect the joint bays. In this case, a ‘joint-hopping’ approach can be used.

The following procedure was agreed upon with the end client:
– Perform a 1-hour withstand test.
– If the HV AC withstand test is successful, perform “joint-hopping” with 3 teams and energize for PD tests

This will expose the cable to a longer duration of energization above U₀, but is more economical than deploying 35 PD measurement units with an operator each and testing them simultaneously.

In this joint-hopping approach, three teams are deployed to different joint bays. Each team is equipped with an HFCT, a calibrator, and a PD measurement device. The teams then connect their HFCTs in the link boxes and perform a sensitivity check at different center frequencies. After injecting calibration signals at the near end of the cable and at different joint bays, while measuring the calibrator signals at multiple joint bays simultaneously, the team decided to use 3 MHz and 8 MHz as the center frequencies, each with a bandwidth of 650 kHz.

The distance between the teams varied during testing since the cable was not along a single street. Instead, it crossed multiple streets, quarters and areas with varying levels of population and traffic. Mean distance between two joint bays was around 650 m, but the traveling distance from one joint bay to another could vary in the city from 3 minutes to over 1 hour, depending on routing and traffic.

Access to the joint bays was often next to roads and on sidewalks but sometimes on the road itself or in parking lots. Occasionally, cars were parked on top. If their owners could not be found within 15 minutes, the team would move on to the next joint bay. If at least two teams were ready to test PD at the joints, the cable was energized for about 5 minutes to 330 kV. During that time, one measurement was taken at 3 MHz and one at 8 MHz for a minimum of 1 minute each.

Measurement Results

Sensitivity Check
The sensitivity check was performed for 3 MHz and 8 MHz for all HFCTs by running a calibrator cable through the HFCT to determine the k factor. For both center frequencies and all HFCTs the k factor was around 5. This was expected, as the transfer function of the used HFCTs is almost linear in that area (see Fig. 4 as example).

Measurement of Partial Discharge (PD) Signals
The IEC 60270 standard defines how to calculate the charge value (Q) of repetitive partial discharges. The utilized measurement system shows the Q_IEC value for the last PDs in a screenshot. Usually, this value is used to describe the discharge values of PD or background noise respectively.

The sensitivity check does not constitute a calibration procedure according to IEC 60270 nor are chosen center frequencies of 3 MHz and 8 MHz within the recommended range. As such, the measured charge values will not be described as “Q_IEC” values. The measurement system calculates the charge by the same principle, but the value is called Q weighted (QWTD) instead.

Verification of Measurement Center Frequencies
One way to ensure that all possible PD in a cable system can be detected is to assure that sensors at each joint are able to detect signals from the respective adjacent joints if the signals are well above background noise. A calibrator signal with 100 pC was injected at three neighboring joints simultaneously, in this case joint bay 5, 6, and 7. The PD measurement at each joint was recorded.

Fig. 11 shows the PRPD at joint 7. At 100 pC the calibrator signals from joint 7 can be seen as 6 dots within the 20ms time frame of the PRPD. This was the signal to determine the k factor. Also the calibrator signal from joint 6, around 650 m away, could be detected at around 4 pC, and the calibrator signal from joint 5, around 1300 m away, could be detected at around 1.8 pC. This shows that with the chosen measurement center frequencies it is possible to detect PD from adjacent joints, even over relatively long distances.

Figs. 12 and 13 show seven phase resolved partial discharge diagrams (PRPDs) each, from the end termination to joint 28, with both used center frequencies of 3 MHz and 8 MHz. The PRPDs show massive external discharges with different patterns, which were created by the long and disadvantageous HV source setup (as shown in Fig. 10). The external corona discharge signals are attenuated over distance, reducing the detectable and calculated charge value from joint to joint.

When utilizing a higher center frequency, a higher part of the signals frequency spectrum is measured. Especially on long cables, higher frequencies are attenuated more strongly than lower frequencies. This results in a stronger attenuation for the measurement with 8 MHz in relation to the measurement with 3 MHz.

In Fig. 12 the corona discharges are still visible above background noise after 6 joints, around 3.9 km respectively. With 8 MHz, the corona discharges are barely visible after 4 joints, around 2.6 km.
In both cases, it is clear that the sensor at a certain joint is able to detect PD signals from adjacent joints, if the PD are well above background noise and of sufficiently low frequency content at the point of origin to propagate in a detectable manner to adjacent accessories.

In comparison with the calibrator signals in Fig. 11, one could assume that calibrator signals attenuate stronger than PD signals. The reason for this effect lies in the way, the calibrator signals were couples into the cable.

When a calibrator signal is injected into an HFCT, only part of the energy will couple into the cable screen. This already reduced energy will travel in both directions, roughly halving the energy again. This even smaller signal will be attenuated while traveling to the next joint, where it can be picked up by another HFCT.
Different signal propagation paths of PD signals within the cable system also explain the difference of the PRPD of the end termination and joint bay 33 in Figs. 11 and 12.

Part of the energy of the external corona discharge created at the HV setup will travel towards ground at the end termination. This part can be detected by the HFCT connected to the cable screen ground at the end termination. Based on the impedance of this ground, different parts of the signal’s spectrum behave differently.

The ”other” part of the corona discharge signals energy will travel “into” the cable and propagate along the cable to the adjacent joints and can be detected at the joint bays.

The different behaviour of propagation of different frequency contents gives the impression, that for 3 MHz the corona discharge is higher at Joint Bay 33 than at the end termination where the signals are created. This is not the case. The PRPD symbolize the split up of the discharge energy, with one part going directly to ground and the other part “into” the cable.

For 8 MHz, the discharge values at Joint Bay 33 show smaller values than for the end termination. Either a larger part of the relevant frequency spectrum travelled towards ground at the end termination, or the attenuation of that part of the frequency spectrum is already dominant enough to reduce the measured and calculated charge values at Joint Bay 33.

It is important to note that the energy levels associated with externally occurring corona are higher than those typically associated with internal PD occurring within an HV or EHV cable accessory. Signals created by internal PD will be less powerful and will not be visible over multiple joints and distances of over 3.9 km, at 3 MHz for this cable testing setup. Thus, it was not possible to assess, if internal PD occurred at the end termination and the first couple of joints. Testing from the other end of the cable was not feasible due to aforementioned reasons.

The future owner of the cables system was aware of the situation. Based on the fact that the cable system was energized well above U0 multiple times and was able to withstand the elevated voltage stress without failure, the cable was later tested with U₀ and permanent monitoring for multiple days before being put into service.

STRI

Sweden

GCC Electrical Testing Laboratory

Kingdom of Saudi Arabia

See more Laboratories

Summary

Long cables do not behave like a lumped capacitance. As such, the calibration procedure according to IEC 60270 is not feasible. When testing PD on joints, the transfer function of the HFCT sensor used do not represent the frequency response of the full PD measurement setup.

The full PD measurement setup, a combination of the cable under test + the sensors + leads to the sensor + location of the sensors creates a complex network of impedances. These impedances are cause for a unique signal response.

Measurement center frequencies can not be pre-selected but must be determined at site to assure in a cross-check, that PD signals, or at least calibrator signals, from adjacent joints can be detected.

Joint hopping can be a form of testing long cables. This should be executed by multiple teams at the same time to perform the cross-check.

PD signals will attenuate over distance travelled in the cable. The higher the chosen measurement center frequency, the higher the observed attenuation of signals.

References
[1] CIGRE TB 728, May 2018, B1 Technical Brochure: “On-site Partial Discharge assessment of HV and EHV cable systems “, WG B1.38
[2] CIGRE TB 841, Sept. 2021, B1 Technical Brochure: “After laying tests on AC and DC cable systems with new technologies”, WG B1.38
[3] IEC 60270, 2015, “High-voltage test techniques – Partial discharge measurements”
[4] IEC 62067, 2006, “Power Cables Above 150 kV and their Accessories for Rated Voltages Above 150 kV (Um = 170 kV) up to 500 kV (Um = 550 kV) – Test methods and requirements”
[5] M. Fenger J. Levine, “Sensitivity Assessment for HV & EHV Field Partial Discharge Measurements via Laboratory Testing”, IEEE Conference Record of the 2008 International Symposium on Electrical Insulation, June 2012 
”
[6] N. Oussalah, Y. Zebboudj & S. A Boggs, “Partial Discharge Pulse Propagation in Shielded Power Cable and Implications for Detection Sensitivity”, IEEE Electrical Magazine, Vol 23. Issue 6, pp.. 5 – 10, Nov/Dec 2007.

The post Case Study of Partial Discharge & Commissioning Testing of Long (+20 km) 400 kV XLPE Cables appeared first on .

Conductor Galloping Under Wind, Ice & Snow

Galloping is large amplitude, low frequency, wind-induced oscillation of overhead lines. In most cases, there is an ice accretion on the conductor that modifies its normal cross-sectional shape such that it becomes aerodynamically unstable. Amplitudes are mainly vertical and typically range from +0.1 to 1.0 times the sag of the span while frequencies usually range from 0.15 to 1.0 Hz. Driving wind can vary between 8 to 72 km/h (5 to 45 mph) at a 10 to 90 degree angle to the line and can be unsteady in velocity or direction.

Guangzhou MPC Power International Co. Ltd.

China

Shaanxi Taporel Electrical Insulation Technology

China

See more suppliers of FRP Rods & Tubes

Glaze ice accreted on overhead conductors is more commonly considered in analysis and testing studies of galloping than are other types of accretion. Thin layers of glaze ice on a conductor have a rounded shape (see Fig. 5).

Elevated sections of the Beauly-Denny route where galloping occurred in the past on the pre-existing 132 kV line have experienced wet snow and rime ice accretions. Rime ice does not create a conductor shape susceptible to galloping. However wind-driven wet snow packed onto the windward sides of conductors forms a hard, tenacious deposit with a sharp leading edge (see Fig. 6) that is highly unstable aerodynamically. Bundled conductors are especially susceptible in this regard since their high torsional strength prevents accreted snow weight from twisting the conductor to reduce the concentration of the accretion into the sharp leading edge. Shape of conductor ice and snow accretion thereby affects conductor galloping movement.

Predicting Movement of Conductor Galloping

Knowledge of galloping conductor movement has come mostly from field observation and utility guidelines on how to observe and record galloping events have been developed. These include reporting format as well as instructions for camera and video recording. Plotting movement recorded by multiple photos of galloping conductors with glaze ice shows mostly vertical motion (see Fig. 7). A (galloping) envelope around recorded points has an elliptical shape (see Fig. 8).

Lissajous ellipses can be calculated for conductors with glaze ice in relatively flat, open areas (see Fig. 9).

Field observation of galloping for glaze ice covered conductors were used here instead of calculations and predicted peak-to-peak galloping amplitudes of 10 m for the new line (see Fig. 10).

The galloping envelope was also changed to be circular. Wind tunnel studies on conductors with glaze ice versus wet snow accretions have shown the wet snow profile to be more aerodynamically unstable and that more wind energy transfers to the conductor. Moreover, wet snow galloping occurs over a wider range of wind conditions. For example, galloping studies on bundled conductors with wet snow in Japan showed large horizontal movement (see Fig. 11) and support using such a circular galloping envelope.

Wet snow circular galloping envelopes for the new 400 kV line would overlap if interphase spacers were not used (see Fig. 12). Also, horizontal movement with wet snow would only increase this overlap.

Required galloping envelope separation is based on voltage to flashover between phases or between phase and earth (see Fig. 13).

The envelope separation to avoid flashovers at 400 kV is interpolated to be 1.3 m phase-to-phase and 0.9 m phase-to-earth.

Potential Effects of Conductor Galloping

Movement of conductors such as in galloping can result in:

1. contact between phase conductors or between phase conductors and overhead ground wires, resulting in electrical outages and conductor burning;

2. conductor failure at support point due to the violent stress caused by galloping;

3. possible structural damage; and

4. excessive sag due to overstressing conductors.

Vertical galloping loads on suspension structures can be 2 times static vertical loads from ice-covered conductors while horizontal galloping loads on dead-end structures and conductors can be 2.8 times static ice related tension loads. Structure and conductor fatigue strengths, not static strength, therefore must be considered under galloping. Application of interphase spacers would reduce dynamic loads in proportion to the square of the amplitude of motion.

Mitigating Conductor Galloping

Use of interphase spacers was found to be the most cost effective solution to reduce galloping amplitude. For example, shortening spans would reduce amplitude but intended span length in some instances would need shortening by 50% thus requiring an additional tower. The option of increasing vertical phase spacing was reviewed as well. But studies on conductor tension versus galloping amplitude indicated that tension changes would not adequately reduce galloping amplitude in the case of this new line.

Effectiveness of application of interphase spacers has also been confirmed not only by experience with the pre-existing 132 kV Beauly-Denny transmission line but also by results from an international Interphase Spacer Survey conducted by CIGRE.

Documented field observations have also confirmed reduced galloping motion if interphase spacers are installed (see Fig. 15). This data was then used to predict a 50% decrease in galloping amplitude with application of interphase spacers on the new 400 kV line’s phase conductors. Reduced circular galloping envelopes gave the separation needed between these envelopes to prevent flashovers between phase conductors (see Fig. 16).

Span Considerations for Conductor Galloping

Interphase spacers were to be used for suspension and dead ended spans on the new line. Galloping in suspension spans can couple to adjacent spans as suspension insulators swing while dead ended spans on the new line would not couple to adjacent spans. Location of maximum galloping amplitude varies with span length for common shapes of galloping mode (see Fig. 17).

Field tests of bundled conductors in Japan showed that galloping in dead ended spans is often a two-loop mode of oscillation. Moreover, bundles have large amplitudes of galloping when torsion and vertical oscillations are in-phase.

Four interphase spacers per span (see Fig. 18) were used on the new line to maintain required conductor spacing during observed mixed mode galloping in the field, with top and bottom phases having single loop galloping and middle phase having two loop galloping (see Fig. 19).

Interphase Spacer Design

Interphase spacers for the new line were designed with hinge joints for phase conductor spacing ranging from 8.5 to 10 m. Hinged designs have been used for years for long interphase spacers to prevent excessive bending of the insulator (see Fig. 20). Interphase spacers do not need to resist the conductors moving together. Rather, they serve to reduce maximum galloping amplitude in the span by preventing the phase conductors from moving apart at locations where they are installed.

The composite insulator used for these interphase spacers (see Fig. 21) is lightweight, non-brittle and has a mechanical strength of 120 kN (SML). Moreover, its silicone rubber housing features excellent hydrophobicity, making it suitable even in high pollution environments. In addition, to ensure reliability of this important line, the galvanized forged steel end fittings are Charpy compliant for cold temperature impact loads.

Electrical characteristics of interphase spacers are based on line or phase-to-phase voltage and in this case are equipped with corona rings at both ends. Interphase spacer BIL is 1.2 times system BIL. Flexible bonding straps are used at the two-bundle spacer connections and also at hinge locations since the interphase spacer is alternately compressed and extended by conductor movement. In addition, semi-conducting rubber inserts are used in the two-bundle spacers for the twin 37.26 mm diameter ‘Araurcaria’ conductors.

Validating 400 kV Line Interphase Spacer Design

Electrical testing of these interphase spacers included dry lightning impulse (Fig. 22), wet switching impulse (Fig. 23) as well as RIV noise and corona extinction tests (Fig. 24), according to both CSA and IEC standards. Where applicable, the spacer was installed during these tests. The 2-bundle spacer was tested as well, following IEC 61854 and user specifications. Evaluation of performance also included longitudinal slip of the conductor in the clamp and indentation of outer conductor strands when the clamp was tightened. Electrical resistance of the rubber inserts used in the clamp was also measured to ensure these were semi-conducting.

Interphase Spacer Installation

Appropriate packaging was developed to support and protect the interphase spacers during transport. Cleaning, handling and packaging of composite insulators are being covered in IEEE 987, a Guide for Application of High Voltage Composite Insulators for Overhead Electric Power Lines. Interphase spacers can be installed with lines de-energized or using various live-line work methods (Fig. 25).

Other Interphase Spacer Applications

Interphase spacers are used on distribution lines as well as on single or bundled conductor transmission lines. Their role is to reduce conductor movement during wind induced vertical galloping or horizontal swinging or whenever snow drops from a lower conductor such that it rises rapidly toward the conductor above. Reduced conductor movement allows for reduced phase spacing, thereby lowering structure height as well as required right-of-way width. Interphase spacers can even be offered with turnbuckles to allow length adjustment for varied phase spacing.

TECNALIA Electrical Labs

Spain

Powertech Labs Inc.

Canada

See more Laboratories

Summary

Experience with the pre-existing 132 kV transmission line has shown that wet snow would likely also accrete on the bundled conductors of the new Beauly-Denny 400 kV transmission line. Combined with high winds, this would inevitably cause galloping. Circular galloping envelopes were used for modeling wet snow accretion. The composite insulator interphase spacers designed for this project serve to maintain necessary separation between phase conductors during galloping events.

References
[1] CIGRE (2007) Technical Brochure 322, Task Force B2.11.06: State of the Art of Conductor Galloping
[2] Edwards, A.T. & Ko, R.G. (1979): IEEE Symposium on Mechanical Oscillations of Overhead Conductors, Interphase Spacers for controlling galloping of overhead conductors.
[3] CIGRE (1995) Electra No. 162, WG 22.11: Field Observations of Overhead Line Galloping: Galloping Reporting Forms
[4] Carreira, A.J. (2008): IEEE, DEIS Feature Article, Vol. 24, No. 6, Controlling Conductor Motion with Interphase Spacers in Regions of Contamination
[5] Havard, D.G. (2013): Beauly-Denny 400 kV Overhead Line Studies Review
[6] CIGRE (1992) GIDGE Electra No. 143, WG 22.11: Results of the Questionnaire on Interphase Spacers
[7] Havard, D.G. (2003): 5th International Symposium on Cable Dynamics, Dynamic Loads on Transmission Line Structures during Galloping
[8] U.S. DEPARTMENT OF AGRICULTURE, RURAL UTILITIES SERVICE (RUS), ELECTRIC STAFF DIVISION (2015): Bulletin 1724E-200, Design Manual for High Voltage Lines
[9] EPRI (2009): EPRI Transmission Line Reference Book, Wind-Induced Conductor Motion, Second Edition
[10] Nigol, O. & Clarke, G.J. (1974): IEEE conference paper, C74 016-2, Conductor Galloping and Control Base on Torsional Mechanism.
[11] Koutselos, L.T. & Tunstall (1988): International Workshop on Atmospheric Icing of Structures, Paris, Further Studies of the Galloping Instability of Natural Ice Accretions on Overhead Conductors
[12] Edwards, A.T. & Madeyski, A. (1956): AIEE Trans. Vol 75 part 3, A Progress Report on the Investigation of Galloping Transmission Line Conductors
[13] Morishita S., Tsujimoto, K., Yasui M., Mori N., Inoue T., Shimojima K., Naito K. (1984): CIGRE conference paper 22-04, Galloping Phenomena of Large Bundle Conductors, Experimental Results of the Field Lines
[14] Havard, D.G. (1998): 8th International Workshop on Atmospheric Icing of Structures, Analysis of Galloping Conductor Field Data
[15] Pon, C.J., Havard, D.G., Edwards, A.T. (1982): Ontario Hydro Research Division Report No. 82-216-K
[16] Pon, C.J., Havard, D.G. (1994): Canadian Electrical Association, Report on R&D Project 133 T386, Field Trials of Galloping Control Devices for Bundle Conductor Lines

The post Case Study for Application of Composite Interphase Spacers appeared first on .

This edited past contribution to INMR by engineers at Manitoba Hydro in Canada presented a method to optimize structure design based on application of transmission line surge arresters.

Guangzhou MPC Power International Co. Ltd.

China

DALEKOVOD OSO

Croatia

See more suppliers of Arresters

The Manitoba-Minnesota Transmission Project (MMTP) was the province’s second 500 kV AC line connecting to the U.S. power grid. This transmission line runs 575 km and connects Dorsey Station in Canada to Iron Range Station in the United States. The objective of the project was to deliver contracted power supply to the U.S. and also to ensure reliable energy supply to Manitoban customers in the event of prolonged drought or unforeseen equipment failures.

The first 500 kV international power line in Manitoba – M602F – was designed during the 1970s. Based on design and weather information at the time, a minimal approach distance (MAD) of 10 feet was used at all locations. Two safety incidents have since been recorded during live line work on towers and live line maintenance is therefore currently not permitted at structure window locations along this line.

The MMTP has 60% series compensation to maintain a 2000 A continuous current rating and this 1440 MVar series capacitor was among the world’s largest at 500 kV. This size of series capacitor combined with long line length results in very high switching surge voltage, affecting both tower design and future live line maintenance. To address this, Manitoba Hydro conducted a detailed switching surge study to help optimize design of the line’s towers.

Switching Overvoltage Analysis

PSCAD studies confirmed that the large series capacitor rating and line length would result in high switching surge voltages and require significantly larger phase-phase and phase-ground spacing as well as wider right-of-way. This would make it nearly impossible to utilize portions of existing right-of-ways identified for this project. For example, results showed that a typical series capacitor design with a bypass breaker (i.e. the least cost solution) would result in a 3.5 p.u. switching overvoltage and violation of breaker TRV. Simulation studies also showed that a special series capacitor design with fast bypass and an external damping resistor could mitigate higher breaker TRV and maintain switching overvoltage below 3.0 p.u.

Transmission line surge arresters (TLSAs) are designed to limit voltages between phase conductors and tower structure and prevent flashover. Initially designed to mitigate outage rate due to lightning on transmission lines, TLSAs have been applied in North America for many years and shown excellent results. More recently, TLSAs have also been applied to limit high switching overvoltages. However, unlike for lightning related applications where arresters might typically be installed on consecutive structures, arresters to control switching surges are needed only at specific locations along a line and installed on all phases at these locations. Moreover, these arresters typically require one energy class lower than what is needed for arresters installed at line ends in substations.

Transient simulations confirmed that a compact transmission line tower that matches current limits of approach for live-line maintenance, namely 2.5 p.u. voltage, can be built assuming high switching surges are mitigated using TLSAs along with design enhancements identified for the series capacitor. As a result, Manitoba Hydro engineers recommended TLSAs for application on the new 500 kV line to mitigate high switching surge voltages and facilitate future live-line maintenance work.

Five locations out of the line’s total 524 towers were selected for surge arrester installation and numerous PSCAD simulations were performed to determine these locations (see Fig. 1).

Tower Top Geometry & Right-of-Way

Tower top geometry is designed to meet all required electrical clearances based on internal and external standards and determined primarily by: power frequency voltage; switching overvoltage; and lightning overvoltage. As shown in Fig. 2, if switching overvoltage is limited to about 2.0 per unit for 500 kV system voltage, clearances required for lightning and switching overvoltage are almost the same. However, most 500 kV systems have a switching surge factor larger than 2.0 p.u. and switching surge overvoltage is therefore the dominant factor in design of tower top geometry. Moreover, in the case of the MMTP, clearance under power frequency voltage was not deemed a governing factor due to the light pollution in Manitoba. Thus, tower top clearance was governed mainly by switching overvoltage.

The tangent suspension insulators commonly used on transmission lines are subject to swing due to wind. Sufficient clearances between the live end of the insulator string and any grounded component of the support structure must be maintained to provide safe and reliable operation of the line under wind. Based on past studies and review of historical meteorological data, three wind conditions were considered in design of the MMTP tower top geometry. The first was a high wind condition, expected to occur only rarely while the system is under power frequency voltage. The second was moderate wind, assumed to coincide with a switching surge overvoltage. The last was a nominal wind condition, expected to occur frequently and therefore most likely to coincide with live line work.

All wind conditions were applied only to I-string insulators at cross-arm locations since V-string insulators, within tower windows, were assumed to have limited movement during these wind conditions. CIGRE Technical Brochure 348 discusses how to ensure adequate clearance and prevent flashover between conductors and structure during still air as well as swing positions. High wind is defined as the 50-year return period design condition which will rarely occur while the system is under power frequency voltage only. The moderate wind condition is defined as wind pressure during 99% of the time (CIGRE 1%), i.e. the insulator string is exposed to this type of wind condition frequently enough to coincide with a switching surge overvoltage. The 40 km/h wind condition is defined as the maximum wind speed during which live line procedures can be safely executed by qualified personnel.

Switching surge clearance and MAD both depend on maximum switching surge factor of the system. As mentioned, there were three possible switching surge factors based on different system configurations. Table 2 summarizes the electrical clearances required for switching surge voltage and MAD.

A higher switching surge factor would require larger electrical clearance and wider tower, which was likely to be more costly. To further examine each design, three different designs of tower top geometry were created in PLS-TOWER based on different switching surge factors. The most compact tower design would be based on a 2.5 p.u. switching surge factor, which is set as the base case due to it being of lowest cost.

In addition to higher material costs, a wider tower usually requires wider right-of-way (ROW). Several factors influence ROW width for a transmission line including conductor swing out, audible noise, electromagnetic field, etc. Detailed analysis indicated minimum ROW widths of 80 m, 86 m and 92 m respectively required for 2.5, 3.0 and 3.5 p.u. switching surge factors. Significant land would be required for tower designs using a higher switching surge factor. For example, in Manitoba, a 3.0 p.u. switching surge factor would require an additional 316 acres while 631 more acres of land would be needed for a 3.5 p.u. switching surge factor.

Nearly 50% of the MMTP line traverses existing Manitoba Hydro owned 80 m wide ROW and any increase in land requirement would significantly impact overall project cost. Moreover, a wider ROW is less eco-friendly with the potential to affect built infrastructure. Reducing the ruling span from 470 m to 420 m was considered as one option to maintain the existing 80 m wide ROW for 3.0 and 3.5 p.u. switching surge factors. However, reducing ruling span would have meant adding 38 more towers to the 524 towers along MMTP.

Consideration was also given to contamination of hot sticks since this had been identified as a significant challenge for live line maintenance. Indeed, two such incidents were recorded during live line maintenance work on an existing Manitoba Hydro 500 kV line with 2.5 p.u. switching surge factor. Given that 2.5 p.u. switching surge factor has proven challenging for live-line maintenance work, larger clearance requirements and longer hot sticks would be required for 3.0 and 3.5 p.u. switching surge factors. This would make live-line maintenance work even more challenging or nearly impossible.

Applying TLSAs to reduce switching surge factors allows design of a more compact line with lower material costs, easier construction and narrower ROW. Table 3 summarizes cost comparison between three alternatives.

Design of Transmission Line Surge Arrester

The structure of an arrester is simple for most standard applications and consists of a stack of cylindrical MOVs (non-linear resistors made of ceramic material), placed inside a housing. Modern metal-oxide varistors consist of approximately 90% zinc oxide, a semi-conductor, and about 10% additives that act as doping elements. Grain size of the raw materials is only about 1 µm, necessary to achieve high homogeneity when wet-mixing ingredients and important towards quality of the end product. The second important component is the housing that provides mechanical and dielectric strength, connection to the energy system and earth and protection against environmental stresses. Over the past decade, polymer-housed surge arresters have become more popular due to their compact design, lower weight, more efficient production and better performance.

The most suitable design for line surge arrester application is the cage design, whereby the stack of MOVs is surrounded by FRP rods that form a stable cage. In this case the active part becomes part of the mechanical support system made possible by the compressive strength of the MOVs. During manufacture, a stretching force is applied to the FRP cage and the varistor stack is clamped between the metal end fittings. Resulting mechanical strength of the arrester is based on pre-stress applied during production. After the cage pre-manufacture, an HTV silicone rubber is molded directly onto the varistors, the FRP rods and the end fittings. Modern cage design arresters reach high mechanical strengths and can be used up to 550 kV (station class) and therefore for all line surge arrester applications. Table 4 lists technical ratings for the non-gapped line arrester (NGLA) selected for the MMTP.

Conclusions

Switching surge overvoltage on transmission lines can be effectively reduced utilizing TLSAs, which also help optimize tower design and right-of-way width. In the case of the 500 kV MMTP, TLSAs allowed reducing the switching surge factor from 3.5 p.u. to 2.5 p.u. when combined with other design enhancements identified for the series capacitor. A lower switching surge factor also helps transmission line designers optimize tower design to make it more compact and reduce required land. Significant cost savings are achieved in comparison with other options.

Bibliography

[1] EPRI, “Outline of Guide for Application of Transmission Line Surge Arresters—42 to 765 kV” Project ID 1012313
[2] CIGRE Technical Brochure 440, “Use of Surge Arresters for Lightning Protection of Transmission Lines” Working Group C4.301
[3] CSA standard C22.3 No. 1-15 “Overhead Lines”
[4] CIGRE Technical Brochure 348, “Tower Top Geometry and Mid Span Clearances” Working Group B2.06
[5] EPRI, “500-kV Transmission Line Design” Project ID 1022363
[6] P. Bunov; L. Klingbeil; M. Schubert; B. Gossler; D. Biswas; James Hunt; R. Thallam; A. J. F. Keri, “Transmission line arresters application for control of switching overvoltages on 500-kV transmission line” 2014 IEEE PES T&D Conference and Exposition

The post Optimizing Structures Using Transmission Line Surge Arresters appeared first on .

Most insulation on existing lines and at substations has been dimensioned according to standards based on historical experience with service parameters such as precipitation, wind, humidity, temperature, etc. These are the factors that most influence global dust patterns as well as deposition rate and type of pollution settling on insulator surfaces. Yet with projected rising sea levels, more desertification and higher wind activity, it is precisely these variables that are now changing at an unprecedented rate. There may a tendency to believe that adding creepage is all that is necessary to provide an added margin of security to insulation in the face of changing pollution conditions. But this is not always true. Shed shape and geometry affect how an insulator ‘self cleans’ while distance between sheds impacts icing behaviour as well as potential bridging during heavy rain. Moreover, at DC there is limited margin to add more creepage without affecting insulation distances.

At the same time, the relative proportions of ESDD and NSDD levels on surfaces impact how insulation performs as well as the onset of problems such as corrosion from excessive leakage current. NSDD is also important in electrical performance of insulators because it affects the nature of surface wetting. A surface with a heavy but inert dust deposit will stabilize whatever conductive pollution there is and promote repeated development of partial discharges and dry bands. The settled layer of dust can also absorb sulphur dioxide and water vapor directly from air. This adds to the risk that hygroscopic salts will scavenge enough water from humid air to ‘self-wet’, thereby forming a continuous conductive surface that reduces insulator flashover performance, even if there is no rain or fog.

This edited past article contributed to INMR by Dr. William Chisholm, looked at dust and pollution accumulation on insulators with a view to more correctly identifying actual service conditions and better specifying insulators.

Newell (A PGC Company)

United States

Guangzhou MPC Power International Co. Ltd.

China

See more suppliers of Insulators for Substation Equipment

Site Pollution Severity According to IEC 60815

The key to properly applying standards such as IEC 60815 is establishing the correct site pollution severity before insulators are specified. If site pollution is underestimated, the standard provides little value in avoiding discharge and flashover problems downstream. Fortunately, there are many guides to help power engineers assess the site pollution severity (or SPS) of any service environment. Perhaps the best such practice has involved sampling the pollution from exposed insulators as part of a systematic long-term program to develop a countrywide pollution map.

Assessing electrically conductive deposits forms a fundamental part of these studies and selection of leakage distance based on measuring equivalent salt deposit density (ESDD) is well established. However, IEC 60815 also calls for an estimate of non-soluble deposit density (NSDD) in the accumulated pollution.

Effect of NSDD on Electrical Strength of Insulators

NSDD deposits often far outweigh ESDD levels that accumulate on exposed insulator and metal surfaces (e.g. in China by some 5 to 1). When in contact with metal and combined with moisture, this non-soluble dust can lead to different problems, such as accelerating metal corrosion. For example, one way to put NSDD into proper perspective is to compare its weight (in mg/cm²) with the amount of water on a surface under conditions that lead to corrosion. At the ‘critical relative humidity’, where water is absorbed directly onto a surface containing salt pollution, the resulting water layer thickness weighs roughly 0.001 mg/cm². This increases to 0.1 mg/cm² at 100% relative humidity, then to 1 mg/cm² when covered with dew and finally to 10 mg/cm² when wetted by rain.

NSDD is also important in electrical performance of insulators because it affects the nature of surface wetting. A surface with a heavy but inert dust deposit will stabilize whatever conductive pollution there is and promote repeated development of partial discharges and dry bands. The settled layer of dust can also absorb sulphur dioxide and water vapour directly from air. This adds to the risk that hygroscopic salts (including chlorides and sulphates) will scavenge enough water from humid air to ‘self-wet’, thereby forming a continuous conductive surface that reduces insulator flashover performance, even if there is no rain or fog.

Indeed, the role of NSDD has been recognized for years in contamination testing using the clean-fog method. Test standards require that, whatever electrical conductivity is used in the pre-contamination slurry, it must always have the same 40 g/l concentration of kaolin clay. This yields a repeatable clean fog test result and, according to work in 1996 by Prof. R. Matsuoka, the resulting low NSDD level of 0.05-0.07 mg/cm² has only minimal influence on insulator selection according to IEC 60815. At higher levels (e.g. up to 1 mg/cm²), however, the effect of NSDD on electrical strength becomes progressive. For example, test results analyzed in 2009 by the author show that a 5:1 increase in ESDD (from 0.04 to 0.2 mg/cm²) causes flashover performance to drop by 30 to 40%. A 7:1 increase in NSDD (from 0.14 to 0.95 mg/cm²), by contrast, results in a reduction of performance from 20 to 25%.

The important performance advantages of silicone rubber surfaces over glass and porcelain – whether as polymeric insulators or RTV silicone coatings – relate mainly to the ability of these high surface energy materials to break up continuous water films. This ensures that there is no direct electrical path from the end fittings across the insulator surface. The according-to-some ‘unresolved’ or open question in application of silicone materials is whether this performance advantage can be maintained under all circumstances. There may be some critical loading level above which inert dust accumulation overwhelms the ability of the silicone material to produce a water-beading film of low molecular weight (LMW) entities that are responsible for hydrophobicity transfer.

Estimating Global Dust Deposit Rates

NSDD can be estimated using some generalized multiplier based on experience, such as 5 times ESDD, as suggested for China, or 10 times ESDD, as in the Middle East. However, it is usually far better to obtain independent estimates of both non-soluble and soluble pollution. In this regard, satellite observations seem to offer a novel and economical way to gather useful NSDD data covering large areas of the globe and also to study how the NSDD/ESDD ratio changes across different regions. Authors of a 2005 article in Science used three studies that matched satellite optical depth estimates for dust deposition with locally recorded concentrations of iron. They then proposed a map of average annual dust deposit density, which highlights areas with a rate of 20 g/m²/year deposit rate in north Africa, the Middle East and north-west China.

Dust Accumulation Model

Dust is removed from the atmosphere by dry or wet deposition. A simple model for accumulation of NSDD on insulator surfaces would therefore be as follows:

• The dust flux is uniform throughout the year

• Days without precipitation allow the dust to accumulate

• Days with precipitation wash all dust away

This model would clearly be better for upper surfaces than those facing the ground. The rate of increase of NSDD, with a heavy dust flux of 10 g/m²/year in Fig. 3, would be 0.0027 mg/cm² per day. To reach the level of 1 mg/cm², at which the influence of NSDD becomes very strong, would therefore require (1/0.0027) days – roughly a full year without rain. Unfortunately, there is no well-defined computer model yet to establish how air flow past insulator surfaces influences rate of increase of NSDD on bottom surfaces, or how it reaches an equilibrium after long-term service.

Dust Accumulation Observations

There are relatively few points of confirmation to guide application of dust deposit rate to estimate NSDD levels on insulators. For example, along the route of a 1000-km transmission line in Russia, NSDD levels from 40 test stations were reported to be in the range from 0.02 to 0.14 mg/cm², with a median value of 0.04 mg/cm². The global map shows annual dust deposit rates on the order of 1-2 g/m²/year, which when converted gives 0.1-0.2 mg/cm² per year. In Iran, NSDD levels on test insulators were found to be in the range 0.2 to 0.8 mg/cm2 in the Bushehr region, compared to values of 0.5 to 3 mg/cm² in the Hormozgan region. These were then plotted on the IEC 60815 classification chart for SPS, along with the measured values of ESDD. Close inspection of the Jickells map for this region revealed that annual dust deposit rates also differed (i.e. 10-20 g/m²/year for Bushehr and 20-50 g/m²/year for Hormozgan). In both cases, proximity to the sea ensured that ESDD levels exceeded the high level of 0.1 mg/cm²/year.

Use of logarithmic scales on both axes in the IEC chart is deliberate since such distributions are more helpful when selecting insulators. They faithfully capture the wide dispersion in the test results, whereas simple averages and standard deviation values for normal distributions do not. Scatter in NSDD values proved the same or less than variation in corresponding ESDD measurements. Israel Electric confirmed data from the global dust map and also suggested that an NSDD/ESDD ratio of 10:1 is indeed appropriate for the Middle East. The Hormozgan observations above support this refinement. After long-term service on a ±600 kV DC line near the Itaipu dam in Brazil, measured NSDD levels exceeded 2 mg/cm². Given local ESDD of 0.4 mg/cm², a 5:1 ratio seemed to apply here. As discussed, this ratio has also been proposed for many insulator applications in China.

Differing NSDD/ESDD ratios may also be appropriate for other regions and exposure conditions. For example, Ontario, Canada is an area with low annual levels according to the global dust map (i.e. 0.2 to 0.5 g/m²/year). In a season of winter measurements with frequent intervals of natural rain, the observed NSDD/ESDD ratio was less than 3:1. Statistically, the standard deviation of the natural logarithm of the NSDD values (sln NSDD =1.5) was three times higher than the corresponding ESDD value (sln ESDD = 0.5). This contrasts with pollution levels in Iran, where there was less variation in NSDD than in ESDD. On the other hand, local sources of non-soluble pollution deposits, such as cement plants, can lead to extreme levels of NSDD on all surfaces. For example, heavy deposits accumulated over a period of 27 years on insulators near a cement plant in Indonesia, which has about the same dust deposit density as Canada and should otherwise have low pollution levels due to frequent rain. The ESDD level of 0.7 mg/cm2 would be otherwise be considered ‘medium’ without the NSDD, but ‘very heavy’ with NSDD (being 26-29 mg/cm² higher than the maximum level shown in the IEC 60815 classification chart). In this unusual case, the overall NSDD/ESDD ratio was about 400.

ICMET Craiova – National Institute For Research, Development And Testing In Electrical Engineering

Romania

PSW – Siemens Energy Testing Laboratories Berlin

Germany

See more Laboratories

The post Climate Crisis Highlights Need to Monitor Pollution Severity & NSDD appeared first on .