Comparative study of AC and DC inclined plane tests on silicone rubber ( SiR ) insulation

Currently there are no international standards for evaluating the tracking and erosion resistance of DC polymeric insulation under contaminated conditions. Researchers often modify the existing AC inclined plane test standards such as the IEC-60587 to accommodate DC voltage conditions but this has been reported to give various inconsistences. This paper presents comprehensive experimental results on inclined plane tests of silicone rubber (SiR) insulation at 3.5 and 4.5 kV AC and positive DC using intravenous (IV) system as the pollutant supply. The leakage currents (LC) were recorded throughout the entire tests. In addition, various physiochemical tests namely, Fourier Transform Infrared analysis, thermo-gravimetric analysis, scanning electron microscopy and energy dispersive spectroscopy were performed on the aged and unaged samples. Results show that DC LC is bigger (about three times) than that under AC for the same equivalent voltages. Furthermore, DC LC variations are less random and the average magnitudes increase with duration of voltage application compared with AC. The physiochemical analyses show that 3.5 kVrms AC and 3.5 kV DC aged samples have comparable chemical characteristics albeit with electrode corrosion elements detected on the DC aged samples. Under 4.5 kV DC the degradation becomes significantly more severe and unrepeatable. It is therefore concluded that at 0.3 ml/min, pollutant flow rate, 3.5 kV positive DC and 3.5 kVrms AC are comparable as test voltages for inclined plane accelerated ageing of SiR insulation.


Introduction
Among other possible outcomes in the evolutionary development of electrical insulation technology as reviewed by Li et al. [1], silicon rubber composites (SiR) are a popular contemporary option.This insulation technology however has been in existence for at least 50 years and mainly in outdoor high-voltage applications.Despite the various advantages of SiR over the traditional ceramic and glass insulation materials, long term performance under DC voltage is still not fully understood [2].Presently the biggest challenge with respect to the performance of outdoor SiR insulation is the vulnerability to tracking and erosion degradation under contaminated conditions.For AC applications, the IEC-60587 [3] standard on inclined plane tests is a well-established technique to evaluate the tracking and erosion resistance of outdoor polymeric insulation.In China, a procedure has recently been established for DC insulation tests [4].Internationally however, no standard on DC inclined plane tests exists and yet it is necessary to have an international standard in that regard [2].
Currently, quality evaluation and ranking of insulation material for DC applications, is done using modified AC inclined plane test procedures.The modifications in the test procedures include use of the equivalent root mean square (RMS) voltage and increasing the creepage distance and/or varying the pollutant flow rates.These adaptations are done to try and account for and normalise the differences in insulation degradation dynamics under AC and DC electric stress.As expected, the adapted DC inclined plane tests have generally been found to be more aggressive.Furthermore, the correlation between the accelerated ageing and service ageing has not yet been established [2].Although over the past few years, many studies on the inclined plane test under AC and DC have been carried out, they have mostly not been adequately comprehensive.The relationship between electrical activity and chemical changes have not been adequately explored and documented.The absence of a suitable testing technique to evaluate the tracking and erosion resistance of DC polymeric insulators coupled with the increased interest in high-voltage DC power technology all over the world make the development of DC inclined plane test a priority research area.The intention of this work therefore is to experimentally compare and contrast AC and DC inclined plane tests on SiR insulation seeking to confirm and/or add more insights into this knowledge domain.

DC versus AC inclined plane tests: a literature survey
The inclined plane test setup, which uses a liquid contaminant on inclined plane test specimen, is designed to accelerate ageing by encouraging the formation of dry-bands and surface discharges in order to monitor surface erosion and tracking [3].The IEC-60587 standard on AC inclined plane tests specifies three RMS voltages and corresponding pollutant flow rates.The voltages are 2.5, 3.5 and 4.5 kV.The corresponding flow rates in millimetre per minute (ml/min) are 0.15, 0.3 and 0.6 [3].
Generally, the RMS equivalent of DC voltage is used in the adapted DC inclined plane test [5][6][7]; however, there are studies that have used different voltages as shown in Table 1.The unintended consequences of the variations are difficulties on how best to compare AC and DC polymeric insulation performances with the aid of the well-established AC inclined plane test setup.
Another often overlooked but yet critically important parameter is the stiffness of the supply voltage.A weak voltage source can significantly influence the results of the inclined plane test through a considerable voltage drop during discharges.Many articles on inclined plane tests however, do not comment on the source voltage stiffness and this could be contributory to results inconsistences.
Some of the other challenges with DC inclined plane tests are conflicting experimental results in the literature regarding the polarity effect.There is no consensus on which polarity gives more degradation.
From Table 1, it is evident that generally, leakage current (LC), mass loss and time to track or failure are the parameters commonly used by researchers to evaluate the quality of SiR insulators using the inclined plane test setup.Another observation made in the literature is that the results are presented in relative qualitative terms (either 'higher' and 'lower or better') and not as quantitative values.Despite the inconsistencies however, a common thread across various research findings is the greater severity of the ageing under DC compared with AC as manifested in LC magnitude and the subsequent material damage.What is still yet to be determined are the DC and AC voltage levels and/or corresponding pollutant flowrates that give the same degree of degradation.Consequently, different researchers have proposed different equivalent DC voltages of the corresponding RMS AC voltage.As an example, Bruce et al. [15] found that the equivalent DC inclined plane test level is 90% of the corresponding AC.On the other hand Bossi et al. [16] recommended that the DC flashover voltages of SiR insulators are between 75 and 85% of the corresponding AC voltage.
It is also notable that most of the work on the inclined plane test of SiR insulation is done using the constant tracking voltage (CTV) method instead of the initial tracking voltage (ITV).This is mostly informed by the simplicity of the CTV procedure as there is no need for continually varying the source voltage.
The effect of electrode corrosion on DC SiR degradation is another area of great concern and yet is not fully understood.Generally, it is reported that DC causes more electrode corrosion than AC [2,5] and that this may exacerbate ageing of the test specimen.For the case of positive DC, the electrolytic erosion byproducts of the anode modify the pollutant conductivity and enhance the degradation severity.This phenomenon has also been cited as a contributing factor to relatively large scatter in the repeatability of DC inclined plane test results [2].It is therefore apparent that there are no consistent answers to the questions about which parameters to use and to what extent they need to be adjusted to enable AC inclined plane test procedure to be adapted for DC voltages.
The present paper therefore, adds knowledge on the comparative characteristics of DC and AC inclined plane tests and especially regarding the voltage ranges and pollutant flow rates in which comparisons are sensible.An important objective of the present work is to investigate the threshold voltage beyond which the accelerated degradation of DC and AC inclined plane tests becomes significantly different.

Experimental work
The IEC-60587 inclined plane test was implemented to evaluate the DC and AC resistance of SiR outdoor insulation to tracking and erosion as illustrated in Fig. 1.
The system consists of a supply voltage, current limiting resistors, the pollutant drip system, test specimen and accessories for the current measurements.These accessories include HV probes, fast blow fuse for protection purposes, shunt resistance for current measurement and data logger for recording and storing the LC values.
Conventionally, peristaltic pumps are used for the pollution supply in the inclined plane test setup in order to achieve desired controllable flowrates.In the present work, however, an adjustable intravenous (IV) drip system is used as a replacement for the conventional peristaltic pumps to get the desired contaminant flow in accordance with the standard.The IV systems were calibrated by measuring the liquid flow per minute and it could be fine-tuned to an acceptable accuracy level of ±10% as stipulated in IEC60587 standard [3].The main subsystems and procedures of the inclined plane test system are described in a previous paper [17] by the same authors.
The altitude of the test location was 1740 m above sea level.The laboratory test environment has natural ventilation.Furthermore, the test setup was not in a restricted enclosure and as a result the steam build-up easily diffused away naturally and no build-up of gaseous by-products was observed during the test.
The RMS AC voltages of 3.5 and 4.5 kV and corresponding flow rates of 0.3 and 0.6 ml/min were used in the tests.The DC voltages were chosen to be equal to the RMS AC voltages with the reasoning that the electrical damage in terms of tracking and erosion is energy-dependant and is determined by the RMS values [10].For simplicity and limiting the number of variables, the CTV method was adopted for this experiment.Only positive polarity DC was considered in this work as it has generally been shown that positive polarity DC causes greater damage than negative DC of the same magnitude [5,8,10,12,14,15].The IEC-60587 standard allows testing of at least five samples either simultaneously or individually [3].In the present work however, one sample at a time  was tested as the voltage source was not strong enough to stress five samples simultaneously.
To ensure repeatability and consistency, each test was repeated three times and in each case new samples and electrodes were used.LC measurements and material analyses were used to comparatively characterise the extent of SiR ageing under DC and AC stresses as presented in the following subsections.

Results: accelerated ageing at 3.5 kV rms AC and 3.5 kV DC test voltages
The measured LC, and material characterisation results of SiR test specimen aged at 3.5 kV rms and 3.5 kV DC at 0.3 ml/min pollutant flow rate are comparatively analysed and discussed in this section.

LC analysis
The LCs for each of the samples were recorded using a data logger at a sampling rate of 5 kHz throughout the six hour test period.The 3.5 kV AC and DC LCs are presented in Fig. 2. The AC LC in Fig. 2a does not show any relatively significant increasing or decreasing trend in the six hour duration of the test.The 3.5 kV DC however, gave an average LC of about three times bigger than that of AC.Moreover, unlike consistently intermittent AC LC, the DC LC gradually increased exponentially with ageing time as shown in Fig. 2b.The continuous increase of the average LC under DC is more apparent in Fig. 3 where the hourly average LCs of the 3.5 kV rms AC and 3.5 kV DC test specimens are plotted on the same axes.In contrast the hourly AC LC remained practically constant for the entire duration of the test.
The degree of insulation surface erosion due to tracking is directly proportional to the LC magnitude [5,12] and in that regard it can be inferred that the 3.5 kV DC voltage is more deleterious than AC as also reported elsewhere in the literature [12][13][14].
A further statistical analysis of the LC was carried out.This includes a quantification of the duration when the sample is arcing (in a conductive state) and not-arcing (non-conductive state).A LC of 1 mA and above is assumed to be associated with arcing, therefore, all the currents below 1 mA are regarded as being nonconductive [14].The percentage of non-arcing (off) state was calculated to be 48% for the DC condition and 69% for the AC samples.This is in agreement with the observed physical damage where the AC samples which arced for 31% of the test duration showed less damage (as presented in more detail later) compared with the DC samples which arced for 52% of the total exposure time.The non-arcing incidences under AC are attributed to the zero crossings of the current in contrast to DC that is constant.The nonarcing currents significantly lower the hourly average leakage at AC in comparison to DC as shown in Fig. 3.
In addition to LC measurements, material analysis also reveals useful information on the physiochemical changes caused by the accelerated ageing of the insulation.The next section presents the Fourier Transform Infrared Spectroscopy (FTIR) analysis results of the 3.5 kVrms AC and 3.5 kV DC aged samples.

FTIR spectroscopy analysis of the 3.5 kV rms AC and 3.5 kV DC inclined plane accelerated aged samples
FTIR analysis is a useful technique in detecting physiochemical changes of materials by analysing infrared (IR) absorption bands of the materials before and after ageing [6].The IR spectrum is a plot of transmitted (or absorbed) frequencies against the intensity of transmission (or absorption).Certain chemical bonds absorb specific frequencies of radiated energy and by measuring the transmitted and absorption frequencies of test specimen, specific bonds can be determined.FTIR technique is most useful in providing information about the presence or absence of specific functional groups in the insulation.
The spectra for all the samples were recorded from 500-4000 cm −1 .The reference material in each case is the unaged sample.The IR spectra of new and aged samples are shown in Fig. 4. From Fig. 4, it is evident that there are distinct absorption peaks.Comparing the aged and unaged, both AC and DC samples show a decrease of the OH group from both ATH and silanol (Si-OH) in the region of 3373-3618 cm −1 .In a study by Hillborg [18], it is reported that dehydration of the ATH filler is characterised by a decrease in the absorption peaks of OH groups bonded with ATH in the interval 37,003,200 cm −1 .It is known that ATH filler is added to the base polymer of SiR in order to make it more heat resistant.However, above 200°C, ATH loses its bond water leaving behind aluminium oxide (Al 2 O 3 ) [19].It is suggested that both the accelerated ageing at 3.5 kV rms AC and 3.5 kV DC produce intensive localised heat that causes the loss of moisture in the SiR samples.
In the region of 2962.52 cm −1 , there is a decrease in the intensity of the C-H peak.The formation of tracking due to the arc of the aged samples, modify the structure of the SiR polymer by shrinking the chain of the polymer resulting in the harder spot on the surface of the SiR which causes less vibrations of the C-H bond, and hence the decrease in the C-H stretch intensity.
The region 2507.34-2009.73cm −1 , shows disappearances of three broad bands from all the aged samples due to Si-H stretch.These disappearances can be attributed to the heat and arc generated under both types of voltages; 3.5 kV rms AC and 3.5 kV DC.Since there are only few Si-H bonds in the structure of the polymer, the bond is easily broken because of the high electron density of the silicone atom which has the tendency of attracting the lone electron from the hydrogen thereby making the bond weak and susceptible to breakage.
Around 1259.45 and 786.91 cm −1 there is not much change in the peak intensity of the Si-C bond.This is due to the fact that Si-C forms the skeleton of the structure of the polymer.
At a wave number of 1008.72 cm −1 , there is a slight decrease in the peak intensity caused by the Si-O stretch vibration.This decrease is not very significant since Si-O-Si constitutes the major skeleton of the polymer.However, in the region of 729.05 cm −1 , the peak of the Si-O rocking vibration has disappeared from all samples.Since the heat and arc modify the structure of the aged SiR by making it more compact, rocking vibration becomes difficult for the Si-O bond.
It is also remarkable that there is no distinction between the 3.5 kV rms AC and 3.5 kV DC samples in terms of absorption peaks.Therefore, it can be concluded that in as much as the chemical changes are concerned, 3.5 kV rms AC and 3.5 kV DC samples at 0.3 ml/min pollutant flow rate, had the same degradation although the 3.5 kV DC samples showed relatively more tracking damage compared with the corresponding AC samples.
Another useful material analysis technique is thermogravimetric analysis (TGA) which evaluates the thermal stability of materials as explained in the next section in comparing 3.5 kV rms AC and 3.5 kV DC aged samples.

Thermal stability analyses of the 3.5 kV rms and 3.5 kV DC inclined plane accelerated aged samples
TGA is a technique in which the mass of a material is determined as a function of temperature or time and therefore evaluates the thermal stability of materials.Deterioration due to tracking and erosion of SiR insulation depends on the thermal decomposition of the material and therefore the point at which thermal decomposition takes place can be predetermined using the TGA technique.A thermally stable material is a material in which there is no observable mass change when heated [20].
The TGA procedure entails subjecting the test specimen to a controlled temperature in a controlled atmospheric environment.Generally, the mass loss of the test specimen is read out directly in units of weight per cent of the original sample quantity [20].The TGA results may be presented by (i) mass versus temperature (or time) curves, referred to as TGA curve or (ii) rate of mass loss versus temperature curve, referred to as differential thermogravimetric (DTG) curves.While TGA gives the change of weight loss of a sample with respect to temperature in a controlled atmosphere, DTG data indicates the rate of degradation of materials with temperature and it is given as the first order derivative weight loss (%) per °C on the y-axis and temperature on the X-axis [21].
The curves of TGA and DTG of the unaged and 3.5 kV rms AC and 3.5 kV DC aged test specimens are presented in Fig. 5.The TGA curves in Fig. 5a show all samples lost weight (decompose) at two stages.The first stage at roughly 220°C corresponds to the release of the water of hydration from the ATH filler material.The second stage refers to the degradation of low molecular weight, cross-linked silicone elastomer and silica [21,22].The weight loss profiles as a function of temperature are the same for the aged and unaged as shown in Fig. 5a.However the rate of change of weight is a function of the nature of ageing of the material as shown in Fig. 5b.
The first minimum peak (weight loss) of the DTG curves in Fig. 5b at 340°C corresponds to the release of the water of hydration from ATH particles according to Abdollahian et al. [23].With reference to Table 2, the aged samples of 3.5 kV rms AC and 3.5 kV DC showed equal rate of weight loss but that is twice higher than that of unaged samples.The difference is because the ageing process would have already reduced the ATH filler and therefore the rate of loss of the remaining ATH filler would be higher than in unaged.The second weight loss of Fig. 5b corresponds to the SiR decomposition that occurs above 400°C, which is the safe limit temperature of organic materials [24].It is also apparent from the second weight loss that all the aged and unaged samples decompose at roughly the same rate with the DC sample showing a marginally higher rate as quantified in the last column of Table 2.
Same analyses (LC, FTIR and TGA) done for the 3.5 kV rms AC and 3.5 kV DC aged samples were repeated for the 4.5 kV rms AC

Results: accelerated ageing at 4.kV AC and DC test voltages
The measured LC, and material characterisation results of SiR test specimen aged at 4.5 kV rms and 4.5 kV DC at 6 ml/min pollutant flow rate are comparatively analysed and discussed in this section.

LC analysis
The 4.5 kV rms AC and 4.5 kV DC LC plots are shown in Figs.6a  and b respectively.The AC LC at 4.5 kV rms was higher than that at the 3.5 kV rms tests as expected.It is remarkable to note that for all the three samples at 4.5 kV DC, the presence of LC stopped at about 1½ hours after the initial voltage application.This happened as tracking stopped on the region near the ground (cathode) electrode.A more aggressive activity around the ground electrode had removed the carbonaceous coating thereby exposing a whitish hydrophobic surface which stopped the tracking.The maximum recorded LC at 4.5 kV DC was 30 mA before the arcing/tracking completely stopped.Gorur et al. [25], also found that samples degraded more when the majority pulses of LCs were in the range of 15-30 mA.Another study by Kumagai and Yoshimura [26] found that LCs in the range of 20-25 mA appeared to be the most effective current pulses for causing tracking and erosion.Fig. 7 shows an image of the sample taken after the test.The test was allowed to run for the whole 6 h duration but once the erosion of the DC sample near the ground electrode was established, arcing could not be initiated.
The hourly average LC at 4.5 kV rms AC and 4.5 kV DC presented in Fig. 7 shows the same trends as those of the 3.5 kV rms AC and 3.5 kV DC.Generally, the hourly average AC LCs do not show any noticeable change from the first hour to last hour of the test.Conversely, the hourly average DC LC shows a consistent increasing trend.
According to the IEC-60587 standard LC failure criterion of 60 mA, all the samples passed the test but with regard to tracking failure criterion, all the DC samples would have failed the test within the first hour of the voltage application.A summary of the LC and its effects on the test specimens is given Table 3.
At 4.5 kV DC, the inclined plane accelerated ageing process becomes significantly irregular in that it stops prematurely.The procedure in the IEC-60587 standard does not give the same results   under the same conditions for AC and DC SiR insulation.Samples that would pass the test under AC would fail the test under DC applications.

Thermal stability analyses of the 4.5 kV rms AC and 4.5 kV DC inclined plane accelerated aged SiR test specimen
The curves of TGA and DTG of 4.5 kV rms AC and 4.5 kV DC are shown in Figs.8a and b.The analysed specimens consist of the AC sample, the specimen cut off the region adjacent to the cathode and another cut from the carbonised tracking region of the 4.5 kV DC sample.
As with the 3.5 kV (both DC and AC), the mass loss for the aged and unaged samples were the same as shown in Fig. 8a.The rates of mass loss however are different as shown in Fig. 8b.For the aged samples, the rate of mass loss of the 4.5 kV AC aged sample was higher than the 4.5 kV DC as also indicated by the corresponding values in Table 4. Similar trends of the TGA and DTG curves can be found in the literature [22,23].The sharp peaks at the start and end points of the graphs in Fig. 8b are due to the sensitivity of the TGA instruments, which took some time to settle to a stable operation point.

FTIR analysis for the 4.5 kV AC and DC inclined plane accelerated aged samples
The IR spectra of 4.5 kV AC and DC samples are shown in Fig. 9.It can be noticed that there are differences in absorption peaks   between the whitish erosion part adjacent to the ground electrode of the DC sample and the rest of the aged samples.The material taken from the whitish portion of the DC sample has the same absorption peaks as those of the new sample at 3373-3618 cm −1 .This region lies in the ATH band [27].It can be inferred that the ATH filler material did not lose its bond water because the ATH particles in the region 3373-3618 cm −1 retained their characteristic peaks.It was also observed that whitish portion of the 4.5 kV DC sample was hydrophobic.The indication is that either chemical decomposition did not take place or the erosion removed the top degraded layer thereby exposing unaged layer with the original chemical signature.Most polymer materials contain carbon atoms which, due to tracking can form carbonised conducting paths.If however, conducting carbon tracks could be removed without major erosion, the flow of LC would be inhibited [28].It is evident that with the exception of the peaks due to the OH group, the rest of the absorption peaks of 4.5 kV aged samples are similar to those observed under 3.5 kV aged samples and therefore the same explanation would apply.
In summary, the FTIR analysis results suggest that 3.5 kV rms AC and 3.5 kV DC tracking cause similar chemical degradation but at higher voltages the phenomenon changes.

Results: energy dispersive spectroscopy (EDS) of the inclined plane SiR test specimens
The spectral analysis of the samples using EDS were performed and the results showing the elemental compositions are presented in Table 5.The atomic composition of the reference sample consists of carbon (C), aluminium (Al), silicone (Si) and oxygen (O).The aluminium originate from the ATH filler material.With the exception of 4.5 kV AC, all the other aged samples showed a decrease in their atomic concentration of aluminium with respect to the virgin sample.
It is important to note that only the DC samples with the exclusion of the white erosion portion, show the presence of potassium, calcium and a high concentration of chromium and iron compared with the AC aged samples.This is possibly from the electrolytic corrosion of the metallic electrodes under the influence of the DC voltage and the pollutant.The corrosion effect of DC was also confirmed by Heo et al. [6], who reported that DC causes more electrode corrosion than AC.The presence of metallic elements on the DC samples proves what many researchers have reported that under DC, electrodes corrode.Although the increased erosion under DC tracking is attributed to the absence of current zero crossings; the metallic electrodes corrosion by-products could increase the pollutant conductivity resulting in increased LC and corresponding erosion.
An important conclusion from the EDS results as presented in Table 5 is that the chemical composition of both 3.5 kV rms AC and 3.5 kV DC aged samples are different from the unaged samples.However, the differences in chemical composition between the 3.5 kV rms AC and 3.5 kV DC aged are marginal which suggest that these ageing stresses are comparable.

Results: scanning electron microscopy (SEM) of the inclined plane SiR test specimens
The surface morphology of the samples was investigated using an FIE Quanta 250 FEG SEM.Electron SEM images provide qualitative estimates of the type and extent of the degradation [29].The SEM images showing the material structural changes on the surface of the specimens are presented in Fig. 10.
To quantify the roughness of the surface in each image, a standard edge detection algorithm was applied to detect large changes in depth of the surface.Roughness can be characterised as a surface with many large, rapid changes in depth, which in the micrographs appears as large, rapid changes in the value of the pixels.As with most edge detection techniques, a Gaussian blur was applied to each image before edge detection to remove noise, including any small changes in pixel values that may be incorrectly detected as edges.A Canny edge detection algorithm was then   implemented and the threshold varied to reduce the amount of unimportant features detected.These edges were counted within the image to quantify how many edges were detected and thus how many large rapid changes in depth occur on the surface.All the images were subjected to the same algorithm with the same Gaussian blur kernel size and the Canny edge detector using the same thresholds in all cases to enable accurate comparisons.In Fig. 10, each original image is accompanied with the corresponding edge analysed image.The corresponding edge count values for each image are also indicated.The more the edges, the rougher the surface.It is evident that despite a higher resolution of the image of the unaged material, the edge count is much lower (a three digit count) compared with those of the aged samples (4 digits).Although much higher, the 3.5 kV DC aged sample has an edge count closest to that of the 3.5 kV rms AC -aged sample.
An overall important conclusion from all the physiochemical analyses is that the 3.5 kV rms AC and 3.5 kV DC aged samples show a relatively similar surface morphology as well as chemical signature.All the material analysis tools used in this work (TGA, FTIR, EDS and SEM) have indicated that the 3.5 kV rms AC and 3.5 kV DC ageing effects are in the same order although at 3.5 kV DC the degradation is relatively more severe.In various round robin tests involving major HV test laboratories in the world, the results as reported in a Cigre technical brochure [2] also give evidence of similarities between 3.5 kV rms AC and 3.5 kV DC aged SiR insulation.
It can therefore be argued that when comparative analysis of AC and DC inclined plane tests are desired, the 3.5 kV gives the closest correlation both from electrical and physiochemical perspectives.At higher voltages (4.5 kV) the AC and DC degradation behaviour becomes inconsistent and unreliable as a repeatable test procedure.

Consolidated theoretical analysis and discussion of the mechanisms in the context of the obtained experimental results
The tracking and degradation mechanism observed in the present work under the inclined plane setup is illustrated in Fig. 11 and is discussed in this section based on models in the literature [29][30][31].
As the pollutant droplet rolls down from the top electrode along the inclined insulator surface, it creates a semi-conductive wet filament.Due to the presence of the ions in the wet filament layer, LC flows thereby heating up the wetted track.The heat initiates already, the insulation surface degradation process.It therefore follows that if the voltage is constant (i.e.DC), the current flow is consistent and more heat is generated in a unit time compared with AC voltage.Furthermore as the electrode corrode under the influence of DC voltage, the metallic ions dissolve in the pollutant and further increase the conductivity of the semi-conductive tracks.These are the metallic elements detected in the DC-aged specimens using the EDS analysis as presented in Section 6. Due to the LCgenerated-heat, portions of the wet track dry up thereby creating dry-bands.
The dry-bands between the conductive wet spots on the insulator surface essentially become a series of capacitors.Electric field enhancement occurs around the wet spots and within the dry bands.If the resultant electric fields exceed the withstand strength of the air at the altitude, spot discharges occur around the wet spots and across the dry bands; dry band arcing ensues.The process continues cascaded across the electrode gap.The discharges manifest as increased LC that can be measured on the external circuit supplying voltage to the electrodes.Under AC voltage, since the voltage changes cyclically in magnitude from peak to zero, the spot discharges and dry-band arcing may not always be consistent and certain in comparison to the DC conditions.Consequently the LC values are smaller and more random under AC as measured in this present work given in Sections 4.1 and 5.1.Each wet spot discharge and dry-band arcing impart intensified energy such as UV and heat in localised regions on the insulator surface.
The more there are arcing dry-bands, as in the case of DC voltage, the more the destructive energy imparted onto the insulation surface.The degradation as quantified through various physiochemical analyses presented in this paper confirm that under the same conditions, degradation under DC voltage is more severe

Conclusion
This work was carried out with the aim of improving the understanding of the fundamental aspects related to the electrical performance of outdoor HV DC SiR insulation in polluted environments.A critical analysis of the available literature on the subject has been presented and contradictions identified and discussed.A study comprising of experimental tests and comprehensive material analyses yielded the following deductions: • The use of IV drip systems is a reliable alternative in carrying out the inclined plane tests as per the standards.• With 0.3 ml/min pollutant flow rate, the 3.5 kV AC rms and 3.5 kV DC inclined plane tests gave relatively similar results qualitatively (electrically and physio chemically) although at DC the degradation was quantitatively more severe.• Above 3.5 kV DC, the inclined planed tests results become erratic and therefore unsuitable as a repeatable standardised procedure.• If the IEC-60587 standard is to also apply on DC, the LC limit of 60 mA should be the only failure criterion.

Fig. 2 Fig. 3
Fig. 2 Typical LCs measured during the inclined plane accelerated ageing of SiR insulation at 3.5 kV and 0.3 ml/min pollutant flow rate (a) Typical peak LC at 3.5 kV rms AC, (b) Typical peak LC at 3.5 kV DC

Fig. 4
Fig. 4 FTIR of 3.5 kV AC and DC aged specimen

Fig. 5
Fig. 5 Thermal stability analysis graphs of SiR insulation at 3.5 kV (a) TGA curves at 3.5 kV rms AC and 3.5 kV DC aged and unaged samples, (b) DTG curves at 3.5 kV rms AC and 3.5 kV DC aged and unaged samples

Fig. 6
Fig. 6 Typical LCs measured during the inclined plane accelerated ageing of SiR insulation at 4.5 kV and 6 ml/min pollutant flow rate (a) Typical peak LC at 4.5 kV rms AC, (b) Typical peak LC at 4.5 kV DC

Fig. 7
Fig. 7 Hourly average LC at 4.5 kVrms AC and an image of the 4.5 kV DC-aged sample

Fig. 8
Fig. 8 Thermal stability analysis graphs of SiR insulation at 4.5 kV (a) TGA curves at 4.5 kV rms AC and 4.5 kV DC aged and unaged samples, (b) DTG curves at 4.5 kV rms AC and 4.5 kV DC aged and unaged samples

Fig. 9
Fig. 9 FTIR of 4.5 kV rms AC and 4.5 kV DC inclined plane aged SiR insulation specimen

Fig. 10
Fig. 10 Summary of SEM images of the samples

Table 1 Summary of the literature on DC inclined plane tests
254].et al.[14]3.25kVAC, +2 kV DC & −2.5 kV DC Generally, +ve DC LCs is highest, followed by -ve DC and then AC.Mass loss is lower at AC than DC.Positive DC has a higher mass loss than negative DC.Time to track is longest for AC than for DC.

Table 3
Summary of the tracking and erosion caused by AC and DC LCs