**4. Discussion**

The proposed experimental procedures containing a cycle of 8-h voltage application and 16-h interruption seem suitable to evaluate partial discharge degradation of hydrophobic material like silicone rubber because hydrophobicity recovers almost to the initial value during 16 h. It reflects a practical usage condition of polymer insulators that hydrophobicity recovers if factors a ffecting hydrophobicity are removed.

Consistent results are obtained in 50- and 100-cycle tests. In both tests, surface erosion is larger in the order of samples D and E, samples B and C, and sample A. Progress of partial discharge degradation of silicone rubber, especially samples D and E, is accelerated in the case of 2 mm gap spacing compared with that of 1 mm in both tests.

Figure 13 shows change in Ra of five kinds of samples up to 50 cycles measured with the surface roughness meter, which are obtained by carrying out experiments 3 times under the condition of 1 mm air gap. Each sample shows almost the same performance in any experiment. Also Ra is larger in the order of samples D and E, samples B and C, and sample A in any experiment. It is considered the proposed experimental procedures give acceptable repeatability of results.

Surface erosion of a sample increases gradually by partial discharge with the number of cycle. In samples D and E, decrease in hydrophobic Si-CH3 and Si-O-Si of main chain is confirmed by FTIR analysis. This is a possible reason for low contact angle, sever surface erosion, and consequently short cycles to breakdown. Samples D and E contain ATH, which is considered e ffective to enhance resistance to tracking and erosion by discharge because heat is absorbed by releasing the water of hydration from ATH molecule when the temperature of an ATH filled polymer reaches ~200 degrees [1,14]. This role of ATH seems e ffective for partial discharge because decrease in ATH is suggested through FTIR analysis in the present study. Nevertheless, surface erosion of samples D and E containing ATH is serious compared with sample A without ATH. This is attributed to intensified partial discharge activities in the air gap for samples D and E because their higher permittivity generates higher localized electric field in the air gap of the electrode system. Lower resistivity of samples D and E may also relate to shorter lifetime to breakdown.

**Figure 13.** Repeatability of change in Ra with number of cycle obtained for 1 mm gap spacing. (**a**) samples A, C, and E; (**b**) samples B and D.
