LTD次级多层薄膜绝缘的击穿概率分析

Breakdown Probability for Multi-layer Film Insulation of LTDês Secondary-turn

LTD次级多层薄膜绝缘子层数达几百、上千层,利用厚几十μm的高本征击穿强度薄膜制作强度更高或者更紧凑的低感绝缘子是这种绝缘的发展方向。在新型绝缘子研制中,希望以较小规模(层数)试样的击穿实验数据预测较大规模绝缘子在一定场强下的击穿概率。为此基于Weibull型半经验分布函数模型,利用一组较小规模多层绝缘试样的击穿实验数据进行拟合得到模型参数,导出分析预估多层薄膜绝缘子击穿概率的公式,进而建立了分析预估较大规模多层绝缘击穿概率的计算方法。多层薄膜绝缘的击穿强度随层数增多而降低,击穿概率随层数增多而增大,但击穿强度降低和击穿概率增大的速率随层数增多而逐步减小。如将1 000层绝缘分为10组,每组100层薄膜,仅需对其中一组薄膜每增加10层进行一次击穿实验(共10次),即可计算得到1 000层绝缘的击穿概率。本方法适用于稍不均匀电场多层薄膜绝缘在一定平均场强下的击穿概率计算。

The glycerin-impregnated multi-layer polymer-film insulation is often used as the high voltage insulator for the secondary turn of LTD（Linear Transformer Driver）, and the layer number of film for this insulator can be more than several hundreds and even a thousand. As the development of insulating materials, polymer films of tens of micrometers of high intrinsic breakdown strength become cheap and easily available, for example, the electric strength of PP（polypropylene） film used in high voltage capacitors has reached 500 V/μm（500 k V/mm or 5 MV/cm） or higher. As a result, it is feasible that technical improvement for the LTD＇s multi-layer film insulation makes it possible to more compact insulators of lower inductance. Consequently, we derived a Weibull-type distribution to apply to the prediction of breakdown probability for multi-layer film insulation by semi-empirical analysis. The distribution parameters can be obtained from a smaller scale samples of multi-layer film insulation, and the prediction of breakdown probability for larger scale samples can be fulfilled using a simplified computation method. The electric strength reduces and breakdown probability increases with the increase of film layers. However, the rate of electric strength reduction and breakdown probability augmentation becomes slow with the increase of film layers. An example indicates that for an insulation of N times n layers, this method is valid for n less than 10 with N unlimited. For N=100, if breakdown test is performed every 20 layers, then the breakdown probability for the insulation of 1000 layers can be computed with n=10, helpful in the development of new insulators.