Trial calculation of the AC test voltage using a new cumulative failure probability model

In the condition settings of a power-frequency withstand voltage test for substation equipment, an approach based on insulation reliability evaluation using Weibull distribution is available. In this method, temporary overvoltage is evaluated using an “independence model”. This “independence model” assumes no remaining influence of voltage application (i.e. the model obeys the failure probability distribution starting at time 0 each time the voltage is applied). However, the authors have conducted insulation characteristics tests by applying a voltage multiple times at intervals and clarified that the influence of voltage application remains to a certain extent (i.e. the effect assumed in the “accumulation model” appears) as well as the degree of influence depending on the conditions for oil-immersed transformer. The present paper proposes an evaluation method with the degree of influence of voltage application history taken into consideration. In the proposed method, the portions following an “independence model” and an “accumulation model” are respectively assumed and combined after multiplying the relevant coefficient representing the respective degrees of influence. According to the trial calculation based on the proposed method, the required withstand voltage decreased by several percentage points, while it was confirmed that the existing method is on the safer side as a means of evaluating the insulation reliability of temporary overvoltage.