On the statistical testing of solid insulation systems

Solid insulating materials are not self-restoring due to electrical discharges; a separate test piece is needed for each individual test. Even with no discharge observed, polarization and conductivity processes can produce space charges that exist for a long time and that can affect the results of a subsequent test on the same test piece. The solution is to use one test piece per stress. Since solid test pieces are relatively expensive, a small number of test pieces is recommended. Consideration of the choice of sample size should be exercised with great care. The statistical nature of high-voltage testing of electrical insulation has long been recognized. A brief overview of the application of statistical methods to establish the insulation strength and lifetime of solid dielectrics is reported. Accelerated aging tests on solid dielectrics are statistical tests that have to be rigorously analyzed in order to justify any valid conclusions. The application of Weibull statistics for the description of dielectric breakdown is presented. The breakdown probability of the test voltages is a function of the test method, of their parameters, of the nature of the breakdown probability function, and of the assumed physical processes. Proper test methods and their parameters can be selected to determine the breakdown voltages and time to breakdown.