Using of statistical tools within optimalization of design of material for high-voltage applications

Paper deals with evaluation of chosen properties of newly developed modification of insulation material, composite based on glass-mica-epoxy material. The main reason for this analysis was to optimize the specific design of material. For improving some of mechanical and electrical properties the currently used material was modified by incorporation of nonvowen nanolayers into its structure before the curing. Basic material (without any nanolayer) and two variants of material with incorporated nanolayer (with one and two layers) of the same size were prepared in two different generations. Several nanolayers with unknown and not identical density were obtained and used for the production of the first generation of samples. The second generation was prepared from homogeneous nanomaterial with known density. Process of manufacturing was identical for samples of both generations (with the same glass-mica-epoxy composite). Tools, which can be used to evaluate material design and to determine material wearout failure, are investigated in this paper. Graphical and computation statistical tests were used for evaluation of measured data. Box plot, probability plot were used from the group of graphical methods and t-test for mean value (factor analysis of variance) and F-test for homogeneity of variances (Bartlett's test) from the group of computation methods. Data from the measurement of dielectric strength, impact strength and resistivity were used for the statistical evaluation. On the basis of statistical analyses, risky samples were identified and will be additionally analysed. These information can help with production of the next generation of samples.