An Accurate Experimental Approach for Deriving Equivalent Thermal Conductivity of Impregnated Electrical Windings

This paper presents an experimental approach for accurate derivation of equivalent thermal conductivity of anisotropic impregnated electrical windings. The proposed method employs a custom-built heat flow metering system for the analysis of cuboidal materials samples, which allow for the material anisotropic properties to be experimentally derived. Both theoretical fundamentals and experimental data from tests on 8 differing materials samples are discussed in the paper demonstrating the effectiveness of the proposed method. A comparison between various resins and winding geometries is made concluding that a rectangular winding with Epoxylite resin demonstrates the highest equivalent thermal conductivity in all planes. For sample planes with high thermal conductivities (i.e., 180W/m.K) a measured accuracy down to 1.81% was achieved. As predicted by numerical methods, sample planes with low thermal conductivity (i.e., 0.2W/m.K) had a much higher propensity for error. Further to these, an impact of accuracy of the thermal conductivity data on the winding temperature distribution is illustrated for a case study electrical machine demonstrator. The theoretical predictions show a significant effect, i.e., here, an increased winding to housing thermal resistance up to 17% when using measured thermal conductivity data for the proposed test setup.