Temperature rises in electrical machines as related to the properties of thermal networks

The conditions are reviewed under which an electrical machine may usefully be regarded, for the purpose of relating temperature rises to losses, as a thermal network of lumped conductances and capacitances, having an electrical analogue. Although the thermal resistivities are distributed, it is shown that lumped-resistance networks may be found that reproduce the relationships between the inputs of heat to the parts and the mean temperature rises over the volumes of the parts. Sources of non-linearity are considered, and it is shown that in many cases a linear model is adequate. The effect of transport of heat by the flow of a cooling medium cannot be introduced into a simple network model. The equations are linear, but do not have the properties of symmetry and reciprocity that are characteristic of static networks. Examples of such equations are given and some comments are made on the general form of the relationship between temperature rises and heat inputs, and the uses of such relationships. Consideration is given to methods of testing to discover the coefficients of such equations, or the constants of an analogous electrical network where this exists, and methods that make use of intermittent loading are described. A novel timing device for use in such testing, a novel slip-ring contact for use with temperature detectors on rotating parts, and some special features of resistance-type temperature detectors are described. The results of measurements on a 25 h.p. mill-type motor are given both for forced-ventilated and for totally-enclosed conditions, and the appropriate equations, and where possible the equivalent thermal networks, are deduced. The measurements also include information about the temperature distribution in various parts, both for continuous and for variable loading. One purpose of the work was to provide a basis for a new approach to the problem of the estimation of temperature rises with variable loads and speeds. The ground-work for this is in- cluded, but the specific problem is discussed in a separate paper.