Prediction and measurement of the heat transfer coefficient in a direct oil-cooled electrical machine with segmented stator

Abstract

The heat transfer coefficient (HTC) is a critical parameter that is required for accurate thermal modeling of electrical machines. This is often achieved from empirical correlations of ideal geometries or computational fluid dynamics (CFD) simulations. This paper presents a novel technique using double-sided thin film heat flux gauges for measuring the HTC from a direct oil-cooled electrical machine with segmented stator. While thin film gauges are often used in transient measurements of the HTC on gas turbine components, their application to electrical machines has been largely unexplored. This is the topic of this paper. Due to the large viscosity of the coolant, the transient technique was found to be inadequate and a steady-state adaptation for oil-cooled machines was developed. This paper explores the challenges linked with this measurement technique when applied to oil-cooled machines, and develops new nondimensional correlations of the Nusselt number with Reynolds number. These correlations are applicable to machines with different geometries, flow, and coolant properties. The experimental results were compared to CFD simulations and existing pipe flow correlations. It is shown that these underpredict the HTC by approximately 60% at Re = 20. The discrepancy gradually decreases to around 10% at Re = 200.