Improving Combined Flow and Thermal Network Accuracy for Radially Air-Cooled Generators by Considering the Nonlinear Resistance Characteristics of T-Junction Flow

Abstract

This work aims to improve the combined flow and thermal network (CFTN) accuracy for radially air-cooled generators by investigating the possible error stemming from the flow part. The nonlinear resistance characteristics of the T-junction dividing flow are first time considered for such cooling arrangement, which is achieved by employing the Gardel equations and the blend equation as the governing equations. Compared with the CFD model results as the benchmark, the nonlinear CFTN predicts an accurate flow rate in the radially extended airducts, making only 9% root mean square (rms) error for all the airducts. On the contrary, the linear CFTN fails to obtain a reliable flow distribution, as the rms error is as large as 34%. As a result, the nonlinear CFTN improves the accuracy of the maximum winding temperature by 6.4% over the linear CFTN and gives a more reasonable temperature gradient in the axial direction. Furthermore, the relative difference between the linear and nonlinear CFTNs is found to be even larger in the following conditions: large cooling air flow rate, wide airgap, low slot fill factor, and large airduct number. The proposed nonlinear CFTN modeling approach is also validated against experimental measurements for a prototype. The modeled temperatures match well with the measurements at different axial positions. According to the aforementioned findings, the nonlinear CFTN modeling approach is considered to be a reliable thermal analysis tool for the radial air-cooling arrangement.