Thermal analysis and air flow modelling of electrical machines

Abstract

Thermal analysis is an important topic that can affect the electrical machine performance, reliability, lifetime and efficiency. In order to predict the electrical machine thermal performance accurately, thermal analysis of electrical machines must include fluid flow modelling. One of the technologies which may be used to estimate the flow distribution and pressure losses in throughflow ventilated machines is flow network analysis, but suitable correlations that can be used to estimate the pressure losses in rotor ducts due to fluid shock is not available. The aim of this work is to investigate how the rotation affects the pressure losses in rotor ducts by performing a dimensional analysis. Apart from the additional friction loss due to the effects of rotation, other rotational pressure losses that appear in a rotor-stator system are: duct entrance loss due to fluid shock and combining flow loss at the exit of the rotor-stator gap. These losses are analysed using computational fluid dynamics (CFD) methods. The CFD simulations use the Reynolds-averaged Navier Stokes (RANS) approach. An experimental test rig is built to validate the CFD findings. The investigation showed that the CFD results are consistent with the experimental results and the rotational pressure losses correlate well with the rotation ratio (a dimensionless parameter). It shows that the rotational pressure loss generally increases with the increase in the rotation ratio. At certain operating conditions, the rotational pressure loss can contribute over 50 % of the total system loss. The investigation leads to an original set of correlations for the pressure losses in air ducts in the rotor due to fluid shock which are more suitable to be applied to fluid flow modelling of throughflow ventilated machines. Such correlations provide a significant contribution to the field of thermal modelling of electrical machines. They are incorporated into the air flow modelling tool that has been programmed in Portunus by the present author. The modelling tool can be integrated with the existing thermal modelling method, lumped-parameter thermal network (LPTN) to form a complete analytical thermal-fluid modelling method.