This thesis focuses on the inverse thermal process of determining power losses of electrical machines from temperature measurements. The feasibility of this endeavour is investigated and the inverse mapping of losses with analytical and numerical thermal models of a 37 kW cage induction motor are demonstrated. Accurate lumped thermal networks as well as 2D and 3D finite element models of a 37 kW motor are built as forward thermal models for this purpose. Attention is given to managing error in measurement so that the temperature-to-source mapping can be accurate. The inverse mapping was achieved with two methods - least squares and conjugate gradient. The results of the inverse mapping indicate that inverse thermal approaches hold promise in estimating losses in machine domains where direct measurements are cumbersome, such as the stator yoke or tooth regions. Lumped and distributed loss components could be recovered reliably using both inverse methods. Furthermore, a water-cooled calorimeter was designed and built to measure the total power loss of the induction motor. With only a few measured temperatures of the motor, the inverse thermal process is able to segregate the total losses into its constituent components.
|Translated title of the contribution||Inverse Thermal Analysis of Electrical Machines|
|Publication status||Published - 2019|
|MoE publication type||G5 Doctoral dissertation (article)|
- electrical motor
- heat transfer
- inverse thermal problem