Summary

Accurate prediction of temperature distribution in an electrical machine at the design stage is becoming increasingly important. This prediction of temperatures enables a designer to optimize the design and thereby effective cost savings. In spite of the importance of knowing the temperature inside rotating electrical machines the available literature lacks methodologies that allow obtaining such information in a satisfactory manner and making use of the state of art knowledge related to heat transfer and fluid flow. In the present work the thermal design of induction motors is reviewed and a lumped thermal model is presented. The lumped thermal model is a combination of global thermal parameters and distributed parameters at different parts of the machine. The knowledge of losses (heat sources) involved in the different parts of the machines is essential for the construction and analysis of the thermal model. Adopting standard procedures in motor analysis, the losses are evaluated through electrical tests and are distributed in both stator and rotor. Next, employing the thermal model a sensitivity analysis is performed indicating that the most critical parameter in cooling the motor is the convective heat transfer coefficient on its external surface. Measurements and calculations explore the non uniformity of the local external heat transfer coefficient and its impact on the motor temperature. Tests on three different rotations showed variations in the external heat transfer coefficient around 20% as it changes the rotation of 900 rpm to 1200 rpm and around 40% changing from 1200 rpm to 1800 rpm. Some suggestions are presented to reduce the motor operation temperature, including the use of guide vanes to better direct the airflow over the frame. The effect of correction of guides air flow over the external surface allowed to observe an average increase of 5% in the external heat transfer coefficient in the rear, 40% in the central and 20% in the front of the motor compared to the results obtained in the reference condition.

Material for download