Consideration of loss distribution to evaluate the hotspot temperature when up-rating generators

Some exiting generators may have more generation capability than their nameplate rating if they hold large security margin. However, before changing the limit of a generator one must do a thorough evaluation of its behavior in different conditions to extrapolate the effect of any increase. With the help of simulation models to calculate magnetic and stray losses of a machine, it is possible to couple loss distribution with the calculated I²R losses and the measured ventilation losses in a thermal model and calculate the location and the magnitude of the hotspot. These models must be validated by in situ measurements to make sure they correspond to reality. The temperature of RTD is easily accessible, but most of the time insufficient to validate detailed models. This paper presents some of the limitation of the RTD to detect axial or radial temperature distribution. Unfortunately, the copper temperature can not be measured directly, but it can be calculated by numerical models and this was done for two generators in characteristic conditions. The results are compared with temperature and airflow measurements. The models show that up-rating should not be made without understanding the exact behavior of generator as a function of load.