Predicting Airflow Distribution in A Radially Air-Cooled Generator by Flow Network Method

摘要

This paper uses flow network (FN) method to predict airflow distribution in a generator with stacked lamination and radially-extended airducts which are commonly-seen in wind power generators and hydro power generators. It is found that conventional FN modelling approach does not predict accurate flow distribution in different airducts, since airflow is rather complex in such cooling structure. To improve the model accuracy, the FN model is modified by introducing Gardel equations as the governing equations for T-junction dividing flow. The results of both original and modified FN models are compared with CFD modelling results. It is shown that the modified FN model not only gives much more consistent airflow distribution but also predicts more reasonable pressure drop. The flow rate difference between the modified FN model and the CFD model is within 10% for all the airducts and within 5% for 8 out of 9 airducts.