Assessment of Hydraulic Network Models in Predicting Reverse Flows in OD Cooled Disc Type Transformer Windings

Predicting liquid flow distribution in the winding is crucial for transformer thermal design and hydraulic network models are usually used to guarantee reasonable liquid flow distribution. This paper is a continuation of the research effort in using computational fluid dynamics (CFD) to calibrate and improve the network modelling approach. The possibility is assessed of hydraulic network models in predicting the occurrence of reverse flows in pump-driven oil directed (OD) cooling modes. CFD modelling shows that the occurrence of reverse flows requires both high Reynolds number at the winding pass inlet and the “real” pass inlet velocity profile. To reflect the effect of pass inlet velocity profile on flow distribution, a winding pass is added beneath dividing and merging junctions in deriving minor pressure loss correlations. In addition, hybrid minor loss correlations are used to incorporate the effect of pass inlet velocity profile and to mitigate the inaccuracy of minor pressure loss segregation introduced by the inclusion of a winding pass beneath the junctions. Even with the above mentioned improvements, a conventional network model with one node for each junction still cannot predict reverse flows because of its architecture limit, hence a 3-node junction network model is then proposed. It is demonstrated that the newly developed 3-node junction hydraulic network model can predict the occurrence of reverse flows in OD cooled disc-type transformer windings.