Thermal Network and Random Forest Model Application in Heat Transfer of Large Nuclear Generator End Region

Abstract

The end ventilation structure of large nuclear generator is complex. The cooling medium will be affected by the disturbance effect, turbulence effect and local pressure change in the flow process, which makes the convective heat transfer coefficient have the characteristics of nonlinearity and difficult to calculate. A 1266 MW nuclear generator is taken as an example to study the effect of cooling medium flow on convection heat transfer coefficient and temperature distribution in the end domain. A random forest prediction model of end convective heat transfer coefficients was constructed to accurately predict the convective heat transfer coefficients of end ventilation holes and ventilation ducts. A three-dimensional thermal network model of the stator end is also constructed, which is combined with the prediction results of random forest convection heat transfer coefficients to quickly calculate the temperatures of the stator core, tooth pressure plate, pressure ring, and other components. Compared with the experimental results, this combined model can significantly reduce the calculation time cost while ensuring the accuracy of the prediction results. It can replace the cumbersome parametric experiments for large generators, and provides an efficient analysis method for the study of heat transfer under complex ventilation paths of large generators.