Development of a numerical correlation for heat transfer coefficients in steam turbines inner chambers

Nowadays, the growing need for advanced solutions to limit the environmental impact in thermal power generation has given rise to significant changes in steam turbines design and operability. In this context, improvement in efficiency and more flexibility in different operating conditions are the major challenges. An increased flexibility is related to rapid load changes, fast start-up and frequent shutdown that could lead to high material thermal stresses in the unit. Hence, in order to achieve the rise in efficiency and to preserve the lifetime of the turbine, a detailed knowledge of the thermal behavior of the turbine is required. This work is focused on the investigation of flow and heat transfer phenomena in the steam chambers between stator blade carriers. Correlative 1D models are usually used in the industrial practice to estimate heat transfer coefficients to be used as boundary conditions in detailed FEM simulation of the Whole Engine Modelling (WEM). However, the flow pattern in this chamber is complex and this approach could represent an oversimplification, therefore computational fluid dynamic (CFD) analyses have been performed with the aim to obtain a more accurate estimation. All the analyses have been carried out with ANSYS CFX with a quasi-2D axis-symmetric approach. This methodology, despite the need for some simplifications and assumptions to reduce the problem from 3D to quasi-2D, ensures less expensive simulations in terms of computational resources when compared to a fully 3D. After a numerical setup validation in terms of mesh sensitivity, a parametric CAD has been generated and tested. The goal is to analyze how geometrical parameters (such as the axial distance between the stator blade carriers or the height of the chamber) of the chamber will influence heat transfer coefficients, considering different operating conditions. A design of experiment (DOE) with ANSYS Workbench platform has been carried out in order to perform automatically a set of simulations. Finally, the results obtained have been post-processed to extract a sufficiently accurate correlation for the average Nusselt number applicable for almost all the steam chambers in the turbine even in different operating conditions.