In an axial flow turbine, almost one-third of the total losses are caused by secondary flows, and the non-axisymmetric endwall profiling has been a major tool for years to reduce the secondary flow loss. This paper presents the non-axisymmetric endwall profile construction and optimization both for the stator hub and shroud of a high pressure turbine in the presence of an axisymmetric rotor. The flow simulation in the turbine was conducted by using steady RANS. The perturbation law of non-axisymmetric endwall was based on Bezier curves, and the commercially available optimization software NUMECA Fine/Design 3D was used to design the non-axisymmetric endwall. A genetic algorithm based on the artificial neural network was used as the optimization method. The objective function was aimed at maximizing the stage isentropic efficiency. The change in mass flow rate was kept less than 0.5% (relative) so that efficiency might not be influenced by the mass flow through the variation of the throat area. From the design point of view, the stator hub endwall was optimized at design conditions firstly, but the shroud endwall was kept constant, which resulted in an increase of stage efficiency because flow angles at stator exit were changed. The flow structures in the passage of stator were compared pre and post optimization by using 3-D streamlines in the vicinity of the endwall. Subsequently, the shroud endwall was optimized using the optimized non-axisymmetric hub as initial design. Due to hub and shroud endwall perturbation, the cross passage gradient and entropy were reduced, and the turbine stage efficiency at design conditions was calculated and the improvement in the efficiency was noticed. In addition, the improved hub and shroud contour were considered for off-design conditions as well, and efficiency was even more increased over a considerable off-design regime than at the design point condition.

This content is only available via PDF.
You do not currently have access to this content.