In this paper, an optimization system was applied to design the non-axisymmetric endwalls for the stator of a high pressure axial flow turbine. This optimization system combines the endwall parameterization, 3D Navier-Stokes flow field calculation and genetic algorithm based on artificial neural network, which has the advantages of flexible geometry representation and automatic design of the optimal non-axisymmetric endwalls.
And, the 3D steady flow field calculation was carried out to analyze the detailed behavior of complex flow structures pre and post optimization and to examine the influences of the optimized endwalls on the stage performance as well. The results of investigation show that the optimized non-axisymmetric endwalls can significantly decrease the flow loss in the stator, but also affect other aerodynamic parameters at the stator exit, especially the flow angle, and then the flow loss at the rotor exit caused by both the passage vortex in the rotor passage and the tip leakage vortex were increased by changing the incidence angle of the rotor due to the non-axisymmetric endwalls. Finally, the stage performance of the HP turbine is not improved as expected.