Redesign of a Two and a Half Stage Subsonic Turbine Using a New Viscous Inverse Design Method

The midspan section of a multi-stage subsonic turbine that is built and tested at the University of Hannover is redesigned using a new inverse blade design method where the blade walls move with a virtual velocity distribution derived from the difference between the current and the target pressure loading on the blade surfaces. The prescribed design variables are the blade loading and thickness distribution. This new inverse method is fully consistent with the viscous flow assumption and is implemented into the time accurate solution of the Reynolds-Averaged Navier-Stokes equations that are expressed in an arbitrary Lagrangian-Eulerian (ALE) form to account for mesh movement. A cell-vertex finite volume method of the Jameson type is used to discretize the equations in space; time accurate integration is obtained using dual time stepping. An algebraic Baldwin-Lomax turbulence model is used for turbulence closure. The mixing plane approach is used to couple the blade row regions of the multistage. The CFD analysis formulation is first assessed against the multi-stage turbine experimental data. The method is then used to redesign the second and third stators of the 2.5 stage turbine so as to reduce the blade suction side diffusion. The results show that by carefully tailoring the target pressure loading, some improvement can be achieved in the turbine performance.