Low pressure exhaust hood of steam turbine is used to connect the last stage turbine and the condenser. To further improve the recovery capability of low pressure turbine exhaust hood, an aerodynamic optimization system has been developed with Matlab platform. The system includes four modules: parametric geometry modeling, structured meshes generator, commercial aerodynamic simulator and Kriging surrogate based optimizer. The diffuser geometry profile in the exhaust hood is parameterized using cubic Bezier curve, and the control points of these curves are considered as design variables. The exhaust hood performance is optimized by changing those design variables to maximizing the average static pressure coefficient. The block-structured meshes are generated with ICEM-CFD Hexa. Aerodynamic performance evaluations of the hood are carried out by the three dimensional Reynolds-averaged Navier-Stokes computational fluid dynamics solver-CFX. Mesh generation and aerodynamic analysis are done automatically, which are driven by script commands in batch mode. Kriging model is used as surrogate, which establishes a global mapping between design variables and objective variable. In order to balance the exploration and exploitation with Kriging surrogate, Expected Improvement (EI) sample criteria is adopted to update Kriging surrogate. The proposed optimization framework drastically reduces the number of calling time-consuming CFD. Two aerodynamic optimization test cases are performed with the system. The aerodynamic performance of the original and optimal exhaust hood will be compared, while the flow filed in the both exhaust hood will be illustrated.

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