Global optimization of aerodynamic design and knowledge discovery method of an exhaust hood

To improve the aerodynamic performance and explore design guidelines for an exhaust hood, a robust and efficient design optimization and data mining method, which combines meta-based global optimization algorithm with a 3rd-order Bezier curve-based 3D parameterized method, Reynolds-Averaged Navier-Stokes (RANS) solver technique and data mining technique based on variance analysis, is proposed for the aerodynamic optimal design of an exhaust hood. By fully taking into account the interaction between the last turbine stage blades and exhaust hood, design optimization and knowledge discovery of a low-pressure exhaust hood are carried out for maximizing the static pressure recovery coefficient. The static pressure recovery coefficient of the optimal exhaust hood is improved from 0.165 to 0.516. The interactions among design variables and objective function are illustrated using the data mining technique combined with detailed aerodynamic analysis. The research results indicate that the diffuser outlet width, the outer hood width, the outer flow guider's height and the outer flow guider's outlet angle have significant effects on the performance of the exhaust hood. The proposed design optimization and data mining method for the exhaust hood provide a basis for the design of high-performance exhaust hood. © 2015, Editorial Office of Journal of Xi'an Jiaotong University. All right reserved.