Optimization of the design of a heavy-duty gas turbine intake air volute using the surrogate model method

Abstract

This study introduces an optimization design approach for gas turbine intake air volutes using surrogate models, with the goal of minimizing intake air loss and improving the efficiency of the gas turbine unit. A CFD-based parameterization method is combined with an optimized Latin hypercube sampling technique to generate 200 sets of sample data, from which three surrogate models are constructed. The NSGA-II algorithm is then applied to optimize the support vector regression (SVR) model that exhibits the best performance, ultimately identifying the optimal design solution within the specified design range. The study also investigates the influence of various design parameters on the target parameters, providing valuable insights for parameter design. The optimized target parameters are compared with those of the original model, and the effects of optimization under varying operating conditions are analyzed. The results reveal that the control length of the contraction section is a key design parameter significantly affecting the target parameters. The optimization effect is most pronounced under full-load operation. The optimal design leads to reductions in inlet mean velocity uniformity and total pressure loss coefficient by 84.14% and 82.69%, respectively, compared to the original design. This optimization method effectively enhances the aerodynamic performance of the intake air volute.