عنوان مقاله [English]
Gas turbines have many applications in different industries. The axial turbine is one of the most challenging components of gas turbines for industrial and aerospace applications. With the ever-increasing requirement for high aerodynamic performance blades, two and three dimensional aerodynamic shape optimization is of great importance. In this paper, an automatic design procedure is presented for single point optimization of an axial flow turbine stator and rotor cascades. In this method, the genetic algorithm, the blade geometry generator, a computational mesh generator and the flow field solver are related. The objective function is the total pressure loss of the flow passing through the stator or rotor blade cascades. Particular modifications are performed with a limited number of optimization parameters, by changing stator and rotor blade thickness distribution. The stator and rotor airfoils are regenerated by adding a smooth perturbation of Wagner shape functions to the thickness distributions. Because of the symmetric geometry of the turbine blade cascade, periodic boundary conditions are used for simulations. Furthermore, three-dimensional and turbulent flow field investigations are numerically performed employing a compressible Navier-Stokes solver and the k-e (RNG) turbulence model. The experimental results of initial stator cascade are used for validation of numerical results. The experimental investigation is performed in the Gas Turbine Laboratory of Sharif University of Technology. The maximum deviation of numerical results from cascade test data is 1.14 percent. This optimization strategy resulted in a reduction of 1.5% total pressure loss in the rotor and 3.0% in the stator, for a prescribed incidence angle, while the cross sectional area of the modified stator and rotor blades increased, 1.3% and 2.0%, compared with the initial ones, respectively. It should be noted that for the rotor blade cascade, a multi-point optimization is required, based on a comparison of the original and modified loss-incidence angle chart results.