Effect and optimization of geometric parameters and arrangement on film cooling performance of fan-shaped holes based on generalized regression neural network

Inadequate design of film cooling holes can significantly reduce film cooling effectiveness of turbine blades, compromising high temperature resistance. By comparing experimental data, the most suitable turbulence model is selected to calculate film cooling effectiveness on the leading edge of turbine blades. A high-precision nonlinear mapping relationship between the geometric parameters and arrangement of fan-shaped holes, and film cooling effectiveness is established by using machine learning which is trained by 1200 cases. The genetic algorithm is employed to search for optimal parameters. Sensitivity and correlation analyses are conducted to assess the influence degree of the geometric parameters and arrangement on film cooling effectiveness and these mechanisms are analyzed in depth. Findings reveal that the injection angle exerts the most significant effect on film cooling effectiveness, with significant effects observed for the forward expansion angle and the angle with respect to stagnation line. Interactions among parameters play a crucial role in influencing film cooling effectiveness and cannot be disregarded. In the design space, the optimal configuration comprises the injection angle of 30°, the forward expansion angle of 10°, the streamwise torsion angle of 10°, the pitch of holes of 12 mm and the angle with respect to stagnation line of 30°.