Multidisciplinary sensitivity analysis for turbine blade considering thickness uncertainties

This work presents an approach for sensitivity analysis of turbine cooling blade with surface thickness uncertainties, combining mesh deformation method, neural network model and multidisciplinary analysis. Normally, for even tiny shape changes, conventional geometry-based method failed easily during the auto-processing analysis. Therefore, mesh deformation method was utilized to capture the tiny size changes in the multidisciplinary analysis for both the fluid and the structure meshes. The neural network model is constructed by design of experiments to reduce the computational cost. Sensitivity analysis of the multidisciplinary system of blade is performed by numerical difference algorithm with the neural network model. Results showed that the proposed method was effective and practical in engineering.