Film cooling technology is significant for aero-turbines development. However, uncertainty on operating and manufacturing conditions would put the high-pressure turbines at the risk of unexpected problems. To extend the service life of the gas turbines, it is necessary to conduct an uncertainty analysis of the effects of the geometry parameters at different blowing ratios on film cooling performance. A classical flat-plate film cooling model under the multiple-row superposition conditions with trench is investigated in this study. The uncertainty inputs included the hole location, depth, and width of the trench and compound angle, while the lateral-averaged film cooling effectiveness was seen as the output. To reduce the computational cost, a Multilayer Perceptron (MLP) model based on supervised learning is built to model the nonlinear regression between the film cooling parameters and the film cooling effectiveness. The computational Fluid Dynamics (CFD) method was utilized to provide the training dataset. After rigorous validation and comparison, the MLP model shows higher accuracy and performance than other classical methods. Then the statistical flow and thermal characteristics of the film cooling outputs and variance were analyzed by the Monte Carlo simulation method. Results of sensitivity analysis show that the uncertainty of compound angle has the greatest effect at both low and high blowing ratios, followed by the location of the first row, trench width, and depth. Besides that, the hole pitch upstream gains more influence than those downstream. Also, both at low and high blowing ratios, the uncertain interval due to compound angle is the biggest, and the uncertain interval due to the hole location of the last row is the smallest. In the design and manufacturing process, it is believed that the uncertainty of compound angle, the hole location of the first row, depth and width of the trench should be paid extra attention to.
Read full abstract