Fast prediction and uncertainty analysis of film cooling with a semi-sphere vortex generator using artificial neural network

Abstract

The advancement of aircraft engines relies heavily on film cooling technology. To enhance the film cooling efficiency in high-pressure turbines, many passive flow control methods have been explored. Downstream of the cooling hole, a semi-sphere vortex generator (SVG) decreases the lateral dispersion of the coolant and increases the efficiency of film cooling. To better understand the influence and uncertainty of SVG parameters such as the compound angle, size, and location, a supervised learning-based artificial neural network model is developed to identify the nonlinear mapping between the input parameters and the horizontal-averaged film cooling efficiency. Training data are generated by computational fluid dynamics. The model is quite accurate and stable after sufficient testing and validation. Through Monte Carlo simulations, the framework is used to analyze the thermal and flow characteristics of the film cooling efficiency. The radius of the SVG dominates the film cooling effectiveness at low blowing ratios, whereas at comparatively large blowing ratios, the angular placement of the SVG downstream of the cooling hole is the most important element. The angular position of the SVG has a much stronger impact than the distance at both low and high blowing ratios between the cooling hole and the SVG.