Taguchi and ANN model optimization for uni-array vortex generator incorporated rectangular channel for high thermo-hydraulic performance

In this study, various vortex generator (VG) geometries-delta (D), envelope (M), and fishtail (F)-are numerically analyzed under different ambient conditions using ANSYS 2022R2 to evaluate their thermohydraulic effectiveness in a rectangular channel. Ambient heat transfer coefficients (HTCs) of 5, 15, and 25 W/m2K are considered. The Taguchi design-of-experiments method is applied, and ANOVA and signal-to-noise (S/N) ratios are used to identify optimal performance factors. No previous work has designed a new uni-array channel utilizing a machine-learning ANN model to optimize thermohydraulic performance, particularly for three distinct vortex generator shapes-delta, envelope, and fishtail-under three different ambient operating conditions. The results show that the heat transfer coefficient increases by 2.797, 2.777, and 2.834 for DVG, MVG, and FVG, respectively, at h = 25W/m2K, compared to the inlet value. For h = 15W/m2K, the corresponding increases are 2.801, 2.783, and 2.838, while for h = 5W/m2K, the increases are 2.785, 2.766, and 2.566 for DVG, MVG, and FVG, respectively. Pressure drops by 84.94 %, 92.42 %, 91.39 %, and 89.44 % for smooth, DVG, MVG, and FVG channels. The thermal enhancement factor (TEF) increases up to 2.65 times for VGs at h = 25 W/m2K, with the fishtail VG showing the highest TEF values of 3.278, 3.011, and 2.614 for h = 25, 15, and 5 W/m2K. Additionally, the factors A3 (fishtail VG) and B1 (h = 5 W/m2K) demonstrate the highest signal-to-noise ratio values among the nine runs. The ANN model achieves R2 values of 1, 0.99994, and 0.99962, demonstrating high predictive accuracy. In conclusion, fishtail VG exhibits the best thermohydraulic performance among all the VGs studied in different ambient situations.