Constructal theory of aerodynamic design in NACA airfoils for search and rescue UAVs using integrated CFD and ANN approaches.

Constructal design of the NACA (National Advisory Committee for Aeronautics) airfoil was to reduce the drag force and increase the lift force for search and rescue (SAR) missions. This series of NACA airfoils is defined by a 4-digit NACA mptt, which indicates the camber, the location of maximum camber, and the thickness. Three degrees of freedom can be used to optimize aircraft airfoils using constructal theory. The CFD (Computational Fluid Dynamics) 2-D simulation was conducted in ANSYS FLUENT version (15), and the SST k-ω model turbulence equations were used to solve the incompressible Reynolds-averaged Navier-Stokes (RANS) equations. Simulations are performed using MATLAB's Neural Network ANN (Artificial Neural Network), which enables a robust surrogate-based algorithm for the design study. The highest Cl/Cd ratio indicates the trade-off between lift and drag. An airfoil with a high Cl/Cd (lift-to-drag coefficient) ratio produces more lift than drag, thereby enhancing aerodynamic performance through comparative analysis of three degrees of freedom (m, p, t) at low Reynolds numbers ([Formula: see text] and [Formula: see text]). NACA 4412 makes a better Cl/Cd ratio, 1.38 at 4[Formula: see text]-12[Formula: see text], than NACA 4418, about 26% higher at the same AoA 4[Formula: see text]-12[Formula: see text]. An increase in Reynolds number led to higher Cl/Cd ratios, 1.38 at angles of attack (AoAs) of [Formula: see text], compared to NACA 4418, about 26% higher at the same AoAs of [Formula: see text]. across all profiles, indicating higher aerodynamic efficiency. It is particularly applicable to search-and-rescue UAVs, which can operate at varying speeds and altitudes depending on mission requirements. The ANN model has determined the optimal AoA and Reynolds number that maximizes the Cl/Cd ratio of the NACA 4412 airfoil. The CFD results are validated against ANN results, which are based on experimental results used to train the artificial intelligence algorithm to predict ANN results from the present study.