Optimisation of drag coefficient of a car with integrated canards

The objective of this study is to investigate the impact of canards geometric design on the airflow field at the rear of a sedan, with the aim of reducing the drag coefficient of the vehicle. Four key design variables were analysed: canard angle (θ), length (L1), radius (R), and thickness (H). Using these variables, Latin hypercube sampling generated 50 DOE points, which were simulated in ANSYS Fluent® to calculate the drag coefficient (CD). A comprehensive performance comparison was performed. Ultimately, a backpropagation neural network (BPNN) coupled with a genetic algorithm (GA) was adopted as the optimal method for canard design optimisation. A random forest model identified θ as the most influential factor on CD. The optimised design achieved a 21% reduction in drag, minimised the rear vortex region, and significantly accelerated the optimisation process.