For commercial transport vehicles, achieving maximum volume poses a challenge in optimizing aerodynamics, often resulting in bulky shapes. Studies reveal that aerodynamic drag constitutes approximately 80% of total drag at high speeds (130 km/h), compared to 50% at lower speeds (50 km/h). The constrained optimization space, coupled with regulatory constraints, renders modifying the geometry of commercial transport vehicles through geometric optimization or the addition of aerodynamic appendages extremely challenging, if not impossible. Consequently, there is potential for employing active flow control to minimize the aerodynamic drag of these vehicles.
In the experimental environment, optimal control laws for compressed air jets positioned at the sides and rear base of the model are identified using machine learning-based approaches, specifically Deep Reinforcement Learning (DRL). The objective is to achieve drag reduction through this method.
Planar and Stereo PIV setup