Reducing base drag on road vehicles using pulsed jets optimized by hybrid genetic algorithms

TL;DR

This study addresses base-drag reduction for flat-backed road vehicles by optimizing four pulsed jets via an experiment-in-the-loop Hybrid Genetic Algorithm (HyGO) that includes an energy-penalty in the cost function. The method explores a seven-parameter actuation space with open-loop policies, combining global GA search and local Nelder–Mead refinement, and uses repeated measurements to manage experimental uncertainty. The optimized control achieves about $10\%$ drag reduction at $Re_W \approx 78{,}300$, primarily by strong, low-frequency bottom-jet actuation that suppresses the main wake shedding and by higher-frequency top/side jets modulating shear-layer instabilities; wake-field analyses show significant wake stabilization and pressure recovery on the lower base. The results demonstrate that model-free, experiment-driven optimization can discover robust, energy-efficient wake-control strategies that align with findings from closed-loop DRL approaches, highlighting potential for practical deployment in realistic vehicle geometries.

Abstract

Aerodynamic drag on flat-backed vehicles like vans and trucks is dominated by a low-pressure wake, whose control is critical for reducing fuel consumption. This paper presents an experimental study at $Re_W\approx 78,300$ on active flow control using four pulsed jets at the rear edges of a bluff body model. A hybrid genetic algorithm, combining a global search with a local gradient-based optimizer, was used to determine the optimal jet actuation parameters in an experiment-in-the-loop setup. The cost function was designed to achieve a net energy saving by simultaneously minimizing aerodynamic drag and penalizing the actuation's energy consumption. The optimization campaign successfully identified a control strategy that yields a drag reduction of approximately 10%. The optimal control law features a strong, low-frequency actuation from the bottom jet, which targets the main vortex shedding, while the top and lateral jets address higher-frequency, less energetic phenomena. Particle Image Velocimetry analysis reveals a significant upward shift and stabilization of the wake, leading to substantial pressure recovery on the model's lower base. Ultimately, this work demonstrates that a model-free optimization approach can successfully identify non-intuitive, multi-faceted actuation strategies that yield significant and energetically efficient drag reduction.

Figures (9)

• Figure 1: Schematic of the experimental setup. (a) Side view of the wind tunnel test section showing the bluff body model. (b) Rear view of the model's base, indicating the locations of the 31 static pressure taps (blue circles ) and the four pairs of pulsed jet actuators (). (c) Diagram of the auxiliary systems, including the pneumatic circuit for actuation, the force balance and load cell for drag measurements, and the pressure data acquisition system.
• Figure 2: Schematic of the PIV planes, illustrating the field of view $-0.03\leq|x|/W\leq\ 1.82$ and $-0.112\leq|z|/W\leq 1.433$ and the lateral positioning. A total of nine planes were acquired symmetrically distributed about the base's centerline, which include two planes aligned with the model's side walls ($|y|/W=0.5$), two aligned with the actuators' ($|y|/W=0.43$), four positioned in the vertical lines defined by the pressure probes ($|y|/W=0.29$ and $|y|/W=0.15$), and one plane in the centerline ($|y|/W=0$).
• Figure 3: Evolution of the optimization process across generations, showing: (a) the total cost, $J$; (b) the drag cost, $J_a$; and (c) the penalization, $J_b$. Within each generation, individual solutions are represented by circular markers, which are sorted and coloured according to their total cost $J$ in ascending order from left to right. The yellow star denotes the best-performing individual in the optimization, $\min(J)$, placed within the generation it appears.
• Figure 4: Analysis of the optimization, illustrating the relationship between aerodynamic performance ($J_a$) and actuation cost ($J_b$). (a) Distribution of all evaluated individuals in the objective space, plotting drag cost, $J_a$, against penalization, $J_b$. The data points are colored by their total cost, $J$, and the dashed lines () represent iso-contours of constant $J$. The solid black line () indicates the non-dominated, or Pareto, front, with the yellow star marking the location of the overall fittest individual. (b) Evolution of the Pareto front across successive generations, indicated by color. The stars () denote the best-performing individual discovered up to that respective generation.
• Figure 5: Time-averaged ($\overline{p}$) and fluctuating ($p'$) base pressure contours for four characteristic cases. From left to right, the columns correspond to: (i) the non-actuated baseline flow; (ii) continuous steady-jet actuation; (iii) the optimal individual achieving the minimum total cost, $\min(J)$; and (iv) the individual achieving the maximum drag reduction, $\min(J_a)$. Each image limits represent entire base surface.