Flow control over a circular cylinder using vortex generators: Particle image velocimetry analysis and machine learning-based prediction of flow characteristics

Controlling the flow around circular cylinders is crucial to mitigate vortex-induced vibrations and prevent structural damage in a range of applications, such as marine and offshore engineering, tall buildings, long-span bridges, transport ships, and heat exchangers. In this study, we aimed to control the turbulent flow structure around a circular cylinder by placing vortex generators (VGs). We examined the flow structure using particle image velocimetry (PIV). This enabled quantitative data acquisition, intuitive flow visualization, and drag coefficient determination from PIV data. We developed artificial neural network (ANN) models that successfully predict both mean and instantaneous flow characteristics for different scenarios. Our findings show that using VGs extended the wake and increased vortex formation lengths, while reducing velocity fluctuations and the drag coefficient. A minimum drag coefficient of 0.718 was achieved with VGs oriented at α = 60° & β = 60°, reducing the drag by 35.3% compared to the bare cylinder. The drag coefficient exhibited a substantial inverse correlation with both wake and vortex formation lengths. This study is significant for controlling flow structures, providing detailed insights into the near-wake region, and highlighting the potential applications of machine learning in fluid dynamics. •Turbulent flow around a circular cylinder was controlled using vortex generators.•Flow structure was investigated with particle image velocimetry, providing quantitative data and intuitive flow visualization.•Vortex generators are highly effective on suppression of turbulence characteristics.•Drag coefficient exhibited a strong inverse correlation with wake length and vortex formation length.•The drag coefficient decreased by a 35.3% with VGs at ɑ = β = 60°.•Artificial neural network models accurately predicted both instantaneous and mean flow characteristics, highlighting machine learning's potential in fluid dynamics.