This study aims to explore the impact of surface modifications on a circular cylinder with ridges of diverse configurations. Dynamic mode decomposition (DMD) is used for flow field analysis and deep learning gated recurrent unit (GRU) framework is used to predict variations in surface pressure.
The research entails wind tunnel testing of a cylinder with ridges of different diameters and angular position in a flow field of Re = 1.01 × 105 and the surface pressure is measured by a Scanivalve pressure scanner. DMD is used to scrutinize pressure field data, elucidating significant dynamic modes and their influence on aerodynamic characteristics. The GRU model is trained with an experimental surface pressure data set, 80% for training, 15% for testing and 5% for validation phase to predict the surface pressure distribution of the model.
The study inferred that the ridge ratio and angular position of the ridge on a cylindrical surface significantly influence the drag coefficient (Cd), while the DMD reveals alterations in dynamic mode frequencies and amplitudes; In addition, the GRU model accurately predicts time-dependent aerodynamic characteristics, assessed through K-fold cross-validation, demonstrating superior predictive accuracy with minimal root mean squared error.
This study uses DMD for flow field analysis and the GRU model to predict the surface pressure distribution, presenting a more efficient and time-saving alternative to conventional methodologies.
