Pattern recognition by Recurrence Analysis in the flow around a bluff body

The identification of scales, patterns and structures in a turbulent flow, starting from a high-fidelity experimental or numerical database, is an aspect of primary importance for the understanding of the dynamics of transition to turbulence and the energy cascade. In this frame, the present study represents a preliminary effort to evaluate a time series analysis tool, the so-called Recurrence Analysis (RA), for the identification of the dominant features of a relatively complex flow. The test case is represented by the turbulent flow around a rectangular cylinder with a chord-to-thickness ratio $C/D = 5$, for a Reynolds number value $Re=3000$. The problem at hand has already been tackled by means of a well-resolved Direct Numerical Simulation, whose results highlighted the presence of a multiplicity of scales and structures. Due to this interesting combination of features, the case appears as a promising benchmark for the development of a novel tool for pattern recognition. To this aim, the system dynamics are condensed to pointwise observations at various abscissas along the flow. The analysis is aimed at verifying whether or not it is possible to isolate regular structures that: i) represent characteristic features of the flow and ii) can be used to distinguish the various phase of the shear-wake development along the flow. Results, cast in the form of Recurrence Plots (RP), reveal that the main scales of the flow, reflected in the sampled time series, are associated with well-defined recurrent patterns. This encouraging outcome leaves room for further utilization of the technique for the description of transitional and turbulent flows in thermo-fluid dynamics.