Feature Extraction from Turbulent Channel Flow of Moderate Reynolds Number via Composite DMD Analysis

In this contribution, we described a Dynamic Mode Decomposition (DMD) analysis of a turbulent channel flow database at a moderate friction Reynolds number Re τ ≈ 950. More specifically, a composite-based DMD analysis was conducted, employing hybrid snapshots assembled by skin friction C f ( t k ) and either instantaneous Reynolds stress ( u ′ v ′ ( x → ; t k ) ) or streamwise velocity fluctuation ( u ′ ( x → ; t k ) ) fields. The DMD modes thus obtained were sorted according to its relevance to the C f : less than 2% of the modes suffice to reconstruct accurately either the streamwise velocity or the Reynolds stress profiles near the wall. Furthermore, we aim to extend our preliminary work on the analysis of the turbulent database, by considering snapshots encompassing a larger spatial subdomain and covering a longer temporal span. However, this study involved data matrices significantly larger than that one, which the memory footprint of this problem exceeds a typical workstation. Accordingly, we have resorted to the parallel, memory distributed DMD algorithm as a reinforcement. With this enhanced composite DMD algorithm, flow features of moderate and even large turbulent channel problems could be identified and characterized.