Multi-resolution DMD characterization of non-stationary dynamics in the PRECCINSTA burner

This work assesses the capabilities of a multi-resolution implementation of the Dynamic Mode Decomposition ( mrDMD ) in the identification of coherent structures when non-stationary flow dynamics are present. Two different datasets obtained through Large Eddy Simulation ( LES ) of the PRECCINSTA burner flow have been analyzed. These datasets were computed at the equivalence ratios ϕ = 0 (inert flow, cold case) and ϕ = 0 . 75 (lean conditions), both at Reynolds number Re = 75 000. The results demonstrate the potential of mrDMD in recovering a time-frequency DMD representation of the flow dynamics when they are dominated by a small number of well-defined coherent structures ( e.g. , the nonreactive flow). The difficulties encountered when mrDMD is applied to strongly non-stationary dynamics (the reactive flow), are also illustrated. • mrDMD is an extension of the classic DMD that defines a recursive sequence of divisions of the temporal samples into halves, and then applies DMD independently to the resulting temporal blocks. This enhances the analysis capability of DMD by allowing the temporal evolution of the modes. • Multi-resolution Dynamic Mode Decomposition (mrDMD) is applied here to the analysis of the flow dynamics in two datasets, corresponding to LES of one nonreactive and one fully-premixed lean methane-air reactive flow in the swirl-stabilized PRECCINSTA burner. • The results demonstrate the potential of mrDMD in recovering a time-frequency DMD representation of the flow dynamics when they are dominated by a small number of well-defined coherent structures (e.g. the present nonreactive flow). The difficulties encountered when mrDMD is applied to strongly non-stationary dynamics, like the present reactive flow, are also illustrated.