Dynamic mode decomposition of a direct numerical simulation of a turbulent premixed planar jet flame: convergence of the modes

Abstract

Dynamic Mode Decomposition (DMD) is a technique that enables investigation of unsteady and dynamic phenomena by decomposing data into coherent modes with corresponding growth rates and oscillatory frequencies. Because the method identifies structures unbiased by energy, it is particularly well suited to exploring dynamic processes having phenomena that span disparate temporal and spatial scales. In turbulent combustion, DMD has been previously applied to the analysis of narrowband phenomena such as combustion instabilities utilising both experimental and computational data. In this work, DMD is used as a tool to analyse broadband turbulent combustion phenomena from a three-dimensional direct numerical simulation of a low Mach number spatially-evolving turbulent planar premixed hydrogen/air jet flame. The focus of this investigation is on defining the metric of convergence of the DMD modes for broadband phenomena when both the temporal resolution and number of data snapshots can be varied independently. The residual is identified as an effective, even if imperfect, metric for judging convergence of the DMD modes. Other metrics – specifically, the convergence of the mode eigenvalues and the decay of the amplitudes of the modes – fail to capture convergence of the modes independently but do complete the information needed to evaluate the quality of the DMD analysis.