DYNAMIC MODE DECOMPOSITION OF PIV MEASUREMENTS FOR CYLINDER WAKE FLOW IN TURBULENT REGIME

Dynamic Mode Decomposition (DMD) is a new post-processing technique that can extract from snapshots dynamic informations relevant for the flow. Without explicit knowledge of the dynamical operator, the DMD algorithm determines eigenvalues and eigenvectors of an approximate linear model. DMD can be viewed as a non linear generalization of global stability classically used for a linearized system. This algorithm can be used to determine the eigen-elements of the Koopman operator, an infinite dimensional linear operator associated with the nonlinear system. The ability of DMD to extract dynamically relevant features of the flow has been tested on an experimental PIV dataset of a turbulent cylinder wake flow.