Granger causality in wall-bounded turbulence

1st Multiflow Summer Workshop

Tissot, Gilles^1 ; Lozano-Durán, Adrian^2 ; Cordier, Laurent^1 ; Jiménez, Javier^2 ; Noack, Bernd R.^1

Institut PPRIME, CEAT, 43 route de l'aérodrome, 86036 Poitiers, France^1

School of Aeronautics, Universidad Politécnica, Madrid 28040, Spain^2

关键词: Dynamic mode decompositions;    Energy re distributions;    Logarithmic layers;    Physical mechanism;    Predictive modeling;    Turbulent channel flows;    Wall bounded turbulence;    Wall-bounded turbulent flows;   

Others  :  https://iopscience.iop.org/article/10.1088/1742-6596/506/1/012006/pdf
DOI  :  10.1088/1742-6596/506/1/012006

来源: IOP

【 摘 要 】

Granger causality is based on the idea that if a variable helps to predict another one, then they are probably involved in a causality relationship. This technique is based on the identification of a predictive model for causality detection. The aim of this paper is to use Granger causality to study the dynamics and the energy redistribution between scales and components in wall-bounded turbulent flows. In order to apply it on flows, Granger causality is generalized for snapshot-based observations of large size using linear-model identification methods coming from model reduction. Optimized DMD, a variant of the Dynamic Mode Decomposition, is considered for building a linear model based on snapshots. This method is used to link physical events and extract physical mechanisms associated to the bursting process in the logarithmic layer of a turbulent channel flow.