Space-time reconstruction of finely resolved velocities of turbulent flows from low resolution measurements etc15 Tracking Number 50Presentation: Session: Vortex Dynamics 2 Room: Room B Session start: 15:00 Tue 25 Aug 2015 Linh Van Nguyen nvlpfiev@gmail.com Affifliation: Univ Lille Nord de France, F-59000 Lille, France; Laboratoire de Mecanique de Lille (LML), CNRS-UMR 8107, F-59650 Villeneuve d'Ascq, France Jean-Philippe Laval jean-philippe.laval@univ-lille1.fr Affifliation: Laboratoire de Mecanique de Lille (LML), CNRS-UMR 8107, F-59650 Villeneuve d'Ascq, France Pierre Chainais pierre.chainais@ec-lille.fr Affifliation: Ecole Centrale Lille, CRIStAL-CNRS UMR 9189, INRIA SequeL, F-59650 Villeneuve d'Ascq, France Topics: - Vortex dynamics and structure formation, - Wall bounded flows Abstract: A model is developed to reconstruct the High-Temporal-High-Spatial (HTHS) resolution velocities of a wall bounded turbulent flow combining two sources of measured data: the Low-Temporal-High-Spatial (LTHS) and the High-Temporal-Low-Spatial (HTLS) resolution measurements. Direct Numerical Simulation (DNS) database of a turbulent channel flow is used. LTHS and HTLS are subsampled (in time or space) from the fully resolved data. A fusion model is constructed in Bayesian framework using a Maximum A Posteriori (MAP) estimation. By maximizing the posterior probability of desired HTHS conditional on LTHS and HTLS, the model estimates the most probable fields knowing the measurements. Reconstructed velocities are compared to the reference to qualify the fusion model and to those by conventional Linear Stochastic Estimation (LSE) and interpolation methods to demonstrate its robustness. |