Conf’luence Jörn Sesterhenn – Institut für Strömungsmechanik und Technische Akustik,Technische Universität Berlin, Germany
Mercredi 11 Septembre à 10 h 30 – Amphithéâtre Nougaro
Adjoint-based data assimilation adapts the boundary and initial conditions of a numerical simulation to exactly reproduce a specific experimental flow realization, see Fig. 1. The result is a full solution to the Navier-Stokes equations, ready for further analysis. Since the full state and smooth derivatives are available, it combines the advantages of measurement and simulation. Noisy data is filtered by the damping of the physical viscosity. The Navier-Stokes equations thus act as a filter on the data.
The specific aim of this work is to assimilate a subsonic jet flow at low Reynolds numbers (Re=7000). The latter constraint is necessary since the method is only feasible when a DNS is possible, unless the assimilation was based on an LES.
To generate the desired low Reynolds number flow, a vacuum chamber is used, see Fig. 2. Various flows at different Reynolds and Mach number combinations can be examined by using different nozzles, pressure, and pressure ratios. The velocity components of the flow are measured with PIV and will be use as the target for the adjoint-based data assimilation.