Unsteady Simulation of Synthetic Jet in a Crossflow

Abstract

The purpose is to address some issues concerning the computation of the interaction of a synthetic jet with a turbulent boundary layer. The main question arising is, What accuracy can be expected according to the grid resolution and the turbulence model? With a view of comparison, large-eddy simulation (LES) and unsteady Reynolds-averaged Navier-Stokes (URANS) simulations on three grids are presented. The computational configuration corresponds to the case number 2 of the CFD Validation Workshop of Synthetic Jets (CFDVAL2004) held in March 2004 at the NASA Langley Research Center. The incoming flow is a canonical turbulent boundary layer with a momentum-thickness-based Reynolds number Re θ equal to 4275. The synthetic jet has a Strouhal number based on the jet diameter and the jet velocity of 0.019. The ratio of the jet maximum velocity to the crossflow one is 1.45. The ratio of the jet diameter to the boundary-layer thickness is one-third. The whole cavity is computed, and an alternating blowing/suction condition is imposed on its lower surface to model the diaphragm displacement. The effects of mesh size, time step, turbulence modeling, and boundary conditions have been carefully documented. The variation of these numerical parameters has only a moderate influence on the computation/experiment agreement, which remains very satisfactory in all cases. For example, even the URANS computation on the coarsest grid with only 220,000 cells is able to reproduce the synthetic jet dynamics and provides good results for the mean and phase-averaged velocity profiles. Nevertheless, in this case turbulence stresses are not computed with sufficient accuracy, and a LES with realistic inflow boundary conditions on a fine mesh is necessary to recover the right stresses behavior.