Turbulence model upgrades for swirling flows

The ability to accurately model axisymmetric, turbulent swirling jet flows over a variety of inflow conditions is evaluated. The deficiency of the standard k-epsilon turbulence model in predicting mixing rates in flows with streamline curvature is well known. A relatively straightforward modification to this model is made based on a local value of the flux Richardson number which accounts for the azimuthal velocity and its variation. To evaluate the effectiveness of this modification two different experimental data sets are used to compare the computational results against. All calculations were performed using the structured. density based, CRAFT CFD (R) code utilizing a preconditioning methodology. Both cases have initial swirl distributions that are equivalent to a solid-body rotation profile, and have swirl numbers that are low enough to remain below the vortex breakdown regime. They also have non-swirling jet data available for the same geometries and operating conditions which allows the increased jet mixing rate of swirling jets over purely axial jets to be confirmed. All calculations showed a significant improvement of centerline velocity decay as well as downstream radial velocity profiles when the Richardson number correction was activated. For the case with turbulence data, the centerline decay of turbulent kinetic energy was also much improved. An important result that was discovered was the extreme sensitivity of the downstream evolution of the jet to the specification of the initial k and epsilon profiles, highlighting the critical need for a comprehensive experimental characterization of all flow properties at the jet exit.