PANS turbulence model: investigation of computational and physical closure issues in flow past a circular cylinder

Abstract

Partially Averaged Navier-Stokes (PANS) turbulence method provides a closuremodel for any degree of velocity field altering - ranging from completely resolved Di-rect Numerical Simulation (DNS) to completely averaged Reynolds Averaged Navier-Stokes (RANS) method. Preliminary investigations of PANS show promising re-sults but there still exist computational and physical issues that must be addressed.This study investigates the performance of the PANS method for ow past acylinder at a Reynolds number of 140,000. The cylinder ow is a benchmark owproblem for which there are signicant experimental results available for validation ofPANS. First, we examine if RANS convergence criteria and discretization schemes -which are meant for robust, nearly steady-state calculations - are adequate for PANS,which is inherently unsteady and may contain delicate ow features. For the range offk values tested here, it is determined that the standard RANS residual value and the2nd order spatial discretization scheme are appropriate for PANS. The physical clo-sure investigations begin with the validation of turbulent transport models: the ZeroTransport Model, the Maximum Transport Model and the Boundary Layer Trans-port Model. The implementation of the PANS ku-ey!u model is also performed andcompared against the standard PANS ku-ey model. All these studies yield interest-ing insights into the PANS models. This study concludes with an investigation of alow Reynolds number correction for the PANS ku-wu!u model which yields excellent improvement.