Abstract: PANS Modeling for Variable-Density and Materials Mixing Problems
: Variable-density and materials mixing problems are characterized by their complex physics, which often features regions of laminar, transitional, and turbulent flow and numerous instabilities and coherent structures. These phenomena pose challenges to the modeling and simulation of such a class of transient flows. Whereas direct numerical simulation (DNS) and large-eddy simulation (LES) methods are accurate but excessively expensive for practical problems, the caveats of Reynolds-Averaged Navier-Stokes equations closures simulating transient flows limit the accuracy of their computations. Thus, developing efficient (accuracy vs. cost) formulations for variable-density and materials-mixing problems is imperative. For this reason, we recently extended the framework of the partially-averaged Navier-Stokes equations (PANS) bridging turbulence model to variable-density and materials mixing problems. This class of turbulence models aims to efficiently predict complex problems by only resolving the flow physics not amenable to modeling. The remaining phenomena can be represented through an appropriate parameterization. This presentation summarizes our PANS work on variable-density and materials mixing problems, focusing on mixing and oceanic flows. The results demonstrate that PANS can match the accuracy of LES and DNS methods at a fraction of their computational cost. The potential of the model to prevent error canceling cases is also illustrated.