Inspecting Interactions of Discretization, Filter Formulation, and Stabilization in LES: Lessons from the Taylor-Green Vortex

摘要

The Taylor-Green Vortex is used as a testbed for observing the interplaying roles of discretization, modeling, and filter formulation in the algorithmic design of an LES method. Specifically, scale-similarity type closures are considered and inspected, with the Approximate Deconvolution Method demonstrating inherent stability properties relative to the Bardina-type Scale-Similarity model. Reference solutions derived from the filtered DNS computations are used to make both a priori and a posteriori assessments of the LES method. In addition, stabilization in the form of filter-based artificial dissipation is briefly presented as a viable way of enforcing scheme robustness while maintaining suitable accuracy relative to the reference LES field.