ABSTRACT
For uniform measure δ Gaussian filter, Part I derives the totally analytical aFNS theory closing rigorously space filtered Navier–Stokes (NS) partial differential equation (PDE) system absent a Reynolds number (Re) assumption. aFNS theory state variable is scaled O(1; δ2; δ3) via classic fluid mechanics perturbation theory which also identifies the O(δ2) elliptic PDE system. Filter penetration of domain bounding surfaces requires O(1) PDE system inclusion of boundary commutation error (BCE) integrals. For the O(1; δ2), PDE system to be bounded domain well-posed requires derivation of domain encompassing nonhomogeneous Dirichlet boundary conditions (DBC). Resolution of BCE and DBC requirements is theorized via O(δ4) approximate deconvolution (AD) Galerkin differential definition weak form algorithms. Amenable to any space-time discretization, detailed is aFNS theory insertion in the optimal Galerkin weak form CFD algorithm, finite element linear tensor product basis implemented. Coupled Galerkin CFD/BCE/DBC code a posteriori data reported herein validate theory resolution algorithms including accuracy/convergence assessments.
Acknowledgments
During the dissertation project, the first author served as HPC Graduate Assistant in the Joint Institute for Computational Sciences (JICS), a collaboration between the US DOE Oak Ridge National Laboratory and the University of Tennessee/Knoxville (UTK). Dissertation project coordination was through the UTK College of Engineering CFD Laboratory.