Enhanced-discretization selective stabilization procedure (EDSSP)

The enhanced-discretization selective stabilization procedure (EDSSP) provides a multiscale framework for applying numerical stabilization selectively at different scales. The EDSSP is based on the enhanced-discretization, multiscale function space concept underlying the enhanced- discretization successive update method (EDSUM). The EDSUM is a multi-level iteration method designed for computation of the flow behavior at small scales. It has a built-in mechanism for transferring flow information between the large and small scales in a fashion consistent with the discretizations resulting from the underlying stabilized formulations. This is accomplished without assuming that the small-scale trial or test functions vanish at the borders between the neighboring large-scale elements of the enhanced-discretization zones. This facilitates unrestricted movement of small-scale flow patterns from one large-scale element to another without any constraints at the border between the two elements. The enhanced-discretization concept underlying the EDSUM can also facilitate using different stabilizations for equations or unknowns corresponding to different scales. In this paper we propose a version of the EDSSP where the SUPG and PSPG stabilizations are used for unknowns corresponding to both the large and small scales but the discontinuity-capturing stabilizations are used for unknowns corresponding to only the small scales. We also propose a version where a linear discontinuity-capturing is used for the small-scale unknowns and a nonlinear discontinuity-capturing is used for the large-scale unknowns. We evaluate the performances of these versions of the EDSSP with test problems governed by the advection-diffusion equations.