flux_functor.h File Reference

Convective flux functor. More...

#include <Kokkos_Core.hpp>
#include "common.h"
#include "boundary.h"
#include "riemann_solver.h"

Go to the source code of this file.

Classes
struct	UpwindSensorFunctor
	Upwind fraction of each cell for the hybrid convective flux: 1 where the Ducros sensor restricted to compressions (Ducros et al. 1999; Bhagatwala & Lele 2009), theta = (div u)^2 / ((div u)^2 + |omega|^2) for div u < 0, exceeds the threshold (shocks), else the floor. gradients(c, 1 + k, i) = d u_k / d x_i. More...
struct	ConvectiveFluxFunctor< T_riemann_solver >
	Integrates the convective flux over every face into face_flux, the rate of change it causes in cells_of_face(:, 0) (cells_of_face(:, 1) receives its negative). Boundary faces use the boundary ghost state as the right state. Characteristic boundaries take the low-Mach correction like interior faces: without it, the different dissipation of the boundary cell's two sides holds the outlet pressure of a steady shear flow away from its target. More...
struct	FaceFluxSumFunctor
	Sums the face fluxes of each cell into its RHS in faces_of_cell order, so that the result does not depend on thread scheduling. More...

Functions
KOKKOS_INLINE_FUNCTION void	low_mach_correction (rtype *W_l, rtype *W_r, const rtype gamma, const rtype M_cut)
	Low-Mach correction of Thornber et al. (J. Comput. Phys. 227, 2008): scales the jump of the reconstructed velocity across a face by z = min(1, max(M_L, M_R, M_cut)), keeping its mean. Upwind fluxes otherwise damp velocity jumps at the sound speed, a dissipation that does not vanish as M -> 0; with the correction it scales with the flow speed. The cutoff M_cut keeps some acoustic damping in gas nearly at rest, like the cutoff Mach number of preconditioned all-speed schemes (Weiss & Smith, AIAA J. 33(11), 1995). Supersonic faces (z = 1) are untouched.
KOKKOS_INLINE_FUNCTION const rtype *	quad_normal (const Kokkos::View< rtype *** > &quad_normals, const uint32_t f, const uint8_t q, const rtype *n_face, rtype *n_q)
	Unit normal at quadrature point q of face f: the point's own on curved meshes (quad_normals, (face, q, dim)), else the face's n_face.
KOKKOS_INLINE_FUNCTION rtype	face_upwind_fraction (const Kokkos::View< rtype * > &upwind, const BoundaryData &boundaries, const int32_t c0, const int32_t c1, const uint32_t i_face)
	Upwind fraction of a face of the hybrid convective flux: the larger of its cells' (boundary faces other than walls and symmetry planes, whose mirror states make the central flux exact, take the Riemann flux).

Detailed Description

Convective flux functor.

Author

Matthew Bonanni (mbona.nosp@m.nni0.nosp@m.01@gm.nosp@m.ail..nosp@m.com)

Version

0.2

Date

2024-11-26

Copyright

Copyright (c) 2024 Matthew Bonanni

Function Documentation

face_upwind_fraction()

KOKKOS_INLINE_FUNCTION rtype face_upwind_fraction	(	const Kokkos::View< rtype * > &	upwind,
		const BoundaryData &	boundaries,
		const int32_t	c0,
		const int32_t	c1,
		const uint32_t	i_face
	)

Upwind fraction of a face of the hybrid convective flux: the larger of its cells' (boundary faces other than walls and symmetry planes, whose mirror states make the central flux exact, take the Riemann flux).

low_mach_correction()

KOKKOS_INLINE_FUNCTION void low_mach_correction	(	rtype *	W_l,
		rtype *	W_r,
		const rtype	gamma,
		const rtype	M_cut
	)

Low-Mach correction of Thornber et al. (J. Comput. Phys. 227, 2008): scales the jump of the reconstructed velocity across a face by z = min(1, max(M_L, M_R, M_cut)), keeping its mean. Upwind fluxes otherwise damp velocity jumps at the sound speed, a dissipation that does not vanish as M -> 0; with the correction it scales with the flow speed. The cutoff M_cut keeps some acoustic damping in gas nearly at rest, like the cutoff Mach number of preconditioned all-speed schemes (Weiss & Smith, AIAA J. 33(11), 1995). Supersonic faces (z = 1) are untouched.

quad_normal()

KOKKOS_INLINE_FUNCTION const rtype * quad_normal	(	const Kokkos::View< rtype *** > &	quad_normals,
		const uint32_t	f,
		const uint8_t	q,
		const rtype *	n_face,
		rtype *	n_q
	)

Unit normal at quadrature point q of face f: the point's own on curved meshes (quad_normals, (face, q, dim)), else the face's n_face.