|
MFC
Exascale flow solver
|
|
MFC
Exascale flow solver
|
Shared Riemann-solver module state and the per-sweep setup, state-buffer population, viscous source flux, and finalization helpers. More...
Functions/Subroutines | |
| real(wp) function | f_elastic_signal_speed (c, g, tau, rho) |
| Elastic signal speed of Rodriguez et al. JCP (2019): the acoustic speed stiffened by the shear modulus and the normal elastic stress. Callers subtract it for the left-going wave and add it for the right-going one. | |
| real(wp) function | f_low_mach_zcoef (vel_l_rms, vel_r_rms, c_l, c_r) |
| Low-Mach parameter of Thornber et al. JCP (2008): the larger of the two face Mach numbers, capped at one so the correction switches itself off once the flow is no longer low speed. | |
| real(wp) function | f_low_mach_pcorr_hll (vel_l_rms, vel_r_rms, c_l, c_r, rho_l, rho_r, s_m, s_p) |
| Low-Mach pressure correction added to the HLL and Lax-Friedrichs fluxes, which restores the pressure jump that the dissipation of those fluxes over-damps at low Mach number. Zero unless low_Mach == 1. | |
| real(wp) function | f_low_mach_pcorr_hllc (vel_l_rms, vel_r_rms, c_l, c_r, rho_l, rho_r, s_l, s_r, vel_l_norm, vel_r_norm) |
| The same correction for the HLLC flux, where the star state supplies the pressure jump directly and the correction scales with the mass flux through the acoustic waves instead. Zero unless low_Mach == 1. | |
| subroutine | s_apply_low_mach_velocity (vel_l_rms, vel_r_rms, c_l, c_r, vel_l_norm, vel_r_norm) |
| The alternative low-Mach treatment of Thornber et al. JCP (2008) selected by low_Mach == 2: rather than correct the flux, blend the wave-normal velocities towards their mean before the wave speeds are computed, which is why this mutates its arguments and must be called ahead of s_L, s_R and s_S. The tangential velocities and vel_L/R_rms are deliberately left untouched. | |
| subroutine | s_compute_average_state (rho_l, rho_r, vel_l, vel_r, h_l, h_r, gamma_l, gamma_r, qv_l, qv_r, rho_avg, vel_avg_rms, h_avg, gamma_avg, qv_avg) |
| Interface-averaged state that the pressure-based wave-speed estimate reads. avg_state selects between the density-weighted Roe average, which costs eight square roots per face, and the plain arithmetic mean; unlike the other solver switches this one is not implied by the call site, so the dispatch stays here. | |
| subroutine | s_compute_chemistry_average_state (rho_l, rho_r, t_l, t_r, ys_l, ys_r, r_species, h_il, h_ir, cp_il, cp_ir, vel_avg_rms, gamma_avg, c_sum_yi_phi) |
| Roe-averaged reacting-mixture quantities: replaces gamma_avg with the mixture Cp/Cv and builds the c_sum_Yi_Phi term s_compute_speed_of_sound_avg needs. vel_avg_rms must be the full squared magnitude - its Phi_avg and vel_sum terms cancel to leave the Roe sound speed, and only do so for the full magnitude. | |
| subroutine | s_compute_viscous_source_flux (vell_vf, dvell_dx_vf, dvell_dy_vf, dvell_dz_vf, velr_vf, dvelr_dx_vf, dvelr_dy_vf, dvelr_dz_vf, flux_src_vf, q_prim_vf, norm_dir, ix, iy, iz) |
| Dispatch to the subroutines that are utilized to compute the viscous source fluxes for either Cartesian or cylindrical geometries. For more information please refer to: 1) s_compute_cartesian_viscous_source_flux 2) s_compute_cylindrical_viscous_source_flux. | |
| subroutine | s_populate_riemann_states_variables_buffers (ql_prim_rsx_vf, dql_prim_dx_vf, dql_prim_dy_vf, dql_prim_dz_vf, qr_prim_rsx_vf, dqr_prim_dx_vf, dqr_prim_dy_vf, dqr_prim_dz_vf, norm_dir, ix, iy, iz) |
| Populate the left and right Riemann state variable buffers based on boundary conditions. | |
| subroutine | s_initialize_riemann_solver (flux_src_vf, norm_dir) |
| Set up the chosen Riemann solver algorithm for the current direction. | |
| subroutine | s_compute_cylindrical_viscous_source_flux (vell_vf, dvell_dx_vf, dvell_dy_vf, dvell_dz_vf, velr_vf, dvelr_dx_vf, dvelr_dy_vf, dvelr_dz_vf, flux_src_vf, q_prim_vf, norm_dir, ix, iy, iz) |
| Compute cylindrical viscous source flux contributions for momentum and energy. | |
| subroutine | s_compute_cartesian_viscous_source_flux (dvell_dx_vf, dvell_dy_vf, dvell_dz_vf, dvelr_dx_vf, dvelr_dy_vf, dvelr_dz_vf, flux_src_vf, q_prim_vf, norm_dir) |
| Compute Cartesian viscous source flux contributions for momentum and energy. | |
| subroutine | s_calculate_shear_stress_tensor (vel_grad_avg, re_shear, divergence_v, tau_shear_out) |
| Compute shear stress tensor components. | |
| subroutine | s_calculate_bulk_stress_tensor (re_bulk, divergence_v, tau_bulk_out) |
| Compute bulk stress tensor components (diagonal only). | |
| subroutine | s_compute_interface_reynolds (alpha_k, re_k, re_size_loc1, re_size_loc2) |
| Compute the shear and volume Reynolds numbers of one Riemann state by inverse-weighting the fluid Reynolds numbers with the volume fractions. | |
| subroutine | s_compute_hypoelastic_interface_energy (nf, alpha_l, alpha_r, damage_l, damage_r, tau_e_l, tau_e_r, g_l, g_r, e_l, e_r) |
| Accumulate the hypoelastic stress contribution to the energies of the left and right Riemann states: mix the shear modulus over the fluids, add the elastic energy of each stress component (doubled for the shear components) on each side whose mixture modulus is non-negligible, then scale the returned moduli by the continuum damage state when damage is modeled (energy uses the undamaged modulus; the damaged moduli feed the callers' wave speeds). The elastic shear stresses are loaded from the state buffers by the caller, which reuses them for the stress fluxes and elastic wave speeds. The G > verysmall per-side gate is a deliberate maintainer ruling that replaces HLL's former hard-coded G > 1000 stability floor, retiring its "TODO take out if statement if stable without". | |
| real(wp) function | f_compute_hllc_star_momentum_flux (rho_l, rho_r, vel_l_norm, vel_r_norm, s_m, s_p, s_s, xi_l, xi_r, xi_m, xi_p, dir_flg_norm) |
| Compute the advective part of the HLLC star-state momentum flux in the wave-normal direction (pressure excluded), used to assemble the geometrical source flux of the cylindrical and azimuthal sweeps. | |
| subroutine | s_finalize_riemann_solver (flux_vf, flux_src_vf, flux_gsrc_vf, norm_dir) |
| Reshape and copy the Riemann-solver flux buffers back to the physical-space output arrays for the selected sweep direction, finalizing the Riemann solve. Two variants are emitted from one template so the shared unpermute logic cannot drift apart: the plain routine also copies the advection flux_src set and the grid_geometry==3 z-sweep geometric source flux, while the _hatR variant unpermutes the hat_R-anchored flux_hatR_rs* set of the fused dual-pass HLLD solve (called between the two RHS assemblies) and is a strict subset: flux_src is anchor-independent (already finalized with the hat_L set) and its geometric source flux only exists for the axisymmetric y-sweep. | |
| subroutine | s_finalize_riemann_solver_hatr (flux_vf, flux_gsrc_vf, norm_dir) |
Variables | |
| real(wp), dimension(:,:,:,:), allocatable | vel_src_rsx_vf |
| real(wp), dimension(:,:,:,:), allocatable | mom_sp_rsx_vf |
| real(wp), dimension(:,:,:,:), allocatable | re_avg_rsx_vf |
| real(wp), dimension(:), allocatable | gs_rs |
| real(wp), dimension(:,:), allocatable | res_gs |
| real(wp), dimension(:,:,:,:), allocatable | flux_rsx_vf |
| The cell-boundary values of the fluxes (src - source) that are computed through the chosen Riemann problem solver, and the direct evaluation of source terms, by using the left and right states given in qK_prim_rs_vf, dqK_prim_ds_vf where ds = dx, dy or dz. | |
| real(wp), dimension(:,:,:,:), allocatable | flux_src_rsx_vf |
| real(wp), dimension(:,:,:,:), allocatable | flux_gsrc_rsx_vf |
| The cell-boundary values of the geometrical source flux that are computed through the chosen Riemann problem solver by using the left and right states given in qK_prim_rs_vf. Currently 2D axisymmetric for inviscid only. | |
| real(wp), dimension(:,:,:,:), allocatable | nc_iface_vel_rsx_vf |
| real(wp), dimension(:,:,:,:), allocatable | flux_hatr_rsx_vf |
| Dual-pass HLLD second flux set: the hat_R-anchored fluxes (and, for axisymmetric runs, the hat_R interface velocities) written by the same fused solve that fills flux_rsx / nc_iface_vel_rsx with the hat_L-anchored values. Allocated only when hypo_nc_mode_dual_pass. | |
| real(wp), dimension(:,:,:,:), allocatable | nc_iface_vel_hatr_rsx_vf |
| real(wp), dimension(:,:,:,:), allocatable | flux_gsrc_hatr_rsx_vf |
Indical bounds in the s1-, s2- and s3-directions | |
| type(int_bounds_info) | is1 |
| type(int_bounds_info) | is2 |
| type(int_bounds_info) | is3 |
| type(int_bounds_info) | isx |
| type(int_bounds_info) | isy |
| type(int_bounds_info) | isz |
Shared Riemann-solver module state and the per-sweep setup, state-buffer population, viscous source flux, and finalization helpers.
| real(wp) function m_riemann_state::f_compute_hllc_star_momentum_flux | ( | real(wp), intent(in) | rho_l, |
| real(wp), intent(in) | rho_r, | ||
| real(wp), intent(in) | vel_l_norm, | ||
| real(wp), intent(in) | vel_r_norm, | ||
| real(wp), intent(in) | s_m, | ||
| real(wp), intent(in) | s_p, | ||
| real(wp), intent(in) | s_s, | ||
| real(wp), intent(in) | xi_l, | ||
| real(wp), intent(in) | xi_r, | ||
| real(wp), intent(in) | xi_m, | ||
| real(wp), intent(in) | xi_p, | ||
| real(wp), intent(in) | dir_flg_norm ) |
Compute the advective part of the HLLC star-state momentum flux in the wave-normal direction (pressure excluded), used to assemble the geometrical source flux of the cylindrical and azimuthal sweeps.
| [in] | rho_r | Left and right densities |
| [in] | vel_r_norm | Left and right wave-normal velocities |
| [in] | s_s | Clamped left/right and contact wave speeds |
| [in] | xi_p | Star-state compression factors and upwind selectors |
| [in] | dir_flg_norm | Direction flag of the wave-normal direction |
Definition at line 3249 of file m_riemann_state.fpp.f90.
| real(wp) function m_riemann_state::f_elastic_signal_speed | ( | real(wp), intent(in) | c, |
| real(wp), intent(in) | g, | ||
| real(wp), intent(in) | tau, | ||
| real(wp), intent(in) | rho ) |
Elastic signal speed of Rodriguez et al. JCP (2019): the acoustic speed stiffened by the shear modulus and the normal elastic stress. Callers subtract it for the left-going wave and add it for the right-going one.
Definition at line 535 of file m_riemann_state.fpp.f90.
| real(wp) function m_riemann_state::f_low_mach_pcorr_hll | ( | real(wp), intent(in) | vel_l_rms, |
| real(wp), intent(in) | vel_r_rms, | ||
| real(wp), intent(in) | c_l, | ||
| real(wp), intent(in) | c_r, | ||
| real(wp), intent(in) | rho_l, | ||
| real(wp), intent(in) | rho_r, | ||
| real(wp), intent(in) | s_m, | ||
| real(wp), intent(in) | s_p ) |
Low-Mach pressure correction added to the HLL and Lax-Friedrichs fluxes, which restores the pressure jump that the dissipation of those fluxes over-damps at low Mach number. Zero unless low_Mach == 1.
| [in] | vel_r_rms | Left and right squared velocity magnitudes |
| [in] | c_r | Left and right sound speeds |
| [in] | rho_r | Left and right densities |
| [in] | s_p | Clamped left and right wave speeds |
Definition at line 630 of file m_riemann_state.fpp.f90.
| real(wp) function m_riemann_state::f_low_mach_pcorr_hllc | ( | real(wp), intent(in) | vel_l_rms, |
| real(wp), intent(in) | vel_r_rms, | ||
| real(wp), intent(in) | c_l, | ||
| real(wp), intent(in) | c_r, | ||
| real(wp), intent(in) | rho_l, | ||
| real(wp), intent(in) | rho_r, | ||
| real(wp), intent(in) | s_l, | ||
| real(wp), intent(in) | s_r, | ||
| real(wp), intent(in) | vel_l_norm, | ||
| real(wp), intent(in) | vel_r_norm ) |
The same correction for the HLLC flux, where the star state supplies the pressure jump directly and the correction scales with the mass flux through the acoustic waves instead. Zero unless low_Mach == 1.
| [in] | vel_r_rms | Left and right squared velocity magnitudes |
| [in] | c_r | Left and right sound speeds |
| [in] | rho_r | Left and right densities |
| [in] | s_r | Left and right wave speeds |
| [in] | vel_r_norm | Left and right wave-normal velocities |
Definition at line 683 of file m_riemann_state.fpp.f90.
| real(wp) function m_riemann_state::f_low_mach_zcoef | ( | real(wp), intent(in) | vel_l_rms, |
| real(wp), intent(in) | vel_r_rms, | ||
| real(wp), intent(in) | c_l, | ||
| real(wp), intent(in) | c_r ) |
Low-Mach parameter of Thornber et al. JCP (2008): the larger of the two face Mach numbers, capped at one so the correction switches itself off once the flow is no longer low speed.
| [in] | vel_r_rms | Left and right squared velocity magnitudes |
| [in] | c_r | Left and right sound speeds |
Definition at line 582 of file m_riemann_state.fpp.f90.
| subroutine m_riemann_state::s_apply_low_mach_velocity | ( | real(wp), intent(in) | vel_l_rms, |
| real(wp), intent(in) | vel_r_rms, | ||
| real(wp), intent(in) | c_l, | ||
| real(wp), intent(in) | c_r, | ||
| real(wp), intent(inout) | vel_l_norm, | ||
| real(wp), intent(inout) | vel_r_norm ) |
The alternative low-Mach treatment of Thornber et al. JCP (2008) selected by low_Mach == 2: rather than correct the flux, blend the wave-normal velocities towards their mean before the wave speeds are computed, which is why this mutates its arguments and must be called ahead of s_L, s_R and s_S. The tangential velocities and vel_L/R_rms are deliberately left untouched.
| [in] | vel_r_rms | Left and right squared velocity magnitudes |
| [in] | c_r | Left and right sound speeds |
| [in,out] | vel_r_norm | Left and right wave-normal velocities, blended in place |
Definition at line 740 of file m_riemann_state.fpp.f90.
| subroutine m_riemann_state::s_calculate_bulk_stress_tensor | ( | real(wp), intent(in) | re_bulk, |
| real(wp), intent(in) | divergence_v, | ||
| real(wp), dimension(num_dims, num_dims), intent(out) | tau_bulk_out ) |
Compute bulk stress tensor components (diagonal only).
Definition at line 3018 of file m_riemann_state.fpp.f90.
| subroutine m_riemann_state::s_calculate_shear_stress_tensor | ( | real(wp), dimension(num_dims, num_dims), intent(in) | vel_grad_avg, |
| real(wp), intent(in) | re_shear, | ||
| real(wp), intent(in) | divergence_v, | ||
| real(wp), dimension(num_dims, num_dims), intent(out) | tau_shear_out ) |
Compute shear stress tensor components.
Definition at line 2975 of file m_riemann_state.fpp.f90.
| subroutine m_riemann_state::s_compute_average_state | ( | real(wp), intent(in) | rho_l, |
| real(wp), intent(in) | rho_r, | ||
| real(wp), dimension(num_vels), intent(in) | vel_l, | ||
| real(wp), dimension(num_vels), intent(in) | vel_r, | ||
| real(wp), intent(in) | h_l, | ||
| real(wp), intent(in) | h_r, | ||
| real(wp), intent(in) | gamma_l, | ||
| real(wp), intent(in) | gamma_r, | ||
| real(wp), intent(in) | qv_l, | ||
| real(wp), intent(in) | qv_r, | ||
| real(wp), intent(out) | rho_avg, | ||
| real(wp), intent(out) | vel_avg_rms, | ||
| real(wp), intent(out) | h_avg, | ||
| real(wp), intent(out) | gamma_avg, | ||
| real(wp), intent(out) | qv_avg ) |
Interface-averaged state that the pressure-based wave-speed estimate reads. avg_state selects between the density-weighted Roe average, which costs eight square roots per face, and the plain arithmetic mean; unlike the other solver switches this one is not implied by the call site, so the dispatch stays here.
| [in] | rho_r | Left and right densities |
| [in] | h_r | Left and right total enthalpies |
| [in] | gamma_r | Left and right specific heat ratio functions |
| [in] | qv_r | Left and right reference energies |
| [out] | vel_avg_rms | Squared magnitude of the averaged velocity, summed over all components |
Definition at line 796 of file m_riemann_state.fpp.f90.
| subroutine m_riemann_state::s_compute_cartesian_viscous_source_flux | ( | type(scalar_field), dimension(num_dims), intent(in) | dvell_dx_vf, |
| type(scalar_field), dimension(num_dims), intent(in) | dvell_dy_vf, | ||
| type(scalar_field), dimension(num_dims), intent(in) | dvell_dz_vf, | ||
| type(scalar_field), dimension(num_dims), intent(in) | dvelr_dx_vf, | ||
| type(scalar_field), dimension(num_dims), intent(in) | dvelr_dy_vf, | ||
| type(scalar_field), dimension(num_dims), intent(in) | dvelr_dz_vf, | ||
| type(scalar_field), dimension(sys_size), intent(inout) | flux_src_vf, | ||
| type(scalar_field), dimension(sys_size), intent(in) | q_prim_vf, | ||
| integer, intent(in) | norm_dir ) |
Compute Cartesian viscous source flux contributions for momentum and energy.
Definition at line 2774 of file m_riemann_state.fpp.f90.
| subroutine m_riemann_state::s_compute_chemistry_average_state | ( | real(wp), intent(in) | rho_l, |
| real(wp), intent(in) | rho_r, | ||
| real(wp), intent(in) | t_l, | ||
| real(wp), intent(in) | t_r, | ||
| real(wp), dimension(num_species), intent(in) | ys_l, | ||
| real(wp), dimension(num_species), intent(in) | ys_r, | ||
| real(wp), dimension(num_species), intent(in) | r_species, | ||
| real(wp), dimension(num_species), intent(in) | h_il, | ||
| real(wp), dimension(num_species), intent(in) | h_ir, | ||
| real(wp), dimension(num_species), intent(in) | cp_il, | ||
| real(wp), dimension(num_species), intent(in) | cp_ir, | ||
| real(wp), intent(in) | vel_avg_rms, | ||
| real(wp), intent(out) | gamma_avg, | ||
| real(wp), intent(out) | c_sum_yi_phi ) |
Roe-averaged reacting-mixture quantities: replaces gamma_avg with the mixture Cp/Cv and builds the c_sum_Yi_Phi term s_compute_speed_of_sound_avg needs. vel_avg_rms must be the full squared magnitude - its Phi_avg and vel_sum terms cancel to leave the Roe sound speed, and only do so for the full magnitude.
| [in] | rho_r | Left and right densities |
| [in] | t_r | Left and right temperatures |
| [in] | vel_avg_rms | Squared magnitude of the averaged velocity |
| [in] | ys_l | Per-species gas constants, formed by the caller: nvfortran cannot compile a caller that passes the constant molecular_weights array into a declare-target routine. Species enthalpies and heat capacities, evaluated by the caller. m_thermochem is called from the loop body rather than from here: CCE faults the GPU on that call one routine deeper. |
| [in] | ys_r | Per-species gas constants, formed by the caller: nvfortran cannot compile a caller that passes the constant molecular_weights array into a declare-target routine. Species enthalpies and heat capacities, evaluated by the caller. m_thermochem is called from the loop body rather than from here: CCE faults the GPU on that call one routine deeper. |
| [in] | r_species | Per-species gas constants, formed by the caller: nvfortran cannot compile a caller that passes the constant molecular_weights array into a declare-target routine. Species enthalpies and heat capacities, evaluated by the caller. m_thermochem is called from the loop body rather than from here: CCE faults the GPU on that call one routine deeper. |
| [in] | h_il | Per-species gas constants, formed by the caller: nvfortran cannot compile a caller that passes the constant molecular_weights array into a declare-target routine. Species enthalpies and heat capacities, evaluated by the caller. m_thermochem is called from the loop body rather than from here: CCE faults the GPU on that call one routine deeper. |
| [in] | h_ir | Per-species gas constants, formed by the caller: nvfortran cannot compile a caller that passes the constant molecular_weights array into a declare-target routine. Species enthalpies and heat capacities, evaluated by the caller. m_thermochem is called from the loop body rather than from here: CCE faults the GPU on that call one routine deeper. |
| [in] | cp_il | Per-species gas constants, formed by the caller: nvfortran cannot compile a caller that passes the constant molecular_weights array into a declare-target routine. Species enthalpies and heat capacities, evaluated by the caller. m_thermochem is called from the loop body rather than from here: CCE faults the GPU on that call one routine deeper. |
| [in] | cp_ir | Per-species gas constants, formed by the caller: nvfortran cannot compile a caller that passes the constant molecular_weights array into a declare-target routine. Species enthalpies and heat capacities, evaluated by the caller. m_thermochem is called from the loop body rather than from here: CCE faults the GPU on that call one routine deeper. |
| [out] | gamma_avg | Mixture Cp/Cv, replacing the density-weighted average |
Definition at line 897 of file m_riemann_state.fpp.f90.
| subroutine m_riemann_state::s_compute_cylindrical_viscous_source_flux | ( | type(scalar_field), dimension(num_dims), intent(in) | vell_vf, |
| type(scalar_field), dimension(num_dims), intent(in) | dvell_dx_vf, | ||
| type(scalar_field), dimension(num_dims), intent(in) | dvell_dy_vf, | ||
| type(scalar_field), dimension(num_dims), intent(in) | dvell_dz_vf, | ||
| type(scalar_field), dimension(num_dims), intent(in) | velr_vf, | ||
| type(scalar_field), dimension(num_dims), intent(in) | dvelr_dx_vf, | ||
| type(scalar_field), dimension(num_dims), intent(in) | dvelr_dy_vf, | ||
| type(scalar_field), dimension(num_dims), intent(in) | dvelr_dz_vf, | ||
| type(scalar_field), dimension(sys_size), intent(inout) | flux_src_vf, | ||
| type(scalar_field), dimension(sys_size), intent(in) | q_prim_vf, | ||
| integer, intent(in) | norm_dir, | ||
| type(int_bounds_info), intent(in) | ix, | ||
| type(int_bounds_info), intent(in) | iy, | ||
| type(int_bounds_info), intent(in) | iz ) |
Compute cylindrical viscous source flux contributions for momentum and energy.
Interface velocity ( \(v_1,v_2,v_3\)) (grid directions) for viscous work.
Shear stress vector ( \(\sigma_{N1}, \sigma_{N2}, \sigma_{N3}\)) on N-face (grid directions).
Definition at line 2506 of file m_riemann_state.fpp.f90.
| subroutine m_riemann_state::s_compute_hypoelastic_interface_energy | ( | integer, intent(in) | nf, |
| real(wp), dimension(nf), intent(in) | alpha_l, | ||
| real(wp), dimension(nf), intent(in) | alpha_r, | ||
| real(wp), intent(in) | damage_l, | ||
| real(wp), intent(in) | damage_r, | ||
| real(wp), dimension(6), intent(in) | tau_e_l, | ||
| real(wp), dimension(6), intent(in) | tau_e_r, | ||
| real(wp), intent(out) | g_l, | ||
| real(wp), intent(out) | g_r, | ||
| real(wp), intent(inout) | e_l, | ||
| real(wp), intent(inout) | e_r ) |
Accumulate the hypoelastic stress contribution to the energies of the left and right Riemann states: mix the shear modulus over the fluids, add the elastic energy of each stress component (doubled for the shear components) on each side whose mixture modulus is non-negligible, then scale the returned moduli by the continuum damage state when damage is modeled (energy uses the undamaged modulus; the damaged moduli feed the callers' wave speeds). The elastic shear stresses are loaded from the state buffers by the caller, which reuses them for the stress fluxes and elastic wave speeds. The G > verysmall per-side gate is a deliberate maintainer ruling that replaces HLL's former hard-coded G > 1000 stability floor, retiring its "TODO take out if statement if stable without".
| [in] | nf | Number of fluids to mix the shear modulus over |
| [in] | alpha_r | Left and right volume fractions |
| [in] | damage_r | Continuum damage states (referenced only when cont_damage) |
| [in] | tau_e_r | Left and right elastic shear stresses |
| [out] | g_r | Left and right mixture shear moduli |
| [in,out] | e_r | Left and right state energies |
Definition at line 3144 of file m_riemann_state.fpp.f90.
| subroutine m_riemann_state::s_compute_interface_reynolds | ( | real(wp), dimension(num_fluids), intent(in) | alpha_k, |
| real(wp), dimension(2), intent(out) | re_k, | ||
| integer, intent(in) | re_size_loc1, | ||
| integer, intent(in) | re_size_loc2 ) |
Compute the shear and volume Reynolds numbers of one Riemann state by inverse-weighting the fluid Reynolds numbers with the volume fractions.
| [in] | re_size_loc1 | host copies of Re_size; amdflang reads the declare-target original stale cross-TU |
| [in] | re_size_loc2 | host copies of Re_size; amdflang reads the declare-target original stale cross-TU |
Definition at line 3055 of file m_riemann_state.fpp.f90.
| subroutine m_riemann_state::s_compute_viscous_source_flux | ( | type(scalar_field), dimension(num_vels), intent(in) | vell_vf, |
| type(scalar_field), dimension(num_vels), intent(in) | dvell_dx_vf, | ||
| type(scalar_field), dimension(num_vels), intent(in) | dvell_dy_vf, | ||
| type(scalar_field), dimension(num_vels), intent(in) | dvell_dz_vf, | ||
| type(scalar_field), dimension(num_vels), intent(in) | velr_vf, | ||
| type(scalar_field), dimension(num_vels), intent(in) | dvelr_dx_vf, | ||
| type(scalar_field), dimension(num_vels), intent(in) | dvelr_dy_vf, | ||
| type(scalar_field), dimension(num_vels), intent(in) | dvelr_dz_vf, | ||
| type(scalar_field), dimension(sys_size), intent(inout) | flux_src_vf, | ||
| type(scalar_field), dimension(sys_size), intent(in) | q_prim_vf, | ||
| integer, intent(in) | norm_dir, | ||
| type(int_bounds_info), intent(in) | ix, | ||
| type(int_bounds_info), intent(in) | iy, | ||
| type(int_bounds_info), intent(in) | iz ) |
Dispatch to the subroutines that are utilized to compute the viscous source fluxes for either Cartesian or cylindrical geometries. For more information please refer to: 1) s_compute_cartesian_viscous_source_flux 2) s_compute_cylindrical_viscous_source_flux.
Definition at line 982 of file m_riemann_state.fpp.f90.
| subroutine m_riemann_state::s_finalize_riemann_solver | ( | type(scalar_field), dimension(sys_size), intent(inout) | flux_vf, |
| type(scalar_field), dimension(sys_size), intent(inout) | flux_src_vf, | ||
| type(scalar_field), dimension(sys_size), intent(inout) | flux_gsrc_vf, | ||
| integer, intent(in) | norm_dir ) |
Reshape and copy the Riemann-solver flux buffers back to the physical-space output arrays for the selected sweep direction, finalizing the Riemann solve. Two variants are emitted from one template so the shared unpermute logic cannot drift apart: the plain routine also copies the advection flux_src set and the grid_geometry==3 z-sweep geometric source flux, while the _hatR variant unpermutes the hat_R-anchored flux_hatR_rs* set of the fused dual-pass HLLD solve (called between the two RHS assemblies) and is a strict subset: flux_src is anchor-independent (already finalized with the hat_L set) and its geometric source flux only exists for the axisymmetric y-sweep.
Definition at line 3310 of file m_riemann_state.fpp.f90.
| subroutine m_riemann_state::s_finalize_riemann_solver_hatr | ( | type(scalar_field), dimension(sys_size), intent(inout) | flux_vf, |
| type(scalar_field), dimension(sys_size), intent(inout) | flux_gsrc_vf, | ||
| integer, intent(in) | norm_dir ) |
Definition at line 3806 of file m_riemann_state.fpp.f90.
| subroutine m_riemann_state::s_initialize_riemann_solver | ( | type(scalar_field), dimension(sys_size), intent(inout) | flux_src_vf, |
| integer, intent(in) | norm_dir ) |
Set up the chosen Riemann solver algorithm for the current direction.
Definition at line 2087 of file m_riemann_state.fpp.f90.
| subroutine m_riemann_state::s_populate_riemann_states_variables_buffers | ( | real(wp), dimension(idwbuff(1)%beg:,idwbuff(2)%beg:,idwbuff(3)%beg:,1:), intent(inout) | ql_prim_rsx_vf, |
| type(scalar_field), dimension(:), intent(inout), allocatable | dql_prim_dx_vf, | ||
| type(scalar_field), dimension(:), intent(inout), allocatable | dql_prim_dy_vf, | ||
| type(scalar_field), dimension(:), intent(inout), allocatable | dql_prim_dz_vf, | ||
| real(wp), dimension(idwbuff(1)%beg:,idwbuff(2)%beg:,idwbuff(3)%beg:,1:), intent(inout) | qr_prim_rsx_vf, | ||
| type(scalar_field), dimension(:), intent(inout), allocatable | dqr_prim_dx_vf, | ||
| type(scalar_field), dimension(:), intent(inout), allocatable | dqr_prim_dy_vf, | ||
| type(scalar_field), dimension(:), intent(inout), allocatable | dqr_prim_dz_vf, | ||
| integer, intent(in) | norm_dir, | ||
| type(int_bounds_info), intent(in) | ix, | ||
| type(int_bounds_info), intent(in) | iy, | ||
| type(int_bounds_info), intent(in) | iz ) |
Populate the left and right Riemann state variable buffers based on boundary conditions.
Definition at line 1004 of file m_riemann_state.fpp.f90.
| real(wp), dimension(:,:,:,:), allocatable m_riemann_state::flux_gsrc_hatr_rsx_vf |
Definition at line 432 of file m_riemann_state.fpp.f90.
| real(wp), dimension(:,:,:,:), allocatable m_riemann_state::flux_gsrc_rsx_vf |
The cell-boundary values of the geometrical source flux that are computed through the chosen Riemann problem solver by using the left and right states given in qK_prim_rs_vf. Currently 2D axisymmetric for inviscid only.
Definition at line 377 of file m_riemann_state.fpp.f90.
| real(wp), dimension(:,:,:,:), allocatable m_riemann_state::flux_hatr_rsx_vf |
Dual-pass HLLD second flux set: the hat_R-anchored fluxes (and, for axisymmetric runs, the hat_R interface velocities) written by the same fused solve that fills flux_rsx / nc_iface_vel_rsx with the hat_L-anchored values. Allocated only when hypo_nc_mode_dual_pass.
Definition at line 406 of file m_riemann_state.fpp.f90.
| real(wp), dimension(:,:,:,:), allocatable m_riemann_state::flux_rsx_vf |
The cell-boundary values of the fluxes (src - source) that are computed through the chosen Riemann problem solver, and the direct evaluation of source terms, by using the left and right states given in qK_prim_rs_vf, dqK_prim_ds_vf where ds = dx, dy or dz.
Definition at line 360 of file m_riemann_state.fpp.f90.
| real(wp), dimension(:,:,:,:), allocatable m_riemann_state::flux_src_rsx_vf |
Definition at line 360 of file m_riemann_state.fpp.f90.
| real(wp), dimension(:), allocatable m_riemann_state::gs_rs |
Definition at line 505 of file m_riemann_state.fpp.f90.
| type(int_bounds_info) m_riemann_state::is1 |
Definition at line 489 of file m_riemann_state.fpp.f90.
| type(int_bounds_info) m_riemann_state::is2 |
Definition at line 489 of file m_riemann_state.fpp.f90.
| type(int_bounds_info) m_riemann_state::is3 |
Definition at line 489 of file m_riemann_state.fpp.f90.
| type(int_bounds_info) m_riemann_state::isx |
Definition at line 490 of file m_riemann_state.fpp.f90.
| type(int_bounds_info) m_riemann_state::isy |
Definition at line 490 of file m_riemann_state.fpp.f90.
| type(int_bounds_info) m_riemann_state::isz |
Definition at line 490 of file m_riemann_state.fpp.f90.
| real(wp), dimension(:,:,:,:), allocatable m_riemann_state::mom_sp_rsx_vf |
Definition at line 461 of file m_riemann_state.fpp.f90.
| real(wp), dimension(:,:,:,:), allocatable m_riemann_state::nc_iface_vel_hatr_rsx_vf |
Definition at line 419 of file m_riemann_state.fpp.f90.
| real(wp), dimension(:,:,:,:), allocatable m_riemann_state::nc_iface_vel_rsx_vf |
Definition at line 390 of file m_riemann_state.fpp.f90.
| real(wp), dimension(:,:,:,:), allocatable m_riemann_state::re_avg_rsx_vf |
Definition at line 474 of file m_riemann_state.fpp.f90.
| real(wp), dimension(:,:), allocatable m_riemann_state::res_gs |
Definition at line 518 of file m_riemann_state.fpp.f90.
| real(wp), dimension(:,:,:,:), allocatable m_riemann_state::vel_src_rsx_vf |
Definition at line 448 of file m_riemann_state.fpp.f90.