5.3. exec/compressible_stag – compressible staggered-grid FHD

Purpose

Solves the compressible fluctuating Navier–Stokes equations for an ideal-gas mixture of up to MAX_SPEC species, with momentum on the staggered grid. Optional features are bulk chemistry (src_chemistry), mean-field surface chemistry (src_MFsurfchem), particle reservoirs, membranes and turbulence forcing.

Libraries

src_compressible_stag (which provides main_driver), src_compressible, src_chemistry, src_MFsurfchem, src_rng, src_analysis, src_common.

Build options

MAX_SPEC (default 8) and MAX_REAC (default 5) bound the number of species and reactions. DO_TURB = TRUE compiles in turbulence forcing. The scripts build_perlmutter.sh and build_frontier.sh show GPU builds on NERSC Perlmutter (CUDA) and OLCF Frontier (HIP).

Algorithm

Explicit three-stage Runge–Kutta time stepping (RK3stepStag in timeStepStag.cpp). The state is stored as conserved variables cu (density, momentum, total energy, species densities) and derived primitive variables prim. The driver checks that

• nvars = AMREX_SPACEDIM + 2 + nspecies, and

• nprimvars = AMREX_SPACEDIM + 3 + 2*nspecies

and aborts otherwise. Both must therefore be set consistently in the inputs file (e.g. nvars = 9, nprimvars = 14 for four species in 3D), together with ngc = 2 2 2 ghost cells.

Key parameters

Parameter	Meaning
prob_type	Initial condition, e.g. 1 = constant concentration, 2 = Rayleigh–Taylor, 3 = diffusion barrier. Further cases (4, 5 and 100–111) are defined in src_compressible_stag.
algorithm_type	1 = single component, 2 = multispecies.
visc_type	1 = non-symmetric, 2 = symmetric, 3 = symmetric with bulk viscosity (1D/2D runs require 1 or 2).
advection_type	1 = interpolate primitive, 2 = interpolate conserved quantities.
nspecies, molmass, diameter, dof, hcv, hcp	Species properties (CGS units in the examples).
k_B, Runiv, T_init, rho0	Physical constants and initial state.
bc_vel_lo/hi, bc_mass_lo/hi, bc_therm_lo/hi	Velocity (-1 periodic, 1 slip, 2 no-slip), mass (1 wall, 2 concentration, 3 reservoir) and thermal (1 adiabatic, 2 isothermal with t_lo/hi) boundary conditions. Periodic velocity forces all boundary conditions in that direction to be periodic.
transport_type	Transport-coefficient model (1 = Giovangigli, the default); see compressible.
plot_means, plot_vars, plot_covars, plot_cross, cross_cell	Statistics written to plotfiles.

Example inputs

inputs_regression_* cover periodic, slip/adiabatic, temperature gradient, concentration and reservoir boundaries in 1D and 3D. The test_* subdirectories contain chemistry and surface-chemistry test cases.