Inputs, Configuration & Outputs

CompFlowLab simulations are configured through input files and produce solution data stored in a specified working directory. This section describes the required inputs, their structure, and the generated outputs.

Inputs

1. Command-Line Arguments

CompFlowLab requires two inputs when executed from the command line:

• Working directory
  Directory where all simulation outputs will be stored. It is recommended to use the test case folder.
• Input file (.inp)
  A configuration file that defines the simulation setup, including physical parameters, numerical settings, and ROM options.

Example of Sod Shock Tube:

python -O ./main.py ./examples/sod_shock_tube ./examples/sod_shock_tube/input_file.inp

2. Input File Structure

The .inp file contains all parameters required to run a simulation. It is organized into logical sections, each controlling a specific part of the simulation.

This section provides a complete reference for all parameters available in the CompFlowLab input file. Parameters are grouped by section and include type, allowed values, and descriptions.

Solver

Parameter	Type	Description	Allowed Values
solver_mode	str	Main solver	FOM, PROM, AROM, QROM, NLROM

• FOM: full order model
• PROM: linear projection ROM
• AROM: adaptive ROM
• QROM: quadratic ROM
• NLROM: non-linear ROM

Time Integration

Parameter	Type	Description	Allowed Values
dt	float	Physical time step	>0 (ex: 1e-6)
num_steps	int	Number of time steps	>0 (ex: 3000)
time_scheme	str	Time integration scheme	FDF, SSPRK2, SSPRK3, BDF

• FDF: Forward Difference Euler - Explicit
• SSPRK2: Strong Stability Preserv. RK2 – Explicit
• SSPRK3: Strong Stability Preserv. RK3 – Explicit
• BDF: Backward Difference Euler – Implicit

The BDF method uses finite differences to compute the Jacobian in reacting flow cases, which might make it very stiff and hard to converge if the time step is too large.

Space Discretization

Parameter	Type	Description	Allowed Values
x_initial	float	Start of Domain	>0
x_final	float	End of Domain	>x_initial
cell_number	int	Number of uniform computational cells	>0

Inlet Boundary Conditions

Parameter	Type	Description	Allowed Values
rho_inlet	float/str	Density left boundary	constant/periodic/extrapolate/wall
vel_inlet	float/str	Velocity left boundary	constant/periodic/extrapolate/wall
press_inlet	float/str	Pressure left boundary	constant/periodic/extrapolate/wall
temp_inlet	float/str	Temperature left boundary	constant/periodic/extrapolate/wall
mass_frac_inlet	float/str	Mass fraction left boundary	constant/periodic/extrapolate/wall

Outlet Boundary Conditions

Parameter	Type	Description	Allowed Values
rho_outlet	float/str	Density right boundary	constant/periodic/extrapolate/wall
vel_outlet	float/str	Velocity right boundary	constant/periodic/extrapolate/wall
press_outlet	float/str	Pressure right boundary	constant/periodic/extrapolate/wall
temp_outlet	float/str	Temperature right boundary	constant/periodic/extrapolate/wall
mass_frac_outlet	float/str	Mass fraction right boundary	constant/periodic/extrapolate/wall

• constant: specified constant value will be imposed.
• extrapolate: use the nearest inside value at the boundary.
• periodic: left boundary = right side, and right boundary = left side.
• wall: zero net velocity

CompFlowLab uses ghost cells to apply boundary conditions and does not automatically assign values to them. It completely depends on the user’s CFD knowledge in imposing physically valid boundary conditions.

Initial Condition

Initial conditions can be specified in two different ways:

(1) Using a pre-generated profile saved in the same format as a CompFlowLab primitive variable output file (.npy).

(2) Defining a few intervals, with the location and primitive variables specified for each interval.

Parameter	Type	Description	Allowed Values
ic_path	str	Initial condition profile path	-
x_interval_ic	float	Primitive variables specified for each domain interval	-

If specifying variables using intervals, CompFlowLab always requires values for mass fractions. For non-reacting flows, dummy values can be provided to mass_frac_ic.

Gas Model

Parameter	Type	Description	Allowed Values
gas_model	str	Gas model	Air, path to .yaml or .inp
flux_scheme	str	Flux scheme	2nd Order Roe
limiter	bool	Flux limiter activation	True/False
limiter_method	str	Limiter Method	minmod, barth
viscous	bool	Viscous terms	True/False
numpy_vector	bool	Use NumPy vectorization	True/False

Visualization

Parameter	Type	Description	Allowed Values
visual	bool	Plot visuals while running	True/False
save_visual	bool	Save plots	True/False
variable#	str	Variable to plot	see below

Available variables to plot:

• Density [kg/m3]
• Velocity [m/s]
• Pressure [Pa]
• Temperature [K]
• MF: Mass Fraction of first species [-]
• Heat Release [W/m3]

Saving

Parameter	Type	Description	Allowed Values
save_interval	int	Results save interval	>0 (ex:10)

Profiling

Parameter	Type	Description	Allowed Values
profiling	bool	Profile performance while running	True/False

The profiling results will be saved at the working directory as .prof format. It is recommended to use snakeviz to visualize the profiling results.

Injection

Parameter	Type	Description	Allowed Values
injection	bool	Injection model	True/False
injection_state_dir	str	Path to premixed injection gas state	-

Injection section is only applicable for 1D RDE test case and must be set as False for any other test case.

The injection_state_dir must point to a .npy file which follows the same format as CompFlowLab primitive variable output file.

ROM Setup

These properties will be read if any ROM methods are used.

Parameter	Type	Description	Allowed Values
rom_method	str	ROM projection method	galerkin, lspg
nl_rom_model	str	Non-linear ROM method	rbf, gpr
arom_method	str	Adaptive ROM method	direct, svd, isvd, ojas, grouse, past
pod_energy	float	POD residual energy	>0 (ex: 99.9)
hyper	bool	Use hyper-reduction	True/False
hyper_method	str	Hyper-reduction method	QDEIM, GappyPODE, ECWS, FGS
sampling_rate	float	Rate of cells sampled in hyper-reduction in percentage	>0 (ex: 2)
init_training_win	int	Adaptive ROM initial window size	>0 (ex:10)
unsampled_update_freq	int	Adaptive ROM unsampled cells update frequency	>0 (ex:10)
training_data_dir	str	Path to directory of training snapshots	-
training_start_iter	int	Start index of training snapshots	>0 (ex: 0)
training_step_iter	int	Step index of training snapshots	>0 (ex: 10)
training_end_iter	int	End index of training snapshots	>0 (ex: 5000)
rom_basis_generate	bool	Generate POD basis from scratch	True/False
rom_basis_dir	str	Path to directory of precomputed basis	-

ROM Methodology

Further details about Reduced Order Modeling under Methodology section.

Outputs

1. Field Data

Unsteady field data represents the time evolution of an unsteady field at the time step iteration interval specified by save_interval in input file. Each field data is written in a NumPy binary file containing a single NumPy array with two dimensions. The first dimension is the number of variables, the second is the number of cells in the computational mesh.

Primitive variable indices are as following:

Index	Variable	Unit
0	density	[kg/m3]
1	velocity	[m/s]
2	pressure	[Pa]
3	temperature	[K]
4:	species mass fractions (only in reacting flows)	[-]
-1	heat release rate (only in reacting flows)	[W/m3]

Conservative variable indices are as following:

Index	Variable	Unit
0	mass	[kg/m3]
1	momentum	[kg*m/s/m3]
2	total energy	[J/m3]
3:	species mass	[kg/m3]

2. Visualization

During simulation runtime, plotted visuals will be saved under working_dir/plots if save_visual=True in input file.

This visualization serves only as a basic tool to inspect results during the initial stages. It is recommended that users implement their own customized post-processing scripts to visualize results according to their specific requirements.

When CompFlowLab is run on Linux, interactive plots cannot be displayed during the simulation due to Linux environment limitations. They can only be accessed after the visualizations are saved (i.e., save_visual=True).