Skip to content

Latest commit

 

History

History
273 lines (188 loc) · 16.9 KB

parameters.rst

File metadata and controls

273 lines (188 loc) · 16.9 KB

List of Parameters

As part of the upgrade from Incompact3d 2.0 to the now called Xcompact3d 4.0, the parameters file was reconfigured to support NAMELIST I/O, which produces format-free input for groups of variables.

For a complete view about how the .i3d files are handled, besides to the default value applied to each parameter when not defined by the user, take a look at parameters.f90.

BasicParam

  • p_row & p_col define the domain decomposition for (large-scale) parallel computation. Notice that the product of both should be equal to the number of computational cores where Xcompact3d will run, and p_row = p_col = 0 executes the code in auto-tuning mode. More information can be found at 2DECOMP&FFT).

  • nx, ny & nz are the number of mesh points in each direction; Because of the Fast Fourier Transforms, limitations are in place. Basically you need to pick a combination of power of prime numbers. If the boundary conditions are not periodic (nclXX not equal to zero), then you need to add an extra mesh nodes. For instance 1025 (non periodic boundary conditions) instead of 1024 (periodic boundary conditions).

  • xlx, yly & zlz are the domain size, normalised with the reference length for the simulation (as an example, for a cylinder case, xlx=20D where D is the diameter of the cylinder). You can also decide to run simulations with non-normalised quantities, you just need to be consistent!

  • itype sets the flow configuration, each one is specified in a different Case-<flow-configuration>.f90 file. They are:

    • 0 - User Custom Configuration;
    • 1 - Turbidity Current in Lock-Release;
    • 2 - Taylor-Green Vortices;
    • 3 - Periodic Turbulent Channel;
    • 4 - Periodic Hill;
    • 5 - Flow over a Cylinder;
    • 7 - Mixing Layer;
    • 9 - Turbulent Boundary Layer;
    • 10 - Atmospheric Boundary Layer;
    • 11 - Uniform flow;
    • 12 - Sandbox configuration;
    • 13 - Differentilly heated cavity;
    • 14 - Pipe flow;
    • 15 - Periodic turbulent boundary layer.

You can modify the Case-<flow-configuration>.f90 file to change the inlet and/or initial conditions. Except for the Atmospheric Boundary Layer, quantities are normalised with a reference velocity, a reference length and a constant density, all equal to 1 (as an example, for the cylinder, the reference velocity is the freestream velocity equal to 1, the diameter equal to 1 and a constant density equal to 1; as a results the Reynolds number is equal to 1/nu).

  • istret controls mesh refinement in y direction only (it is not possible to refine the mesh in more than one direction):

    • 0 - No refinement (default);
    • 1 - Refinement at the center;
    • 2 - Both sides;
    • 3 - Just near the bottom.

    More details about the refinement function (which could be changed, but it would have to be represented in the spectral space by few modes only) can be found in Laizet, S., & Lamballais, E. (2009), High-order compact schemes for incompressible flows: A simple and efficient method with quasi-spectral accuracy, Journal of Computational Physics, 228(16), 5989-6015. The refinement can be control with the parameter beta.

  • beta is the refinement parameter. Large positive beta will lead to an almost uniform mesh, small positive beta will lead to very stretched mesh. Best option to find a suitable beta for your simulation is test and trial errrors!

  • iin defines perturbation at the initial condition:

    • 0 - No random noise (default);
    • 1 - Random noise with amplitude of init_noise;
    • 2 - Random noise with fixed seed (important for development and debugging) and amplitude of init_noise.

    Note

    The exactly behavior may be different according to each flow configuration.

  • inflow_noise Random number amplitude at inflow boundary, expressed as a % of the reference velocity (as an example 0.125 correspond to 12.5% of the reference velocity. It is advise to use small values between 0. and 0.1.

  • re Reynolds number, defined using the reference length scale and reference velocity, hence Re=1/nu.

  • dt Time step, to be inputed manually, depending on the mesh resolution, accuracy of the spatial finite-difference schemes and temporal scheme. By experience, the optimal time step can be found by test and trial errors, as opposed to try to compute the CFL number.

  • ifirst First iteration of the simulation. Do not forget to update if you are using a restart/checkpointing file.

  • ilast Last iteration of the simulation.

  • numscalar Number of scalar in the simulation. When using passive scalars, you can have more than one scalar field at the same time (basically you can solver the same transport equations with different initial conditions.

  • iscalar Enables scalar field(s). It is defined to 1 automatically when numscalar > 0.

  • iibm Flag for Immersed Boundary Method:

    • 0 - Off (default);
    • 1 - On with direct forcing method, i.e., it sets velocity to zero inside the solid body.
    • 2 - On with alternating forcing method, i.e, it uses Lagrangian Interpolators to define the velocity inside the body and imposes no-slip condition at the solid/fluid interface, see Gautier, R., Laizet, S., & Lamballais, E. (2014), A DNS study of jet control with microjets using an immersed boundary method, International Journal of Computational Fluid Dynamics, 28(6-10), 393-410 [no longer supported].
    • 3 - On with alternating forcing method, i.e, it uses Cubic Spline Interpolators to define the velocity inside the body and imposes no-slip condition at the solid/fluid interface. Allows for moving objects, see Giannenas, A. E., & Laizet, S. (2021), A simple and scalable immersed boundary method for high-fidelity simulations of fixed and moving objects on a Cartesian mesh, Applied Mathematical Modelling, 99, 606-627.

  • ilmn Enables Low Mach Number methodology when set to 1, basically solving the compressible Navier-Stokes equations in the low Mach number limit. If you want to solve the incompressible Navier-Stokes equations, ilmn should be equal to 0. See Bartholomew, P., & Laizet, S. (2019), A new highly scalable, high-order accurate framework for variable-density flows: Application to non-Boussinesq gravity currents, Computer Physics Communications, 242, 83-94.

  • ilesmod Enables Large-Eddy simulations (LES) methodologies:

    • 0 - Off.
    • 1 - Smagorinsky
    • 2 - WALE - suitable for wall-bounded flows
    • 3 - dynamic Smagorinsky - avoid as the filtering procedure is expensive
    • 4 - ILES - prefered options for LES

Please note that we will eventually remove all explicit LES models from the code as our ILES approach is cheaper with the same quality of results, if not better. See Dairay, T., Lamballais, E., Laizet, S., & Vassilicos, J. C. (2017), Numerical dissipation vs. subgrid-scale modelling for large eddy simulation, Journal of Computational Physics, 337, 252-274 and Mahfoze, O. A., & Laizet, S. (2021), Non-explicit large eddy simulations of turbulent channel flows from Reτ= 180 up to Reτ= 5,200, Computers & Fluids, 228, 105019.

  • nclx1, nclxn, ncly1, nclyn, nclz1 & nclzn define the velocity's boundary condition:

    • 0 - Periodic boundary conditions;
    • 1 - Free-slip boundary conditions, with two options depending on the velocity components: symmetry or anti-symmetry via the parameter npaire. Basically, a free-slip periodic boundary condition in the vertical direction corresponds to u=w=constant (similar to a symmetry boundary condition for which the gradient of u and w in the vertical direction is equal to zero), and to v=0 (similar to an anti-symmetry boundary condition);
    • 2 - Dirichlet boundary conditions.

    Note that the fractional step method in the code does not need explicit boundary conditions for the pressure field.

  • ivisu enables I/O for 3D snapshots if equal to 1 (every ioutput time step);

  • ipost enables online postprocessing if equal to 1.

  • gravx, gravy & gravz are the three components of the unitary vector pointing in the gravity's direction, only experimented for gravity currents.

  • cpg is a logical parameter for the momentum source term (Turbulent Channel case only). True / False for an imposed pressure gradient / flow rate.

  • ifilter & C_filter to filter the solution [set to 0 & 0 except for ABL and wind turbines cases. ifilter activates the filtering (with different direction combinations given by different values, e.g., 1 is for all directions, 2 for x-z filtering etc) and C_filter is the filter constant (e.g., C_filter=0 filters at 2Delta, C_filter=0.5 does nothing)

  • iturbine case-specific parameter for the wind farm simulator with 1: Actuator line, 2: actuator disk

NumOptions

  • ifirstder Scheme for first order derivative:

    • 1 - 2nd central explicite finite-difference schemes;
    • 2 - 4th central explicite finite-difference schemes - under development;
    • 3 - 4th compact finite-difference schemes - under development;
    • 4 - 6th compact finite-difference schemes - prefered option

  • isecondder Scheme for second derivative:

    • 1 - 2nd central explicite finite-difference schemes;
    • 2 - 4th central explicite finite-difference schemes - under development;
    • 3 - 4th compact finite-difference schemes - under development;
    • 4 - 6th compact finite-difference schemes - prefered option for DNS & Explicit LES
    • 5 - hyperviscous 6th compact finite-difference schemes - for Implicit LES

    For more details about these different options, and in particular for the customized hyperviscous schemes, please have a look at:

    • Lamballais, E., Fortuné, V., & Laizet, S. (2011). Straightforward high-order numerical dissipation via the viscous term for direct and large eddy simulation. Journal of Computational Physics, 230(9), 3270-3275.
    • Dairay, T., Lamballais, E., Laizet, S., & Vassilicos, J. C. (2017), Numerical dissipation vs. subgrid-scale modelling for large eddy simulation, Journal of Computational Physics, 337, 252-274.

  • itimescheme Time integration scheme:

    • 1 - Forwards Euler;
    • 2 - Adams-Bashforth 2;
    • 3 - Adams-Bashforth 3 - recommended option when using immersed boundary methods;
    • 4 - Adams-Bashforth 4 - not implemented yet;
    • 5 - Runge-kutta 3 - recommended option when not using immersed boundary methods;
    • 6 - Runge-kutta 4 - not implemented yet;
    • 7 - Semi-implict Crank-Nicolson + Adams-Bashforth 3 - does not work yet for all cases;
    • 8 - Semi-implict Crank-Nicolson + Runge-kutta 3 - does not work yet for all cases.

  • iimplicit Time integration scheme for the Y-diffusive term:

    • 0 - Explicit, default;
    • 1 - Euler implicit - does not work yet for all cases;
    • 2 - Crank-Nicolson - does not work yet for all cases;

  • nu0nu Ratio between hyperviscosity/viscosity at nu;

  • cnu Ratio between hypervisvosity at k_m=2/3\pi and k_c= \pi.

For more details about these two parameters, please have a look at:

  • Lamballais, E., Fortuné, V., & Laizet, S. (2011). Straightforward high-order numerical dissipation via the viscous term for direct and large eddy simulation. Journal of Computational Physics, 230(9), 3270-3275.
  • Dairay, T., Lamballais, E., Laizet, S., & Vassilicos, J. C. (2017), Numerical dissipation vs. subgrid-scale modelling for large eddy simulation, Journal of Computational Physics, 337, 252-274.

Default values are cnu=0.44 and nu0nu=4 which are ideal in a DNS context.

  • ipinter
    • 1 - conventional sixth-order interpolation coefficients as described in Lele 1992;
    • 2 - optimal sixth-order interpolation coefficients designed to be as close as possible to spectral interpolators - recommended option;
    • 3 - aggressive sixth-order interpolation coefficients designed to add some numerical dissipation at small scales but they could result in spurious oscillations close to a wall.

InOutParam

  • irestart Reads initial flow field if equals to 1. This is when you are using the restarting/checkpointing procedure (have a look at the subroutine restart;

  • icheckpoint Frequency for writing backup file (every icheckpoint time steps). Note that Xcompact3d will only keep on the disc the latest restart file;

  • ioutput Frequency to generate the 3D snapshots (every ioutput time steps);

  • nvisu Size for the 3D snapshots to be written on the disc (every nvisu mesh nodes). By default, use nvisu=1 which corresponds to 3D snapshots of size nx x ny x nz;

  • iprocessing Frequency for online postprocessing - not supported anymore as the guideline is to use our Python post-processing framework;

  • ninflows For precursor simulations for atmospheric boundary layers - will evolve soon!

  • ntimesteps For precursor simulations for atmospheric boundary layers - will evolve soon!

  • inflowpath For precursor simulations for atmospheric boundary layers - will evolve soon!

  • ioutflow For precursor simulations for atmospheric boundary layers - will evolve soon!

  • output2D - not supported anymore (will be removed eventually). Keep default value to zero

    • 0 - 3D, default
    • 1 - 2D, averaged over X
    • 2 - 2D, averaged over Y
    • 3 - 2D, averaged over Z

  • nprobes - not supported anymore (will be removed eventually). Keep default value to zero

Statistics

  • wrotation Rotation speed (Rosby number) of a source term in the equations to trigger turbulence (Channel Flow only);
  • spinup_time number of time steps after which the rotation source term is removed (Channel Flow only);
  • nstat Size arrays for statistic collection (every nstat mesh nodes). By default, use nstat=1 which corresponds to 3D statistic arrays of size nx x ny x nz;
  • initstat Time step when collection of statistics starts.

ProbesParam

  • flag_all_digits When False (default), 6 digits are recorded. Set to True to record 16 digits - not supported anymore (will be removed eventually);
  • flag_extra_probes Default is False. Set to True to monitor the velocity / pressure / scalars gradients - not supported anymore (will be removed eventually);
  • xyzprobes Array of size (3,nprobes) containing the location of the probes - not supported anymore (will be removed eventually).

ScalarParam

  • sc Schmidt numbers;

  • ri Richardson numbers;

  • uset Settling velocities;

  • cp Initial concentrations;

  • nclxS1, nclxSn, nclyS1, nclySn, nclzS1 & nclzSn define the scalar's boundary condition:

    • 0 - Periodic;
    • 1 - Odd or Even (default, no-flux);
    • 2 - Dirichlet.

  • scalar_lbound & scalar_ubound are the Scalar bounds;

  • sc_even True (default) if the scalar is even. False if it is odd;

  • sc_skew Default is False. True to activate the skew-symmetric convection for a scalar;

  • alpha_sc, beta_sc, g_sc are used only when iimplicit > 0. They define the boundary condition for the scalar at the top and bottom walls;

  • Tref

  • iibmS Flag for the scalar treatment at the Immersed Boundary Method (pre-release):

    • 0 - Off (default);
    • 1 - On with direct forcing method, i.e., it sets scalar to zero inside the solid body;
    • 2 - On with alternating forcing method, i.e., it uses Lagrangian Interpolators to define the scalar inside the body and imposes zero value at the solid/fluid interface;
    • 3 - On with alternating forcing method, but now the Lagrangian Interpolators are set to impose no-flux for the scalar field at the solid/fluid interface (only recommended if the normal vectors to the object's faces are aligned with one of the coordinate axes).

LESModel

jles, smagcst, walecst, maxdsmagcst, iwall

WallModel

smagwalldamp

Tripping

itrip,A_tr,xs_tr_tbl,ys_tr_tbl,ts_tr_tbl,x0_tr_tbl

ibmstuff

cex,cey,ra,nobjmax,nraf,nvol,iforces

ForceCVs

xld, xrd, yld, yud

LMN

dens1, dens2, prandtl, ilmn_bound, ivarcoeff, ilmn_solve_temp, massfrac, mol_weight, imultispecies, primary_species, Fr, ibirman_eos

ABL

z_zero, iwallmodel, k_roughness, ustar, dBL, imassconserve, ibuoyancy, iPressureGradient, iCoriolis, CoriolisFreq, istrat, idamping, iheight, TempRate, TempFlux, itherm, gravv, UG, T_wall, T_top

CASE

  • tgv_twod Flag used to initialize the Taylor-Green Vortices case with a 2D / 3D initial condition
  • pfront

ALMParam

ialmrestart,filealmrestart,iturboutput,NTurbines,TurbinesPath,NActuatorlines,ActuatorlinesPath,eps_factor,rho_air

ADMParam

Ndiscs,ADMcoords,C_T,aind,iturboutput,rho_air