Fast: Flexible Aerodynamic Solver Technolgy

Preamble

Fast is a common/front module for all Fast series solvers.

Fast is only available for use with the pyTree interface. You must import the module:

import Fast.PyTree as Fast

List of functions

– Actions

Fast.PyTree.setNum2Base

Set numeric data dictionary in bases.

Fast.PyTree.setNum2Zones

Set numeric data dictionary in zones.

Fast.PyTree.loadTree

Load a single tree.

Fast.PyTree.load

Load all updated trees.

Fast.PyTree.saveTree

Save a single tree.

Fast.PyTree.save

Save all updated trees.

Contents

Actions

Fast.PyTree.setNum2Base(a, numb)

Set the numb dictionary to bases. Exists also as in place version (_setNum2Base) that modifies a and returns None.

Parameters:

  • a (Base, pyTree) – input data

  • numb (dictionary) – numerics for base

the keys of numb dictionary are:

Example of use:

# - setNum2Base (pyTree) -
import Converter.PyTree as C
import Converter.Internal as Internal
import Fast.PyTree as Fast

a = C.newPyTree(['Base'])

numb = { 'temporal_scheme': 'explicit',
         'ss_iteration': 30,
         'modulo_verif':20 }

Fast._setNum2Base(a, numb)
Internal.printTree(a)
Fast.PyTree.setNum2Zones(a, numz)

Set the numz dictionary to zones. Exists also as in place version (_setNum2Zones) that modifies a and returns None.

Parameters:

  • a (Zone, Base, pyTree) – input data

  • numz (dictionary) – input data

the keys of numz dictionary are:

  • ‘scheme’: numerical scheme for convective flux reconstruction. Possible values are

  • ‘slope’: Slope reconstruction method. Possible values are

    • ‘o1’ (first order MUSCL reconstrution, only valid for roe scheme)

    • ‘minmod’ (second order MUSCL reconstrution with minmod limiter, only valid for roe scheme)

    • ‘o3’ (third order MUSCL reconstrution, see p50 https://tel.archives-ouvertes.fr/pastel-00834850/document)

    • ‘o3sc’ (third order MUSCL reconstrution with slope limiter, see Lugrin scheme)

    • ‘o5’ (fifth order MUSCL reconstrution)

    • ‘o5sc’ (fifth order MUSCL reconstrution with slope limiter)

    • default value is ‘o3’

  • ‘senseurType’: Sensor mode for the ‘senseur’ scheme. Possible values are

    • 0 (correction for speed only)

    • 1 (correction for speed, density and pressure)

  • ‘motion’: Motion mode for moving walls. Possible values are

  • ‘time_step’: time step value

    • default value is 1e-4

  • ‘time_step_nature’: time step nature. Possible values are

    • ‘global’

    • ‘local’

    • default value is ‘global’

  • ‘cfl’: Courant-Friedrichs-Lewy condition (only for time_step_nature’=’local’)

    • default value is 1

  • ‘epsi_newton’: Newton stopping criteria based on Loo norm

    • default value is 0.1

  • ‘inj1_newton_tol’: Newton tolerence for BCinj1 inflow condition

    • default value is 1e-5

  • ‘inj1_newton_nit’: Newton iterations for BCinj1 inflow condition

    • default value is 10

  • ‘psiroe’: Harten correction parameter

    • default value is 0.1

  • ‘prandtltb’: turbulent Prandtl number (only active for ‘model’=’NSTurbulent’)

    • default value is 0.92

  • ‘ransmodel’: RANS turbulence models. Possible values are

  • ‘SA_add_RotCorr’: add rotation correction to the SA model (SA-R)

    • 0

    • 1 (toggle rotation correction)

    • default value is 0

  • ‘SA_add_LowRe’: add low Reynolds correction to the SA model (SA-LRe)

    • 0

    • 1 (toggle low-Re correction)

    • default value is 0

  • ‘DES’: toggle the turbulence models for DES simulations. Possible values are

  • ‘DES_debug’: toggle data extraction for DES post-analysis. Possible values are

    • 0

    • 1 (save delta and fd functions in the FlowSolution#Centers node)

    • default value is ‘none’

  • ‘sgsmodel’: LES turbulence models. Possible values are

  • ‘extract_res’: toggle residual extraction during simulations. Possible values are

    • 0 (don’t save)

    • 1 (save div(F_Euler-F_viscous) in the FlowSolution#Centers node)

    • 2 (save dqdt + div(F_Euler-F_viscous) in the FlowSolution#Centers node)

    • default value is 0

  • ‘source’: toggle a source term for CFD simulations. Possible values are

    • 0

    • 1 (read a source terme in the FlowSolution#Centers node. The conservative variables centers:Density_src, centers:MomentumX_src, centers:MomentumY_src, centers:MomentumZ_src and centers:EnergyStagnationDensity_src are used.)

    • default value is 0

  • ‘ratiom’: maximum cut-off for mut/mu

    • default value is 10000

Example of use:

# - setNum2Zones (pyTree) -
import Converter.PyTree as C
import Converter.Internal as Internal
import Fast.PyTree as Fast
import Generator.PyTree as G

a = G.cart((0,0,0), (1,1,1), (10,10,10) )
t = C.newPyTree(['Base', a])

numz = { 'scheme': 'ausmpred',
         'time_step_nature': 'global',
         'time_step': 0.001 }

Fast._setNum2Zones(t, numz)
Internal.printTree(t)
Fast.PyTree.loadTree(fileName='t.cgns', split='single', graph=False)

Load a tree from a file (solution tree t, connectivity tree tc, or statistics tree ts). The file can be a single CGNS file (e.g. ‘t.cgns’), or can be split with one file per processor (e.g. ‘t/t_1.cgns’, ‘t/t_2.cgns’, etc.).

When executed in sequential mode, the function returns a complete tree.

When executed in mpi mode, the function returns a partial tree on each processor.

Warning

in MPI mode, the tree must already be distributed with the correct number of processors (with ‘proc’ nodes within the zones).

If graph is true, create and return the communication graph for Chimera and abutting transfers (for MPI use only). The dictionary is of the form of: graph={‘graphID’:[], ‘graphIBC’:[], ‘procDict’:[], ‘procList’:[]}.

Warning

to create the communication graph, the loaded tree must be the connectivity tree (tc.cgns), not the solution tree (t.cgns).

Parameters:

  • fileName (string) – name of the file to load

  • split (string) – file format: ‘single’ or ‘multiple’

  • graph (boolean) – true for returning communication graph

Returns:

t or tc or ts or (tc, graph)

# - loadFile (pyTree) -
import Fast.PyTree as Fast
import Converter.PyTree as C
import Generator.PyTree as G
import Converter.Mpi as Cmpi

# Save a file
a = G.cart((0,0,0),(1,1,1),(11,11,11))
a = Cmpi.setProc(a,0)
b = G.cart((10,0,0),(1,1,1),(11,11,11))
b = Cmpi.setProc(b,1)
t = C.newPyTree(['Base',a,b])
Fast.saveTree(t, fileName='t.cgns', split='single')

# Load it back
Fast.loadTree(fileName='t.cgns', split='single')

Note

since Fast 4.1, loadFile and loadTree have been merged.

Fast.PyTree.saveTree(t, fileName='restart.cgns', split='single', compress=0)

Save a tree to a file (solution tree t, connectivity tree tc, or statistics tree ts). The file can be a single CGNS file (e.g. ‘t.cgns’), or can be split with one file per processor (e.g. ‘t/t_1.cgns’, ‘t/t_2.cgns’, etc.). The tree is also cleared of unnecessary flow fields and temporary work arrays.

Warning

in MPI mode, the tree must already be distributed with the correct number of processors (with ‘proc’ nodes within the zones).

Before saving the file, the flow fields can be compressed using the compress parameter.

Parameters:

  • t (tree) – tree to save

  • fileName (string) – name of the file to load

  • split (string) – file format: ‘single’ or ‘multiple’

  • compress (integer) – compression parameter: 0 (none), 1 (cart only), or 2 (all)

# - saveFile (pyTree) -
import Fast.PyTree as Fast
import Converter.PyTree as C
import Generator.PyTree as G
import Converter.Mpi as Cmpi

a = G.cart((0,0,0),(1,1,1),(11,11,11))
a = Cmpi.setProc(a,0)
b = G.cart((10,0,0),(1,1,1),(11,11,11))
b = Cmpi.setProc(b,1)
t = C.newPyTree(['Base',a,b])
Fast.saveTree(t, fileName='t.cgns', split='single')

Note

since Fast 4.1, saveFile and saveTree have been merged.

Fast.PyTree.load(fileName='t.cgns', fileNameC=None, fileNameS=None, split='single')

Load all trees at once (solution tree t, connectivity tree tc, and/or statistics tree ts), using multiple calls of the Fast.PyTree.loadTree function. The files can be single CGNS files (e.g. ‘t.cgns’), or can be split with one file per processor (e.g. ‘t/t_1.cgns’, ‘t/t_2.cgns’, etc.).

Warning

in MPI mode, the trees must already be distributed with the correct number of processors (with ‘proc’ nodes within the zones).

If fileNameC is not None, the connectivity tree is loaded and the communication graph is automatically created.

If filenameS is not None, the statistics tree is loaded.

Parameters:

  • fileName (string) – name of the solution file to load

  • fileNameC (string) – name of the connectivity file to load

  • fileNameS (string) – name of the statistics file to load

  • split (string) – file format: ‘single’ or ‘multiple’

Returns:

(t, tc, ts, graph)

# - load (pyTree) -
import Fast.PyTree as Fast
import Converter.PyTree as C
import Generator.PyTree as G
import Converter.Internal as Internal
import Connector.PyTree as X
import Converter.Mpi as Cmpi

# Create case
if Cmpi.rank == 0:
    a = G.cart((0,0,0),(1,1,1),(11,11,11))
    a = Cmpi.setProc(a,0)
    b = G.cart((10,0,0),(1,1,1),(11,11,11))
    b = Cmpi.setProc(b,1)
    t = C.newPyTree(['Base',a,b])

    t = X.connectMatch(t, dim=3)
    t = Internal.addGhostCells(t,t,2,adaptBCs=0)
    C.convertPyTree2File(t, 't.cgns')
    tc = C.node2Center(t)
    tc = X.setInterpData(t, tc, loc='centers', storage='inverse')
    C.convertPyTree2File(tc, 'tc.cgns')
Cmpi.barrier()

t,tc,ts,graph = Fast.load('t.cgns', 'tc.cgns', split='single')

#print(t)
#print(tc)
#print(ts)
#print('procDict = ', graph['procDict'])
#print('graphID = ', graph['graphID'])
Fast.PyTree.save(t, fileName='restart.cgns', tc=None, fileNameC='tc_restart.cgns', ts=None, fileNameS='tstat_restart.cgns', split='single', compress=0)

Save all trees at once (solution tree t, connectivity tree tc, and/or statistics tree ts), using multiple calls of the Fast.PyTree.saveTree function. The files can be single CGNS files (e.g. ‘t.cgns’), or can be split with one file per processor (e.g. ‘t/t_1.cgns’, ‘t/t_2.cgns’, etc.). The trees are also cleared of unnecessary flow fields and temporary work arrays.

Warning

in MPI mode, the trees must already be distributed with the correct number of processors (with ‘proc’ nodes within the zones).

If tc is not None, the connectivity tree is saved.

If ts is not None, the statistics tree is saved.

Before saving the file, the flow fields can be compressed using the compress parameter.

Parameters:

  • t (tree) – updated solution tree to save

  • fileName (string) – name of the solution file

  • tc (tree) – updated connectivity tree to save

  • fileNameC (string) – name of the connectivity file

  • ts (tree) – updated statistics tree to save

  • fileNameS (string) – name of the statistics file

  • split (string) – file format: ‘single’ or ‘multiple’

  • compress (integer) – compression parameter: 0 (none), 1 (cart only), or 2 (all)

Example of use:

# - save (pyTree) -
import Fast.PyTree as Fast
import Converter.PyTree as C
import Generator.PyTree as G
import Converter.Mpi as Cmpi

a = G.cart((0,0,0),(1,1,1),(11,11,11))
a = Cmpi.setProc(a,0)
b = G.cart((10,0,0),(1,1,1),(11,11,11))
b = Cmpi.setProc(b,1)
t = C.newPyTree(['Base',a,b])
tc = C.node2Center(t)
Fast.save(t, fileName='t.cgns', tc=tc, fileNameC='tc.cgns', split='single')

Index