$$$$ EQEX NOTICE CHAT 11/09/12 21:16:04 7124
DATE 11/09/12
Operateur EQEX Voir aussi : NAVI
--------------
Description : Creates a table containing the data required
for the explicit or implicit solution of one or
several P.D.E. performed in the EXEC procedure
This table has must contain a INCO table
containing the unknowns
Syntax :
RV = EQEX
|-> 'ZONE' nomz 'OPER' nomo >>
| 'INCO' > --|
|-------------------------------<------------------|
|
|-> 'CLIM' nomii 'TYPCL' val |
e |
|-------<--------|
|
|--------<--------------------|
|
|-> 'DUMP' ----|
|
|--------<------|
|
|-> 'ALFA' alfa ------------|
|
|----------------<------------|
|
|-> 'ITMA' itma ------------|
|
|
|
other instruction ---<--------|
V
;
Comment : If a table RV is specified behind EQEX this table will be
augmented / modified by the following parameters.
It allows to circumvent the limit of 9 lines allowed
for an instruction.
List of instructions :
DUMP
ALFA
ITMA
FIDT
NISTO
NOMVI
TPSI
TFINAL
OMEGA
NITER
OPTI
ZONE
CLIM
DUMP instruction
----------------
This instruction is useful for checking out the set of data.
The decoding flow of EQEX is printed.
ITMA instruction
----------------
itma (integer) specifies the maximum number of time steps (default 1)
ALFA instruction
----------------
alfa (floating) tolerance on the time step (default 1.)
FIDT instruction
----------------
fidt (integer) time steps printing frequency (default 20)
NISTO instruction
-----------------
nisto (integer) plotting frequency of a particular unknown
(default 20)
NOMVI instruction
-----------------
nomvi (mot) name of velocity field used for the pressure computation
in the semi explicite algorithm (default 'UN').
TPSI instruction
----------------
tpsi (float) initial time
TFINAL instruction
------------------
tfinal (float) Final time (default 1.e30)
NITER instruction
-----------------
niter (integer) internal iteration number to a time step needed
to solve a non linearity (defaut 1)
OMEGA instruction
-----------------
omega (float) relaxation factor (0 < omega < 1)
(defaut 1.)
OPTI instruction
----------------
It allows to precise the general options of an operator and therefore is
placed just before ZONE instruction
list of options
Formulations EFM1 Finite Element (one point quadrature + correction)
(a la Gresho) only available with the semi explicit
algorithm.
EF Finite Element Galerkin or Petrov-Galerkin
VF Finite Volume
Upwinding CENTREE only with EF/EFM1 IMPL/EXPL/SEMI 2D/3D
SUPGDC only with EF/EFM1 IMPL/EXPL/SEMI 2D/3D
SUPG only with EF/EFM1 IMPL/EXPL/SEMI 2D/3D
SUPGH only with EF/EFM1 IMPL/EXPL/SEMI 2D/3D
CNG only with EF SEMI 2D/3D
PSI only with operators TSCA EFM1 EXPL 2D
JOHNSON only with operators TSCA EFM1 EXPL 2D
UPWIND only with VF EXPL 2D
GODUNOV only with VF EXPL 2D
VANLEER only with VF EXPL 2D
VLH only with VF EXPL 2D
HUSVL only with VF EXPL 2D
HUSVLH only with VF EXPL 2D
AUSM only with VF EXPL 2D
CMD Tuning coefficient for upwinding.
Only SUPG,SUPGDC,SUPGH in EF are concerned
Default value 0.2 .
Time formulation
IMPL implicit
EXPL explicit
SEMI beta semi implicit (0 Crank Nicolson
SEMI 0.5 CNG -> generalized Crank Nicolson (Donea)
BDF2 Backward 3 points,time integration. Second
order accurate.Unconditionally stable.
Only DFDT is concerned.
Transport formulation
CONS conservative form
NOCONS non conservative form
only with operators KONV,NS,NSKE,TSCA
NODIV in the case of non conservative form
the stabilisation term 1/2 T Div U is
ommited
specifics options to some operators
Underlying spaces for primal ou dual unknowns
INCOP spg
INCOD spg
spg has to be taken in the following list :
SOMMET FACE CENTRE CENTREP1 CENTREP0 MSOMMET
Mass Matrix type
MMPLEINE consistant mass matrix CM
MMDIAGO lumped (diagonal) mass matrix LM
MMPG Petrov-Galerkin mass matrix
Turbulence model RNG K - epsilon
RNG only with NSKE EFM1 EXPL
A-L-E formulation for operator NS EFM1 EXPL
ALE
Formulation IDEUL
EULER only with VF
EULERMS only with VF
Formulation of the viscous stesses in the momentum equation
MUVARI corresponds to the 'Laplacian Vector'
FTAU Formulation in viscous stress
MUCONS the viscosity gradients are neglected over
the element (default)
ZONE instruction
----------------
nomz : MMODEL object (type NAVIER-STOKES) or DOMAINE table
on which the operator will apply.
nomo : word type object, name of the discretization operator to
be executed. To be chosen from the list below :
FIMP FPU FROT LAPN TSCA NS NSKE ECHI DFDT KONV MDIA ...
arg1 arg2 ...etc : operator arguments (See concerned operator)
Once read, the arguments are stored in the table associated
with the operator in the reading order at the ARG1 ARG2 ... etc
indices
nomi1 nomi2 ... etc : names of indices in the table of unknowns
containing the unknowns on which the operator will apply.
For the time being, this table will have to be created
separately and stored in the table created by EQEX (RV) at the
INCO index.
example :
RV.'INCO'=TABLE 'INCO' ;
RV.'INCO'.'UN'=kcht$bellvectsommet(0.0.)(u1etu2);RV.'INCO'.'TN'=kcht$bellscalsommet1.t2;
CLIM instruction
-----------------
This instruction makes it possible to specify the boundary
conditions of imposed value type that will be applied to each
unknown;
nomii : is a word referring to an unknown that must already
be included in the list nomi1 nomi2 ... etc
TYPCL : is a key word specifying the component on which the
condition will be applied
to be chosen from :
UIMP vectorial unknown according to ox
VIMP vectorial unknown according to oy
WIMP vectorial unknown according to oz
TIMP scalar unknown
COMMENTS :
----------
A table is created for each operator ; its name is that of the
operator preceded by the number of order of the operator (OPER
instruction). This table contains the name of tbe operator, the
mesh object on which it applies, the coefficients which are
associated with it (base ARG1 ARG2 .. etc ), and the list of the
names of unknowns on which it applies.
These names are the indices (MOT type) of the INCO table in which
the unknowns are ordered. The INCO table will be created and filled
by the user, then it will be stored in the table created by EQEX.
Some general parameters are initialized by EQEX.
ITMA (1)
FIDT (20)
ALFA : tolerance coefficient on the time step 0 < alfa < 1.
1 by default (usually 0.5 to 0.8)
LISTOPER contains the list of the operators called.
FURTHER INFORMATION
-------------------
If you wish to output history files, you may add (manually)
a RV input to rv. 'HIST' contains a table constructed by the
KHIS operator. (Refer to this operator and to example bc30.dtc)
An entry of TABLE type is also created : RV.'METHINV'.
It defines the inversion method used to solve linear systems.
The default is a direct solver (Crout) which is robust but
memory-consuming.
It is possible to use an iterative solver (conjugate gradient
alike) instead, but you NEED to know what you're doing.
The RV.'METHINV' table's entry syntax is described in KRES
operator's help.
On output RV.'METHINV'.'CONVINV' contains the residual's norm
history (LISTREEL type object) as a function of the number of
iterations performed.