$$$$ TRANGEOL NOTICE CHAT 11/09/12 21:18:23 7124 DATE 11/09/12 Operateur TRANGEOL Voir aussi : ------------------ RES1 RES2 = TRANGEOL modarcy TRANS1 (TRANS2); CALLED by TRANSGEN - NOT ADVISED TO NON DEVELOPPING USERS Solves : POROSITE * DC/DT = DIV (DIFFUSIVITE GRAD C - CONVECTION C) + SOURCE where C is the CONCENTRATION, might have many composants (multispecies uncoupled). But physical caracteristiques such as convection diffusion etc.. have to be common. Only le source term and limit conditions can be given by components. limit conditions Neumann DIrichlet Mixtes flux total DIFFUSIVITE = DIFFUSIVITE input + Dispersivity calculated automated num. diffusion (can be desactivated with EFMH). Resolution on 1 time step. Discrztization VF or EFMH Solver 'KRES' ------------------------------------------------------------------- |modarcy modele (MMODEL created by MODL) DARCY | ------------------------------------------------------------------- ------------------------------------------------------------------- | TRANS1 Table contains : | |---------------------------------------------------------------- | |'DIFFUSIVITE' | | effective diffusivity - CHAMPOINT - COMPONANTS | | K11 K21 K22 K31 K32 K33 at the center of elements | | | |'POROSITE' Porosity value (Champoint, Comp | | 'SCAL', or FLOTTANT) | | | | DELTAT time step | | | |'CONVECTION' velocity at the faces of the mesh. It is the | | convective flux multiplied by the normal vector on | | that face (DOMA modele NORMALE) and divided by | | the length of the face (doma modele SURFACE). | | | | It is actually the projection of the velocity flied| | on the direction perpendicular to the face. | | (Type CHPO Face, comp. | | VX VY VZ). The interest lies in the fact that | | the velocity is independant of the orientation | | of the normal to the faces which was not the case | | for the fluxes. | | | |'VITELEM' Velocity at the center of the elements (Type CHPO | | centre, comp. VX VY VZ). Used uniquely if | | DECENTREMENT is true or some dispersion is given | | via the ALPHAL and ALPHAT entries | | | |'ALPHAL' longitudinal dispersivity coef (CHPO | | comp. SCAL) - 0 if not there | | | |'ALPHAT' transverse dispersivity coef (CHPO | | comp. SCAL) - 0 if not there | | Rque : if ALPHAL or ALPHAT is present, both have to| | be present . | | | | | |---------------------- | |Conditions initiales : | |---------------------- | | | |'CONCENTRATION' concentration at the beginning of the time step | | (quantity of element per unit of water volume) | | (Type CHPO Centre, Comp 4 charact. max) | | The concentration may have many composants | | The resolution is done independantly for each | | componant | | | |-------------------------------------- | |limit conditions / chargements : | |-------------------------------------- | ------------------------------------------------------------------- | | | 'CLIMITES' table containing the following indices : | | | |'TRACE_IMPOSE' Concentration's trace imposed values. | | (Type CHARGEMENT of CHPO Face, Comp concentration) | | | |'FLUX_IMPOSE' Imposed surface-integrated flux's values | | (Type CHARGEMENT of CHPO Face, Comp concentration -| | default 0.) | | | |'FLUXTOT_IMP' Imposed surface integrated total flux' value | | (Type CHARGEMENT de CHPO Face, comp. | | concentration ) | | | |'MIXTES' Table : - index C contains values of the mixed flux| | surface integrated | | (Type CHARGEMENT de CHPO Face, | | comp. idem concentration default 0.)| | - indices A and B are reals | | | | mixed conditions is writen : | | C = A * diffusif flux + B * Concentration | ------------------------------------------------------------------ | | |'SOURCE' Source term per cell and per time unit (ex : | | filiation). Values at index i are between times i-1| | and i. (Type CHARGEMENT of CHPO Center, Comp concen| | tration - default 0.) | | | | | |-------------------- | |Donnees numeriques : | |-------------------- | | | | | | 'LUMP' FAUX SI pas de mass lumping, VRAI sinon. | | VRAI seulement sur des maillages de rectangles et | | parallelepipedes rectangles et tenseur de dissusion| | orthotrope. Permet de rendre les schemas monotone | | pour la diffusion-instationnaire - OBLIGATOIRE | | | | 'DECENTREMENT' VRAI si diffusion numerique pour Peclet = 2, | | permet | | de stabiliser (en explicite) voire rendre monotone | | le schema de convection. | | FAUX si schema sans convection, ou en implicite et | | absence d'oscillations - plus precis | | OBLIGATOIRE | | | | 'TYPDISCRETISATION' 'VF' si VF et 'EFMH' si EFMH | | | | | |'THETA_DIFFUSION' Coefficient de relaxation compris entre 0. et 1| | (theta-method diffusion) ('FLOTTANT' - OBLIGATOIRE)| | | |'THETA_CONVECTION' Idem pour la convection | | ('FLOTTANT', OBLIGATOIRE) | ------------------------------------------------------------------- | Solveurs : - OPTIONNELS |----------- | | | 'METHINV' TABLE OF PARAMETRES of the solver KRES, cf 'KRES' | | can be partially filled | | 2 important indices : | | 'TYPINV' 1 for direct 3 for BiCGSTAB | | 'PRECOND' 3 for ILU0, 5 for ILUT | | advice TYPINV = 1 in 2D or small calc., 3 otherwise| | advice PRECOND = 3, of no convergence = 5 | | DERNIER CONSEIL : si message du type convergence | | breakdown, Pivot nul ... mettre la tolerence | | BCGSTOL a la precision machine 1.D-300, cf notice | | de KRES | | METHINV is OPTIONNAL | | | | 'SOLVEUR' type of solver 1 direct | | 2 grad conjugate | | 3 BiCGSTAB | | 4 GMRES | | crunch the choice in METHINV if double INPUT | | OPTIONNAL | | | | 'PRECONDITIONNEUR' type of precond. 0 NIL | | 1 diag. | | 2 D-ILU | | 3 ILU0 | | 4 MILU | | 5 ILUT | | 6 ILUT2 | | OPTIONNAL- crunch choice in METHINV if double| | input. | | | |----------- | | recalculations - OPTIONNAL | |----------- | | 'ABANDON' IF VRAI resultats are nil when concentration | | is below 'SEUILCALC' | | A cheap calculation is done | | THe calculation is normal if the conc. | | gets again above the threashold (source again...)| | | | 'SEUILCALC' If ABANDON is VRAI, it gives the threashold | | for calculations | | | |----------- | | pilote -OPTIONNAL | |----------- | | 'RECALCUL' IF VRAI all matrices are recalculated | | | | 'INITIALI' IF vrai we acte as if it's the first call | | RECALCUL set to VRAI and TRANS2 unused | | ie all preconditioning objects are recalculated | | | | 'MODIFICATI' TABLE - more detailed than 'RECALCUL' | | indices : | | 'POROSITE' = VRAI if recalc due to variation of | | POROSITE | | 'CONVECTI' = VRAI if recalc due to variation of | | la CONVECTIOn | | 'DELTAT' = VRAI if recalc due to variation of | | DELTAT | | 'DIFFUSIV' = VRAI if recalc due to variation of | | DIFFUSIVITE | | | | If all these indices are not there, we reinitialize | | REINIT > RECALCUL > MODIFICATI | ------------------------------------------------------------------ ------------------------------------------------------------------- | TRANS2 - OPTIONNAL | ------------------------------------------------------------------ | Table optionnal which contains all preconditionners | | such as matrices of discretization, surfaces of elements | | etc... calulated by previous call | | | | TRANS2 est donc en general le resultat RES2 d'un precedent | | | | Not giving TRANS2 is equal to carry on a first calc | | Identical to REINIT = VRAI | | | | TRANS2 only interesting for constant time steps (EFMH) | | or relatively linear calc (en EFMH et VF) | |_________________________________________________________________| OUTPUT RES1 table identical to TRANS1 but with the new concentration. L'indice INITIALI set TO FAUX (created if not there in TRANS1). indices FLUXDIFF et FLUXCONV contenaininf CHPO of flux convective and diffusive surface integrated are added RES2 table contenaining preconditionners for accelarated calc. USE first call : RES1 RES2 = TRANGEOL MODARCY TRANS1 ; after : RES1 RES2 = TRANGEOL MODARCY RES1 RES2 ; fixe point -picard : think !!!
© Cast3M 2003 - All rights reserved.
Disclaimer