Operands ========= Operands BASE_PRIMAL and BASE_DUAL --------------------------------- .. code-block:: text ♦ BASE_PRIMAL = base1, [empi_mode] ♦ BASE_DUAL = base2, [empi_mode] Name of the modes_empi result data structures to be analyzed to generate the interpolation points. The two bases base1 and base2 are calculated using the operator DEFI_BASE_REDUITE [:ref:`U4.67.01 `]. The calculation of the base 1 is based on a temperature field or a displacement field. The calculation of the base 2 is based on a flow field or a constraint field. Operand MAILLAGE ----------------- .. code-block:: text ♦ MAILLAGE = email, [mesh] The command will enrich an already existing mesh concept with the new groups of nodes and meshes defined by DOMAINE_INCLUS, GROUP_NO_INTERFACEet GROUP_NO_ENCASTRE. The MAILLAGE keyword is therefore mandatory. Operand NOM_DOMAINE -------------------- .. code-block:: text ♦ NOM_DOMAINE = name_d Specify the name of the group of elements corresponding to RID. Operand GROUP_NO_INTERF ------------------------ .. code-block:: text ♦ GROUP_NO_INTERF = name_i We specify the name of the group of nodes contained in the interface between RID and the rest of the domain. Operand NB_COUCHE_SUPPL ------------------------ .. code-block:: text ◊ NB_COUCHE_SUPPL = nb_layer, [I] By default, the command builds group RID by selecting the meshes attached to the magic points (application of d EIM, see [:external:ref:`R5.01.05 `]). When using the NB_COUCH_SUPPL keyword, you can enlarge the RID by taking nb_layerof additional elements around the initial RID. By default, nb_layer =0. Operand DOMAINE_MINI/DOMAINE_MAXI ---------------------------------- ◊ DOMAINE_MINI = _F ( .. code-block:: text ◊ GROUP_MA = lgma1, [l_gr_mesh] ◊ GROUP_NO = lgno1, [l_gr_node] ) This optional factor keyword allows you to put part of the mesh in the RID even if the algorithm for finding magic points by d EIM (see [:external:ref:`R5.01.05 `]) does not allow it. This is very useful in particular to "force" the integration of some of the boundary conditions in RID. You can either add knots or add stitches. ◊ DOMAINE_MAXI = _F ( .. code-block:: text ◊ GROUP_MA = lgma1, [l_gr_mesh] ) Symmetrically, it is possible to restrict the size of the reduced domain RID by the DOMAINE_MAXI keyword. This option is useful when creating incremental models such as multi-pass welding or excavation. Operands CORR_COMPLET/GROUP_NO_ENCASTRE/NB_COUCHE_ENCASTRE ---------------------------------------------------------- .. code-block:: text ***** CORR_COMPLET = /' NON '[DEFAUT] /' OUI ' If CORR_COMPLET = 'OUI' { **♦** GROUP_NO_ENCASTRE = big ◊ NB_COUCHE_ENCASTRE = /0 [DEFAUT] /nb_layer, [I] } .. code-block:: text To improve the quality of the hyper-reduction results, it is possible to correct the hyper-reduced calculation by a detailed finite element calculation with CORR_COMPLET =' OUI '. To do this, it is necessary to define a group of nodes used to link RID and the rest of the model with the GROUP_NO_ENCASTRE keyword. This group of nodes makes it possible to impose the Dirichlet boundary conditions necessary for the corrected problem to be well defined. By default (NB_COUCHE_ENCASTRE =0), this group of nodes is defined within the reduced domain (keyword NOM_DOMAINE) by taking the first nodes contained in the domain and linked to the nodes defined by GROUP_NO_INTERF. It is possible to make the domain bigger by using the NB_COUCHE_ENCASTRE keyword. Because of the construction of these groups of nodes, we necessarily have NB_COUCHE_ENCASTRE less than or equal to NB_COUCHE_SUPPL.