J modeling ============== Characteristics of modeling ----------------------------------- This is a functional test case and computer non-regression of the calculation of the a posteriori error indicator implemented in thermal technology (cf. [:ref:`R4.10.03 `]). It excavates a spatial error map on which will be based, in a Python loop, the refinement/deraffination tool HOMARD encapsulated in MACR_ADAP_MAIL (cf. [:ref:`U7.03.01 `]). The calculation of this error indicator map is carried out, via the option 'ERTH_ELEM' of the post-processing operator CALC_ERREUR, on a EVOL_THER (provided to the keyword RESULTAT) coming from a previous thermal calculation (linear or not, transient or stationary, isotropic or stationary, isotropic or orthotropic, via THER_LINEAIRE or orthotropic, via or THER_NON_LINE, cf. necessary environment, configuration and scope of use [ :ref:`R4.10.03 `] §6.2/4). This calculation requires first the use of the option 'FLUX_ELNO' of CALC_CHAMP which determines the values of the heat flow vector at the nodes (cf. example of use [:ref:`R4.10.03 `] §6.5). The indicator consists of fifteen components per element and for a given moment. In this test case, the fifteen components are calculated but the refinement/deraffination procedure is based only on component ERTABS, which represents the total absolute spatial error (cf. [:ref:`R4.10.03 `] §6.3). In order to be able to post-process via POST_RELEVE or GIBI, we need to extrapolate fields by element to fields at nodes by element. The addition of the 'ERTH_ELNO_ELEM' option (after the call to 'ERTH_ELEM_TEMP') allows this purely computer transformation to be carried out. For a given moment and for a given finite element, it only duplicates the fifteen components of the indicator on each node of the element. This modeling is therefore as much an example of the use, in a PYTHON loop, of possible "indicator calculation"/"mesh refinement/deraffination" combinations, as well as a test case of non-regression of the options' ERTH_ELEM_TEMP 'and' ERTH_ELNO_ELEM 'and of their adherence to the remeshing process. This test case uses the characteristics of modeling I and its associated mesh (TETRA10 + TRIA6). Tested values --------------- We test the computer non-regression of the component ERTREL (total relative spatial error) of the error indicator with respect to versions V6.2.1 of the SGI and SUN platforms of *Code_Aster* and of version V4.3 of the software HOMARD. The relative tolerance is therefore severe: 5.10—6%. .. csv-table:: "**Identification**", "**Aster**", "**Tolerance**" "Value of ERTRELsur the mesh MA1avant remeshing", "4.15918735 10—5", "5.10—8" "Value of ERTRELsur the NO4avant remeshing node", "4.15918735 10—5", "5.10—8" "Value of ERTRELsur the M1mesh after remeshing", "5.48408914 10—6", "5.10—7"