can be used in PHOENICS in numerous ways. Indeed there are so many ways in which the grid can be required to move that to provide for means of describing them generally from the SATELLITE is not practicable.
Instead, therefore, PHOENICS provides a few examples of how Fortran-coding can be used for the creation of moving grids.
These are to be found in the BFC-option GROUND coding, for which the file name is GXBFGR.HTM.
The configuration of the grid is specified by ascribing values to the corner coordinates XC, YC and ZC in Cartesian space. This is the user's task, because only he knows what he wants.
Then calls to the internal-to-EARTH subroutine BGEOM are made which result in the computation of the corresponding internode distances, cell-face areas, cell volumes, angles and direction cosines which PHOENICS needs. The user does not have to attend to these matters.
If, as is usual, the fluid can penetrate the cell faces, the convection fluxes which enter the balance equations of PHOENICS depend on the grid velocity as well as on the fluid velocity. PHOENICS calculates the grid-movement contribution by way of the "swept volume" of the cell face. This depends on the following twenty-four 24 quantities:-
The swept-volume calculation is effected by a call to the EARTH sub-routine, VOLUM, which is to be seen in Group 8 of GREX3; and it leads to values of the arrays CONI, CONJ and CONK, for which three- dimensional storage should be provided by appropriate settings in the Q1-file.
A difficulty about this practice is that VOLUM always returns a positive quantity; it is therefore necessary to determine independently whether the swept volume tends to increase the volume of the cell in question or to decrease it.
In the former case, the relevant CON (eg CONI for an east face) must be given a negative sign; otherwise the sign should be positive. The explanation is that the CONs have the same signific-ance as those based upon the fluid velocity in respect of tendency to increase the mass of material in a cell.
A positive u-velocity at an east face gives a positive CON; and a positive cell-wall velocity has the opposite tendency.
Examples of how the sign may be fixed are to be found in GREX3, where the determination is made to depend upon the values of RSG10, 8 and 12 (in version 1.6) and on BFGRJ, BFGRH and BFGRI in versions 2.0 and 2.1.
However, the practice shown there still leaves something to be desired, because it requires that all the CONI's (for example) at a given z-slab shall have the SAME sign. In general, both positive and negative CONIs must be expected.
Of course, when the grid movement is an adjustment made during the course of a steady-state calculation, as happens when, for example the movement is made in order to fit grid lines to streamlines or shock waves, no adjustment to the convection fluxes is necessary.
Users wishing to avail themselves of the moving-BFC facility are advised to study library cases B996 and B997, and also the subroutine GXBFGR.