KIVA-3(V) Interface



The ICEM/KIVA-3(V) interface writes the grid and boundary condition information in a TAPE17 file according the KIVA-3(V) specifications.

Creating the KIVA-3(V) Input File

The translator writes the KIVA-3(V) input file using structured domain files (domain.1 to domain.n), the boundary condition file (boco) and the topology file (topo_mulcad_out). The multi-blocks structured grid is first transformed into an unstructured grid file in the interface. This unstructured grid file is different from the one which would be created with the utility "structured -> Unstructured". The difference is that all 3D elements must be written as HEXA-8 with 8 distinct nodes even if they are degenerated. As a result, points on singularities are represented by more than one node. To create the KIVA-3(V) input file, select the translator "KIVA-3(V)" in the "Output" menu from the ICEM CFD/CAE Manager. After clicking "Write output" you can specify options for the KIVA-3(V) interface: The interface generates the KIVA-3(V) file located in the Transfer Shell. You may open a shell window in this directory by selecting the menu items "Utility" and "Transfer Shell".

Aligning the IJK axes with the origin

For KIVA-3(V) to read the boundary conditions correctly, it is imperative that the IJK axes of the structured domain be correctly aligned with the configuration's origin. To realign the axes in the ICEM CFD Mesh Editor, do the following:

Defining boundary conditions for KIVA-3(V)

For a grid generated with MULCAD/PADAMM, the "Unspecified Bound. Cond." menu in MULCAD is used to define vertex flags, cell face boundary conditions and regions. For a grid generated with the HEXA-Mesher, the boundary conditions are defined in the property menu. Please refer to the General Remarks section for more details. In P-Cube, the NAME and ZONE fields serve the same purpose. The settings required to correctly define boundary conditions and properties are described in the following sections.

Cell Face Boundary Condition (BCL, BCF, BCB):

In MULCAD, set FLAG to FACE to indicate that a "Cell face boundary condition" is being defined, then set ATTRIB to one of the choices allowed (see list below). In the DDN-Mesher interface, set string1 to BCx and string2 to one of the choices below. In P-CUBE, enter one of the allowed choices in the NAME field. The cell face boundary conditions are only recognized on surfaces.

    MULCAD      DDN-Mesher      
    ------      ----------     
    1.FLAG      1.string1       Enter the word BCx
   12.ATTRIB    2.string2       Enter one of the face type below


        Face Type               Description             BCx value

        FLUID                   between 2 fluid cells    4.0
        MOVE                    moving parts             1.0
        SOLID                   walls                    2.0
        PERIF                   periodic front           5.0
        PERID                   periodic derriere        6.0
        INFL                    velocity inflow          7.0
        OUTFL                   continuative outflow     8.0
        PRESI                   pressure inflow          9.0
        PRESO                   pressure outflow        10.0
Note that BCx can also equal 3.0 if the face is degenerated into an axis. In such case, the interface automatically assigned a BCx of 3.0. By default the interface assigns type SOLID to all non-degenerated wall face, type AXIS to all degenerated wall face and type FLUID for all interior faces. The faces located between two ghost cells always get a BCx of 0.0.

Moving surface specification (IDFACE):

This additional boundary condition is only available for the KIVA-3V file format. The KIVA-3 format allows only one type of moving surface, the piston face. In KIVA-3V however, a mesh may also display other types of moving surfaces such as intake or exhaust valves. Since different moving surfaces may have different motion, an additional parameter, IDFACE, is defined for each face. Note that this option is not available for P-CUBE generated meshes.

IDFACE may take the following values:

IDFACE is specified with the vertex flag (FV). Note that only the faces and edges boundary conditions are considered for IDFACE .

Vertex Flag (FV):

In MULCAD, set FLAG to FV to indicate that a "Vertex flag" is being defined, then set ATTRIB to one of the choices allowed (see list below). In the DDN-Mesher interface, set string1 to FV and string2 to one of the choices below. In P-CUBE, enter one of the allowed choices under NAME and select either USER2 or USER3 for the boundary condition type record. The USER2/USER3 flag will permit the translator to distinguish between vertex flag settings and cell face boundary condition settings. The vertex flag (FV) can be assigned to either domains (blocks), subfaces (surfaces) or edges.

    MULCAD      DDN-Mesher      
    ------      ----------     
    1.FLAG      1.string1       Enter the word FV
    2.NR        3.integer       Option to specify IDFACE 
   12.ATTRIB    2.string2       Enter one of the flag type below

        Flag in ICEM            Description                 FV 

        GHOST                   inactive vertices           0.0

        FACE                    vertices on piston face     2.0

        BOWL                    bowl vertices, 
                                not on piston face          3.0

        SQSH                    vertices in the squish 
                                region                      4.0

        DOME                    vertices within the head 
                                but not touching the head   5.0

        HEAD                    vertices touching the head  6.0

        FLUID                   vertices not described in 
                                the above categories        1.0

Note that surface tagging overwrites block tagging and edge tagging overwrites surface tagging. The default value for FV is 1.0 for live nodes and 0.0 for ghost nodes.

Region (IDREG):

In MULCAD, set FLAG to REG and enter the region number under entry NR. In the DDN-Mesher Interface, set string1 to REG and enter the region number under integer. In P-CUBE, use the ZONE number to specify the region number. The region number can be defined on the domains (blocks) only. If no regions are defined by the user, all domains are part of region number one. If some domains but not all are given a region number, an extra region is created by the interface with a distinct number and all the remaining domains are included in this extra region.

    MULCAD      DDN-Mesher
    ------      ----------
    1.FLAG      1.string1       Enter the word REG
    2.NR        3.integer       Enter the region number
The ghost cells are given a region number equals to zero by the interface so this region number should not be used for other regions.

Cell Flag (F):

This flag is equal to 1.0 or 0.0 depending if the cell is live (fluid) or inactive (dead). Unless otherwise specified, all cells generated by the mesher are set to live. If a region needs to be turned off, simply set its vertex flag to GHOST, as shown above.

Note that an extra layer of ghost cells is always created by the translator on the left, bottom and front side of the model.

Option to split the grid:

This option allows to cut the grid in two parts, translate one of the parts and fill the gap created by this split with a number of layers of cells.

The blocks to translate are tagged with the word MOVE. Note that the translation may only be purely in x, or purely in y, or purely in z.

    MULCAD      DDN-Mesher
    ------      ----------
    1.FLAG      1.string1       Enter the word MOVE
The user must specify the "split-plane" by tagging the appropriate face(s) with the word SPLIT. All surfaces with this tagging are copied nlayers time to fill the gap created by the translation of part of the mesh.
    MULCAD      DDN-Mesher
    ------      ----------
    1.FLAG      1.string1       Enter the word SPLIT
The distance to translate the selected blocks and the number of layers of cells to generate are specified in the KIVA-3(V) translator GUI.

Content of the KIVA-3(V) file (TAPE17)

where: (Return)

Related References


KIVA - Los Alamos National Laboratory

University of Wisconsin-Madison, Engine Research Center, KIVA Page