Corner point grid

From opm
Jump to: navigation, search
See also Wikipedias page on corner-point grids

The dune-cornerpoint module provides an interface to dune-grid.

Corner-point grid is a grid format for 3D-geometries. The grid format is sometimes referred to as a pillar grid, because it is based on a set of pillars/coordinate lines running from the top to the bottom of the model. Grid cells are defined by eight corners/nodes which lay pairwise on four neighbouring pillars. The corners of neighbouring grid blocks are specified independently, enabling the specification of grids with faults and gaps. Cornerpoint grids can be read by most reservoir simulators.

Corner-point grid is a grid format for 3D-geometries. The grid format is sometimes referred to as a pillar grid, because it is based on a set of pillars/coordinate lines running from the top to the bottom of the model. Grid cells are defined by eight corners/nodes which lay pairwise on four neighbouring pillars. The corners of neighbouring grid blocks are specified independently, enabling the specification of grids with faults and gaps. Cornerpoint grids can be read by most reservoir simulators.

The grid blocks in a corner-point grid is ordered in a logically Cartesian fashion. Each grid block is indexed by an ijk-triplet, where the integers i, j, and k are associated with the three spatial dimmensions x, y, and z, respectively. Therefore, a corner-point grid relates to three coordinate systems; the logical (integer i,j,k indicies), the local (x,y,z coordinates), and the global (x,y,z coordinates) coordinate system. The positive z direction is always downwards (depth). Properties in the grid are specified with the i-index varying fastest, then j, and finally k.

For the corner-point grids non-neighboring connections are supported, meaning that grid blocks that are not neighboring in ijk-space can be defined as neighboring. This feature allows for representation of faults with significant throws.

Keywords specifying grid

In the ECLIPSE implementation, which is the "standard", the grid is defined by the following keywords (see the ECLIPSE manual for details):

  • SPECGRID: Three numbers separated by spaces, nx ny nz, number of cells in each direction. Two additional optional items are ignored by the upscaling codes.
  • MAPAXES: Six numbers (map points in global x,y coordinates) defining the transformation from global to local coordinates (ignored by the upscaling codes)
  • COORD: 6 numbers for each of the pillars, x, y and z in local coordinates for top and bottom point of pillars. (nx+1)*(ny+1)*6 numbers
  • ZCORN: For all the eight corners of each cell, the z-values are specified, in a peculiar way (see ECLIPSE manual for details)

Keywords for properties

Grid properties are defined by a number of keywords, the following are used by the upscaling code:

  • ACTNUM: Either 0 or 1 to denote if the cell is to be treated as active or not. Ignored by Sintef upscaling code.
  • SATNUM: Integer specifying the rock type.
  • PERMX: Float specifying the absolute permeability (in x-direction) for each cell
  • PORO: Float specifying porosity for each cell

Locally defined keywords for elastic properties

The code for upscaling of elastic moduli allows the user to provide cell-level elastic properties in the eclipse files. To accomplish this, the following locally defined additional keywords are supported. The keywords must come in pairs as listed here:

  • YOUNGMOD and POISSONMOD
  • LAMEMOD and SHEARMOD
  • PWAVEMOD and SHEARMOD
  • BULKMOD and SHEARMOD