MULTX, MULTY and MULTZ are grid-block-based transmissibility multipliers used by Eclipse. They have the same format as any other grid-block property (e.g. PERMX) and can be included in a TransGen run in exactly the same fashion.
A MULTX value assigned to a cell (I, J, K) is applied to the transmissibilities between this cell and all cells with which it forms connections in the X+ direction (i.e. those with the indices (I+1, J, K*), where K* can take any value). Hence if no fault is present on the X+ side of cell (I, J, K), the multiplier is only applied to the single connection into cell (I+1, J, K), but if a fault is present, MULTX is applied to any connection (neighbour or non-neighbour) formed from (I, J, K) in this direction.
MULTY and MULTZ are analogous and apply to connections in the Y+ direction and Z+ directions respectively.
MULTX and MULTY can have an influence on faulted connections. The figures below shows XZ cross-sections of a simulation model.
Figure (a) shows cells with very low MULTZ values of 0.001 shaded (i.e. low transmissibility due to dipping shales) and cells with high MULTZ values of 1.0 in white.
Fig (b) shows the same values, but with the cells indexed to MULTX.
Figure (c) shows that between them, these 2 sets of multipliers define the 2 continuous sedimentological transmissibility baffles.
Figure (d) shows a faulted version of this model, with the (2D) cell indices labelled for grid-blocks that will be influenced by the MULTX values.
If MULTX is included in the TransGen run file, the "faulted" (i.e. including fault-rock) and "unfaulted" (i.e. the simulator default which excludes fault-rock) transmissibilties across all faulted connections will be calculated including the MULTX values. (See
TransGen Transmissibility Calculations in the
Technical Description of what TransGen does for details).
The EDITNNC, TRANX and TRANY output files will contain values (for non-neighbour and neighbour connections) representative of both the fault rock present on the connection and the MULTX and MULTY values.
Since the MULT value is included in both the faulted and unfaulted transmissibility calculations, the transmissibility multipliers for non-neighbour connections (as output in the EDITNNC file) are the same irrespective of whether the MULT values were included in TransGen or not. However, the transmissibilities output in the TRAN files (for neighbour connections) will be different.
For instance, the connection between cells (4,3) and (5,2) is a non- neighbour connection and Eclipse will calculate its transmissibility including the MULTX value applied to cell (4,3). The EDITNNC value calculated by TransGen will simply lower this (already low) value as a function of the fault rock present. The Eclipse calculation of the neighbour connection between cells (4,3) and (5,3), however, is overwritten by the contents of TransGen's output TRANX file, which there is calculated as a function of both MULTX and fault rock.
This is a simple, but not necessarily ideal solution. In most cases, the fault geometry breaks up the continuity of the baffles, as would be expected. In the example in Figure (d), however, one connection (between cells (1,6) and (2,5)) will be influenced by the baffle where perhaps it shouldn't, as the displacement on the fault is less than a grid-block height, and the fault offset is in the opposite sense to the dip direction of the surface.
Sedimentological baffles represented in this way cannot be used directly in TransGen fault seal calculations (e.g. to form a continuous shale smear on the connection between blocks (3,3) and (4,4)). However, a bit of imagination with the CELLPROP plugin (to assign effective Vshale values partly as a function of the MULTX, MULTY and MULTZ), the TGFSP keyword, and the PERM and THICK plugins would probably allow this kind of effect to be modelled.
NOTE: Sedimentological baffles should probably not be modelled in this way; i.e. as a continuous surface with a constant low transmissibility multiplier. The reason for this is exactly the same as the reason that fault rock should not be modelled as continuous surfaces with a constant low transmissibility multiplier. A constant transmissibility multiplier in a heterogeneous reservoirs implies that the properties of the baffle are heterogeneous (irrespective of whether it is a thin zone of low permeability fault rock or a thin low permeability shale drape), and this heterogeneity has no geological significance but is simply an artefact of modelling geological features using constant transmissibility multipliers.
MULTZ has no influence on the transmissibility calculations in TransGen. It is a recognized keyword and, if used, does not need to be specified using TGNEWKEY. In this respect it is treated by TransGen in exactly the same way as PERMZ.
