MULTREGT NNC

gistfile1.txt

  +----------------+
  | Region:1       |
  | I:0            +----------------+
  | J:0            | Region:2       |
  +----------------+ I:1            |
  | Region:1       | J:0            |
  | I:0            +----------------+
  | J:1            | Region:2       | 
  +----------------+ I:1            | 
                   | J:1            |    
                   +----------------+
                   
  
Two regions, with numerical value 1 and 2. There is a fault in the grid creating a NNC between cells (0,1) and (1,0), so in total there are three connections between region 1 and 2 with finite transmissibility:

Region 1     Region 2     is NNC
--------------------------------
(0,0)     |  (1,0)    |   No
(0,1)     |  (1,0)    |   Yes
(0,1)     |  (1,1)    |   No
--------------------------------

The fifth column in the MULTREGT keyword can be used control which of these connections should have their transmissibility modified. Setting the NNC column to 'NNC' would imply that only the connection (0,1) -> (1,0) would be affected, setting the NNC column to 'NONNC' would only affect the connections (0,0) -> (1,0) and (0,1) -> (1,1); the default value is 'ALL' which would affect all the connections.

In the opm-parser we do not have access to the NNC relationships, it is therefor not clear to me how we should handle the NNC part of this scenario. The possible solutions I see now are:

 - We implement a an API in opm-parser/EclipseState like getNNCTransMult(i,j,k,face) which can be used by the grid preprosessing    routine to ask for NNC transmissibility multpiliers *when* the grid processing layer has determined that this is indeed a NNC.

 - We move the detection of NNC's to the parser.
 
- ???