Methods and systems for flow-based coarsening of reservoir grid models

1. A method, comprising:

performing, by a computer processor, a first reservoir simulation of a reservoir region of interest using a first grid model to produce a first plurality of streamlines;

obtaining, by the computer processor, model data for the reservoir region of interest;

generating, by the computer processor, a multilevel coarsening mask describing a plurality of coarsening levels that correspond to different flow values among flow property data based on the first plurality of streamlines;

generating, by the computer processor, a first coarsened grid model using the model data and the multilevel coarsening mask;

performing, by the computer processor, a second reservoir simulation of the reservoir region of interest using the first coarsened grid model to produce a second plurality of streamlines;

performing, by the computer processor, a comparison of the first plurality of streamlines with the second plurality of streamlines;

adjusting, by the computer processor, the multilevel coarsening mask based on the comparison to produce an updated multilevel coarsening mask;

updating, by the computer processor, the first coarsened grid model based on the model data and the updated multilevel coarsening mask to produce a second coarsened grid model; and

performing, by the computer processor, a third reservoir simulation of the reservoir region of interest using the second coarsened grid model.

2. The method of claim 1 ,

wherein the multilevel coarsening mask is determined using a lookup table describing a plurality of coarsening values based on predetermined ranges of flow rate property values, and

wherein the multilevel coarsening mask is an integer mask.

3. A method of claim 1 ,

wherein the multilevel coarsening mask is an adjusted property mask comprising a smoothed grid region, and

wherein the smoothed grid region is produced by performing an edge smoothing operation to an expanded grid region.

4. A method of claim 1 ,

wherein the first coarsened grid model is generated using a plurality of local grid refinement and coarsening (LGR) object statements that are determined using a decomposition algorithm and one or more binary masks.

5. A method of claim 1 ,

wherein the first coarsened grid model is based on a coarsening scenario generated using a combinatorial algorithm.

6. A computer system, comprising:

a processor; and

a memory coupled to the processor, the memory comprising instructions, when executed by the processor, are configured to perform a method comprising:

performing a first reservoir simulation of a reservoir region of interest using a first grid model to produce a first plurality of streamlines;

obtaining model data for the reservoir region of interest;

generating a multilevel coarsening mask describing a plurality of coarsening levels that correspond to different flow values among flow property data based on the first plurality of streamlines;

generating a first coarsened grid model using the model data and the multilevel coarsening mask;

performing a second reservoir simulation of the reservoir region of interest using the first coarsened grid model to produce a second plurality of streamlines;

performing a comparison of the first plurality of streamlines with the second plurality of streamlines;

adjusting the multilevel coarsening mask based on the comparison to produce an updated multilevel coarsening mask;

updating the first coarsened grid model based on the model data and the updated multilevel coarsening mask to produce a second coarsened grid model; and

performing a third reservoir simulation of the reservoir region of interest using the second coarsened grid model.

7. The computer system of claim 6 ,

wherein the multilevel coarsening mask is determined using a lookup table describing a plurality of coarsening values based on predetermined ranges of flow rate property values, and

wherein the multilevel coarsening mask is an integer mask.

8. The computer system of claim 6 ,

wherein the multilevel coarsening mask is an adjusted property mask comprising a smoothed grid region, and

wherein the smoothed grid region is produced by performing an edge smoothing operation to an expanded grid region.

9. A non-transitory computer readable medium storing instructions executable by a computer processor, the instructions being configured to perform a method comprising:

performing a first reservoir simulation of a reservoir region of interest using a first grid model to produce a first plurality of streamlines;

obtaining model data for the reservoir region of interest;

generating a multilevel coarsening mask describing a plurality of coarsening levels that correspond to different flow values among flow property data based on the first plurality of streamlines;

generating a first coarsened grid model using the model data and the multilevel coarsening mask;

performing a second reservoir simulation of the reservoir region of interest using the first coarsened grid model to produce a second plurality of streamlines;

performing a comparison of the first plurality of streamlines with the second plurality of streamlines;

adjusting the multilevel coarsening mask based on the comparison to produce an updated multilevel coarsening mask;

updating the first coarsened grid model based on the model data and the updated multilevel coarsening mask to produce a second coarsened grid model; and

performing a third reservoir simulation of the reservoir region of interest using the second coarsened grid model.

10. The non-transitory computer readable medium of claim 9 ,

wherein the multilevel coarsening mask is an adjusted property mask comprising a smoothed grid region, and

wherein the smoothed grid region is produced by performing an edge smoothing operation to an expanded grid region.

11. The non-transitory computer readable medium of claim 9 ,

wherein the first coarsened grid model is generated using a plurality of local grid refinement and coarsening (LGR) object statements that are determined using a decomposition algorithm and one or more binary masks.

12. The non-transitory computer readable medium of claim 9 ,

wherein the first coarsened grid model is based on a coarsening scenario generated using a combinatorial algorithm.