Graph network fluid flow modeling

1. A method for modeling fluid flow dynamics in a fluid system, comprising:

providing an initial fluid system model comprising:

a plurality of nodes, each node characterized by one or more node fluid system parameters; and

a plurality of edges between two of the plurality of nodes, each edge characterized by one or more edge fluid system parameters; and

using the initial fluid system model, determining an updated fluid system model using a history-matching process;

wherein the fluid system comprises a subterranean fluid reservoir and one or more wells;

wherein the plurality of nodes comprise:

a plurality of well nodes, each well node corresponding to a well and characterized by a well productivity index (PI); and

a plurality of cell nodes, each cell node corresponding to a portion of the reservoir and characterized b a cell ore volume (PV);

wherein the plurality of cell nodes comprise, for each well:

one or more inner cell nodes that do not communicate with cells associated with another well; and

one or more outer cell nodes that communicate with cells associated with another well; and

wherein the fluid system is characterized by a number of inner cells and a number of outer cells.

2. The method of claim 1 , wherein the history-matching process is selected from the group consisting of: an ensemble smoothing multiple data assimilation (ES-MDA) process, a gradient descent process, a stochastic optimization process, an ensemble Kalman filter, a manual search, and any combination thereof.

3. The method of claim 1 , wherein the history-matching process is an ensemble smoothing multiple data assimilation (ES-MDA) process.

4. The method of claim 3 , wherein the ensemble smoothing multiple data assimilation (ES-MDA) process is a history-matching process that performs parameter updating two or more times.

5. The method of claim 1 , wherein:

the plurality of edges comprise:

a plurality of well-cell edges between one well node and one cell node, wherein each well-cell edge is characterized by a well-cell edge well productivity index;

a plurality of cell-cell edges between two cell nodes, wherein each cell-cell edges are characterized by one of:

a transmissibility (T), if the cell-cell edge is between cells associated with the same well; or

an interpartition transmissibility (T x ), if the cell-cell edge is between cells associated with two different wells; and

determining the updated fluid system model using a history-matching process includes:

determining, for one or more edges, an updated transmissibility (T) or an updated interpartition transmissibility (T x );

determining, for one or more well nodes, an updated well productivity index (PI); and

determining, for one or more cell nodes, an updated cell pore volume (PV).

6. The method of claim 5 , wherein, for each well, the number of inner cells is independent from the number of outer cells.

7. The method of claim 5 , further comprising:

forecasting production from the subterranean fluid reservoir based, at least in part, on results of the history-matching process.

8. The method of claim 5 , further comprising:

performing a well connectivity analysis, based, at least in part, on results of the history-matching process.

9. The method of claim 5 , further comprising:

controlling production from the subterranean fluid reservoir, based, at least in part, on results of the history-matching process.

10. The method of claim 5 , further comprising:

controlling a flooding operation of the subterranean fluid reservoir, based, at least in part, on results of the history-matching process.

11. The method of claim 5 , further comprising:

designing an integrated subsurface and surface network, based, at least in part, on results of the history-matching process.

12. A system for modeling fluid flow in a fluid system, comprising:

one or more processors; and

a memory comprising a plurality of non-transitory executable instructions that, when executed, cause the one or more processors to:

provide an initial fluid system model comprising:

a plurality of nodes, each node characterized by one or more node fluid system parameters; and

a plurality of edges between two of the plurality of nodes, each edge characterized by one or more edge fluid system parameters; and

using the initial fluid system model, determine an updated fluid system model using a history-matching process;

wherein the fluid system comprises a subterranean fluid reservoir and one or more wells:

wherein the plurality of nodes comprise:

a plurality of well nodes, each well node corresponding to a well and characterized by a well productivity index (PI); and

a plurality of cell nodes, each cell node corresponding to a portion of the reservoir and characterized by a cell pore volume (PV);

wherein the plurality of cell nodes comprise, for each well:

one or more inner cell nodes that do not communicate with cells associated with another well; and

one or more outer cell nodes that communicate with cells associated with another well; and

wherein the fluid system is characterized by a number of inner cells and a number of outer cells.

13. The system of claim 12 , wherein the history-matching process is an ensemble smoothing multiple data assimilation (ES-MDA) process.

14. The system of claim 13 , wherein the ensemble smoothing multiple data assimilation (ES-MDA) process is a history-matching process that performs parameter updating two or more times.

15. The system of claim 12 , wherein:

the plurality of edges comprise:

a plurality of well-cell edges between one well node and one cell node, wherein each well-cell edge is characterized by a well-cell edge well productivity index;

a plurality of cell-cell edges between two cell nodes, wherein each cell- cell edges are characterized by one of:

a transmissibility (T), if the cell-cell edge is between cells associated with the same well; or

an interpartition transmissibility (Tx), if the cell-cell edge is between cells associated with two different wells; and

determining the updated fluid system model using a history-matching process includes:

determining, for one or more edges, an updated transmissibility (T) or an updated interpartition transmissibility (T x );

determining, for one or more well nodes, an updated well productivity index (PI); and

determining, for one or more cell nodes, an updated cell pore volume (PV).

16. The system of claim 15 , wherein, for each well, the number of inner cells is independent from the number of outer cells.