Method and system for determining predicted reservoir simulation solutions using artificial intelligence

A system, comprising:

• a plurality of wells in a geological region of interest;
• a plurality of control systems coupled to the plurality of wells;
• and
• a reservoir simulator comprising a computer processor, wherein the reservoir simulator is coupled to the plurality of control systems, the reservoir simulator being configured to perform a method comprising:
• obtaining grid model data regarding a geological region of interest, wherein the grid model data comprises porosity data, lithological data, and permeability data;
• obtaining a plurality of reservoir simulation parameters for a reservoir simulation of the geological region of interest, wherein the plurality of reservoir simulation parameters comprises well data regarding the plurality of wells;
• obtaining a plurality of reservoir equations for the reservoir simulation;
• determining first simulation data for a first time step of the reservoir simulation based on the grid model data, the plurality of reservoir simulation parameters, and the plurality of reservoir equations;
• determining second simulation data for a second time step of the reservoir simulation based on the grid model data, the plurality of reservoir simulation parameters, and the plurality of reservoir equations;
• determining a first plurality of adaptive weights using a first machine-learning model, the first simulation data, and the second simulation data;
• determining first predicted simulation data for a third time step of the reservoir simulation using the first plurality of adaptive weights, the first simulation data, the second simulation data, the grid model data, and the plurality of reservoir simulation parameters,
• wherein a respective adaptive weight among the first plurality of adaptive weights adjusts a contribution of respective simulation data for determining the first predicted simulation data;
• determining whether the first predicted simulation data satisfies predetermined criterion, wherein the predetermined criterion is a convergence criterion corresponding to a reservoir solution to the plurality of reservoir equations converging to a predetermined limit;
• and
• determining, in response to determining that the first predicted simulation data fails to satisfy the predetermined criterion, updated simulation data for the third time step of the reservoir simulation based on the first predicted simulation data, the grid model data, the plurality of reservoir simulation parameters, a search method, and the plurality of reservoir equations.