Reservoir simulation systems and methods to dynamically improve performance of reservoir simulations

A reservoir simulation system, comprising:

• a storage medium for storing data associated with simulations of a reservoir;
• and
• one or more processors operable to:
• obtain one or more input variables for generating a reservoir simulation of a reservoir, the reservoir simulation comprising a plurality of reservoir simulation runs, and the reservoir simulation being associated with at least one of one or more internal parameters and one or more external parameters that control performance of the reservoir simulation;
• generate the reservoir simulation based on the one or more input variables;
• during the reservoir simulation, determine a variance of computation time for processing one or more processes of the reservoir simulation;
• in response to a determination that the variance of computation time is less than or equal to a threshold, perform a first sequence of Bayesian Optimizations of the at least one of the one or more internal parameters and the one or more external parameters to improve performance of the reservoir simulation;
• and
• in response to at least one of the determination that the variance of computation time is greater than the threshold, perform a second sequence of Bayesian Optimizations of the at least one of the one or more internal parameters and the one or more external parameters to improve performance of the reservoir simulation,
• wherein the one or more internal parameters are parameters of one or more algorithms used to generate the reservoir simulation, and
• wherein the one or more external parameters are parameters of one or more hardware components used to generate the reservoir simulation.

A machine-readable medium comprising instructions stored therein, which when executed by one or more processors, causes the one or more processors to perform operations comprising:

• obtaining one or more input variables for generating a reservoir simulation of a reservoir, the reservoir simulation comprising a plurality of reservoir simulation runs, and the reservoir simulation being associated with at least one of one or more internal parameters and one or more external parameters that control performance of the reservoir simulation;
• generating the reservoir simulation based on the one or more input variables;
• during the reservoir simulation, determining a variance of computation time for processing one or more processes of the reservoir simulation;
• in response to a determination that the variance of computation time is less than or equal to a threshold,
• during each period of time of a threshold number of periods of time:
• generating one or more reservoir simulation runs of the reservoir simulation, wherein the one or more reservoir simulation runs are controlled by the at least one of the one or more internal parameters and the one or more external parameters;
• determining an objective function of the reservoir simulation based on the at least one of the one or more internal parameters and the one or more external parameters;
• and
• performing a Bayesian Optimization of the at least one of the one or more internal parameters and the one or more external parameters to obtain new values for the at least one of the one or more internal parameters and the one or more external parameters,
• wherein one or more reservoir simulation runs during a next period of time of the threshold number of periods of time are controlled by the new values for the at least one of the one or more internal parameters and the one or more external parameters, and
• in response to a determination that the variance of computation time is 1 greater than the threshold, during each period of time of a threshold number of periods of time:
• for each iteration of a threshold number of iterations:
• generating one or more reservoir simulation runs of the reservoir simulation for a threshold period of time, wherein the one or more reservoir simulation runs are controlled by the one or more internal parameters and the one or more external parameters;
• determining an objective function of the reservoir simulation based on the at least one of the one or more internal parameters and the one or more external parameters;
• and
• performing a Bayesian Optimization of the at least one of the one or more internal parameters and the one or more external parameters to obtain new values for the at least one of the one or more internal parameters and the one or more external parameters,
• wherein one or more reservoir simulation runs in a next iteration of the threshold number of iterations are controlled by the new values for the at least one of the one or more internal parameters and the one or more external parameters.