Dynamic engine for a cognitive reservoir system

One or more non-transitory tangible computer-readable storage media storing computer-executable instructions executed by a processor to perform a method for developing reservoir models, the method comprising:

• generating a static model based on static data regarding a reservoir, the static data identifying one or more clusters of rock types and one or more faults in the clusters;
• generating a reservoir graph from the static model using a neural network, the reservoir graph representing each of the one or more clusters as a vertex, wherein each cluster has at least one of one pressure observation point, one perforation pass through, and a spatially continuous voxel set;
• propagating pressure values across a three-dimensional space of the reservoir based on nodal connectivity of neighboring clusters in accordance with the faults indicated by the static data;
• transforming the static data into one or more dynamic models of the reservoir based on the static model and the pressure values;
• evaluating target regions of the reservoir that correspond to volumes of the dynamic models, wherein evaluating each of the target regions includes using a set of one or more rules and fuzzy inference on values of a set of features of a corresponding volume;
• generating a ranked list of the target regions with information of where to drill based on the evaluation, each of the target regions ranked in accordance with an aggregated score assigned to the set of features of the corresponding volume;
• automatically updating the reservoir graph based on nodal connectivity of the inferred values of the corresponding volume of target regions;
• and
• outputting on a user interface the ranked list of the target regions identifying where to drill.

A system for developing reservoir models, the system comprising:

• one or more processors that execute:
• a static modeler comprising one or more neural networks configured to generate a static model of a reservoir based on received static data regarding a reservoir, the static data identifying one or more clusters of rock types and one or more faults in the clusters,
• a dynamic modeler in communication with the static modeler, the dynamic modeler using artificial intelligence to:
• generate a reservoir graph from the static model, the reservoir graph representing each cluster of the one or more clusters as a vertex, wherein each neighboring cluster has at least one of one pressure observation point, one perforation pass through, and a spatially continuous voxel set;
• propagate pressure values across a three-dimensional space of the reservoir based on nodal connectivity of neighboring clusters in accordance with the faults indicated by the static data, and
• transform the static model into one or more dynamic models of the reservoir based on the static model and the pressure values;
• and
• a reasoner that includes a fuzzy inference engine, wherein the reasoner using the fuzzy inference engine to:
• evaluate target regions of the reservoir that correspond to volumes of the dynamic models, wherein evaluating each of the target regions includes using a set of one or more rules and fuzzy inference on values of a set of features of a corresponding volume;
• generate a ranked list of the target regions with information of where to drill based on the evaluation, each of the target regions ranked in accordance with an aggregated score assigned to the set of features of the corresponding volume;
• automatically update the reservoir graph based on nodal connectivity of the inferred values of the corresponding volume of target regions;
• and
• outputting on a user interface the ranked list of the target regions identifying where to drill.