US 12,412,001 B2
Generating well model flow tables for artificial intelligent models
Obiomalotaoso Leonard Isichei, Abqaiq (SA); Ayman Alkhalaf, Dammam (SA); Said Almalki, Dhahran (SA); and Najmul Hasan Ansari, Dhahran (SA)
Assigned to Saudi Arabian Oil Company, Dhahran (SA)
Filed by Saudi Arabian Oil Company, Dhahran (SA)
Filed on Oct. 12, 2021, as Appl. No. 17/499,026.
Prior Publication US 2023/0114355 A1, Apr. 13, 2023
Int. Cl. G06F 30/12 (2020.01); G06F 8/30 (2018.01); G06F 30/27 (2020.01); G06N 20/00 (2019.01)
CPC G06F 30/12 (2020.01) [G06F 8/315 (2013.01); G06F 30/27 (2020.01); G06N 20/00 (2019.01)] 20 Claims
OG exemplary drawing
 
1. A computer-implemented method, comprising:
receiving historical well data corresponding to a well, wherein the historical well data comprises information about past operations of the well;
generating an expanded range of values for parameters used in a well model for the well, wherein the expanded range of values expands a range of values corresponding to the historical well data;
generating, using the historical well data and the expanded range of values, object-oriented object instances representing the historical well data, wherein each object-oriented object instance is an object configured to be used by the well model and comprising a combination of parameters within the expanded range of values;
executing the well model using the object-oriented object instances to generate result objects, wherein each result object is an object-oriented object instance representing a result of the well model;
generating, using the object-oriented object instances and the result objects, a hybrid well model variant of the well model that models combinations of the object-oriented object instances and the result objects;
passing, by an object-oriented application, a data object encapsulating the object-oriented object instances and the result objects to a machine learning model, wherein passing the data object occurs in an absence of database tables;
executing the machine learning model using the data object;
generating, in response to the executing, an electric submersible pump (ESP) optimization dashboard displaying suggestions for production improvements for a well; and
implementing user-selected suggestions to change settings on ESPs used in the well.