Method and system for predicting hydrocarbon data for unconventional reservoirs using machine learning

Abstract

A method may include obtaining reservoir data, hydraulic fracturing data, and static wellbore data for a geological region of interest. The method may further include obtaining temporal production data for the geological region of interest. The temporal production data may include a predetermined production rate with respect to a predetermined period of time. The method may further include determining various temporal features based on the temporal production data and an extraction process. The extraction process may include a deconvolution function that separates a portion of the temporal features from the predetermined production rate. The method may further include determining, using a machine-learning model, predicted hydrocarbon-in-place (HIP) data for the geological region of interest using the reservoir data, the hydraulic fracturing data, the static wellbore data, and the temporal features. The method may further include transmitting a command to a well control system based on the predicted HIP data.

What is claimed: 
     
         1 . A method, comprising:
 obtaining first reservoir data, first hydraulic fracturing data, and first static wellbore data for a geological region of interest;   obtaining first temporal production data for the geological region of interest, wherein the first temporal production data comprises a predetermined production rate with respect to a predetermined period of time;   determining, by a computer processor, a plurality of temporal features based on the first temporal production data and a first extraction process, wherein the first extraction process comprises a deconvolution function that separates a portion of the plurality of temporal features from the predetermined production rate;   determining, by the computer processor and using a machine-learning model, first predicted hydrocarbon-in-place (HIP) data for the geological region of interest using the first reservoir data, the first hydraulic fracturing data, the first static wellbore data, and the plurality of temporal features; and   transmitting, by the computer processor, a command to a well control system based on the first predicted HIP data.   
     
     
         2 . The method of  claim 1 , further comprising:
 obtaining maturity data regarding in-place organic material within the geological region of interest; and   determining, using a second extraction process, a maturity feature from the maturity data,   wherein the maturity feature is used by the machine-learning model to determine the first predicted HIP data.   
     
     
         3 . The method of  claim 2 ,
 wherein the maturity data comprise a maturity map that describes a plurality of kerogen quantities in the geological region of interest, and   wherein the maturity map is acquired using a plurality of drill cutting samples or a plurality of core samples from a plurality of wells.   
     
     
         4 . The method of  claim 1 , further comprising:
 obtaining a selection of a plurality of wells;   obtaining training data comprising second reservoir data, second hydraulic fracturing data, second static production data, and second temporal production data for the plurality of wells; and   performing a training operation on the machine-learning model iteratively using the training data until second predicted HIP data that is generated by the machine-learning model satisfies a predetermined criterion.   
     
     
         5 . The method of  claim 1 ,
 wherein the first temporal production data comprises gas production rate data,   wherein the first extraction process separates a plurality of predetermined gas rates and a plurality of respective gas time periods using a plurality of exponential decay curves, and   wherein the plurality of temporal features correspond to the plurality of predetermined gas rates and the plurality of respective gas time periods.   
     
     
         6 . The method of  claim 1 ,
 wherein the first temporal production data comprises gas specific density data, carbon dioxide composition data, δ 13  C composition data, methane composition data, liquid phase data, choke size data, or well head pressure data.   
     
     
         7 . The method of  claim 1 ,
 wherein the first predicted HIP data comprises molar ratio data of gas phase.   
     
     
         8 . The method of  claim 1 ,
 wherein the first reservoir data comprises geological data regarding one or more formation layers reservoir fluid data, reservoir pore pressure data, gamma ray log data, density log data, neutron long data, resistivity log data, permeability data, and porosity data, or open fracture data.   
     
     
         9 . The method of  claim 1 ,
 wherein the first hydraulic fracturing data comprises fracturing fluid data for a stimulation operation, injection rate data for a stimulation operation, injection consequence data, and hydraulic fracture geometry data.   
     
     
         10 . The method of  claim 1 ,
 wherein the first static wellbore data comprises well location data, well tubing data, and number of fractures adjacent to a wellbore.   
     
     
         11 . The method of  claim 1 , further comprising:
 determining a sweet spot region in the geological region of interest using the first predicted HIP data; and   determining a stimulation operation based on the sweet spot region and the first predicted HIP data,   wherein the command that is transmitted to the well control system is configured to cause performance of the stimulation operation.   
     
     
         12 . The method of  claim 1 , further comprising:
 determining a well path in the geological region of interest using the first predicted HIP data,   wherein the well control system is a drilling system, and   wherein the command causes the drilling system to perform a drilling operation based on the well path.   
     
     
         13 . The method of  claim 1 ,
 wherein the machine-learning model is an artificial neural network comprising an input layer, a plurality of hidden layers, and an output layer.   
     
     
         14 . A system, comprising:
 a stimulation control system coupled to a wellbore; and   a reservoir simulator coupled to the stimulation control system, wherein the reservoir simulator comprises a computer processor, the reservoir simulator is configured to perform a method comprising:
 obtaining first reservoir data, first hydraulic fracturing data, and first static wellbore data for a geological region of interest; 
 obtaining temporal production data for the geological region of interest, wherein the temporal production data comprises a predetermined production rate with respect to a predetermined period of time; 
 determining a plurality of temporal features based on the temporal production data and a first extraction process, wherein the first extraction process comprises a deconvolution function that separates a portion of the plurality of temporal features from the predetermined production rate; and 
 determining, using a machine-learning model, first predicted hydrocarbon-in-place (HIP) data for the geological region of interest using the first reservoir data, the first hydraulic fracturing data, the first static wellbore data, and the plurality of temporal features, 
   wherein the stimulation control system is configured to perform a hydraulic stimulation operation based on the first predicted HIP data.   
     
     
         15 . The system of  claim 14 , further comprising:
 a user device coupled to the stimulation control system,   wherein the user device is configured to provide a graphical user interface for presenting the first predicted HIP data.   
     
     
         16 . The system of  claim 14 , wherein the method further comprises:
 obtaining maturity data regarding in-place organic material within the geological region of interest; and   determining, using a second extraction process, a maturity feature from the maturity data,   wherein the maturity feature is used by the machine-learning model to determine the first predicted HIP data.   
     
     
         17 . The system of  claim 16 ,
 wherein the maturity data comprise a maturity map that describes a plurality of kerogen quantities in the geological region of interest, and   wherein the maturity map is acquired using a plurality of drill cutting samples or a plurality of core samples from a plurality of wells.   
     
     
         18 . The system of  claim 14 , wherein the method further comprises:
 obtaining a selection of a plurality of wells;   obtaining training data comprising second reservoir data, second hydraulic fracturing data, second static production data, and second temporal production data for the plurality of wells; and   performing a training operation on the machine-learning model iteratively using the training data until second predicted HIP data that is generated by the machine-learning model satisfies a predetermined criterion.   
     
     
         19 . A system, comprising:
 a drilling system comprising a plurality of sensors and a drill string comprising a drill bit, wherein the drilling system is coupled to a wellbore; and   a reservoir simulator coupled to the drilling system, wherein the reservoir simulator comprises a computer processor, the reservoir simulator is configured to perform a method comprising:
 obtaining reservoir data, hydraulic fracturing data, and static wellbore data for a geological region of interest; 
 obtaining temporal production data for the geological region of interest, wherein the temporal production data comprises a predetermined production rate with respect to a predetermined period of time; 
 determining a plurality of temporal features based on the temporal production data and a first extraction process, wherein the first extraction process comprises a deconvolution function that separates a portion of the plurality of temporal features from the predetermined production rate; and 
 determining, using a machine-learning model, predicted hydrocarbon-in-place (HIP) data for the geological region of interest using the reservoir data, the hydraulic fracturing data, the static wellbore data, and the plurality of temporal features, 
   wherein the drilling system is configured to perform a drilling operation for a well path based on the predicted HIP data.   
     
     
         20 . The system of  claim 19 , wherein the method further comprises:
 obtaining maturity data regarding in-place organic material within the geological region of interest; and   determining, using a second extraction process, a maturity feature from the maturity data,   wherein the maturity feature is used by the machine-learning model to determine the predicted HIP data.