Data-driven model for control and optimization of hydrocarbon production

Abstract

A system for controlling and optimizing hydrocarbon production may include sensors that capture sensor data pertaining to wellhead pressure values in a well. The system may also include a multiphase flow meter that captures production data pertaining to multiphase production flow rates of the well. The system may include an access module to access an estimated parameter value associated with a second time and a parameter that pertains to production. The estimated parameter value is predicted by a data-driven model for describing production fluid dynamics of the well, based on the sensor data and the production data obtained at a first time. The system includes a processor to update the data-driven model using a data assimilation algorithm and the production data received at the second time. The processor generates, using the updated data-driven model, an optimal control setting of a control tool for causing an adjustment to a production system.

What is claimed: 
     
         1 . A method comprising:
 accessing an estimated parameter value associated with a second time and a parameter, the parameter pertaining to production from a well, the estimated parameter value being predicted by a data-driven model for describing production fluid dynamics of the well, based on sensor data and production data obtained at a first time;   updating the data-driven model using a data assimilation algorithm and the production data obtained during a production process at the second time, the updating being performed using one or more hardware processors; and   generating, using the updated data-driven model, an optimal control setting of a control tool for causing an adjustment to a production system.   
     
     
         2 . The method of  claim 1 , further comprising:
 accessing the sensor data captured by one or more sensors from the well at the first time, and production data for the well at the first time;   extracting dynamically relevant process data from the sensor data and the production data using a dynamic mode decomposition (DMD) algorithm; and   training the data-driven model, based on the extracted dynamically relevant process data, to predict the estimated parameter value associated with the second time.   
     
     
         3 . The method of  claim 2 , wherein the dynamically relevant process data characterizes a temporal change in a system state. 
     
     
         4 . The method of  claim 1 , further comprising:
 generating an instruction for the control tool to make the adjustment to the production system based on the optimal control setting.   
     
     
         5 . The method of  claim 4 , further comprising:
 executing the instruction during the production process, the executing of the instruction causing the adjustment, by the control tool, to the production system.   
     
     
         6 . The method of  claim 1 , wherein the control tool is a production valve. 
     
     
         7 . The method of  claim 1 , wherein the control tool is a choke assembly. 
     
     
         8 . The method of  claim 1 , wherein the data-driven model is a lower-order, non-linear data-driven model. 
     
     
         9 . The method of  claim 1 , wherein the optimal control setting is generated using an optimization algorithm to maximize hydrocarbon recovery over a particular period of production. 
     
     
         10 . The method of  claim 1 , wherein the parameter is at least one of a multiphase production flow rate, a wellbore pressure, or a temperature. 
     
     
         11 . The method of  claim 1 , wherein the updating of the data-driven model using the data assimilation algorithm and production data obtained during the production process includes:
 adjusting the estimated parameter value to match an actual measurement value obtained during the production process.   
     
     
         12 . A system comprising:
 one or more sensors arranged to capture sensor data pertaining to one or more wellhead pressure values in a well;   a multiphase flow meter arranged to capture production data pertaining to multiphase production flow rates of the well;   an access module configured to access an estimated parameter value associated with a second time and a parameter, the parameter pertaining to production from the well, the estimated parameter value being predicted by a data-driven model for describing production fluid dynamics of the well, based on the sensor data and the production data obtained at a first time; and   one or more hardware processors configured to: 
 update the data-driven model using a data assimilation algorithm and the production data received during a production process at the second time; and 
 generating, using the updated data-driven model, an optimal control setting of a control tool for causing an adjustment to a production system. 
   
     
     
         13 . The system of  claim 12 , further comprising:
 an access module configured to access the sensor data captured by one or more sensors from the well at the first time, and production data for the well at the first time;   wherein the one or more hardware processors are further configured to: 
 extract dynamically relevant process data from the sensor data and the production data using a dynamic mode decomposition (DMD) algorithm; and 
 train the data-driven model, based on the extracted dynamically relevant process data, to predict the estimated parameter value associated with the second time. 
   
     
     
         14 . The system of  claim 12 , wherein the one or more hardware processors are further configured to generate an instruction for the control tool to make the adjustment to the production system based on the optimal control setting. 
     
     
         15 . The system of  claim 14 , wherein the one or more hardware processors are further configured to execute the instruction during the production process, the executing of the instruction causing the adjustment, by the control tool, to the production system. 
     
     
         16 . The system of  claim 12 , wherein the optimal control setting is generated using an optimization algorithm to maximize hydrocarbon recovery over a particular period of production. 
     
     
         17 . The system of  claim 12 , wherein the parameter is at least one of a multiphase production flow rate, a wellbore pressure, or a temperature. 
     
     
         18 . The system of  claim 12 , wherein the updating of the data-driven model using the data assimilation algorithm and production data obtained during the production process includes:
 adjusting the estimated parameter value to match an actual measurement value obtained during the production process.   
     
     
         19 . A non-transitory machine-readable storage medium comprising instructions that, when executed by one or more processors of a machine, cause the machine to perform operations comprising:
 accessing an estimated parameter value associated with a second time and a parameter, the parameter pertaining to production from a well, the estimated parameter value being predicted by a data-driven model for describing production fluid dynamics of the well, based on sensor data and production data obtained at a first time;   updating the data-driven model using a data assimilation algorithm and the production data obtained during a production process at the second time; and   generating, using the updated data-driven model, an optimal control setting of a control tool for causing an adjustment to a production system.   
     
     
         20 . The non-transitory machine-readable storage medium of  claim 19 , wherein the operations further comprise:
 accessing the sensor data captured by one or more sensors from the well at the first time, and production data for the well at the first time;   extracting dynamically relevant process data from the sensor data and the production data using a dynamic mode decomposition (DMD) algorithm; and   training the data-driven model, based on the extracted dynamically relevant process data, to predict the estimated parameter value associated with the second time.