Electric submersible pump operating parameters

Abstract

This disclosure describes methods and systems for determining operating parameters for an electric submersible pump (ESP) and controlling the ESP based on the determined operating parameters. A method involves determining a target production rate for a wellbore; providing the target production rate as input to a neural network that provides as output ESP operating parameters to achieve the target production rate, the ESP operating parameters including a choke size percentage and a motor speed, and the neural network modeling an ESP-equipped wellbore; and controlling an ESP to operate according to the ESP operating parameters.

What is claimed is: 
     
         1 . A method for controlling an electric submersible pump (ESP) installed in a wellbore, the method comprising:
 determining a target production rate for the wellbore;   providing the target production rate as input to a neural network that provides as output ESP operating parameters to achieve the target production rate, the ESP operating parameters comprising a choke size percentage and a motor speed, and the neural network modeling an ESP-equipped wellbore; and   controlling the ESP to operate according to the ESP operating parameters.   
     
     
         2 . The method of  claim 1 , wherein the input to the neural network further comprises a real-time data comprising: an oil production rate, a water cut, an intake pressure, an ESP motor load, and an upstream-downstream (US/DS) differential pressure (DP). 
     
     
         3 . The method of  claim 1 , wherein the neural network is trained based on historical production data. 
     
     
         4 . The method of  claim 1 , wherein generating the neural network comprises:
 obtaining historical production data comprising data points associated with respective wells, each data point comprising at least one of oil production rate, a water cut, an intake pressure, an ESP motor load, or an upstream-downstream (US/DS) differential pressure (DP);   splitting the historical production into training data and testing data; and   iteratively training the neural network using the training data to generate the neural network.   
     
     
         5 . The method of  claim 4 , wherein iteratively training the neural network comprises:
 creating the neural network;   defining initial hyperparameters for the neural network;   training the neural network based on the training data;   determining, based on at least one performance indicator, whether the training of the neural network is complete;   if the training the neural network is complete, deploying the neural network; and   if the training the neural network is incomplete, returning to training the neural network using the training data to generate new hyperparameter values for the neural network.   
     
     
         6 . The method of  claim 5 , wherein determining, based on at least one performance indicator, whether the training of the neural network is complete comprises:
 determining whether the at least one performance indicator satisfies a respective threshold.   
     
     
         7 . The method of  claim 5 , wherein the at least one performance indicator comprises at least one of a correlation coefficient (CC), a root mean squared error (RMSE), or an average absolute percentage error (AAPE). 
     
     
         8 . The method of  claim 4 , wherein the initial hyperparameters comprise a number of neuron layers, a number of neurons per layer, and a seed number for the neural network. 
     
     
         9 . A system comprising:
 one or more processors configured to perform operations comprising:
 determining a target production rate for a wellbore equipped with an electric submersible pump (ESP); 
 providing the target production rate as input to a neural network that provides as output ESP operating parameters to achieve the target production rate, the ESP operating parameters comprising a choke size percentage and a motor speed, and the neural network modeling an ESP-equipped wellbore; and 
 controlling the ESP to operate according to the ESP operating parameters. 
   
     
     
         10 . The system of  claim 9 , wherein the input to the neural network further comprises a real-time data comprising: an oil production rate, a water cut, an intake pressure, an ESP motor load, and an upstream-downstream (US/DS) differential pressure (DP). 
     
     
         11 . The system of  claim 9 , wherein the neural network is trained based on historical production data. 
     
     
         12 . The system of  claim 9 , wherein generating the neural network comprises:
 obtaining historical production data comprising data points associated with respective wells, each data point comprising at least one of oil production rate, a water cut, an intake pressure, an ESP motor load, or an upstream-downstream (US/DS) differential pressure (DP);   splitting the historical production into training data and testing data; and   iteratively training the neural network using the training data to generate the neural network.   
     
     
         13 . The system of  claim 12 , wherein iteratively training the neural network comprises:
 creating the neural network;   defining initial hyperparameters for the neural network;   training the neural network based on the training data;   determining, based on at least one performance indicator, whether the training of the neural network is complete;   if the training the neural network is complete, deploying the neural network; and   if the training the neural network is incomplete, returning to training the neural network using the training data to generate new hyperparameter values for the neural network.   
     
     
         14 . The system of  claim 13 , wherein determining, based on at least one performance indicator, whether the training of the neural network is complete comprises:
 determining whether the at least one performance indicator satisfies a respective threshold.   
     
     
         15 . The system of  claim 13 , wherein the at least one performance indicator comprises at least one of a correlation coefficient (CC), a root mean squared error (RMSE), or an average absolute percentage error (AAPE). 
     
     
         16 . The system of  claim 12 , wherein the initial hyperparameters comprise a number of neuron layers, a number of neurons per layer, and a seed number for the neural network. 
     
     
         17 . A non-transitory computer storage medium encoded with instructions that, when executed by one or more computers, cause the one or more computers to perform operations comprising:
 determining a target production rate for a wellbore equipped with an electric submersible pump (ESP);   providing the target production rate as input to a neural network that provides as output ESP operating parameters to achieve the target production rate, the ESP operating parameters comprising a choke size percentage and a motor speed, and the neural network modeling an ESP-equipped wellbore; and   controlling the ESP to operate according to the ESP operating parameters.   
     
     
         18 . The non-transitory computer storage medium of  claim 17 , wherein the input to the neural network further comprises a real-time data comprising: an oil production rate, a water cut, an intake pressure, an ESP motor load, and an upstream-downstream (US/DS) differential pressure (DP). 
     
     
         19 . The non-transitory computer storage medium of  claim 17 , wherein the neural network is trained based on historical production data. 
     
     
         20 . The non-transitory computer storage medium of  claim 17 , wherein generating the neural network comprises:
 obtaining historical production data comprising data points associated with respective wells, each data point comprising at least one of oil production rate, a water cut, an intake pressure, an ESP motor load, or an upstream-downstream (US/DS) differential pressure (DP);   splitting the historical production into training data and testing data; and   iteratively training the neural network using the training data to generate the neural network.