Sand monitoring and control system
A method can include receiving real-time downhole time series sensor data during production of fluid from a well in fluid communication with a formation reservoir; transforming the real-time downhole time series sensor data to values for a set of predefined model features; detecting a downhole sand event using the values as input to a trained neural network model; and issuing a signal responsive to detection of the downhole sand event.
1 . A method comprising:
receiving real-time downhole time series sensor data during production of fluid from a well in fluid communication with a formation reservoir;
transforming the real-time downhole time series sensor data to values for a set of predefined model features, wherein at least one of the predefined model features is a product of at least two different types of sensor data including pressure sensor data and temperature sensor data;
detecting a downhole sand event using the values as input to a trained neural network model; and
responsive to detecting the downhole sand event, adjusting operation of pump equipment disposed in the well.
2 . The method of claim 1 , wherein the real-time downhole time series sensor data comprise one or more of pressure data and temperature data.
3 . The method of claim 1 , wherein at least one of the predefined model features does not depend on an amplitude offset of the real-time downhole time series sensor data.
4 . The method of claim 1 , wherein at least one of the predefined model features is a derivative with respect to time.
5 . The method of claim 1 , wherein two of the predefined model features are derivatives with respect to time.
6 . The method of claim 5 , wherein the derivatives with respect to time comprise a first derivative of pressure data with respect to time and a first derivative of temperature data with respect to time.
7 . The method of claim 1 , wherein the real-time downhole time series sensor data are acquired using at least one sensor of pump equipment disposed in the well.
8 . The method of claim 7 , wherein the pump equipment comprises an electric submersible pump.
9 . The method of claim 1 , wherein receiving the real-time downhole time series sensor data includes:
receiving, during production of fluid, first real-time downhole time series sensor data from a pressure sensor positioned in a well, the pressure sensor associated with the pump equipment disposed in the well; and
receiving, during the production of fluid, second real-time downhole time series sensor data from a temperature sensor positioned in the well, the temperature sensor associated with the pump equipment disposed in the well.
10 . The method of claim 1 , wherein adjusting operation of the pump equipment includes controlling operation of at least one field component.
11 . The method of claim 10 , wherein the at least one field component comprises an adjustable choke.
12 . The method of claim 10 , wherein the at least one field component comprises an electric submersible pump.
13 . The method of claim 10 , wherein the at least one field component comprises a gas lift component.
14 . The method of claim 1 , wherein the trained neural network model comprises a convolution neural network model.
15 . The method of claim 14 , wherein the transforming comprises transforming the real-time downhole time series sensor data to a multidimensional array.
16 . The method of claim 15 , wherein the multidimensional array comprises a pixel array.
17 . A system comprising:
a processor;
memory accessible to the processor; and
processor-executable instructions stored in the memory to instruct the system to:
receive real-time downhole time series sensor data during production of fluid from a well in fluid communication with a formation reservoir;
transform the real-time downhole time series sensor data to values for a set of predefined model features, wherein at least one of the predefined model features is a product of at least two different types of sensor data including pressure sensor data and temperature sensor data;
detect a downhole sand event using the values as input to a trained neural network model; and
responsive to detecting the downhole sand event, adjust operation of pump equipment disposed in the well.
18 . One or more computer-readable storage media comprising processor-executable instructions to instruct a computing system to:
receive real-time downhole time series sensor data during production of fluid from a well in fluid communication with a formation reservoir;
transform the real-time downhole time series sensor data to values for a set of predefined model features, wherein at least one of the predefined model features is a product of at least two different types of sensor data including pressure sensor data and temperature sensor data;
detect a downhole sand event using the values as input to a trained neural network model; and
responsive to detecting the downhole sand event, adjust operation of pump equipment disposed in the well.