Multi-objective completion parameters optimization for a wellbore using Bayesian optimization
A system for determining completion parameters for a wellbore includes a sensor and a computing device. The sensor can be positioned at a surface of a wellbore to detect data prior to finishing a completion stage for the wellbore. The computing device can receive the data, perform a history match for simulation and production using the sensor data and historical data, generate inferred data for completion parameters using the historical data identified during the history match, predict stimulated area and production by inputting the inferred data into a neural network model, determine completion parameters for the wellbore using Bayesian optimization on the stimulated area and production from the neural network model, profit maximization, and output the completion parameters for determining completion decisions for the wellbore.
1. A system comprising:
a sensor positionable at a surface of a wellbore to detect data prior to finishing a completion stage for the wellbore;
a computing device to receive data detected by the sensor, the computing device comprising:
a processor; and
a non-transitory computer-readable medium on which is code that is executable by the processor to perform operations, the operations comprising:
performing a history match for stimulation and production using the data detected by the sensor and historical data;
generating inferred data for completion parameters using the historical data identified during the history match;
predicting stimulated area and production by inputting the inferred data into a neural network model;
determining completion parameters for the wellbore using Bayesian optimization on a cost, the stimulated area, and the production from the neural network model; and
outputting the completion parameters for determining completion decisions for the wellbore.
2. The system of claim 1 , wherein the neural network model is a deep neural network (DNN) that is trained using surface data and production data from wellbores other than the wellbore.
3. The system of claim 1 , wherein the data detected by the sensor includes positional and size information about fractures in the wellbore and sources for creating the fractures.
4. The system of claim 1 , wherein the completion decisions include a number of holes to perforate, a location of holes to perforate, and a diameter for one or more perforations.
5. The system of claim 1 , wherein the non-transitory computer-readable medium includes code for determining the completion parameters for the wellbore using Bayesian optimization on the cost, stimulated area, and production from the neural network model by determining completion parameters for the stimulated area and for production and cost considerations.
6. The system of claim 1 , wherein the inferred data includes values for parameters in addition to the parameters associated with the data from the sensor.
7. The system of claim 6 , wherein the non-transitory computer-readable medium includes code for generating the inferred data by adjusting parameter values based on a difference between values for other parameters included in the data from the sensor.
8. A method comprising:
receiving data from a sensor at a surface of a wellbore;
performing, by a computing device, a history match for simulation and production using the data detected by the sensor and historical data;
generating, by the computing device, inferred data for completion parameters using the historical data identified during the history match;
predicting, by the computing device, stimulated area and production by inputting the inferred data into a neural network model;
determining, by the computing device, completion parameters for the wellbore using Bayesian optimization on a cost, the stimulated area, and the production from the neural network model; and
outputting, by the computing device, the completion parameters for determining completion decisions for the wellbore.
9. The method of claim 8 , wherein the data detected by the sensor includes positional and size information about fractures in the wellbore and sources for creating the fractures.
10. The method of claim 8 , wherein the completion decisions include a number of holes to perforate, a location of holes to perforate, and a diameter for one or more perforations.
11. The method of claim 8 , wherein determining the completion parameters for the wellbore using Bayesian optimization on the cost, stimulated area, and production from the neural network model includes determining completion parameters for the stimulated area and for production and cost considerations.
12. The method of claim 8 , wherein the inferred data includes values for parameters in addition to the parameters associated with the data from the sensor.
13. The method of claim 12 , wherein generating the inferred data includes adjusting parameter values based on a difference between values for other parameters included in the data from the sensor.
14. A non-transitory computer-readable medium that includes instructions that are executable by a processing device for causing the processing device to perform operations comprising:
receiving data from a sensor at a surface of a wellbore;
performing, by a computing device, a history match for simulation and production using the data detected by the sensor and historical data;
generating, by the computing device, inferred data for completion parameters using the historical data identified during the history match;
predicting, by the computing device, stimulated area and production by inputting the inferred data into a neural network model;
determining, by the computing device, completion parameters for the wellbore using Bayesian optimization on a cost, the stimulated area, and the production from the neural network model; and
outputting, by the computing device, the completion parameters for determining a number of holes to perforate, a location of holes to perforate, and a diameter for one or more perforations for the wellbore.
15. The non-transitory computer-readable medium of claim 14 , wherein the neural network model is a deep neural network (DNN) that is trained using surface data and production data from wellbores other than the wellbore.
16. The non-transitory computer-readable medium of claim 14 , wherein the data detected by the sensor includes positional and size information about fractures in the wellbore and sources for creating the fractures.
17. The non-transitory computer-readable medium of claim 14 , wherein the non-transitory computer-readable medium includes code for determining the completion parameters for the wellbore using Bayesian optimization on the cost, stimulated area, and production from the neural network model by determining completion parameters for the stimulated area and for production and cost considerations.
18. The non-transitory computer-readable medium of claim 14 , wherein the inferred data includes values for parameters in addition to the parameters associated with the data from the sensor, wherein the non-transitory computer-readable medium includes code for generating the inferred data by adjusting parameter values based on a difference between values for other parameters included in the data from the sensor.
19. The non-transitory computer-readable medium of claim 14 , further comprising code for performing the history match using a physics-based model.