Machine-learning based system for virtual flow metering
Various aspects described herein relate to a system that utilized deep learning and neural networks to estimate/predict an amount of natural resource production in a well given a set of parameters indicative of physical changes to the well. In one aspect, a virtual flow meter includes memory having computer-readable instructions stored therein and one or more processors configured to execute the computer-readable instructions to receive one or more input parameters indicative of physical changes to at least one well; apply the one or more input parameters to a trained neural network architecture; and determine one or more outputs of the trained neural network architecture, the one or more outputs corresponding to predicted fluid output of the at least one well.
1. A virtual flow meter, comprising:
memory having computer-readable instructions stored therein; and
one or more processors configured to execute the computer-readable instructions to:
receive one or more input parameters indicative of one or more physical changes to an injector well from a user interface, the one or more input parameters including-a rate of injection into the injector well, a change in well pressure of the injector well, a lift rate, a mode of operation, a temperature of the injector well, operational history of the injector well, and completion metadata of the injector well;
apply the one or more input parameters to a trained neural network architecture;
determine one or more outputs of the trained neural network architecture, the one or more outputs corresponding to a predicted fluid output of a producing well; and
indicate, via an output device, the one or more outputs.
2. The virtual flow meter of claim 1 , wherein the one or more processors are configured to execute the computer-readable instructions to train the trained neural network architecture prior to receiving the one or more input parameters.
3. The virtual flow meter of claim 2 , wherein the one or more processors are configured to execute the computer-readable instructions to train the trained neural network architecture using data collected for the injector at least one well over a period of time.
4. The virtual flow meter of claim 3 , wherein the one or more processors are configured to execute the computer-readable instructions to detrend the data by subtracting off a value of an endpoint of a sequence from all other points in the sequence.
5. The virtual flow meter of claim 3 , wherein the one or more processors are configured to execute the computer-readable instructions to normalize the data after detrending using minimum and maximum values of all sequences of the data.
6. The virtual flow meter of claim 2 , wherein the one or more processors are configured to execute the computer-readable instructions to optimize the trained neural network architecture using a dynamic programing optimization method and the one or more outputs.
7. The virtual flow meter of claim 1 , wherein the one or more outputs include an oil production rate and a water production rate.
8. The virtual flow meter of claim 1 , wherein the virtual flow meter is accessible via a terminal by an operator of the injector ell.
9. The virtual flow meter of claim 1 , wherein the injector well and the producing well are a steam assisted gas drainage well pair.
10. A method of operating a virtual flow meter, comprising:
receiving, via a user interface, one or more input parameters indicative of physical changes to an injector well, the one or more input parameters including a rate of steam injection into the injector well, a change in well pressure of the injector well, a lift rate, a mode of operation, a temperature of the injector well, operational history of the injector well, and completion metadata of the injector well;
applying the one or more input parameters to a trained neural network architecture;
determining one or more outputs of the trained neural network architecture, the one or more outputs corresponding to predicted fluid output of a producing the at least one well; and
indicating, via an output device, the one or more outputs.
11. The method of claim 10 , further comprising:
training the trained neural network architecture prior to receiving the one or more input parameters.
12. The method of claim 11 , wherein the trained neural network architecture is trained using data collected for the injector well over a period of time.
13. The method of claim 12 , wherein the training includes detrending the data by subtracting off a value of an endpoint of a sequence from all other points in the sequence.
14. The method of claim 12 , further comprising:
normalizing the data after detrending using minimum and maximum values of all sequences of the data.
15. The method of claim 11 , further comprising:
optimizing the trained neural network architecture using a dynamic programing optimization method and the one or more outputs.
16. The method of claim 10 , wherein the one or more outputs include an oil production rate and a water production rate.
17. The method of claim 10 , wherein the virtual flow meter is accessible via a terminal by an operator of the injector well.
18. The method of claim 10 , wherein the injector well and the producing well are a steam assisted gas drainage well pair.