SYSTEM AND METHOD FOR USING AN ARTIFICIAL NEURAL NETWORK TO SIMULATE PIPE HYDRAULICS IN A RESERVOIR SIMULATOR

• US 20130304679A1
• Filed: 01/31/2011
• Published: 11/14/2013
• Est. Priority Date: 01/31/2011
• Status: Active Grant

• Current Assignee: Landmark Graphics Corporation (Halliburton Company)
• Original Assignee: Landmark Graphics Comporation
• Inventors: Fleming, Graham, Wong, Terry
• Granted Patent: US 10,055,684 B2
• Time in Patent Office: Days
• US Class Current: 706/14
• CPC Class Codes: E21B 43/00 Methods or apparatus for obtaining oil, gas, water, soluble or meltable materials or a slurry of minerals from wells applicable only to water E03B
• CPC Class Codes: G05B 13/027 using neural networks only
• CPC Class Codes: G05B 17/02 electric
• CPC Class Codes: G06N 3/08 Learning methods

Abstract

A method for implementing a reservoir simulator is described. The method comprises developing training data by performing a calculation on an initial set of input data relating to reservoir conditions to obtain corresponding output data; training an artificial neural network (“ANN”) to perform the calculation using the training data; and using the trained ANN to perform the calculation on a second set of input data to obtain corresponding output data for use by the reservoir simulator in performing simulations.

Claims

1. A method for implementing a reservoir simulator, the method comprising:

developing training data by performing a calculation on an initial set of input data relating to reservoir conditions to obtain corresponding output data relating to reservoir conditions;

training an artificial neural network (“

ANN”

) to perform the calculation using the training data; and

using the trained ANN to perform the calculation on a second set of input data relating to reservoir conditions to obtain corresponding output data for use by the reservoir simulator in performing simulations.

2. The method of claim 1 wherein the ANN comprises a proxy model for simulating pipe hydraulics.

3. The method of claim 2 wherein each of the initial set of input data and the second set of input data comprises at least one of a flow rate, fluid composition, inlet pressure, outlet pressure, and fluid temperature distribution.

4. The method of claim 2 wherein the output data comprises a pressure drop.

5. The method of claim 1 further comprising, prior to the training, extending the initial set of input data by extrapolation.

6. The method of claim 1 further comprising, subsequent to the using, retraining the ANN.

7. The method of claim 6 wherein the retraining is performed responsive to a determination that accuracy of the ANN has fallen below an acceptable level.

8. The method of claim 6 wherein the retraining is performed periodically.

9. The method of claim 1 further comprising using the ANN to provide numeric derivatives by perturbing the second set of input data about current conditions.

10. The method of claim 1 wherein, subsequent to a first run of the reservoir simulator employing the ANN, reusing the initial set of input data and corresponding output data in a subsequent run of the reservoir simulator using the ANN.

11. A computer-readable medium having stored thereon instructions executable by a processor to perform a method comprising:

developing training data by performing a calculation on an initial set of input data relating to reservoir conditions to obtain corresponding output data;

training an artificial neural network (“

ANN”

) to perform the calculation using the training data; and

using the trained ANN to perform the calculation on a second set of input data relating to reservoir conditions to obtain corresponding output data for use by the reservoir simulator in performing simulations.

12. The computer-readable medium of claim 11 wherein the ANN comprises a proxy model for simulating pipe hydraulics, wherein each of the initial set of input data and the second set of input data comprises at least one of a flow rate, fluid composition, inlet pressure, outlet pressure, and fluid temperature distribution, and wherein the output data comprises a pressure drop.

13. The computer-readable medium of claim 11 wherein the method further comprises, prior to the training, extending the initial set of input data by extrapolation.

14. The computer-readable medium of claim 11 wherein the method further comprises retraining the ANN responsive to a determination that accuracy of the ANN has fallen below an acceptable level.

15. The computer-readable medium of claim 11 wherein the method further comprises using the ANN to provide numeric derivatives by perturbing the second set of input data about current conditions.

16. The computer-readable medium of claim 11 wherein the method further comprises, subsequent to a first run of the reservoir simulator employing the ANN, reusing the initial set of input data and corresponding output data in a subsequent run of the reservoir simulator using the ANN.

17. A system for implementing a reservoir simulator, the system comprising:

means for developing training data by performing a calculation on an initial set of input data relating to reservoir conditions to obtain corresponding output data;

means for training an artificial neural network (“

ANN”

) to perform the calculation using the training data; and

means for using the trained ANN to perform the calculation on a second set of input data relating to reservoir conditions to obtain corresponding output data for use by the reservoir simulator in performing simulations.

18. The system of claim 17 wherein the ANN comprises a proxy model for simulating pipe hydraulics, wherein each of the initial set of input data and the second set of input data comprises at least one of a flow rate, fluid composition, inlet pressure, outlet pressure, and fluid temperature distribution, and wherein the output data comprises a pressure drop.

19. The system of claim 17 further comprising means for extending the initial set of input data by extrapolation prior to the training.

20. The system of claim 17 further comprising means for retraining the ANN.

21. A system for implementing a reservoir simulator, the system comprising:

a computer-implemented module for developing training data by performing a calculation on an initial set of input data relating to reservoir conditions to obtain corresponding output data;

a computer-implemented module for training an artificial neural network (“

ANN”

) to perform the calculation using the training data; and

a computer-implemented module for using the trained ANN to perform the calculation on a second set of input data relating to reservoir conditions to obtain corresponding output data for use by the reservoir simulator in performing simulations.

22. The system of claim 21 wherein the ANN comprises a proxy model for simulating pipe hydraulics, wherein each of the initial set of input data and the second set of input data comprises at least one of a flow rate, fluid composition, inlet pressure, outlet pressure, and fluid temperature distribution, and wherein the output data comprises a pressure drop.

23. The system of claim 21 further comprising a computer-implemented module for extending the initial set of input data by extrapolation prior to the training.

24. The system of claim 21 further comprising a computer-implemented module for retraining the ANN.