System and method for quantitative analysis of borehole images

• US 10,997,705 B2
• Filed: 10/15/2019
• Issued: 05/04/2021
• Est. Priority Date: 10/25/2018
• Status: Active Grant

• Current Assignee: Chevron USA Incorporated (Chevron Corp.)
• Original Assignee: Chevron USA Incorporated (Chevron Corp.)
• Inventors: Gong, Bo, Keele, Dustin J., Toumelin, Emmanuel, Clinch, Simon
• Primary Examiner(s): Alam, Fayyaz
• Application Number:
  US16/601,818
• Publication Number:
  US 20200134803A1
• Time in Patent Office: 567 Days
• Field of Search: None
• CPC Class Codes: G01V 11/00 Prospecting or detecting by methods combining techniques covered by two or more of main groups G01V1/00 - G01V9/00
• CPC Class Codes: G01V 99/00 Subject matter not provided for in other groups of this subclass
• CPC Class Codes: G06N 20/00 Machine learning
• CPC Class Codes: G06N 3/08 Learning methods
• CPC Class Codes: G06T 2207/20081 Training; Learning
• CPC Class Codes: G06T 2207/20084 Artificial neural networks [ANN]
• CPC Class Codes: G06T 7/0002 Inspection of images, e.g. flaw detection

Abstract

A method is described for ways to generate a Fraction of Sand (F_sand) estimate and net-to-gross (NTG) estimate of sand in a formation using a machine-learning algorithm such as a neural network based on borehole image logs. The method may use the F_sand and other information to estimate hydrocarbons in place in a subsurface formation. The method may be executed by a computer system.

Claims

1. A computer-implemented method comprising:

a. receiving, at one or more computer processors, a borehole image, wherein the borehole image traverses a subsurface formation including sand and non-sand layers;

b. dividing, via the one or more computer processors, the borehole image into multiple divided borehole images, individual divided borehole images traversing an interval of the subsurface formation;

c. selecting, via the one or more computer processors, a subset of the multiple divided borehole images for use as training data;

d. labeling, via the one or more computer processors, the subset of the multiple divided borehole images with sand fraction values based on core analysis of the sand and non-sand layers in the subsurface formation;

e. training, via the one or more computer processors, a neural network using the subset of the multiple divided borehole images labeled with sand fraction values to generate a trained neural network; and

f. predicting, via the one or more computer processors, a fraction of sand (Fsand) in a subsurface volume using the trained neural network.

2. The method of claim 1 further comprising augmenting the training data prior to the labeling, wherein augmenting the training data includes altering the training data to provide additional samples for training the neural network.

3. The method of claim 2 wherein the augmenting the training data includes azimuthal shifts of the subset of the multiple divided borehole images in horizontal direction such that sinusoids representing the sand layers in the subset of the multiple divided borehole images appear with different phases.

4. The method of claim 2 wherein the augmenting the training data includes depth shifts of the subset of the multiple divided borehole images in vertical direction to teach the neural network that location of sand beds is irrelevant.

5. The method of claim 1 further comprising using the Fsand to estimate at least one of cumulative sand and hydrocarbon in place.

6. A computer system comprising:

one or more processors;

memory; and

one or more programs, wherein the one or more programs are stored in the memory and configured to be executed by the one or more processors, the one or more programs including instructions that when executed by the one or more processors cause the computer system to;

a. receive, at the one or more processors, a borehole image, wherein the borehole image traverses a subsurface formation including sand and non-sand layers;

b. divide, via the one or more processors, the borehole image into multiple divided borehole images, individual divided borehole images traversing an interval of the subsurface formation;

c. select, via the one or more processors, a subset of the multiple divided borehole images for use as training data;

d. label, via the one or more processors, the subset of the multiple divided borehole images with sand fraction values based on core analysis of the sand and non-sand layers in the subsurface formation;

e. train, via the one or more processors, a neural network using the subset of the multiple divided borehole images labeled with sand fraction values to generate a trained neural network; and

f. predict, via the one or more computer processors, a fraction of sand (Fsand) in a subsurface volume using the trained neural network.

7. The computer system of claim 6 further including instructions that when executed by the one or more processors cause the computer system to augment the training data prior to the labeling, wherein augmenting the training data includes altering the training data to provide additional samples for training the neural network.

8. The computer system of claim 6 further including instructions that when executed by the one or more processors cause the computer system to use the Fsand to estimate at least one of cumulative sand and hydrocarbon in place.

12. The computer system of claim 7 wherein the augmenting the training data includes azimuthal shifts of the subset of the multiple divided borehole images in horizontal direction such that sinusoids representing the sand layers in the subset of the multiple divided borehole images appear with different phases.

13. The computer system of claim 7 wherein the augmenting the training data includes depth shifts of the subset of the multiple divided borehole images in vertical direction to teach the neural network that location of sand beds is irrelevant.

9. A non-transitory computer readable storage medium storing one or more programs, the one or more programs comprising instructions, which when executed by an electronic device with one or more processors and memory, cause the device to execute:

a. receive, at one or more processors, a borehole image, wherein the borehole image traverses a subsurface formation including sand and non-sand layers;

b. divide, via the one or more processors, the borehole image into multiple divided borehole images, individual divided borehole images traversing an interval of the subsurface formation;

c. select, via the one or more processors, a subset of the multiple divided borehole images for use as training data;

d. label, via the one or more processors, the subset of the multiple divided borehole images with sand fraction values based on core analysis of the sand and non-sand layers in the subsurface formation;

e. train, via the one or more processors, a neural network using the subset of the multiple divided borehole images labeled with sand fraction values to generate a trained neural network; and

f. predict, via the one or more computer processors, a fraction of sand (Fsand) in a subsurface volume using the trained neural network.

10. The non-transitory computer readable storage medium of claim 9 further comprising instructions that when executed by the one or more processors cause the device to augment the training data prior to beginning the labeling, wherein augmenting the training data includes altering the training data to provide additional samples for training the neural network.

11. The computer system of claim 9 further comprising instructions that when executed by the one or more processors cause the computer system to use the Fsand to estimate at least one of cumulative sand and hydrocarbon in place.

14. The non-transitory computer readable storage medium of claim 10 wherein the augmenting the training data includes azimuthal shifts of the subset of the multiple divided borehole images in horizontal direction such that sinusoids representing the sand layers in the subset of the multiple divided borehole images appear with different phases.

15. The non-transitory computer readable storage medium of claim 10 wherein the augmenting the training data includes depth shifts of the subset of the multiple divided borehole images in vertical direction to teach the neural network that location of sand beds is irrelevant.