System and method for facies classification

Patent metadata

Field	Value
Publication number	US-9274249-B2
Application number	US-201213488876-A
Country	US
Kind code	B2
Filing date	Jun 5, 2012
Priority date	Jun 5, 2012
Publication date	Mar 1, 2016
Grant date	Mar 1, 2016

Abstract

A method of automatically interpreting well log data indicative of physical attributes of a portion of a subterranean formation which include some portion of samples with known facies classification to be used as training data, dividing the training data into two subsets, a calibration set and a cross-validation set, using an automated supervised learning facies identification method to determine a preliminary identification of facies in the subterranean formation based on the calibration set, calculating a confusion matrix for the supervised learning facies identification method by comparing predicted and observed facies for the cross-validation set, calculating a facies transition matrix characterizing changes between contiguous facies, and using the preliminary identification, the facies transition matrix, and the confusion matrix, iteratively calculating updated facies identifications.

First claim

I claim:

1. A method of automatically interpreting well log data indicative of physical attributes of a portion of a subterranean formation comprising:
obtaining training data comprising well logs including facies classification information for at least a portion of the well logs;
dividing the training data into two subsets, a training set and a cross-validation set;
using an automated supervised learning facies identification method to determine a preliminary identification of facies in the subterranean formation based on the training set;
calculating a confusion matrix for the supervised learning facies identification method by comparing predicted and observed facies for the cross-validation set;
calculating a facies transition matrix characterizing changes between contiguous facies;
using the preliminary identification, the facies transition matrix, and the confusion matrix, to iteratively calculate updated facies identifications, wherein the facies identifications are used for selection of a location for drilling in the subterranean formation for hydrocarbon resources.

2. A method as in claim 1 , comprising repeating the iteratively calculating, each time using a differing random seed for a random path to create a plurality of realizations of the updated facies identifications.

3. A method as in claim 2 , wherein the plurality of realizations are used to generate a transition probability for each of a plurality of nodes.

4. A method as in claim 3 , further comprising, creating a reservoir probability volume for the subterranean formation.

5. A method as in claim 2 , wherein the plurality of realizations are used to generate a probability for a facies classification for each of a plurality of facies.

6. A method as in claim 1 , wherein the iteratively calculating comprises applying the confusion matrix for each of a plurality of nodes for which P ij <T ij , where P ij is the predicted matrix and T ij is target transition matrix.

7. A method as in claim 6 , wherein the applying is first performed for T ij for which i=j, then P ij is updated before the applying is performed for T ij for which i≠j.

8. A method as in claim 6 , wherein the target transition matrix is defined as T ij =P ij (l-w) O ij w where O ij is an observed matrix, and w=|(P ij −O ij P ij +O ij )| α , wherein α is a user defined constraint.

9. A system for automatically interpreting well log data indicative of physical attributes of a portion of a subterranean formation, the well log data comprising well logs including facies classification information for at least a portion of the well logs and being divided into two subsets, a training set and a cross-validation set, the system comprising: one or more processors configured to execute computer program modules, the computer program modules comprising:
an automated supervised learning facies identification module configured to perform an automated supervised learning method to determine a preliminary identification of facies in the subterranean formation based on the calibration set;
a confusion matrix calculating module configured to calculate a confusion matrix for the supervised learning facies identification method by comparing predicted and observed facies for the cross-validation set;
a facies transition matrix module configured to calculate a facies transition matrix characterizing changes between contiguous facies; and
an updated identification calculating module configured to use the preliminary identification, the facies transition matrix, and the confusion matrix, to iteratively calculate updated facies identifications, wherein the facies identifications are used for selection of a location for drilling in the subterranean formation for hydrocarbon resources.

10. A non-transitory, tangible medium encoded with machine executable instructions for performing a method of automatically interpreting well log data indicative of physical attributes of a portion of a subterranean formation, the well log data comprising well logs including facies classification information for at least a portion of the well logs and being divided into two subsets, a training set and a cross-validation set, the method comprising:
performing an automated supervised learning method to determine a preliminary identification of facies in the subterranean formation based on the training set;
calculating a confusion matrix for the supervised learning facies identification method by comparing predicted and observed facies for the cross-validation set;
calculating a facies transition matrix characterizing changes between contiguous facies; and
using the preliminary identification, the facies transition matrix, and the confusion matrix, to iteratively calculate updated facies identifications, wherein the facies identifications are used for selection of a location for drilling in the subterranean formation for hydrocarbon resources.

Assignees

• Thorne Julian
• Chevron Usa Inc

Inventors

• Thorne Julian

Classifications

• G01V11/00 Primary

  Prospecting or detecting by methods combining techniques covered by two or more of main groups G01V1/00 - G01V9/00 · CPC title

• G01V1/40

  specially adapted for well-logging · CPC title

• G01V2210/66

  Subsurface modeling · CPC title