Representative elementary volume determination via clustering-based statistics
An example method includes acquiring two-dimensional (2D) or three-dimensional (3D) digital images of a rock sample. The method also includes iteratively analyzing property measurements collected throughout the digital images using different subsample sizes to identify a property distribution convergence as a function of subsample size. The method also includes selecting a smallest subsample size associated with the property distribution convergence as a representative elementary area or volume for the rock sample.
1. A method that comprises:
acquiring two-dimensional (2D) or three-dimensional (3D) digital images of a rock sample;
iteratively analyzing property measurements collected throughout the digital images using different subsample sizes to identify a property distribution convergence as a function of subsample size, wherein iteratively analyzing property measurements comprises representing property measurements as a set of data points and grouping the set of data points into clusters by computing a parameterized function representing a log-likelihood and a single property measurement using at least one Gaussian component, such that the clusters are a best representation of the set of data points; and
selecting a smallest subsample size associated with the property distribution convergence as a representative elementary area or volume for the rock sample.
2. The method of claim 1 , wherein computing the parameterized function involves representing a single property measurement using at least two Gaussian components.
3. The method of claim 1 , wherein computing the parameterized function involves representing multiple property measurements using at least one Gaussian component for each of the multiple property measurements.
4. The method of claim 1 , further comprising assigning a property distribution index value to each of the data points in response to the identified property distribution convergence.
5. The method of claim 4 , further comprising spatially assigning a property distribution index value to subsamples in the 2D or 3D digital images and using the distribution index values for subsequent analysis of the rock sample.
6. The method of claim 1 , further comprising maximizing the parameterized function by splitting or merging the at least one Gaussian component.
7. The method of claim 6 , further comprising updating merge criteria in response to a determination that property distribution convergence to a threshold tolerance is not reached.
8. The method of claim 6 , further comprising applying a merge criteria based on a Sturges formulation.
9. The method of claim 1 , further comprising comparing Gaussian components corresponding to different subsample sizes to identify the property distribution convergence as a function of subsample size.
10. A system that comprises:
a memory having software; and
one or more processors coupled to the memory to execute the software, the software causing the one or more processors to:
acquire two-dimensional (2D) or three-dimensional (3D) digital images of a rock sample;
iteratively analyze property measurements collected throughout the digital images using different subsample sizes to identify property distribution convergence as a function of subsample size;
represent property measurements as a set of data points by grouping the set of data points into clusters and computing a parameterized function representing a log-likelihood such that the clusters are a best representation of the set of data points, wherein the parameterized function is computed by representing a single property measurement using at least one Gaussian component; and
select a smallest subsample size associated with the property distribution convergence as a representative elementary area or volume for the rock sample.
11. The system of claim 10 , wherein the software further causes the one or more processors to represent a single property measurement using at least two Gaussian components.
12. The system of claim 10 , wherein the software further causes the one or more processors to represent multiple property measurements using at least one Gaussian component for each of the multiple property measurements.
13. The system of claim 10 , wherein the software further causes the one or more processors to assign a property distribution index value to each of the data points in response to the identified property distribution convergence.
14. The system of claim 10 , wherein the software further causes the one or more processors to spatially assign a property distribution index value to subsamples in the 2D or 3D digital images and to use the distribution index values for subsequent analysis of the rock sample.
15. The system of claim 10 , wherein the software further causes the one or more processors to compare Gaussian components corresponding to different subsample sizes to identify property distribution convergence as a function of subsample size.
16. The system of claim 10 , wherein the software further causes the one or more processors to maximize the parameterized function by splitting or merging the at least one Gaussian component.
17. The system of claim 16 , wherein the software further causes the one or more processors to update merge criteria in response to a determination that property distribution convergence is not reached.
18. The system of claim 16 , wherein the software further causes the one or more processors to apply a merge criteria based on a Sturges formulation.