Clustering-Based REV Determination for Heterogeneous Rock Samples
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for determining the representative elementary volume (REV) in materials analysis struggle with heterogeneous materials, often leading to overestimation, overly restrictive search regions, and failure to accurately represent the material's heterogeneity.
Innovation Solution
A novel clustering method and property distribution analysis are employed to iteratively determine the REV size and location, using a split-and-merge process to identify stable clusters and their parameters, allowing for a more accurate representation of the material's heterogeneity.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If existing REV determination methods are used on heterogeneous materials, then the REV size is determined, but the representation accuracy of material heterogeneity deteriorates
Solution Approach 1:
The method segments the material property distribution into multiple clusters, each representing a distinct region with unique properties. By dividing the heterogeneous material into cluster-based segments rather than treating it as a uniform whole, the method captures local variations in property distribution, thereby improving both measurement precision and reliability of material representation.
Solution Approach 2:
The method changes the parameter representation from single REV size to multiple parameters including REV size, cluster centers, and property distribution characteristics. This parameter expansion allows the method to accurately represent heterogeneous materials by capturing both the scale of representation and the spatial distribution of different material regions.
2Measurement precision
If clustering-based statistics are used for REV determination, then the accuracy of heterogeneity representation improves, but the computational complexity increases
Solution Approach 1:
The method performs preliminary clustering analysis on the property distribution data before final REV determination. By pre-identifying clusters and their characteristics, the method reduces the computational burden of the main REV determination process, as the clustering results can be reused for multiple property analyses without repeating the full computational procedure.
Solution Approach 2:
The method uses cluster centers as representative copies of complex cluster regions. Instead of analyzing every data point within each cluster, the cluster center serves as a simplified representation that captures the essential properties of the entire cluster, significantly reducing computational complexity while maintaining accuracy.
3Reliability
If the REV concept is extended to include number and location of distinctive regions, then the material representation quality improves, but the method complexity increases
Solution Approach 1:
The clustering-based framework serves multiple functions simultaneously: it identifies REV size, determines cluster centers representing distinctive regions, characterizes property distributions, and validates representativeness. This multi-functionality is achieved through a unified statistical approach that handles all these tasks within a single methodological framework, reducing overall method complexity despite the extended capabilities.
Data Source
Figure 1~3
Figure 4(a)~4(d)
Figure 5
AI summary
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.