A New Cluster Analysis-Marker-Controlled Watershed Method for Separating Particles of Granular Soils

Area of Science:

• Geotechnical Engineering
• Computational Mechanics
• Materials Science

Background:

• Accurate particle-level fabric analysis of granular soils from tomography data relies on precise particle separation.
• The conventional marker-controlled watershed method struggles with particle separation under boundary stresses causing particle crushing.

Purpose of the Study:

• To introduce a novel method, the Monash Particle Separation Method (MPSM), for enhanced particle separation in granular soils.
• To address the limitations of existing methods in accurately separating particles subjected to stress-induced crushing.

Main Methods:

• The MPSM automatically determines an optimal contrast coefficient using a cluster evaluation framework for accurate separation.
• It integrates Gaussian mixture model-based cuboid markers into the watershed method to separate challenging particles.

Main Results:

• The MPSM was validated on uniformly graded sand under one-dimensional compression up to 32 MPa.
• Demonstrated superior particle separation capabilities compared to conventional methods, essential for fabric analysis.

Conclusions:

• The MPSM provides the best possible particle separation for granular soil fabric analysis from tomography data.
• This advancement is critical for understanding soil behavior under complex loading conditions.