A parameterized model reconstruction method for multi-mineral porous rocks

Abstract

Rocks, composed of various minerals and defects, exhibit complex deformation and failure processes. Traditional analytical methods and homogenized medium-based models struggle to address such complexity. Developing a numerical model that closely mimics the structure of real rocks rather than only a replication of rock structure is a critical challenge. To address this, this work proposes a parameterized model reconstruction method for generating multi-mineral porous rock models based on rock images. The basic principle of the method is as follows: first, rock minerals and pores are differentiated through grayscale image analysis; then, the optimal ellipse for each pore or mineral grain is fitted based on its boundary; the relationship between the grains and their optimal ellipses is analyzed, and the boundary features are quantified using ellipse parameters; finally, non-regular pores or grains are generated through inversion, ultimately rebuilding a multi-mineral porous rock model that has same mineral content and structural characteristics to the real rock. The proposed method offers several key advantages: realistic replication of pore and mineral shapes, precise control over porosity and mineral content, support for parametric studies, scalability for large models, directional alignment of mineral grains, and broad applicability. To validate the method, various materials were used, including the generation of models with regular and irregular pores, contacting and non-contacting pores, multi-mineral rocks, and directionally arranged mineral grains. The results demonstrate that the proposed method has unique advantages in generating these materials. Finally, the method was applied to preliminary studies in rock wave propagation, uniaxial compression, and ultra-low temperature damage by using a distinct lattice spring model, highlighting its significant potential for both scientific research and engineering applications. The program code can be made available to researchers with reasonable request.