Discrete Element Analysis Of Soil-Rock Mixture Using Digital Image Processing

The soil-rock mixture?S-RM? is widely distributed in China, especially in western region. With the development of large hydropower project, such as The Three Gorges Dam project, S-RM has attracted many researchers' attention in recent years. It is a kind of complicated geomaterials composed of two kind of material?fined grained soil and block stone? with different physical and mechanical characters. S-RM is extremely heterogeneous, discontinuous and nonlinear mixture. Its macroscopic mechanical properties are closely related to the meso-structural characteristics, such as the content of rock mass, the particle shape and spatial distribution structure of stone, etc.In this thesis, the discrete element method based on the technique of digitial image processing is introduced to establish the discrete numerical model of S-RM. Based on the DEM model, the relationship between meso-structural characteristics and the macroscopic mechanical properties of S-RM is studied. In summary, it achieves the following innovative results and conclusions.?1? Through the technique of digitial image processing?DIP?, the shape of block stone is obtained and, furthermore, a database contains various particle shapes is erected. First, pre-treatments of digital image is implemented to get a better quality and binary image of rock. Then, canny algorithm is chosen to acquire the edges' information of rock in SRM. Base on the data gained from DIP, a new approach to extract corners on the boundary was proposed.?2? Two new parameters, area difference ratio S_cp, and angular indices AI, are proposed to discrible the shapeness of block rock. S_cp is calculated by area raito of the region enclosed by paticle and its equivalent ellipse's boundry to the paticle's, describling particle roundness. The value of AI is related to the number and sharpness of corners.?3? The dicrete element model of rock is established by the modify clumped overlapping sphere algorithm?MCOS algorithm?. In the MCOS algorithm, the potential filled circle is allowed to overlap and its center is located in the midperpendicular of the corresponding edge. When the precision condition is meet, the biggest potential filled circle is selected as the final one. The algorithm is simple and effective.?4? A method to build the distinct element model of S-RM is proposed. Based on the database of rock shape, MCOS algorithm and the grading curve of mixture, a discrete numerical model with arbitrary rock shapes and certain grading feature is generated.?5? Based on the DEM model of S-RM, the influence of the content of rock mass, particle shape, particle size of rock and spatial distribution structure of stone to the mechanical characteristics is studied.