Study On Digital Image Processing And Mesoscopic Damage Constitutive Model Of Coal

The way to build a constitutive model which reflects the whole deformation and failure process of coal is always been a great concerned question in the mining engineering fields. Due to the complexity of coal, the restriction of experimental conditions and measuring instruments etc., the theories of coal damage are still in the stage of development at present. Based on micromechanics test and micro structure quantified tests, two types of microscopic damage theories were used to systematically study the entire deformation and failure process of coal in this thesis. Meanwhile, the elastic-brittle damage constitutive model of coal and MCDC model which were comparison analyzed by examples were build.The precision of the quantifiable results of the microscopic of coal has direct influence on the accuracy of the above-mentioned constitutive model. The Digital Image Processing technology (DIP) is an effective method to quantify microscopic structural, and it has the advantages of precise measurement, digital expression etc. However, due to the large disciplines span, the application of the DIP technology in rock mechanics is hysteretic to its own development. The traditional way of image segmentation which cannot satisfy the demand of image processing of coal is generally used. To end this, the image processing theory which is based on partial differential equations and combined with charactertics of images of coal was introduced. The characteristics of coal image was also taken into consideration. Above all, the improved C-V model and LBF model are proposed to describe the information of microscopic structure of coal accurately and comprehensively. The main research results are as follows:1) According to the charactertics of images of coal and rock, the C-V model was improved by introducing image enhancement operator and the LBF model was modified through constructing initial damage evolutional curve in this thesis. When comparing with the traditional DIP technology and the former method which based on partial differential equations, the improved models have more flexibility and a higher division precision characteristics to deal with inhomogeneity gray, weak edges, severe noise pollution and low contrast in target and background etc.2) For the reference quantities of microscopic structure information of coal including the damage variables which is defined by itself, they have significant corresponding relationship with the macroscopic mechanical response. Furthermore, there is a rapid change of them at peak strength which indicates that coal is a kind of brittle or quasi-brittle material. The damage variables defined in this thesis is based on the area and azimuth which can reflect the anisotropy of coal material. The direction of the maximum value is the main direction of damage, so it can be the simplified direction of damage calculation.3) The image texture information can be seemed as the reference quantity of coal damage. The image texture information contains the fractural dimension and characteristic parameters of texture. The fractural dimension is focused on the morphology of goal content, and the characteristic parameters of texture is concentrated on describing the spatial distribution of image. The damage variable which adopts the characteristic parameters as reference quantity comprehensively considers micro-structure morphology information of surface on coal and the relationship of spatial distribution. Meanwhile, there is no need for specific parameters of micro-structures, so it is more convenient. It also has a marked nonlinear relation with axial strain which can be expressed as:DT= aε2+bε+c. Accordingly, the elastic-battle damage constitutive model can be built as follow:4) The presence of large amount of pore, cracks and initial damage cause that there is a long crack compaction stage of stress-strain curve, while few present damage constitutive model considers it. The MCDC model which contains initial damage and the effects of linear elastic damage is constructed in this research, and it can accurately described the compaction stage. The difference of MCDC model between the former meso-damage computation models are as follow: ① the probability distribution model obeys lognormal distribution. It is based on "proportional effect", central limit quantitative and other theories of probability theory, and it is also supported by real statistical regularity of micro-structure of coal and rock; ② the Regularization Heaviside Function is used to calibrate the less of lognormal model.5) The MCDC model organically connects the initial damage and actual damage during linear elastic stage which is differ from existing meso calculation damage model. There is no need for the MCDC model to continually consider the evolution of damage and fit it. It is distinct from the first meso-scopic damage theory model. So the MCDC model is supported by real damage, and has the advantages that the parameters is fewer, the values of parameters is easily to determine, the physical meaning is clear, and the simulation precision is high.