Application of Digit Image Processing Technique to the Analysis of Concrete Damage Process

In the field of solid mechanics, the study of strain localization and completely stress-strain relation has a prominent role. The development of high-performance materials, especially the quasi-brittle materials which is widely used in engineering, also requires a more precise knowledge of the constitutive-relation for the more and more restrictive safety rules. The image processing techniques approach to the mechanics of materials with finite element analysis (FEA) provides a powerful tool to simulate and analysis the damage process of quasi-brittle materials reliably. A good numerical model which can catch the main properties of object is the foundation of numerical simulation. Most of the papers focus on the random model generation algorithms to get the numerical model, which is difficult to evaluate the validity. From a mesoscopic point of view, concrete materials are multi-phase composite materials. The images of sample section are full of useful information, for example, including the space distribution of aggregates, the shape of aggregates, and the areas of aggregates and so on. This information seldom can be caught by the random model generation algorithms credibly. Nowadays the quick development of image processing techniques enhances our ability to deal with real digit images of concrete samples section. The key problem for image processing technique is how to decide the threshold of gray. In this paper, a new method to determine the threshold of gray has been proposed. After getting the photo of really concrete sample, the image can be transferred automatically by program to build numerical model without manual intrusion. Based on the model, different parameters of material are related to aggregate elements, mortar elements and interfacial transition zone (ITZ) elements individually. The results of simulation reveal the damage process of concrete material under uniaxial load with the help of finite element method (FEM).