混凝土断裂过程细观数值模拟-水利工程专业论文.docxVIP

PAGE III万方数据 PAGE III 万方数据 Abstract Concrete is a complex material and it’s mesoscopic structure has a great effect on the mechanical prosperities. It is a basic method to study the mechanical prosperities of concrete by mechanical test. However, the test result is very difficult to reflect the real mechanical prosperities of concrete, because of the limit of test conditions. With the development of computing technology, study of mechanical properties of concrete at mesoscopic level become a hot topic. On mesoscale concrete is regarded as a kind of three-phase material composed by aggregates, mortar and interfacial cohesive zones. The paper is mainly divided into four parts. Firstly, the problem that how to generate aggregate is studied and the random aggregate model of concrete is generated. Then the cohesive model of concrete is generated and the cracking evolution processes of concrete is simulated under uniaxial compression and tension. Then the mesoscopic parameters of concrete are obtained by combining BP neural network with orthogonal test method. Lastly, the mesoscopic model of RCC is generated and the evolution processes of layer crack is simulated. In conclusion, the main content of this paper are as follows: The random aggregate model of concrete. The Monte Carlo method and the block cutting method is used to generate the random polygonal aggregate model of concrete and is successfully extended to generate polyhedron aggregate. In the process of generating aggregate, in order to save time, a efficient algorithm is applied to judging the positional relationship of point, line, surface and body. The cohesive crack model of concrete. An algorithm is used to insert the cohesive element into the interfaces of mortar-mortar and aggregate-mortar. By the method, the cohesive crack model of concrete is generated. The mesoscopic parameters of concrete. The range of mesoscopic parameter is detailed described according to some literatures. The orthogonal test metho