Abstract
Concrete is a composite material that has complex mechanical properties. The mechanical properties of each of its components are different at the mesoscopic scale. Studying the relationship between the macroscopic and mesoscopic parameters of concrete can help better understand its mechanical properties at these levels. When using the discrete element method to model the macro-mesoscopic parameters of concrete, their calibration is the first challenge. This paper proposes a numerical model of concrete using the particle discrete element software particle flow code (PFC). The mesoscopic parameters required by the model need to be set within a certain range for an orthogonal experimental design. We used the proposed model to perform numerical simulations as well as response surface design and analysis. This involved fitting a set of mapping relationships between the macro-micro parameters of concrete. An optimization model was established in the MATLAB environment. The program used to calibrate the mesoscopic parameters of concrete was written using the genetic algorithm, and its macro-micro parameters were inverted. The following three conclusions can be drawn from the orthogonal test: First, the tensile strength and shear strength of the parallel bond between the particles of mortar had a significant influence on the peak compressive strength of concrete, whereas the influence of the other parameters was not significant. Second, the elastic modulus of the parallel bonding between particles of mortar, their stiffness ratio and friction coefficient, and the elastic modulus and stiffness ratio of contact bonding in the interfacial transition zone had a significant influence on the elastic modulus, whereas the influence of the other parameters was not significant. Third, the elastic modulus, stiffness ratio, and friction coefficient of the particles of mortar as well as the ratio of the contact adhesive stiffness in their interfacial transition zone had a significant influence on Poisson's ratio, whereas the influence of the other parameters was not significant. The fitting effect of the response surface design was good.