| This dissertation describes analytical, computational, and experimental investigations of the behavior of concrete under impact loading conditions. Analytically, a three-dimensional, damage dependent constitutive/failure model which accounts for load-induced anisotropy is presented. The model incorporates the effects of microcracks on the material behavior. To model microcracking, a continuum damage mechanics approach is considered. The stress-strain relationship in the model is derived from a damage dependent Helmholtz free energy function. Damage is defined as a vector and its evolution is described using a fracture mechanics approach which accounts for the opening, sliding, and interaction of pre-existing penny-shaped microcracks. The unique features introduced in the present model include a new Helmholtz energy function with a permanent strain term, local-stress-driven microcrack growth in three different cracking modes, and strain rate dependence for a wide range of strain rates under both tension and compression.; Computationally, the constitutive model is implemented in a finite element hydrocode (EPIC-2), and in a driver routine which performs uniaxial stress computations. Under dynamic loading, the evolution equations, coupled with the stress-strain relationship, constitute a stiff system of ordinary differential equations. These equations are efficiently integrated using a diagonally implicit Runge-Kutta numerical integration scheme with time step control. Experimentally, static and dynamic results, including tensile and compressive split Hopkinson bar data and plate impact data, are presented. These results are used to guide formulating the model, calibrating model parameters, and checking the validity of the model after it had been developed. Simulation results showed that the present model captures many of the salient features of the static and dynamic behaviors of concrete including dilatancy, degradation of mechanical properties, and strain rate and pressure effects. Generally, good correlations are obtained between model predictions and experimental data. |
