Concrete is a complex material and can be modeled on various spatial and temporal scales. While simulations on coarse scales are practical for engineering applications, a deeper understanding of the material is gained on finer scales. This is at the cost of an increased numerical effort that can be reduced by the three methods developed and used in this work, each corresponding to one publication. The coarse spatial scale is related to fully homogenized models. The material is described in a phenomenological approach and the numerous parameters sometimes lack a physical meaning. Resolving the three-phase mesoscopic structure consisting of aggregates, the mortar matrix and the interfaces between them allow to describe similar effects with simpler models. This work addresses two computational challenges related to mesoscale modeling. First, aggregate particles take up a high volume fraction and an efficient particle-packing algorithm is required to generate non-overlapping, random esostructures. Enforcing an additional distance between the aggregates is essential to obtain undistorted meshes for finite element simulations, but further complicates the packing problem. An event-driven molecular-dynamics algorithm is applied to this problem that, in contrast to traditional methods, allows movement and a dense arrangement of the aggregates. This allows creating concrete mesostructures with realistic aggregate volume fractions. The second challenge concerns stability problems in mesoscale simulations of concrete fracture. The geometric complexity and the combination of three material laws for each of the phases leads to numerical instabilities, even for regularized material models. This requires tiny time steps and numerous iterations per time step when integrated with a classic backward Euler scheme. The implicit–explicit (IMPL-EX) integration extrapolates internal variables that account for the nonlinear behavior. This linearizes the equations, provides additional robustness and a computational speedup. In combination with a novel time step control method, a three-dimensional mesoscale compression test is accelerated by a factor of 40, compared to an adaptive backward Euler algorithm. The life time of concrete under cyclic loads is commonly predicted with empirical Wöhler lines. They relate the number of endured cycles with the applied load amplitude and can be included in constitutive formulations. They can, however, hardly be generalized to geometries and load configurations other than the ones tested. On a finer temporal scale, fatigue failure is modeled by the accumulation of damage within each loading cycle. This resolves the whole process of failure, includes stress redistributions and size effects and can easily be extended to multiphysics phenomena. The third computational challenge solved here is the efficient temporal integration that would not be feasible in a naive cycle-by-cycle integration of thousands or millions of cycles. The cost of evaluating a single cycle is reduced by reformulating the problem in the frequency space. It is sufficient to equilibrate the structure once for each Fourier coefficient which significantly speeds up this evaluation. The accumulated damage of one cycle is integrated in time using an adaptive cycle jump concept. For a two dimensional void test structure, the combination of both techniques leads to a 25 times faster simulation compared to the full integration. These three main contributions decrease the numerical cost of mesoscale simulations, allow larger and more detailed models and are a basis to deepen the understanding of the complex failure patterns in concrete.