This paper introduces a new approach to model cracking processes in large reinforced concrete structures, like dams or nuclear power plants. For these types of structures it is unreasonable, due to calculation time, to explicitly model rebars and steel-concrete bonds. To solve this problem, we developed, in the framework of the finite element method, a probabilistic macroscopic cracking model based on a multi-scale simulation strategy : the Probabilistic Model for (finite) Elements of Reinforced Concrete (PMERC). The PMERC's identification strategy is case-specific because it holds information about the local behaviour, obtained in advance via numerical experimentations. The Numerical experimentations are performed using a validated cracking model allowing a fine description of the cracking processes. The method used in the inverse analysis is inspired from regression algorithms: data on the local scale would shape the macroscopic model. Although the identification phase can be relatively time-consuming, the structural simulation is as a result, very fast, leading to a sensitive reduction of the overall computational time. A validation of this multi-scale modelling strategy is proposed. This validation concerns the analysis of the propagation of a macrocrack in a very large Double Cantilever Beam specimen (DCB specimen usually used in the framework of Fracture Mechanics studies) containing rebars. Promising results in terms of global behaviour, macrocracking information and important reduction in simulation time are obtained.