Bottom-up modeling of damage in heterogeneous quasi-brittle solids

The theoretical modeling of multisite cracking in quasi-brittle materials is a complex damage problem, hard to model with traditional methods of fracture mechanics due to its multiscale nature and to strain localization induced by microcracks interaction. Macroscale effective elastic models can be conveniently applied if a suitable Helmholtz free energy function is identified for a given material scenario. Del Piero and Truskinovsky (Continuum Mech Thermodyn 21:141-171, 2009), among other authors, investigated macroscale continuum solutions capable of matching-in a top-down view-the phenomenology of the damage process for quasi-brittle materials regardless of the microstructure. On the contrary, this paper features a physically based solution method that starts from the direct consideration of the microscale properties and, in a bottom-up view, recovers a continuum elastic description. This procedure is illustrated for a simple one-dimensional problem of this type, a bar modeled stretched by an axial displacement, where the bar is modeled as a 2D random lattice of decohesive spring elements of finite strength. The (microscale) data from simulations are used to identify the exact (macro-) damage parameter and to build up the (macro-) Helmholtz function for the equivalent elastic model, bridging the macroscale approach by Del Piero and Truskinovsky. The elastic approach, coupled with microstructural knowledge, becomes a more powerful tool to reproduce a broad class of macroscopic material responses by changing the convexity-concavity of the Helmholtz energy. The analysis points out that mean-field statistics are appropriate prior to damage localization but max-field statistics are better suited in the softening regime up to failure, where microstrain fluctuation needs to be incorporated in the continuum model. This observation is of consequence to revise mean-field damage models from literature and to calibrate Nth gradient continuum models.