Dissipation mechanisms of crack-parallel stress effects on fracture process zone in concrete

The effect of crack-parallel stresses on the fracture properties of quasi-brittle materials has recently received significant attention in the fracture mechanics community. A new experiment, the so-called gap test, was developed to reveal this effect. While the finite element crack band model (CBM) with the physically realistic Microplane damage model M7 was quite successful in capturing the damage and fracture during the gap test, some questions remain, particularly the near doubling of the fracture energy at moderate crack parallel compression, which was underestimated by about 30%. Presented here is an in-depth meso-mechanical investigation of energy dissipation mechanisms in the Fracture Process Zone (FPZ) during the gap test of concrete, an archetypal quasi-brittle material. The Lattice Discrete Particle Model (LDPM) is here used to simulate the quasi-brittle material at the mesoscale, which is the length scale of major heterogeneities. The LDPM can capture accurately the frictional sliding, mixed-mode fracture, and FPZ development. The model parameters characterizing the given mix design are first calibrated by standard laboratory tests, namely the hydrostatic, unconfined compression, and four-point bending (4PB) tests. The experimental data used characteristic of the given mix design are calibrated by the Brazilian split-cylinder tests and by gap tests of different sizes with and without crack-parallel stresses. The results show that crack-parallel stresses affect not only the length but also the width of the FPZ. It is found that the energy dissipation portion under crack-parallel compression is significantly larger than it is under tension, which is caused by micro-scale frictional shear slips, as intuitively suggested in previous work. For large compressive stresses, the failure mode changes to inclined compression-shear bands consisting of axial splitting microcracks. Several complications experienced in the numerical modeling of gap tests are also discussed, and the solutions provided.