Stochastically homogenized peridynamic model for dynamic fracture analysis of concrete

Abstract

Dynamic compact tension tests in concrete are investigated using an intermediately-homogenized peridynamic (IH-PD) model, and the results are compared with a fully-homogenized peridynamic (FH-PD) model as well as with experimental data from the literature. In the IH-PD model, concrete is considered as a two-phase material composed of aggregate and mortar, partially homogenized via a stochastic procedure. Both the FH and IH-PD models capture well the experimentally measured reaction-force time-evolution and crack paths for a wide range of dynamic loading rates, without the need for an explicit representation of microstructure geometry of the concrete material or of rate-dependent parameters. The IH-PD model is more accurate than the FH-PD in reproducing the peak loads and produces crack path tortuosity and randomness, similar to what is observed experimentally. These benefits come from the preservation of some micro-scale heterogeneity, stochastically generated to match the concrete’s aggregate volume fraction. We also find that it matters where one measures the reaction force when trying to connect it to fracture and damage progression in the sample: only when measured on the dynamically loaded side of the pre-notch are features in the reaction-force profile matching critical moments like crack initiation, crack branching, and sample final failure. These results provide guidance for future experiments seeking best placement of damage detection sensors and using elastic wave signal analysis in structural health-monitoring of concrete structures.