Computational modeling of size effects in concrete specimens under uniaxial tension (Q841960)

This paper studies the interacting size effects on the nominal strength of concrete structures using a combination of finite element software enabling nonlinear analysis, and probabilistic methods. The paper attempts to identify possible sources of size effect, and model them by using a generalized model taking into account how various sources interact with each other. The authors also discuss the interaction of different material length scales and the effect of such interaction on the strength size effect. In order to illustrate the problem, they present test results obtained on dog-bone specimens with rotating boundary conditions of varying size. Based on the comparison of trends of nominal strength dependence on the structural size, the authors conclude that the suggested numerical model featuring three scaling lengths is capable of capturing the most important fracture mechanisms. The obtained size effect can be captured at a deterministic level by using deterministic length represented by crack band width. The smeared cracking with the crack band models are compared for a fracture-plastic and microplane material models. Further strength dependence on size in large samples is modeled by an autocorrelated random strength field. The statistical length scale is introduced in the form of the autocorrelation length of the field. It is shown that the inhomogeneity of material properties over the structure in the form of an autocorrelated random strength field leads to imperfections which trigger fracturing in highly stressed regions of a structure. By random sampling of the local strength field, the authors model the random scatter of resulting nominal strengths. Finally, the authors suggest simple scaling rules anchored in theoretical dimensional analysis.