Scaling laws and renormalization groups for strength and toughness of disordered materials (Q1317142)

On the basis of the fractal geometry and renormalization group theory, an attempt is made to break down barriers between structural engineers, who pay close attention to macroscopic phenomena of regular fracture, and material scientists who devote greater attention to microscopic phenomena and models presenting an excessively local character. The size effects are studied for the nominal tensile strength \((\sigma_ u)\) and the fracture energy \(({\mathcal I}_ F)\). It is noted that classical Griffith's and Weibull's theories are based on the investigation of the defect of maximum size in the trend estimation of the strength decrease with sample size, whereas the disordered (i.e. macroscopically heterogeneous and/or damaged) nature of the material microstructure is completely disregarded. In the present paper, this strength behaviour is explained by using the fractal concept of self-similarity and the introduction of a material constant, namely the renormalized tensile strength \(\sigma^*_ u\). The established tendency is confirmed by experimentally observed power decrease of the LEFM stress singularity 1/2, and the ligament of a disordered material at peak stress is represented by a fractal space of dimension \(\alpha=2-d_ \sigma\) \((0\leq d_ \sigma<1)\). Hillerborg's proposal assuming a fracture energy increase with the concrete specimen sizes is analyzed from the point of view of the energy dissipation to be occurring in a fractal space of dimension \(\alpha=2+d_{\mathcal I}\) \((0\leq d_{\mathcal I} < 1)\). This trend is studied using a material constant (renormalized fracture energy \({\mathcal I}^*_ F\)) and is confirmed by experimental data, too. It is noted that the ranges of self-similarity for fracture energy and tensile strength do not necessarily coincide. In the consideration of the condition of extreme disorder, two cases are investigated, namely: (i) defect size distribution (DSD) of self-similarity, and (ii) DSD with a larger dispersion. For the first case, it is shown that the size effect vanishes due to the spherical pores and/or nearly constant defect size, but becomes enormous at very large dispersions in the defect size distribution. In the second case, it is ascertained that for sufficiently large sizes the maximum defect is larger than the body itself. Then the author concludes that the DSD of self-similarity corresponds to the maximum disorder. Next, the conditions defining extreme slopes of the strength decrease diagram are obtained. A generalization of the brittleness number defined by the author allows to conclude that the sum of the dimensional decrement (for material ligament) and the dimensional increment (for fracture surface) must be lower than unity. Finally, it is noted that, for very large specimens, \(\sigma_ u\) and \({\mathcal I}_ F\) may appear constant by varying the specimen size, whereas for size scales where random self-similarity holds, the ``universal properties'' of the system \((\sigma^*_ u,{\mathcal I}^*_ F)\) are constant, if they are defined by the ``renormalization'' procedure from parameters couples \((\sigma_ u,d_ \sigma)\) and \(({\mathcal I}_ F,d_{\mathcal I})\), respectively.