Ragip Ince

Lattice modelling of size effect in concrete strength

Abstract This paper uses a recently improved lattice network model to study the size effect in the strength of plain concrete structures. The several improvements made to the lattice network model are: (i) tension softening of the matrix phase is included in the material modelling; (ii) the structural response is modelled by incrementing the deformation rather than the load. This eliminates the need for introducing arbitrary scaling parameters in the beam element failure criteria and; (iii) a square rather than a triangular lattice beam network is found to be adequate for modelling concrete, thus greatly reducing the computational time. The improved square lattice network has been used to simulate the complete load–deformation response of notched three-point bend beams of different sizes with a view to checking the validity of several size effect models available in the literature. Lattice simulation was found to identify microcracking, crack branching, crack tortuosity and bridging, thus allowing the fracture process to be followed until complete failure. The improved lattice model predicted smooth structural response curves in excellent agreement with test results. The simulated nominal strengths also correlated very well with the test results, apart from that for the smallest beams (depth 38.1 mm). However, even in the relatively broad range of sizes (1:8) of the test beams, there was no clear evidence that one size effect model is superior to the other. In fact, rather surprisingly the test data would appear to be equally well described by all the available size effect models. The lattice simulations however indicated a trend which is better predicted by the multifractal scaling model.

The neural network approximation to the size effect in fracture of cementitious materials

Modeling of material behavior generally involves the development of a mathematical model derived from observations and experimental data. An alternative way discussed in this paper, is neural network-based modeling that is a subfield of artificial intelligence. The main benefit in using a neural network approach is that the network is built directly from experimental data using the self-organising capabilities of the neural network. In this paper, size effects in fracture of cementitious materials are modeled with a back-propagation neural network. The results of neural network-based size effect law look viable and very promising.

An Improved Lattice Model for Fracture and Size Effect of Concrete Structures

This paper describes an improved lattice network model to study the fracture process and size effect in the strength of plain concrete structures. The several improvements made to the lattice network model are: (i) tension softening of the matrix phase is included in the material modelling; (ii) the structural response is modelled by incrementing the deformation rather than the load. This eliminates the need for introducing arbitrary scaling parameters in the beam element failure criteria and; (iii) a square rather than a triangular lattice beam network is found to be adequate for modelling concrete, thus greatly reducing the computational time.