Modellig of fracture paremeters of steel fiber reinforced concretes by artifical neural networks

Concrete is a quasi-brittle material and steel fibers are used to increase the ductility and energy absorption capacity of concrete. Effects of properties of steel fibers on concretes were investigated by numerous researches and some parameters were suggested to use in design of Steel Fiber Reinforced Concretes (SFRCs). The fracture parameters are taken as a criterion in the design of SFRCs due to their more ductile behavior. In the scope of the thesis, an experimental study was carried out on the determination of fracture parameters of concrete. In experiments, three different steel fibers which have different type of hooked-ends and strengths were added in three different concretes (C40/50, C50/60 ve C80/95) in 4 different ratios (15, 30, 45, 60 kg/m3). The three point bending tests were conducted on concretes and fracture parameters were determined according to this test. Effects of type of steel fiber hooked-end on different concretes are revealed by the experiments. It can be seen very difficult to determine the fracture parameters of SFRC by an experimental study, because of the several of ingredients and strength class of concrete and also the properties and amounts of steel fibers. For this reason, an artificial neural networks model was created and fracture parameters were successfully predicted by this model. Keywords: Concrete, steel fiber, fracture parameters, artificial neural networks.