Abstract
High-strength concrete (HSC) is currently popular in the design and construction of heavy structures. Being HSC a brittle material, its cracking is sudden and cracks traverse through the aggregate particles thus producing relatively smooth fracture planes which is contrary to the normal-strength concrete (NSC) wherein cracks go around the aggregate particles. The relatively smooth fracture planes in HSC considerably decrease the concrete shear strength by reducing the contribution of the aggregate interlock. However, the international codes for the design of reinforced concrete (RC) structures make no distinction between NSC and HSC and the assessment of shear capacity of slender RC beams without web steel is still based on the experimental data of beams of NSC. In the present research, an effort is made for predicting the shear strength of HSC slender RC beams without stirrups, using regression models and neural networks. A large database of experimental results for HSC slender beams, covering a wide range of influencing parameters, is used for regression analysis as well as for the development of the artificial neural network models. The results of analysis are compared with major code requirements as well as with the researchers' design model. The regression based as well as the neural network based models are presented. New empirical design models for the two approaches are also suggested.