Effectiveness of artificial neural network for forecasting of fracture toughness of concrete specimens

Since fracture energy and fracture toughness are directly linked to failure behavior, crack initiation and propagation assumes a significant importance in the study of fracture mechanics. The current study attempts to develop a prediction model for fracture toughness by employing artificial neural network (ANN). Data sets of previous research work obtained from 210 fracture tests have been used to train, test and validate the ANN Model. Experimental results are further used check the feasibility of the model. The test specimen were concrete beams made using Nano TiO₂ (1, 2, 3 and 4%) and ground granulated blast furnace slag (30%). Compressive strength, maximum aggregate size and water to cement ratio are used as input parameters while fracture toughness, tensile strength, Compressive strength and flexure strength are the measured outputs for experimentation. The experimental results indicate the fracture toughness, tensile strength, compressive strength increase with the increase in the NT percentage by 46.15, 10.22, 16.07 and 14.83% respectively. The ANN predicted values of fracture toughness are in good agreement with the experimental results.