RC Beams under Blast Loads: Numerical Simulation and Machine Learning Modeling.

The use of explosives to target civilian buildings and other structures around the world is becoming a growing problem in modern societies. This paper focuses on RC beams exposed to free air blast loads. The paper first presents a parametric study on the behavior of RC beams subjected to blast loads using finite element simulation and then proposes an Artificial Neural Network (ANN) model to predict that behavior in a simple and easy manner. The ABAQUS program is used to simulate RC beams under blast loads. Experimental data was collected from the literature and used to validate the ABAQUS models. Deflection, reaction forces, ultimate stress, ultimate strain, and failure mode of RC beams are investigated. The considered design parameters in the parametric study are the characteristic compressive strength of concrete (fcu), the transverse reinforcement ratio (ρT%), the longitudinal reinforcement ratio (ρL%), and the scaled distance (Z). In this paper, the proposed ANN model was trained and tested using datasets produced using ABAQUS. The input parameters of the ANN model are TNT weight, standoff distance (D), characteristic compressive strength of concrete, transverse reinforcement ratio, longitudinal reinforcement ratio, width-to-thickness ratio (b/t), and length-to-thickness ratio (L/t). The predicted behavior using the ANN model showed the credibility of the model. The results indicated that L/t, b/t, and Z have significant effects on the behavior of RC beams under blast loads compared with fcu, ρT%, and ρL%, the cracks area increases with the decrease in Z, fcu, and b/t and decreases with L/t decrease. [ABSTRACT FROM AUTHOR]