Prediction and Sensitivity Analysis of Shear Strength of Reinforced Concrete Beams with Deformed Hook Steel Fiber using Backpropagation Neural Network coupled with Garson's Algorithm
Claire Maulion Garduce, Dante Laroza Silva, Kevin Lawrence M. de Jesus
Abstract
Recent studies have brought developments and researches in various ways of reinforcing concrete beams. Famous in the present era is steel fibers which has been used as reinforcement to increase shear resistance of beams and reduced crack widths. Using the following parameters such as beam height (h), effective depth (d), width (bw), cross sectional area of longitudinal reinforcement (As), shear span-depth ratio (a/d), compressive strength of concrete (f'c), fiber volume fraction (Vf), fiber length (Lf), and fiber diameter (df), a shear strength model was developed using a backpropagation neural network. The model with the best R value of 0.99098 and MSE value of 0.03575 is the governing Artificial Neural Network (ANN) shear strength model. Moreover, Garson's Algorithm was utilized in the sensitivity analysis to show and describe the influence of each parameter to the shear strength of reinforced concrete with steel fibers. Performing the algorithm, the relative importance of parameters was ranked based from its significance to shear strength of beams with steel fibers. The ranking of importance is fiber length<beam width<fiber volume fraction<beam height<compressive strength of concrete<shear span to depth ratio<beam effective depth<area of longitudinal reinforcement.