Optimization and prediction of mechanical properties of composite concrete with crumb rubber using RSM and hybrid DNN-HHO algorithm

• Published in: Journal of Building Engineering Vol. 84; p. 108486
• Main Authors: Anjali, R., Venkatesan, G.
• Format: Journal Article
• Language: English
• Published: Elsevier Ltd 01.05.2024
• Subjects: And response surface methodology; Crumb rubber; Deep neural network; Flexural strength; Horse herd optimization; Pull-out strength; Split tensile strength; Deep neural network; Flexural strength; Split tensile strength; Pull-out strength; And response surface methodology; Crumb rubber; Horse herd optimization
• ISSN: 2352-7102; 2352-7102
• DOI: 10.1016/j.jobe.2024.108486

Scrap tires can cause serious health, environmental, and aesthetic issues since they are heavy and non-biodegradable. Due to the restrictions on recycling waste tires, using crumb rubber from used tires in the construction sector is a highly practical alternative. Concrete with crumb rubber (CR) improves flexibility and durability, providing higher mechanical strength. So, this study aimed to apply Response Surface Methodology (RSM) and a hybrid deep neural network-horse herd optimization (DNN-HHO) to optimize the mechanical characteristics of composite concrete with CR. Aluminium (Al), Hydrogen peroxide (H2O2), Sodium Sulfate (Na2SO4) and Sodium Chloride (NaCl) are used as composite materials with CR. Manufactured sand (M-sand) is used in this study because it is eco-friendly, economical and improved concrete quality. The RSM model evaluates the 27 proportions and tests them to assess their mechanical strength and other parameters. The research's conclusions are as follows: Evaluation of 27 proportions reveals that Al + H2O2+2.5R achieves optimal compressive, split tensile, flexural, and pull-out strength of 3.80 MPa,46.08 MPa,4.75 MPa, and 3.67 MPa, respectively. The developed RSM model exhibits strong regression and significant fit, as evidenced by the derived ANOVA. Comparing the actual versus predicted plots, it is observed that all points align closely with the fitted line all outputs, indicating the superior prediction performance of the proposed DNN-HHO model over both DNN and RSM. Furthermore, the regression values in the proposed model are exceptionally surpassing the RSM regression values. These findings affirm the superiority of the hybrid DNN-HHO algorithm in predicting results. •The mechanical properties of concrete with CR under various proportions were studied.•Al + H2O2+2.5R mix proportion achieved highest mechanical strength of the concrete.•SEM analysis established better matrix formed between CR, M-Sand and cement mixtures.•Hybrid DNN-HHO algorithm proved superior as it predicted results than DNN and RSM.•The developed DNN-HHO model achieved best prediction performance with least error.