Evaluation of the compressive strength of polypropylene fiber reinforced high-strength concrete support with AI-based model

• Published in: Innovative infrastructure solutions : the official journal of the Soil-Structure Interaction Group in Egypt (SSIGE) 2023-12, Vol.8 (12), Article 315
• Main Authors: Ahmad, Soran Abdrahman, Ahmed, Hemn Unis, Rafiq, Serwan Khwrshid, Ahmad, Dler Ali
• Format: Article
• Language: English
• Subjects: Earth and Environmental Science; Earth Sciences; Environmental Science and Engineering; Foundations; Geoengineering; Geotechnical Engineering & Applied Earth Sciences; Hydraulics; Practice-oriented Paper
• ISSN: 2364-4176; 2364-4184
• DOI: 10.1007/s41062-023-01292-6

This study extensively examines the effects of introducing polypropylene fibers on the mechanical properties of high-strength concrete and develops predictive models for its compressive strength based on the mix proportions. The research is divided into two main parts. The initial section involves a thorough analysis of how polypropylene fibers influence various high-strength concrete properties, such as slump, compressive strength, splitting tensile strength, flexural strength, and modulus of elasticity, using a synthesis of experimental data. The subsequent section employs two well-known modeling approaches, linear regression (LR) and Artificial Neural Networks (ANN), to establish equations for forecasting compressive strength based on mixture proportions. LR assumes linear relationships and is less accurate for complex, nonlinear data, while ANN is more versatile and accurate for a wider range of tasks. The accuracy of both models depends on data complexity, with ANN generally performing better for nonlinear relationships. For the first time, eight effective variables were employed as input model parameters during the modeling process, including the water-to-binder ratio, cement content, fine and coarse aggregate content, silica fume content, superplasticizer content, fiber content, and specimens ages. The results show that adding polypropylene fibers significantly improve mechanical properties, particularly tensile strength. The ANN model outperforms LR in predicting compressive strength, with specific statistical metrics indicating its superiority. The ANN model exhibited RMSE, MAE, SI, OBJ, and R 2 values of 4.02 MPa, 2.53 MPa, 6.3%, 5.98, and 0.96, respectively, for the training datasets. However, both models have limitations, including occasional overestimation or underestimation and complexity.