Computational AI prediction models for residual tensile strength of GFRP bars aged in the alkaline concrete environment. (15th July 2021)

• Record Type: Journal Article
• Authors: Iqbal, Mudassir; Zhang, Daxu; Jalal, Fazal E.; Faisal Javed, Muhammad
• Is Part Of: Ocean engineering. Volume 232(2021)
• Publication Date: 2021-07-15
• Subjects: FRP -- Seawater and sea sand concrete -- Material properties -- Degradation -- Artificial intelligence; Ocean engineering -- Periodicals; Ocean engineering; Periodicals; 620.4162
• DOI: 10.1016/j.oceaneng.2021.109134
• Languages: English
• ISSNs: 0029-8018

Abstract

Abstract: Based on a comprehensive literature study, the potential variables that can affect the durability of GFRP bars under a harsh alkaline environment especially seawater and sea sand concrete (SWSSC) environment, were investigated and reported in this paper. The study presents a new strategy for finding tensile strength retention (TSR) using empirical models based on the strong non-linear ability of artificial intelligence techniques, i.e., artificial neuro-networking (ANN), gene expression programming (GEP), and adaptive neuro-fuzzy inference system (ANFIS). The diameter of GFRP bars, the volume fraction of glass fibers, the pH value of solutions, the temperature, and the duration of conditioning were considered as input parameters to find TSR of aged GFRP bars. Statistical checks evaluated the trained models, and the results demonstrate that the models provide a reliable estimate of TSR. A simple mathematical prediction formula was developed using the GEP model that can quickly foresee the TSR for aged GFRP bar. In comparison with the GEP model, the ANN model and the ANFIS model provided slightly better results. The parametric study indicates that the large diameter of bars and the high volume fraction of fibers have positive effects on the TSR, while the high temperature and the long duration of conditioning have negative influences. Highlights: The potential variables affecting the degradation of GFRP in alkaline environment were identified from the extensive literature. Three AI models namely ANN, GEP, and ANFIS were developed to predict the residual tensile strength of conditioned GFRP in an alkaline ageing environment. The predictions made by the ANN model has a relatively better correlation with the experimental results than the others. A parametric study shows the effects of identified variables on the degradation of GFRP bars.