Transverse Shear Capacity Predictions of GFRP Bars Subjected to Accelerated Aging Using Artificial Neural Networks

• Published in: Journal of materials in civil engineering 2023-04, Vol.35 (4)
• Main Authors: Fasil, Mohammed, Al-Zahrani, Mesfer M.
• Format: Article
• Language: English
• Subjects: Aging (artificial); Artificial neural networks; Building materials; Civil engineering; Conditioning; Cracking (fracturing); Fiber reinforced polymers; Fiber-matrix interfaces; Glass fiber reinforced plastics; High temperature; Leaching; Mathematical models; Neural networks; Photomicrographs; Prediction models; Regression analysis; Resins; Robustness (mathematics); Scanning electron microscopy; Shear strength; Technical Papers; Transverse shear
• ISSN: 0899-1561; 1943-5533
• DOI: 10.1061/(ASCE)MT.1943-5533.0004686

AbstractThis paper presents the results of three different types of glass fiber-reinforced polymer (GFRP) bars subjected to different harsh environmental conditions. Three types of pultruded GFRP bars, namely, Bar A (13.04 mm dia.), Bar B1 (13.02 mm dia.), Bar B2 (10.19 mm dia.), and Bar C (13.22 mm dia.) were subjected to accelerated aging. GFRP specimens were exposed to four conditionings (alkaline, alkaline with salt, acidic, and water), two temperature regimes (20°C and 60°C), and three time regimes (3, 6, and 12 months). The mechanical strength retention of the bars investigated based on the transverse shear strength (TSS) test, conforming to ASTM D7617M-11 revealed a decline in the shear strength characteristics of specimens proportionally with the exposure time. In general, Bar A, Bar B1, and Bar B2 performed well, but Bar C performed the worst since Bar C exhibited the least mechanical strength retention at elevated temperatures. Scanning electron microscopy (SEM) of the cross section of the conditioned specimens revealed the nature of the evolution of deterioration and the state of glass fibers, polymeric resin matrix, and fiber-matrix interface, validating the decrease in transverse shear capacity of the bars. SEM micrographs showed fiber damage, with debonding occurring at the fiber-matrix interface and cracking of the polymeric resin. Leaching out of glass fibers into the matrix of varying degrees upon conditioning was observed through the X-ray mapping of silicon. Furthermore, strength prediction models developed using multiple linear regression and artificial neural network (ANN) techniques were compared. Coefficients of correlation (R2) of 0.94 and 0.76 and root mean square errors (RMSE) of 7.43 and 12.00 were obtained with models developed using ANN and multiple linear regression, respectively, showing that ANN can be used as a robust tool for GFRP shear strength prediction. Practical ApplicationsWith glass fiber-reinforced polymer (GFRP) bars possessing properties such as corrosion-resistance, high strength, and being light-weight, it is emerging to be a promising alternative to conventional steel rebars, which are often plagued by durability issues such as corrosion and consequent financial losses caused by its repair and maintenance. The present study makes an effort to contribute to the understanding of the durability performance of GFRP bars exposed to some aggressive environments such as alkaline, acidic, and moisture at room and high