Mechanical Behavior Prediction of GFRP Composites Aged at Different Immersion Times Using Artificial Neural Network

In order to determine the progression of damage in the mechanical properties of glass fiber reinforced polymer composites with epoxy (GE) and vinylester (GV) matrix were subjected to seawater aging according to ASTM D5229 standard at 120, 408, 960 and 1320 h until to moisture saturation. Mechanical tests, including tensile, compression, flexural, and short beam shear, were conducted on aged specimens. Tensile strength was affected in both types of laminates, with a reduction of 36.7% in GE and 43.2% in GV, while the tensile modulus remained unchanged. Compressive strength decreased for GE (53.1%) and remained with minor changes for GV. The compressive modulus increased for GE (87.9%) and remained unchanged for GV. Flexural strength decreased 39.0% in GE and 20.5% in GV, while the flexural modulus remained unchanged for both laminates. Short beam shear strength of GE was significantly affected, whereas GV were less affected. These results provide valuable information about the behavior of composite materials subjected to marine environments. The neural network Multi-layer Perceptron (MLP) was used to predict values of tensile, compressive, and flexural strength and were compared with experimental data for seawater aged GE, and GV, as a function of immersion time. A strong agreement between the experimental and predicted values of seawater absorption was observed, indicating the validity of both absorption models for GE and GV laminates. The study and prediction of incremental damage in laminates (GE and GV) could propose new ways of designing marine components, as well as the projection of component maintenance.