Investigation of the compressive strength of PVA fibre-reinforced Engineered Geopolymer Composites (EGC) by experimental tests and predictive modelling

Abstract

Engineered geopolymer composite (EGC) is a type of fibre-reinforced geopolymer composite that can effectively reduce carbon footprint. This project focused on investigating the compressive strength of PVA fibre-reinforced metakaolinbased EGC through experimental methods and predictive modelling. The study was designed with the mix proportions of metakaolin with 10% to 70% slag or river sand as binder materials, and 6 mol/L, 8 mol/L, 10 mol/L and 14 mol/L of sodium hydroxide (NaOH) to prepare the alkaline activator solution. The compressive strength of specimens was analysed and compared. It was found that with the increase of the slag content, organic aluminosilicate gels were more conducive to formation, the compressive strength and curing rate of EGC also increased, reaching a maximum compressive strength of 51.83 MPa in 7 days. However, the enhancement effect decreases after slag content exceeds 50%. Similarly, increasing the NaOH concentration helps accelerate the geopolymerization, chemical bond formation and compressive strength, but the strengthening effect decreases significantly when the NaOH concentration exceeds 8 mol/L. EGC prepared with low slag content and low NaOH concentration can better maintain certain compressive strength for a while after peak load. Based on the experimental results and relevant literature, a three-layer feedforward back propagation (BP) neural network model was established to predict the compressive strength of EGC. After training and testing the models with 5 to 15 hidden neural nodes, the prediction model with 12 hidden nodes has R2 of 0.936, MAP of 3.282, and RMSE of 5.188, making it the best model for predicting compressive strength.