Performance Evaluation and Statistical Modelling of Glass Waste Powder-Geopolymer Concrete Subjected to Dual Curing Regimes: Strength, Durability, and Predictive Analysis
摘要
The study presents a comprehensive experimental and statistical investigation into the performance of geopolymer concrete (GPC) incorporating Glass Waste Powder (GWP) under dual curing regimes: ambient and oven curing. Fly ash (FA) and Ground granulated blast furnace slag (GGBS) were used as primary binders in varying proportions from 80:20 to 20:80 at 15% intervals, with GWP introduced as partial binder replacement at 5%, 10% and 15%, two alkaline to binder (Al/B): 0.45 and 0.50 were used to evaluate their influence on mix performance. Mechanical properties were assessed through compressive, flexural, and split tensile strength tests, while durability was evaluated using abrasion and sorptivity analysis. Results demonstrated that adding 10% GWP significantly improved strength and durability due to the filler effect and reactive silica content, enhanced matrix density, and reduced porosity. A novel aspect of this study is the application of supervised machine learning (ML) algorithms for compressive strength prediction of Glass Waste Powder incorporated Geopolymer Concrete (GWP-GPC). Models were developed using experimentally generated data and pre-processed via min-max scaling, outlier removal, and one-hot encoding of categorical variables. Among the models, Random Forest (RF) exhibited the highest predictive accuracy (R2=0.918). The study validates the beneficial use of GWP in GPC and demonstrates the effectiveness of data-driven modelling for optimizing sustainable construction materials.