| Abstrakt: |
This study conducts a comparative analysis of linear regression (LR), multiple polynomial regression (MPR), and artificial neural network (ANN) to predict the compressive strength of sustainable geopolymer concrete (GPC). The prediction is made based on the experimental investigation that involves a mix design approach for the production of geopolymer concrete (GPC) using different proportions of industrial by-products, namely fly ash (FA) and ground granulated blast slag (GGBS) while taking into account various influential factors. The variables under consideration incorporate the proportions of FA and GGBS, the concentrations of sodium hydroxide (NaOH) solution (10 M, 12 M, and 14 M), the curing temperatures (65 ℃ and 80 ℃), and the age of the concrete specimen. The fresh and mechanical properties of each GPC mix are observed after each curing age. A scanning electron microscope (SEM) is used to investigate the microstructure of GPC by identifying different constituents of concrete (C-S–H, N-(C)-A-S–H). Experimental results demonstrate that the workability of all GPC mixes falls within acceptable limits. Notably, the FA100GGBS400 mix achieves the highest compressive strength (C-S) when activated with 14 M NaOH and cured at 80 ℃, and the SEM analysis also confirms it. The predictive outcomes indicate that the MPR model establishes the most robust relationship between input and output variables, exhibiting a coefficient of determination (R2) value of 0.99. These findings highlight the potential of the MPR model for accurately forecasting the compressive strength of sustainable geopolymer concrete, offering practical applications in the construction industry. |
Full text from SpringerLink