Predictive Modeling of Geopolymer Concrete Properties Incorporating Phosphogypsum and Slag Using Machine Learning Algorithms

The waste products phosphogypsum (PG) and ground granulated blast furnace slag (GGBS) contained alumina and silica which were helpful in the geopolymerisation. This research focused on the mechanical properties of geopolymer concrete by substituting different proportions of PG with GGBS at varying concentrations of sodium hydroxide (NaOH). The investigation examined the strength performance of geopolymer concrete as GGBS replace PG in increments of 5%, 10%, 15%, 20%, and 25%, while NaOH concentrations range from 6M to 16M. This study examined the varying percentage of GGBS affected the compressive strength, flexural strength and split tensile strength of geopolymer concrete. The findings indicated that the most favourable enhancement in mechanical properties occurred when PG was replaced with 20% GGBS, with the maximum compressive strength obtained being 49.67 MPa. The strength increased with the molar concentration of NaOH, reaching its peak at 12M; beyond that point, the strength decreased. Machine learning was used to predict the mechanical properties of geopolymer concrete with exceptional R² values close to 0.99.