Integrated experimental, machine learning and life cycle assessment of basalt and polypropylene fiber-reinforced ferrosilicon slag-based geopolymer concrete

The construction industry faces increasing pressure to reduce greenhouse gas emissions, conserve natural resources, and promote the utilization of industrial by-products, as ordinary Portland cement (OPC)-based concrete remains one of the largest contributors to global carbon emissions. Although geopolymer concrete (GPC) has emerged as a promising sustainable alternative, limited studies have comprehensively investigated the combined use of ferrosilicon slag, aluminum powder, and basalt and polypropylene fiber reinforcement while simultaneously evaluating their mechanical performance, predictive modeling, and environmental impacts. To address these research gaps, this study experimentally investigates the performance of basalt and polypropylene fiber-reinforced ferrosilicon slag-based geopolymer concrete using mechanical testing, scanning electron microscopy (SEM), machine learning (ML), and cradle-to-gate life cycle assessment (LCA). A ferrosilicon slag (FS)–aluminum powder (AP) binder system comprising 90% FS and 10% AP was identified as the optimum matrix and further enhanced by incorporating basalt fibers (BF) and polypropylene fibers (PPF) at volume fractions of 0.25%, 0.50%, and 0.75%. The geopolymer mixtures were activated using sodium hydroxide/sodium silicate solutions with molarities of 6 M and 9 M. Compressive, flexural, and split tensile strengths, together with post-cracking behavior, toughness, and microstructural characteristics, were evaluated to assess the influence of fiber type and dosage. The results demonstrated that the optimum performance was achieved with 0.50% basalt fiber under 9 M activation, exhibiting approximately 30% higher compressive strength, 25–30% higher flexural strength, and 27% higher split tensile strength than the control mix. Polypropylene fibers also enhanced the mechanical performance, particularly toughness and crack resistance, with the PPF-0.50–9 M mixture achieving approximately 22.2% improvement in split tensile strength at 90 days and 20% enhancement in flexural strength. SEM observations revealed denser geopolymer gel formation and stronger fiber–matrix bonding in the 9 M activated mixtures, explaining their superior mechanical behavior. Furthermore, Gradient Boosting Regression (GBR) and Artificial Neural Network (ANN) models accurately predicted the mechanical properties, with GBR outperforming ANN by achieving a higher coefficient of determination (R²) and lower prediction errors (RMSE and MAE). The integrated experimental, machine learning, and environmental assessment demonstrates that basalt and polypropylene fiber-reinforced ferrosilicon slag-based geopolymer concrete is a promising low-carbon construction material that effectively valorizes industrial waste while improving structural performance and providing reliable predictive tools for sustainable concrete mix optimization.