Multi-Scale evaluation and predictive modeling of volcanic Ash–Silica fume aggregate concrete for sustainable construction

Abstract

The construction industry faces increasing pressure to reduce carbon emissions associated with ordinary Portland cement production and the depletion of natural aggregates. This study investigates the development of sustainable structural concrete through partial replacement of cement with volcanic ash (VA) and natural fine aggregate with silica fume based artificial aggregates (SF), combined with machine learning based performance prediction. Sixteen concrete mixes were designed using a full factorial approach with VA replacement levels of 0, 5, 10, and 15% by mass of cement and SF aggregate replacement levels of 0, 5, 15, and 25% by volume of fine aggregate, while maintaining a constant water to cement ratio of 0.46 and a fixed coarse aggregate content of 1300.55 kg per cubic meter. Mechanical properties and durability indicators were evaluated at 28 days. The control mix achieved a compressive strength of 50.65 MPa, elastic modulus of 30.00 GPa, and rapid chloride permeability of 999.02 Coulombs. The mix VA0–SF25 maintained comparable strength at 50.31 MPa, reduced abrasion loss to 9.24, limited porosity to 4.58%, and preserved very low chloride permeability of 1000.91 Coulombs. The mix VA5–SF25 exhibited the highest flexural strength of 6.73 MPa with acceptable durability retention of 85.28%, indicating an optimal balance between sustainability and performance. Higher VA contents of 10 to 15% significantly reduced strength and durability while increasing porosity and permeability. ANOVA confirmed cement replacement as the dominant factor governing performance, with Fisher’s values exceeding 322 and effect sizes approaching unity. Comparative analysis indicated that tree-based models provided more consistent predictive performance under validation.