Evaluation of Iron Stone and Gravel Based Modified Concrete Using Random Forest Algorithms
1 *Egba, J. E., 2 Adeniji, A. A., and 3Adebayo A.O
1,Department of Civil Engineering, University of Ibadan, Ibadan, Nigeria.
2 Department of Civil Engineering, University of Ibadan, Ibadan, Nigeria.
3 Department of Civil Engineering, Ladoke Akintola University of Technology, Ogbomoso, Nigeria
DOI: 10.36108/laujoces/6202.71.0182
Abstract
The rising demand for sustainable and high-performance concrete has increased interest in alternative aggregates such as ironstone, while conventional empirical approaches remain inadequate for accurate prediction of mechanical behaviour. Accordingly, this study evaluates the predictive efficiency of Random Forest models for ironstone-based concrete. The investigation commenced with physicochemical characterization of ironstone using X-ray fluorescence spectroscopy and energy-dispersive spectroscopy. Subsequently, the mechanical properties of ironstone, gravel, and granite aggregates were determined through Aggregate Crushing Value and Aggregate Impact Value tests. Fresh concrete behaviour was assessed using slump and compaction factor tests, followed by evaluation of hardened concrete properties, including compressive strength, splitting tensile strength, flexural strength, and density. Thereafter, predictive models were developed using Random Forest algorithms, and their performance was assessed using standard statistical indicators. The results revealed that ironstone particles were heterogeneous, angular to sub-angular, and porous, with iron oxide, silicon oxide, and aluminium oxide as the dominant constituents. Aggregate Crushing Value and Aggregate Impact Value ranged from 30.23–36.73% and 27.02–31.48%, respectively, while slump and compaction factor varied between 25–47 mm and 0.71–0.87. Compressive strength and tensile strength, ranged from 6–20 N/mm² and1.4–3.1 N/mm. Overall, moderate ironstone replacement levels achieved satisfactory performance; therefore, a replacement range of 25–50% is recommended for balanced structural applications.
Keywords: Ironstone, Sustainable concrete, Mechanical properties; Machine learning modelling, Gene Expression Programming and Random Forest