Feasibility Of Basic Oxygen Furnace Slags As Roadway Materials In Kazakhstan

Abstract

The goal of the Republic of Kazakhstan of becoming an international transit hub within the framework of the global “One Belt, One Road” initiative presents key challenges, particularly the need for sustainable expansion and modernization of the national road infrastructure. This study investigates the technical feasibility and efficiency of Basic Oxygen Furnace (BOF) slag as both aggregate and filler in asphalt concrete (AC) mixtures for road construction in Kazakhstan. The initial step involves an extensive material characterization of fresh and stockpiled BOF slags sourced from a local steel plant. Laboratory testing of BOF slag-modified AC mixtures includes measurements of water absorption, low-temperature cracking resistance, compressive strength at various temperatures, and other key parameters. The BOF slag content in coarse and fine aggregates, as well as in mineral fillers (including blends with conventional MP1 filler), is varied in the experimental design. Additionally, the study investigates the use of performance-enhancing polymer additives, and a poly-phosphoric acid ester-based binder in the BOF-enhanced AC mixtures. A particular emphasis is placed on evaluating sulfur-modified bitumen for the wearing course, as earlier findings of this thesis research suggest that BOF slag alone does not fully address low- temperature cracking issues. It is established that optimized mixes containing 10% BOF slag in the coarse fraction and 50% in the fine fraction, in combination with sulfur-modified bitumen, meet national standards and offer improved performance characteristics. Furthermore, promising results are achieved in the binder course, where higher proportions of stockpiled BOF slag are used as both coarse and fine aggregates, producing compliant mixtures with enhanced cohesion, water resistance, and mechanical strength. Building upon experimental results, the study employs advanced numerical and machine learning techniques. Data-driven optimization is performed using the XGBoost machine learning algorithm and Bayesian optimization via Tree-structured Parzen Estimation (TPE) to predict performance characteristics, identify key variables influencing cracking resistance and durability, and determine optimal mix formulations for the wearing course. In parallel, Finite Element Modelling (FEM) evaluates the structural response of BOF-modified pavement layers under realistic loading conditions. This thesis highlights the need for full-scale pilot projects to validate laboratory and simulation findings, including long-term monitoring of chemical stability, volumetric expansion, and durability. Revisions to Kazakhstan’s technical regulations and quality assurance protocols for BOF slag and sulfur-modified binders are also necessary. Future work should include trial road sections, broader mix design variations, studies of alternative additives, and life-cycle assessments of environmental and economic impacts.