Strength Prediction of Fiber-Reinforced Clay Soils Stabilized with Lime Using XGBoost Machine Learning

Abstract

This article proposes a predictive model for the compressive strength (UCS) of lime-stabilized clay soils reinforced with polypropylene fibers (PPF) using the extreme gradient boosting (XGBoost) algorithm. The research indicates that the developed model is highly effective and can serve as a reliable tool for anticipating the UCS of these specific soils. A comparison between experimental data and model predictions suggests that it can effectively elucidate the impact of the combined effect of lime and PPF on the compressive strength of clay soils, thus avoiding the need for new experiments to formulate new compositions. Furthermore, a parametric analysis reveals the benefits of fiber incorporation, particularly at an optimum lime content of 6% dosage. The results also show that an optimal fiber content of 1.25% and a length of 18 mm are essential for achieving satisfactory results. These findings have significant implications for the planning and implementing fibre treatments, allowing for considerably enhancing soil strength. They provide a solid foundation for more precise and effective interventions in the lime stabilization of clay soils, thus paving the way for more efficient practices in this area of research.