Performance of fiber-reinforced fly ash–metakaolin geopolymer-stabilized reconstituted expansive soil at different moisture conditions

Mechanical properties of underlying geological strata are the key to the structural integrity and long-term survivability of civil infrastructure. Expansive and soft soils often lead to catastrophic structural failures, including excessive settlement and cracking, necessitating advanced stabilization prior to construction. This study examines the effectiveness of a sustainable Geopolymer-based Alkali-Activated Binder (AAB) produced using a precursor mixture of 90% Fly Ash (FA) and 10% Metakaolin (MK) activated by a Sodium Silicate/Sodium Hydroxide (SS/SH) ratio of 1.5 at 8 Molarity and an Activator/Binder (A/B) ratio of 0.75. The binder was incorporated into the soil in the doses of 10, 20 and 30 percent. Unconfined Compressive Strength (UCS) measurements under different moisture conditions (Optimum Moisture Content [OMC], Liquid Limit [LL], and 1.2 LL) showed that the UCS values declined with the increase in water content, but increased with the increase in binder dosage. Nevertheless, the content of binder was high, which led to a significant decrease in failure strain, which is a sign of brittleness. To counter this, Polypropylene (PP) fibers were added at 0.2, 0.4 and 0.6 percent doses, the findings revealed that 0.4 percent was the best fiber content to add to improve ductility and strain capacity. The 30% binder + 0.4% PP fiber mixture at OMC was found to have the highest 28-day UCS of about 7.89 MPa, as well as a better strain capacity and a slower post-peak response. Durability of the treated soil under the wetting-drying cycles tests at OMC confirmed the stability of the treated soil. SEM and XRD microstructural analyses revealed progressive microstructural densification and mineralogical alterations in the stabilized soil, which is in line with the formation of geopolymerization-related reaction products, which led to the strength enhancement. Machine learning models, including Decision Tree, Random Forest, and XGBoost, were developed to predict the compressive strength of the stabilized soil, with the Random Forest model demonstrating the highest predictive accuracy (R2 = 0.965). This paper has shown that the proposed FA-MK geopolymer, which is fibre-reinforced, offers a strong and binder incorporating industrial by-products alternative to stabilizing soft soil. This study does not involve any clinical trials on human participants; hence, clinical trial registration is not applicable. Not applicable.