Assessment of leachate-contaminated clays using experimental and artificial methods

Abstract

The investigation of leachate-contaminated clay (LCC) is essential for landfill engineering assessment and achievement of sustainable development goals. Several static and dynamic laboratory tests, including unconfined compressive strength (UCS), California bearing ratio (CBR), and cyclic simple shear, are conducted. Cyclic simple shear experiments on LCCs were performed to evaluate the damping and shear modulus. The investigated factors are vertical load (VL), leachate content (LC), frequency (F), and shear strain (ShS) for LCC. Forensic-based investigation optimization (FBIO) and equilibrium optimizer algorithm (EOA) were utilized in addition to multiple types of ensemble models, including adaptive boosting (ADB), gradient boosting regression tree (GBRT), extreme gradient boosting (XGB)) and random forest (RF). The comparison of the methods showed that GBRT-FBIO and XGB-EOA models outperformed other models for shear modulus and damping of LCC. The p-value less than 0.0001 shows the significance of the used models in the response surface methodology (RSM) method.