Determination of groundwater buoyancy reduction coefficient in clay: Model tests, numerical simulations and machine learning methods

Abstract

Groundwater plays an essential role in stabilizing underground structures. However, hydrostatic uplift forces from groundwater can create safety hazards. This paper obtained the groundwater buoyancy reduction coefficients of 36 types of clays through model tests and conducted a finite element simulation to obtain the buoyancy reduction coefficients of additional clays with varying soil properties. Machine learning methods, including extreme gradient boosting (XGBoost) and random forest (RF) algorithms, were used to analyze and identify the soil parameters that have a significant impact on the reduction of groundwater buoyancy. It was found that the permeability coefficient and saturation are the primary factors that influence the reduction of groundwater buoyancy. Additionally, the prediction models developed by XGBoost and RF were compared, and their accuracy was evaluated. These research findings can serve as a reference for designing underground structures that can withstand the potential risk of buoyancy in clay.

Keywords

• Clay
• Buoyancy reduction coefficient
• Numerical simulation
• Model test
• Machine learning