Numerical Analysis of Deep Basement Excavation in Soft Soils of Kathmandu Valley

Abstract

This study investigates the soil–structure interaction behavior of a deep excavation support system at the Ramshahpath site using three-dimensional finite element modelling. The primary objective is to evaluate ground deformation, pile response, and structural performance under varying pile diameters, surcharge positions, construction stages, and constitutive soil models. The subsurface soil at the site predominantly consists of fine-grained clayey silt with low to intermediate plasticity. Numerical simulations were carried out using PLAXIS 3D, employing both the Mohr–Coulomb (MC) and Modified Cam Clay (MCC) constitutive models. The results indicate that pile diameter plays a significant role in controlling system stiffness and deformation, with an increase from 0.65 m to 0.90 m reducing lateral pile displacement by up to 18% and surface settlement by more than 30%. The structural response is also strongly influenced by surcharge load position and construction sequence; surcharge applied at the ground surface induces cantilever-type deformation, while deeper loading reduces displacement and promotes a more uniform stress distribution within the soil mass. A comparative analysis of constitutive models shows that the MC model tends to overestimate deformations, predicting lateral wall movements up to 67 mm, whereas the MCC model provides significantly lower and more realistic responses (approximately 4–7 mm), making it more suitable for soft clay conditions. The numerical results were validated using Matlock p–y empirical relationships, published literature, and relevant empirical methods, showing acceptable agreement in deformation trends and confirming the reliability of the finite element modelling approach. Overall, the study emphasizes that increased pile stiffness, appropriate positioning of surcharge loads, and the use of advanced constitutive models are essential for effectively controlling excavation-induced deformation and improving the reliability of predictions in soft soil environments.