For deep excavations in residual soils that are underlain by highly fissured or fractured rocks, it is common to observe the drawdown of the groundwater table behind the excavation, resulting in seepage-induced ground settlement. In this study, finite element analyses are firstly performed to assess the critical parameters that influence the ground settlement performance in residual soil deposits subjected to groundwater drawdown. The critical parameters that influence the ground settlement performance were identified as the excavation width, the excavation depth, the depth of groundwater drawdown, the thickness of the residual soil, the average SPT N value of the residual soil, the location of the moderately weathered rock, and the wall system stiffness. Subsequently, an artificial neural network (ANN) model was developed to provide estimates of the maximum ground settlement. Validation of the performance of ANN model was carried out using additional data derived from finite element analyses as well as with measured data from a number of excavation sites.

No abstract

Due to rapid urbanization, the land resources available for construction are becoming increasingly scarce in many builtup environments, especially for infrastructure development in mountainous terrain. For deep excavations developed for basements of high-rise buildings in cities with mountainous terrains, the excavation activities may have an influence on the stability or performance of existing nearby upper slopes. Based on a case study in Chongqing, this paper numerically investigates the effects of the excavation geometry, the retaining wall system stiffness and the distance between the excavation and slope on the performances including the global factor of safety and retaining wall deflection of the excavation-slope system. Subsequently, simplified ultimate and serviceability limit state response surface models have been developed and implemented into the First-Order Reliability Method to determine the probability that the ultimate or serviceability limit state is exceeded by performing probabilistic analysis on the global factor of safety through setting the threshold maximum wall deflection as an optimization constraint.