Abstract: The construction of urban subway tunnel inevitably produces disturbance to surrounding rock and causes ground surface settlement. Dynamic prediction of ground surface settlement caused by tunnel excavation is an important method to ensure the safety of above-ground buildings and tunnel construction. In view of the difficulty of accurate dynamic prediction of ground surface settlement during tunnel construction, based on the definition of longitudinal excavation coefficient γ , a dynamic prediction model of lateral ground surface settlement is established. The model can accurately describe the variation of the settlement of the same monitoring location with the advancement of the tunnel face, and then realize the dynamic prediction of the ground surface settlement at the construction site. The results show that under certain constraints, this model can be degenerated into Peck model and stochastic medium theory prediction model. The accuracy and applicability of the dynamic prediction model are verified by on-site construction. The tunnel can be divided into three affected segments longitudinally (i.e., intense influence, moderate influence, and mild influence) based on the obtained γ , which well reflected the influence degree of the excavated tunnel face on the same monitoring section at different positions. Through the analysis of the influence of the buildings and isolation piles on the ground surface settlement curve, it can be found that the building and its adjacent ground surface present the characteristics of cooperative deformation and joint bearing. Moreover, installing geological drill isolation piles on the side of the tunnel can reduce the ground surface settlement of that side up to 71.9%. The research results have a certain guiding and reference significance for the on-site construction of the Central Yunnan Water Diversion Project and similar projects.

Key words: settlement prediction model, dynamic prediction, longitudinal excavation coefficient, stochastic medium theory, affected segmentation, on-site monitoring