A novel back-analysis method for improving predictions of time-dependent ground settlements in deep excavations in soft soils

Cheng Chen,Shuai Zhang,Guang-Nian Chen,Yang Lyv,Tao Wu
DOI: https://doi.org/10.21203/rs.3.rs-4426549/v1
2024-01-01
Abstract:Soft soils are widely distributed in the coastal regions of China, and extensive deep excavation projects are underway to meet the substantial demand for infrastructure construction in these regions. The ground settlements induced by excavation in soft soils are typically time-dependent due to the combined effects of soil consolidation and creep. When considering time-dependent behavior, the increased complexity arising from various sources of uncertainties makes it more challenging to provide accurate predictions of ground settlement induced by excavation. This study presents a novel back-analysis method for improving predictions of time-dependent ground settlements in deep excavations in soft soils. A linear correlation structure for model errors among different stages has been established. A new updating procedure is proposed to improve predictions of time-dependent ground settlements. This method combines Bayesian updating of soil parameters with regression analysis of stage-correlated model errors. The proposed linear correlation structure and updating procedure are validated through case studies. The results indicate that the predictions of time-dependent ground settlements can be effectively improved using the proposed method. The proposed method may inform engineers about the benefits of investing resources in excavation time-related measures and help establish effective strategies for managing excavation time during the construction process.
What problem does this paper attempt to address?