Global Sensitivity Analysis of Tunnel-Induced Building Movements by a Precise Metamodel

Abstract

Shield technology has been widely applied in metro construction, and the safety and stability of tunnels and surrounding structures are of utmost importance for a successful tunnel project. Reliable and sufficient epistemic knowledge are required for predicting ground movements and controlling the displacement of adjacent structures. Hence, this paper provides a novel global sensitivity analysis method, namely the extended Fourier amplitude sensitivity test (EFAST) algorithm, to explore the building disturbance induced by tunnel excavation. The Wuhan Yangtze River Metro Tunnel (WYRMT) was used as a case study to verify the applicability of the proposed approach. The deterministic and stochastic finite-element (FE) model were used to develop the approximate relationship between the input and output parameters with a high level of accuracy. Based on the simulated data from the FE model, a metamodel constructed by particle swarm optimization-least squares support vector machine (PSO-LSSVM) was then built and has been used to generate 10,000 groups of data for sensitivity analysis. The input variables with a high global sensitivity were identified as crucial variables that should be well controlled and managed during tunnel excavation, which helps to reduce the building disturbance and safety risk induced by tunneling. Sensitivity analysis can help in enhancing the knowledge of designers and aiding them to optimize the design and management scheme of tunnel excavation when confronting similar tunnels.