Reliability Analysis of Ground Surface Settlement Given Multivariate Spatial Random Field in Shield Tunneling

Abstract

It is well known that geotechnical parameters impact dramatically soil-tunneling mechanics. Random field theory has been widely used for a single geotechnical parameter, including aleatory uncertainty, spatial variability and local singularity. Current challenge is focused on analyzing the spatially correlated soil layers. Coefficient of cohesion, internal friction angle, and Young’s modulus are accounted in reliability analysis of ground surface settlement due to shallow-buried shield tunneling. At first, convert multiple non-stationary soil layers into stationary field using a local detrending method, and define the geostatistical parameters. Then, assume spatial variability of geotechnical parameters into aleatory randomness, and classical response surface method is utilized into reliability analysis. Furthermore, Co-sequential Gaussian discretization is conceived for multivariate spatial random field, in which failure probability of ground surface settlement is calculated directly by subset Monte-Carlo simulation. This approach is applied into the paralleling zone of four shield tunnels of the 5^th and 6^th metro lines linking to Huanhu W Rd station, Tianjin China. Results prove that reliability index of considering geotechnical parameters as random variables is lower than the assumption of multivariate spatial random field, which would support substantially to construction control and design optimization in complex shallow-buried shield tunneling projects.