This paper presents an efficient method of shield tunneling reliability analysis using spatial random fields. We introduced two stochastic methods into numerical simulation. The first one computes the maximal ground surface settlement using classical statistics, in which the response surface method is utilized to calculate the failure probability by first-order second moment. Cohesion, internal friction angle, Young's modulus and mechanical model factor are considered as random variables. The second method is the spatial random fields of aforementioned three key geotechnical parameters. Using these two methods, similar multiple soil layers are converted into a stationary random field by local regression as the first step, and then the process is followed by the spatially conditional discretization of multivariate. Failure probability of maximal ground surface settlement is calculated by a subset Monte-Carlo Algorithm. This approach is applied into the four-overlapping shield tunnels of the 5th and 6th metro lines intersecting at Huanhu W Rd station, Tianjin China. The failure analysis results indicated that classical statistics of geotechnical parameters showing higher variability than spatial random fields, which substantially support the complex shield tunneling project.
展开▼