Probabilistic analysis of shield-driven tunnel in multiple strata considering stratigraphic uncertainty

• Published in: Structural safety 2016-09, Vol.62, p.88-100
• Main Authors: Wang, Xiangrong, Li, Zhao, Wang, Hui, Rong, Qiguo, Liang, Robert Y.
• Format: Article
• Language: English
• Subjects: Shield-driven tunnel; Stochastic geological model; Stratigraphic uncertainty; Uncertainty quantification
• ISSN: 0167-4730; 1879-3355
• DOI: 10.1016/j.strusafe.2016.06.007

•A stochastic geological modeling framework is proposed based on Markov random field.•Stochastic stratigraphic realizations are generated with uncertainty quantification.•A probabilistic approach is built to analysis tunnel performance in multiple strata.•The key design parameters along the entire tunnel alignment can be obtained. Subsurface formations with multiple soil/rock strata are a common geological condition for shield-driven tunnel (i.e., tunnel constructed using shield-driven machines) construction. The excavation face, under such conditions, often encounters a frequently changing stratigraphic configuration that consists of various lithological units. Furthermore, due to a lack of direct and continuous observations of the subsurface region, it is difficult to predict the stratigraphic profile along the entire excavation path with a high degree of certainty. Such a widely changing and uncertain excavation environment may lead to wide variation in the state of stress and deformation of the tunnel structure along the longitudinal direction. This poses a challenge for design engineers in obtaining accurate performance evaluations or reasonable design outcomes for tunnel construction in subsurface ground with multiple strata. This paper aims to address this challenge by presenting a stochastic geological modeling framework for uncertainty quantification of stratigraphic profiles using sparsely located observation information from geotechnical site investigations. In the proposed modeling framework, the underground soil stratigraphic profile is regarded as a Markov random field with specific energy functions, which is able to describe the inherent anisotropic and non-stationary spatial correlation of lithological units in the subsurface stratigraphic structure. By incorporating the developed stochastic geological modeling framework with a finite element simulation of the tunnel excavation, a probabilistic analysis approach is established to evaluate the effects of stratigraphic uncertainty on the structural performance of a shield-driven tunnel.