Geologic Prediction Model for Tunneling

Abstract

Geologic uncertainty in underground construction often leads to design and construction conservatism and thus to inflated costs. This paper presents a general model for the probabilistic prediction of tunnel geology as a basis for developing more effective tunnel design and construction decision support systems. The geologic conditions along the tunnel alignment are modeled by a set of geologic parameters (e.g., rock type, joint density, degree of weathering, etc.), each following a continuous‐space, discrete‐state Markov process. The state probabilities for each geologic parameter are initially based on general geologic information and are later updated to reflect the outcomes and reliability of the location‐specific, nondeterministic observations provided by exploration programs. The resulting posterior geologic parameter profiles are aggregated into a single probabilistic ground class profile that can be used for determining optimal tunnel design and construction strategies. The model is illustrated by an example application.