Probabilistic Prediction of Squeezing in Tunneling Under High-Stress Conditions

Rock squeezing typically occurs in weak overstressed rock masses, producing difficulties during construction associated to excessive convergences and/or to the need for non-standard excavation and support methods. This paper presents a new tool for empirical prediction of squeezing conditions in rock tunnels based on case-histories in tunnels under high stresses. The methodology is based on the use of a linear classifier for statistical analysis of a database of well-documented case histories compiled from the literature. Results show that rock quality (as measured by the Rock Mass Number, N) and depth have a significant influence on squeezing predictions. The newly proposed estimator is also shown to provide unbiased predictions of squeezing behavior; and it also allows to compute estimated probabilities of squeezing.