Development of Two Empirical Correlations for Tunnel Squeezing Prediction Using Binary Logistic Regression and Linear Discriminant Analysis

Squeezing as a large time-dependent deformation can result in irreparable damages for tunneling projects. The accurate prediction of this phenomenon in preliminary stages of tunneling projects has a remarkable role on reducing its destructive effects. In this paper, two new empirical correlations have been presented for squeezing prediction before starting the tunneling project using binary logistic regression (BLR) and linear discriminant analysis (LDA). These correlations have been developed based on a comprehensive database including 220 tunneling case histories. In both correlations, overburden depth (H) and rock mass quality (Q) are the independent variables and squeezing conditions can be predicted as the dependent variable. Quality assessment of these correlations indicated that both equations have high performances for squeezing prediction. In comparison to previously developed empirical equations, proposed equations have led to improvement of prediction capacity. The validation results reveal that LDA and BLR equations are better than the previously developed equations.