Rectangular tunnel heading stability in three dimensions and its predictive machine learning models

Abstract

Tunnel heading stability in two dimensions (2D) has been extensively investigated by numerous scholars in the past decade. One significant limitation of 2D analysis is the absence of actual tunnel geometry modeling with a considerable degree of idealization. Nevertheless, it is possible to study the stability of tunnels in three dimensions (3D) with a rectangular shape using finite element limit analysis (FELA) and a nonlinear programming technique. This paper employs 3D FELA to generate rigorous solutions for stability numbers, failure mechanisms, and safety factors for rectangular-shaped tunnels. To further explore the usefulness of the produced results, multivariate adaptive regression spline (MARS) is used for machine learning of big dataset and development of design equations for practical design applications. The study should be of great benefit to tunnel design practices using the developed equations provided in the paper.