Considering creep in rock tunnelling with a combined support system: Theoretical solutions and machine-learning solver

Abstract

This study provides an alternative theoretical approach to analyse the mechanical behaviour of tunnels excavated in time-dependent geomaterials, considering the sequential excavation of tunnels and installation of rockbolts and elastic liner. In the theoretical analyses, the Bolted Rock Mass (BRM) is modelled as a homogeneous material but with higher stiffness. An innovative Machine-Learning-Based solver (MLB-solver) has been developed to evaluate the reinforcement ability of rockbolts for geotechnical applications. After that, the complex variable method, the Laplace transformation technique, and the extension of the correspondence principle, combined with the compatibility and boundary conditions, have been employed to obtain the analytical solutions for stresses and displacements at the rock-support and support-support interfaces. Furthermore, the analytical predictions have been validated by monitoring data from the Lyon-Turin Base Tunnel. Parametric analyses are conducted to investigate the influence of rockbolt length and installation time on tunnel behaviour. Meanwhile, the proposed theoretical solutions have been also applied to analyse tunnel stability. This study offers a new and efficient method for the preliminary design of supported tunnels.