Abstract
The threat of terrorist attacks on subway systems using explosives has intensified considerably in recent years. Explosions inside subway tunnels may lead to the failure of subway structures and result in further socioeconomic losses. Specifically, the century-old, single-track, cast-iron subway tunnels in cities such as New York and London are very vulnerable to this type of attack. In this study, an explicit dynamic finite-element procedure was developed to carry out extensive numerical simulation to investigate the soil-structure interaction and failure of cast-iron tunnels in saturated soils subject to internal explosions using the medium amount of explosives (80 kg TNT) that might be perpetrated by terrorists. The stress path and damage mode of these tunnels, subjected to internal blast loading, were first investigated based on the simulated damage of cast-iron tunnel lining using a hardening elastoplastic model that considered shear damage. The study aims to better understand the soil-structure interaction and its relation to lining damage of cast-iron subway tunnels under internal blast and to investigate the influences of several critical parameters on lining failure, including compressibility of saturated soil, brittleness of lining materials, specific impulse of blast, and strength of the soil-lining interface. The study found that ground-tunnel interaction was one of the governing factors that determined the damage of tunnel lining under medium internal blast loading by providing the necessary confinement to resist internal blast loading and by absorbing blast energy with its plastic shear deformation. Lining damage was mainly triggered by the tensile hoop stress because of large inertia and dynamic forces in the radial direction of the tunnel, and it may exhibit a progressive pattern in the vibration phase. Soil compression significantly influenced the damage of the tunnel under internal blast loading. The damage was more severe with compressible saturated ground.

This publication has 14 references indexed in Scilit: