The purpose of the work is to provide an experimental benchmark on the seismic behavior of tunnels, with the final aim of calibrating numerical and analytical design methods. A series of plane-strain centrifuge tests with dynamic loading on a model tunnel was, therefore, carried out at the Schofield Centre of the Cambridge University Engineering Department (CUED). Four samples of dry uniform fine sand were prepared at two different densities, in which an aluminum-alloy tube was installed at two different depths. The tube was instrumented with strain gauges to measure hoop forces and bending moments at significant locations. To monitor the amplification of ground motion from the base to the surface, vertical arrays of accelerometers were placed in the soil model and along the box. The instrumentation also included linear variable differential transformers (LVDTs) that measured the soil surface settlement during all test phases. The test procedure and the results are described in this paper, showing the evolution of both accelerations and internal forces along the tunnel lining during the model earthquakes.

Buildings founded in proximity to shallow tunnels under construction may be damaged by the ground displacements induced by tunneling. This is a matter of concern for design, and a variety of protective interventions are currently adopted to prevent such damages. Among these, rows of piles or jet-grouting columns are widely diffused. In this paper, the effectiveness of a simple row of piles is computed by means of three-dimensional (3D) finite-element (FE) analyses, thus allowing the investigation of the relationship between performance and some simple geometrical parameters, such as the spacing among the piles. The results of centrifuge tests are reported and used as a benchmark. The potential damage has been quantified in this work, taking into account both the settlement profile and the horizontal strain induced at the ground surface by tunneling. It is shown that although the settlement reduction is significant only for very small spacing (s=2–3 pile diameters), even largely spaced piles (s=5–6 pile di...

This paper compares the experimental results of a set of centrifuge models of tunnels in sand under seismic loadings with the predictions of finite element dynamic analyses and of simplified methods. In order to characterise the soil behaviour, mobilised shear stiffness and damping ratio of the sand model have been back-calculated from the experimental results according to two different procedures. Starting from the accelerometer measurements, one was based on the transfer functions from surface to base and the other one on the average shear stress–strain cycles along the sand layer. A series of viscoelastic 2D dynamic analyses were performed to simulate the model tests by a linear equivalent approach. The equivalent shear stiffness and damping ratio determined from stress–strain cycles were used as input values for the analyses. The shear stress transfer at the ground-lining interface was back-analysed to calibrate the interface elements used in the numerical code, in order to improve the assessment of the transient changes of hoop force. Finally, the numerical results have been compared to analytical solutions, widely adopted in the design, and to the experimental data in terms of transient increments of internal forces in the lining. Such a comparison indicates that the analytical formulations give a good estimation of the seismic increment of bending moment in the lining and a reasonable lower bound for the transient changes of hoop forces, provided that cyclic shear strains are correctly evaluate

Tunnelling in urban areas can damage the existing buildings and underground services, as a consequence of ground movements induced by excavation. In this paper the potential of a vertical diaphragm wall to modify the displacement field induced by tunnel excavation is investigated. A series of centrifuge tests were performed in which a diaphragm wall was embedded to one side of a tunnel. The location of the wall, its length, thickness and the roughness of the wall–soil interface were varied. In addition, a series of numerical analyses were carried out to supplement the experimental data. It was observed that the field of displacements was modified in different ways, depending on the combination of the above factors, and the sensitivity of the problem to the variation of each factor is discussed. An efficiency parameter is defined that expresses the ability of the wall to modify the reference (no wall) profile of settlements.