Dynamic behaviour of a circular tunnel in the sand: A numerical verification of a centrifuge program

Abstract

As urban development and population growth accelerate, more tunnel construction in seismic zones has been planned. A tunnel's behaviour under earthquake loading must be assessed using proper techniques, in which field tests must evaluate these techniques. However, the field measurements of actual tunnel responses during earthquakes are rare. Centrifuge tests, for example, may be modelled in laboratory settings to simulate the actual tunnel on a much smaller scale. These tests can provide valuable information on how tunnels respond to seismic events and represent a valuable reference for calibrating numerical approaches. Four sets of centrifuge experiments were conducted under the supervision of the ReLUIS research project at the Cambridge University engineering department. This project aimed to develop seismic design procedures through experimental validations of computational methods. The present study aims to numerically model the two series of centrifuge tests under 2D plane-strain conditions on the model scale. A hardening soil model with a small-strain stiffness overlay is implemented to calibrate the Leighton Buzzard sand characteristics. Detailed inspections of input parameters were performed before any simulations, and potential reasons for the possible deviations in results were discussed. Comparisons of numerical simulations and experimental test results reveal that, in most instances, numerically-calculated and experimentally obtained results are in goodagreement. However, some of the obtained results deviate from the experimental ones. Poor calibration of the dynamic characteristics of the sand was primarily cited as the source of these differences. Nevertheless, these calibrations could accurately predict the laboratory's static results. Modifying the soil model's stiffness parameters brings the predicted results much closer to the experimental ones, but some unidentified deviations remain.