Previous studies on laterally loaded piles in clay have mainly focused on flexible and rigid piles. Little attention has been paid to semi-rigid piles (whose pile–soil stiffness lies somewhere between those of rigid and flexible piles), which may behave as either flexible piles or rigid piles, depending on the change in soil stiffness during cycling. This study aims to understand the cyclic lateral response of a repeatedly loaded semi-rigid pile in soft clay and the failure mechanisms of the soil around the pile, through a series of centrifuge model tests and three-dimensional finite element analyses using an advanced hypoplastic clay model. Numerical parametric studies were also performed to investigate the evolution of soil flow mechanisms with increasing pile rigidity. It is revealed that the semi-rigid pile behaved as if it were a flexible pile (i.e., flexural deformation dominated) during the first few cycles, but tended to behave like a rigid pile (i.e., rotational movement prevailed) during subsequent cycles, which progressively softened the surrounding soil. As a result, the mechanisms of soil flow around the semi-rigid pile exhibited an intermediate behaviour combining the mechanisms of both flexible and rigid piles. Three distinctive mechanisms were identified: a wedge-type mechanism near the surface, a full-flow mechanism (within the transverse sections) near the middle of the pile, and a rotational soil flow mechanism (in the vertical symmetrical plane of the pile) near the lower half of the pile. By ignoring the rotational soil flow mechanism, which has a much lower resistance than the full-flow mechanism, the American Petroleum Institute code (published in 2007) underestimated the cyclic bending moment and the lateral pile displacement by 10% and 69%, respectively. Application of jet grouting around the semi-rigid pile at shallow depth significantly altered the soil flow mechanism (i.e., it was a solely wedge-type mechanism around the grouted zone).

Although Toyoura sand has been widely used as a standard material for physical model tests (particularly centrifuge model tests), its path-dependent stiffness degradation curves are rarely reported in the literature. This leads to difficulties in back-analyzing and understanding the physical tests involving significant changes in the direction of stress paths (such as unloading problems). In this study, path-dependent stiffness degradation curves of medium-dense Toyoura sand were determined by carrying out a series of stress-path triaxial tests, with local strain measurements. In the triaxial tests, four typical recent stress histories (changes of path direction = 0, 90, –90, and 180°) were simulated. The measured small strain stiffnesses of Toyoura sand were then used to calibrate an advanced hypoplastic model that accounts for path-dependent soil stiffness at small strains. To justify the validity of the calibrated hypoplastic model and model parameters, a Type A numerical prediction and a centrifuge model test were carried out to simulate a typical unloading problem (i.e., excavation) in medium-dense Toyoura sand. Triaxial test results show that a 180° reversal in the direction of the stress path increased the initial secant modulus of Toyoura sand by eight times. The path dependency of soil stiffness, however, vanished when the deviatoric strain exceeded 0.3%. The ground deformations (resulting from excavation) predicted by the numerical analysis show reasonable agreement with the measured data from the centrifuge test. This suggests the potential applicability of the calibrated soil model and its model parameters in predicting or back-analyzing other physical model tests in Toyoura sand.

Soil loss during tunneling may induce the ground deformation. The loss volume depends on tunneling methods, soil types, workmanship, etc. The movement pattern along the tunnel section can be separated into two parts: one part of which focuses on the planed center of the tunnel, and the other translates in the vertical direction. On the basis of the complex variable theory for elasticity, the corresponding second part movement could be deduced if the first part is predetermined. This paper studies the ground deformations under four possible focusing movement patterns. A comparison between the solutions and measured data is conducted for five different tunnels. It is proposed that the third focusing movement pattern could be presumed for predicting tunneling-induced ground deformation.