Construction of a subway station in Shanghai called for a 31.4–33.1–m-deep excavation, which extended through upper impermeable clayey strata into underlying erodible silt and sand aquifer layers. When the north section of this pit nearly bottomed out on July 23, 2013, a sudden outburst of groundwater took place and the pit base was flooded quickly. In the subsequent two days, the adjacent ground and building successively underwent 30–42-mm settlements in a plunging manner. To prevent aggravation of the situation, a contingency plan was initiated quickly in an attempt to seal off the water flow path. By constructing mixed-in-place and jet-grouting piles respectively along the outer and inner perimeters of the diaphragm wall where leakage occurred, flow channels were successfully blocked. Despite this, the building adjacent to the outburst of groundwater suffered long-term postexcavation settlement. Forensic diagnoses show that this through-wall leaking accident was related to two major factors: (1) ignorance of minor leakages emerging two weeks before without adopting any remedial measure; and (2) a flawed slab connector embedded in the confined aquifer layer. Based on visual inspection and retrospective analysis of field monitoring data, the possible leakage flow path and associated ground subsidence zone were mapped and a settlement history of the affected building was reconstructed. Both contractual and technical lessons learned from this accident will be useful for professionals to mitigate or prevent future incidence of through-wall leakage during excavation.

From 2005 to 2010, a large excavation (approximately $\mathrm{26,000}{\mathrm{m}}^2$ in plane) was conducted in stiff clay deposits for the Gate of the Orient building in Suzhou, China. To reduce project cost and shorten construction duration, composite earth-retaining systems were designed for this excavation, i.e., removal of the uppermost 7.85 m of soil was conducted mostly by the sloped open-cut method, leaving one side supported by soil nailing wall (SNW); subsequent excavation to a depth of 21.5 m was retained by multipropped continuous bored pile (CBP) wall; final excavation of two inner pits to 30.0–30.7 m deep was supported by a SNW along with jet-grouting of basal soils. In spite of this, lateral wall displacements in this case were comparable to those of multipropped excavations in Suzhou. Because of a high phreatic level and confined artesian water, composite dewatering plans were adopted. Because the waterproof curtain did not extend deeply into the underlying confined aquifer, for the sake of saving cost, discharging artesian water inside the pit incurred a dramatic drawdown in the artesian level outside the pit, accompanied by significant ground settlements. It turned out that the methods in the literature for predicting excavation-induced ground settlements were not applicable to this case any more. For a large excavation not following zoned-construction procedure, corner-stiffening behavior was significant. In this case, deformations near the pit middle span were up to 2.5 times those near corners. Beyond recognition, excavation in stiff clay incurred noticeable basal rebound as well, which was up to one-fourth that in soft clays. Generally, excavations in stiff clay caused smaller lateral wall displacement than excavations in soft clays; their maximum lateral wall deflections mostly occurred above the excavation surface instead of equally above and below the excavation surface in soft clays.

The failure of Shanghai Metro Line 4 in 2003 was one of the most striking accidents in the history of subway construction in China. It involved the first breach failure of a cross passage being mined by an artificial ground-freezing method, followed by a massive ingress of water and soil, dramatic ground subsidence, rapid sinking of existing structures, failure of the adjacent dike and floodwall along the Huang-Pu River, flooding of the site, and collapse of buildings and metro tunnels. The accident directly resulted from failure of the cross passage being excavated within a confined aquifer. In the beginning, artesian water broke the frozen mass; then, water and soil gushed into the cross passage and metro tunnels. As a result of massive ground loss, the ground level underwent significant subsidence up to 4 m and those preexisting structures were damaged. Inherently, this event was a typical project management failure, associated with a number of procedural and ethical blunders. To rebuild the destructive failure case, this paper outlines the design and construction for the cross passage along with the geological conditions, describes those preceding events incurring the failure, depicts the catastrophic scene, and introduces the postfailure emergency responses and disaster-relief measures. Finally, technical, procedural, and ethical factors leading to the failure are summarized.

This paper introduces thirteen kinds of typical excavation failures widely encountered in practice in China along with corresponding failure mechanisms and damages to adjacent structures and facilities. Emphasis is placed on investigation of longitudinal slip failure during excavation of subway station, which frequently occurs but rarely receives serious attention. This investigation is based on a sliding event in Hangzhou, China, in which a soft earthen slope about 13 m high inside an uncompleted subway station underwent massive slip failure. Slide debris travelling on sloped terrain caused the detachment of braced steel pipe struts and breakage of interior columns and two concrete struts atop them. Fortunately, an efficient contingency plan was executed right after the event; thus, potential catastrophic collapse of the entire earth supporting system, a domino effect of the massive sliding, was avoided. Postfailure investigations in terms of ground survey, visual inspection, and numerical analyses show that for this three-step earthen slope with an overall slope gradient (including benches) of 1 vertical (V) to 3 horizontal (H) ($1\mathrm{V}:3\mathrm{H}$ ), the sliding took place at the uppermost 6-m slope and there was no sign of sliding in the lower 7 m of slopes ($1\mathrm{V}:2\mathrm{H}$ ) atop firm to stiff clayey strata. The steep local slope gradient ($1\mathrm{V}:1.75\mathrm{H}$ ) of the uppermost slope overlying weak muddy clay was the predominant factor triggering the sliding; probabilistic stability analyses indicate a high slip failure probability of 88.92%. Degradation of soil strength resulting from soil deformation was mitigated by negative excess pore water pressure associated with excavation (unloading); consequently, the uppermost slope did not show signs of potential sliding during the excavation. Following postexcavation dissipation of negative excess pore water pressure accompanied by decreasing of soil shear strength, the uppermost slope underwent an abrupt sliding failure 7 days later. Apart from its recognized effectiveness in limiting lateral wall displacement, quick soil removal was helpful for keeping the soft earthen slope temporarily stable during excavation. To ensure stability of soft earthen slopes underlain by weak subgrades, a slope gradient no steeper than 1V:3H is essential. Based on sensitivity analysis results, optimal slope angles corresponding to strength parameters of soft subsoils are proposed for future excavations. To prevent slip failure of soft earthen slopes, one cost-effective solution is to install recyclable steel sheet pile vertically penetrating through a potential slide at the deepest point of the estimated critical slip surface.

This paper presents the results of statistical analyses on a database comprising 592 deep excavations in Shanghai soft clays from 1995 to 2018. First, the general trends and patterns over time are outlined for seven different excavation methods and seven types of earth-supporting systems. Then, the major design factors are characterized in a statistical manner. Based on analyses of the collected data, the frequency distributions of ${\delta }_{hm}/H_e$ data (${\delta }_{hm}$ = maximum lateral wall displacement; $H_e$ = final excavation depth) corresponding to each excavation method and each earth-supporting system are presented; furthermore, the correlations of ${\delta }_{hm}/H_e$ with the major design factors are explored. In this new database, 90% of ${\delta }_{hm}/H_e$ data featured magnitudes no more than 1.0%, which frequently lie within 0.10%–0.55%, indicating better performance than other databases worldwide. Via extensive comparisons, both the optimum construction method corresponding to each type of earth-supporting system and the optimum earth-supporting system corresponding to each excavation method for controlling excavation performance were determined. Last, the major design and construction defects as well as natural hazards pertaining to abnormally large ${\delta }_{hm}/H_e=1.01-7.27\%$ tracked at 23 individual excavations are identified and summarized. These include excessive excavation duration, long wall exposure lengths, retaining walls toed in soft muddy clays, overexcavation at deep levels, surcharges behind retaining walls, inadequate supporting system stiffness, and intense rainfall; among these factors, walls toed in soft muddy strata are shown to be detrimental to the performance of excavations. For a deep excavation that is to be carried out in sensitive clayey strata, safety redundancy can be enhanced substantially by accounting for the degradation of soil strengths due to construction disturbance and/or rainfall infiltration in design. The findings and lessons from this study will be useful to professionals worldwide in efficiently achieving cost-effective designs and/or making reliable evaluations regarding the performance of deep excavations in soft clays.