It is clearly shown in Figs.
3–
4 that both the sites of Stockpiles 1 and 2 exhibited typical features of global stability failure. To verify this observation, some preliminary stability analyses were performed using the 2D LEM version 2012 software Slope/W (
2012), which is based on the method of slices. Although LEM is unable to calculate stress in soil mass, it has already been proven in practice to be reliable for evaluating slope stability (
Duncan 1996;
Duncan et al. 2008;
Han et al. 2010;
Wright 2013;
Leshchinsky and Ambauen 2015;
Stark et al. 2018;
Tan et al. 2018). The method of Morgenstern and Price (
1965) was used in the analysis, because it satisfies both force and moment equilibrium rather than just satisfying either moment or force equilibrium (
Fellenius 1936;
Janbu 1954;
Bishop 1955). Regarding strength parameters for stability analysis, consolidated undrained (CU) strengths of fine-grained soils are recommended by Vanden Berge and Wright (
2016) for undrained slope stability analysis in practice. Because (1) Shanghai soft clay is a kind of normally consolidated soil, and (2) both excavation and stockpiling were short-term activities at this site, CU strength parameters (Table
1) measured by consolidated undrained direct simple-shear tests (CUDSST) were adopted for analysis, which is also recommended by both the state and local design specifications in China [GB50007 (
Ministry of Construction of China 2002); DG/TJ08-61 (
Shanghai Urban and Rural Construction and Transportation Committee 2010)]. In Slope/W (
2012), the structural elements (flood wall, earth retaining wall, building basement, and PHC piles) were treated as shear reinforcements and their displacements were unable to be simulated.
Soil Stockpile 1
Since multiple cases have been reported for slope failures triggered by excavations or cuts at or near the toes of earthen embankments or waste landfills (
Stark et al. 2010;
Bonaparte 2018), it seemed to be plausible that the excavation might have reduced resisting forces in a global stability failure extending from the stockpile crest through the very soft clays. As marked in Fig.
2, the excavation was 22 m (
) and 32 m (
) to the southern toe and the southern crest of Stockpile 1, respectively, which was far beyond the potential excavation influence zone within
behind the retaining wall. Moreover, as schematically illustrated in Fig.
2, the excavation influence zone did not overlap with the additional stress zone of Stockpile 1. Most importantly, the buried building basement and its underlying piles located between the excavation and Stockpile 1 would have functioned as barriers mitigating potential interaction between the excavation and the stockpile. Hence, it can be inferred that the existence of the building basement and underlying piles would have enhanced the global stability of Stockpile 1 to some extent (e.g.,
Kourkoulis et al. 2011;
Tan et al. 2018). To verify this barrier effect, global stability analyses were conducted using Slope/W (
2012) for two cases with and without the building basement and its underlying piles, respectively. Figs.
9(a and b) present the global stability analysis results for these two cases.
As indicated in Fig.
9(a), in case of no building basement and piles, Stockpile 1 behind Building 7 had a minimum factor of safety,
, of 1.017 against a shallow toe slip failure and a factor of safety (FOS) of 1.4–1.6 against a deep-seated slip failure (bearing-capacity failure). If both the building basement and its underlying piles were considered in analysis, its
slightly increased to 1.027, but its FOS against deep-seated slip failure increased to 1.6–2.0. Because the stability of the stockpile above the ground level was dominated by slope gradient, slope height, soil unit weight, and soil shear strength, the existence of the buried basement and its underlying piles hardly increased the
against a shallow toe slip failure. However, FOS of the stockpile against deep-seated slip failure increased significantly with the presence of the buried basement and its underlying piles. These comparisons proved the previously postulated barrier effect.
Based on these analyses, the 4.6-m-deep excavation can be precluded as a major factor associated with the global stability failure of Stockpile 1 behind Building 7. As recorded by the local weather station (Fig.
10), about half an hour before Building 7 fell over, there was a heavy rainfall lasting from 0:00 a.m. to 5:00 a.m. on June 27, 2009. As recognized in literature (
Stark and Duncan 1991;
Collins and Znidarcic 2004;
Gamez and Stark 2014;
Stark et al. 2017), shear strength of desiccated stiff clay decreased very rapidly to fully softened strength once the clay was soaked. Using filter paper method, osmotic method, and vapor phase technique, Ye et al. (
2006) explored the suction in the desiccated top crust at 1.5–1.8 m BGS in Shanghai. Consistent with the findings in the literature, the measured suction in the crust reduced quickly owing to saturation. Based on direct shear tests on the soil samples from the top crust in Shanghai, Hu and Fu (
2001) investigated the relationship between soil shear strength,
, and water content,
, with respect to four different vertical stresses,
, of 100, 200, 300, and 400 kPa, respectively. As plotted in Fig.
6(c),
decreased rapidly as
increased.
As pointed out by Houston (
2019), rainfall that results in reduction of suction is the most common triggering mechanism of slope failure. Based upon the preceding analyses, it can be postulated that the abrupt deep-seated slip failure of Stockpile 1 behind Building 7 might have been triggered by the heavy rainfall. Intense water infiltration into the ground caused rapid degradation of shear strength of the desiccated crust and increased weight of the stockpile due to saturation; thus, an undrained general shear failure of the subgrade below Stockpile 1 took place, featuring a deep-seated slip failure surface. To verify this postulate, another LEM analysis was conducted for Stockpile 1 behind Building 7, in which reduced strength parameters were used for the top crust. As plotted in Fig.
6(b),
was around 20%–30% for the top crust;
resulting from the weight of Stockpile 1 on the crust was estimated by
, where
was unit weight of the soil stockpile, equal to
, and
was the height of stockpile. As plotted in Fig.
2, the 10-m-high Stockpile 1 with a trapezoid cross section can be simplified as a 5.75-m-high uniform strip load for analysis. Then,
. Because the rainfall lasted for 5 h with precipitation of 23.7 mm (Fig.
10), it can be postulated that the crust was already saturated before the building failure. Therefore, it was reasonable to assume that the crust before the building failure had a similar water content,
, as its underlying saturated soft muddy clay below the phreatic water level, whose
was 40%–60% [Fig.
6(b)]. As plotted in Fig.
6(c), in the case of
(close to
of this case),
reduced from about 90 kPa at
to about 40 kPa at
. In light of this, the strength parameters of the top crust were reduced by 50% as previously introduced in the section entitled “Preliminary Global Stability Assessment” and the saturated unit weights of the stockpile and the crust were adopted in the LEM analysis to account for the rainfall effect. As shown in Fig.
9(c), Stockpile 1 would undergo a deep-seated slip failure and the estimated conceptual failure surface by LEM was in reasonable agreement with the field observation.
As shown in Fig.
4(d), Stockpile 1 behind Building 6 did not undergo slip failure like Stockpile 1 behind Building 7. This inconsistency largely derived from the following two facts: (1) the 10-m-high stockpile behind Building 6 had a smaller size in plane than that behind Building 7 (Fig.
1), and (2) the stockpile behind Building 6 featured a more gentle slope angle (38°) than that (45°) behind Building 7 (Fig.
2). Thus, the stockpile behind Building 7 was more susceptible to slip failure than that behind Building 6. To verify this postulate, a new LEM analysis was conducted for the stockpile behind Building 6, in which the rainfall effect was considered. As presented in Fig.
9(d), the stockpile behind Building 6 would not undergo a deep-seated slip failure after the rainfall, although its
against a shallow toe slip failure was only 1.014.