Engineering geology

Human-induced landslide on a high cut slope: a case of repeated failures due to multi-excavation


Human activity has been recognized as an important triggering factor that leads to many landslides (Turner and Schuster, 1996, Huang and Chan, 2004, Erginal et al., 2008, Ayalew et al., 2009). Human-induced landslides usually occur on cut slopes due to excavation, which can cause many fatalities and strong destruction (Huang and Chan, 2004, Borgatti and Soldati, 2005, Stark et al., 2005, Erginal et al., 2008, Ayalew et al., 2009, Lee and Hencher, 2009, Zhang et al., 2009).

The effect of excavation on the stability of cut slope has been a hot issue. Many studies have found that the failure on cut slopes is related to stress release induced by excavation (Lutton, 1971, Burland et al., 1977, Tsidzi, 1997, Bozzano et al., 2006, Abderahman, 2007, Rupke et al., 2007, Erginal et al., 2008). Some researchers pointed out that the progress of the stress release in stiff clay slopes is characterized by retarded response to excavation (Bozzano et al., 2006, Burland et al., 1977). Meanwhile, for the failure on cut slopes in bedrock, the exposed geological structures and the toe unloading are main causes (Stark et al., 2005, Erginal et al., 2008, Ayalew et al., 2009). For the long-term stability of cut slopes, weathering-induced undercutting can also increase its destabilization (Tsidzi, 1997, Abderahman, 2007), although rainfall is generally a main trigger (Al-Homoud et al., 1997, Stark et al., 2005, Erginal et al., 2008, Lee and Hencher, 2009). Lee and Hencher (2009) reported a case of cut slope instability induced by excavation in Korea. They found that the failure is related to groundwater and rainfall, but the classification of weathering rocks is of foundational importance in the response of cut slope stability to rainfall. In addition, many studies have recognized that geological investigation prior to design is crucial for understanding failure mechanism, which is in turn helpful in reducing the cost of remedial work and constructing geological model for numerical simulations (Lutton, 1971, Tsidzi, 1997, Potts, 2003, Bozzano et al., 2006, Rupke et al., 2007, Lee and Hencher, 2009). In many cases, however, the interest is focused on the long-term stability (i.e. progressive failure) of cut slopes. Little attention, unfortunately, has been paid to examining the abrupt, repeated failures induced by multi-excavation. Such an examination is essential for the understanding of the rapidly reapted failures of cut slope.

In China, landslides due to excavation in various engineering activities are frequently reported on cut slopes. Unfortunately, current knowledge of these landslides is still incomplete in certain key areas (Wang, 1998, Huang and Chan, 2004). Hence, presenting representative case study on cut slope failure is useful for response of emergency planning, regulating development and landslide hazard mitigation in urban regions, transportation corridors and great engineering areas.

In this paper, a case history of repeated failures on a high cut slope due to multi-excavation is introduced firstly. Then, the characteristics and causes of the repeated failures on cut slope are analyzed by geological investigations and numerical modeling. Geological investigation is employed to analyze the features of slopes after each failure, and numerical modeling is conducted to simulate the stress state after each failure by using FLAC3D code. Simulation results agree well with the phenomena observed in field, showing the rationality of the proposed model.

Geological setting and human activities in the landslide area

Geological setting

The study site involves highway G212, 15 km east from Lanzhou Basin in Gansu Province, China (Fig. 1). The strata of the study area range from Cretaceous to Quaternary (Fig. 2). Bedrocks are composed of mudstone, sandstone, sandy conglomerate and sandy clay with shale. The Quaternary materials belong to Holocene epoch, mainly including Malan loess, as well as gravel, sand and loam on the terrains. Structurally, the study area is relatively simple without tectonic line. But there are some structural planes within the bedrocks.

Fig. 1. Location and topographic map of the landslide.
Fig. 2. Geological map of the landslide area.

Cut slope in the study area is hosted on Cretaceous capped with Quaternary Malan loess (Fig. 2). The rocks are mainly composed of sandy conglomerate, intersected by weak and thin interlayer of mudstone. The rocks are characterized by bedding slope, and its inclination is 80°–102° with dip of 32°–42°. The Quaternary Malan loess covers the total slope. The thickness of loess is 3–10 m on the slope and thicker at top of the slope (approximately 20 m). Additionally, there is a set of almost vertical joint in the upper part of the cut slope. The bedrock surface, forming interface of the cut slope, fails almost parallel to the interface, and has approximately the same slope angle as the interface.

Human activities

Human activities were very strong during the construction of highway G212 in this study area. Nine high and steep cut slopes can be observed within a distance of about 5 km along the highway (Fig. 1(a)). Frequent failures of several cut slopes were captured during construction and operation after completion of excavation (Chen et al., 2005, Zhang et al., 2009).

A typical cut slope is considered in this study. The original slope angle of the cut slope was 26°–30°, gentler than the dip of bedding (32°–40°) (Fig. 3(a)). Furthermore, field investigation prior to excavation showed that the original slopes were stable. The cut slope containing 8 benches was initially designed, and the ratio of each cut slope was 1:0.75 (53.3°). The height of the slope was 80 m, and the excavation was divided into 4 stages. Due to repeated failures during excavation, the initial design was modified for many times by increasing the number of benches and reducing the ratio of cut profiles. Finally, the cut slope was reconstructed as a high and steep bedding cut slope, which was about 100 m high with overall inclination of approximately 60° (Fig. 3(f)).

Fig. 3. Schematic map of failure histories of the cut slope.

Analysis of slope stability

Failure history of the slope

Failure history of the slope is presented by geological investigation after each failure from 2003 to 2006 (Fig. 3). The repeated failures include four relatively large failures along the weak bedding plane of mudstone, several small-scale failures and local deformations.

Excavation started near the interface between loess and bedrock in April, 2003. When the cutting reached the 5th bench (elevation 2 132 m) in July, 2003, the first failure occurred at the toe of the 6th bench along the weak bedding plane of mudstone (Fig. 3(b)). The thickness of displaced mass was about 5 m, with a displacement of about 3 m.

After a rapid remedial work, excavation reached the 4th bench (elevation 2 122 m) in December, 2003. Consequently, a greater failure occurred along a relatively deeper bedding plane of mudstone with a depth of 6 m (Fig. 3(c)), and the failure-induced area was 190 m long and 200 m wide. A great number of fissures and tensile cracks were developed in the displaced mass (Fig. 4). The crown of landslide expanded upward to the higher part of natural slope (elevation 2 182 m). Initial design was modified by increasing the number and width of benches (Fig. 3(c)). In addition, construction of concrete paving was conducted on cut slope from the 7th bench.

Fig. 4. Characteristics after the 2nd failure.

The 3rd failure occurred still along the bedding plane of mudstone in March, 2004 (Fig. 3(d)), when excavation was carried out towards the 2nd bench (elevation 2 102 m). This failure formed a sub-vertical scarp in the loess of about 3 m high (Fig. 5(a)). Field investigation also found the evident sliding traces in the superficial loess (Fig. 5). The displaced mass extended to deeper regions on the slope. The damaged area almost crossed the whole original slope. Due to the rainy season, drainage ditch and waterproof curtains were also employed on the cut slope. The damage degree of the slope became more serious, but controlling measures such as installation of anchor system were not adopted due to the lack of funding.

Fig. 5. Characteristics after the 3rd failure.

The 4th excavation was carried out at the elevation of 2 082 m (Fig. 3(e)). Due to the rainfall in June and July, 2004, small-scale damage at the lower part of cut slope, from the top of the 2nd bench to road pavement, occurred in July, 2004 (Figs. 3(e) and 6). After failure, concrete paving was implemented in the whole cut slope by widening bench and reducing the ratio of cut profile (1:1). In addition, retaining wall was adopted above the 2nd bench. The cut slope was eventually completed and the highway was operated at the end of 2005.

Fig. 6. Characteristics after the 4th failure.

The 5th failure occurred in the middle of cut slope in July, 2006. The failure region extended to the top of the slope (Fig. 7), and the sliding surface approached the depth of about 30 m (Fig. 3(f)). Remedial works and maintenance measures were conducted on the damaged cut slope.

Fig. 7. Characteristics after the 5th failure.

Large-scale damage did not occur after the 5th failure of the slope, but local deformation or damage was frequently encountered, such as cracks and exfoliation on the cut slope (Figs. 8(a), 8(b)). In addition, two small-scale damages occurred on the cut slope (Figs. 8(c), 8(d)).

Fig. 8. Deformation and destruction on the cut slope after the 5th failure.

Analysis of repeated failures

The first failure of slope presented a quick response to excavation, which belongs to toppling type. It occurred in the layers of loess and bedrock of the upper cut slope (Fig. 3(b)). The following two failures gradually became large-scale, developing toward the deeper and higher region on the cut slope. It was clear that the responses of the slope to excavation by two failures were rapid. The increase in damage (Figs. 3(b), 3(c)) indicated that the larger excavation led to greater scopes of the exposing bedrock. In addition, the two damages on the cut slope formed a series of arched tensile cracks around the crown of landslide. The tensile cracks propagated toward the upper part of slope to induce a larger failure, which is dependent on the area of bedrock exposure and the scale of excavation on the cut slope. The shallow sliding of loess and field observation of the slope revealed that the failures were characterized by great energy release and high velocity. The 4th failure, triggered by rainfall after the completion of excavation for some time, was significantly different from the previous failures. The damage was slight and only occurred at the toe of cut slope, but the reinforcements on cut slope were responsible for the failure, including waterproof curtains. This is due to the fact that the reinforcements were not timely performed after excavation, which was helpful for the rainfall infiltration into the toe of the cut slope. The 5th failure occurred until one year after all excavations were completed. The delayed behavior of the 5th failure associated with excavation showed a slow stress-release process, while the previously abrupt failures reflected a rapid stress-release process. The following local destruction and deformation may be related to the stress redistribution of the cut slope.

The repeated failures indicated that the excavation had a direct impact on slope stability by exposing the weak bedding plane and unloading the toe of the cut slope. As a result, the sliding resistance of slope was reduced, and the decrease in shear resistance occurred on the weak bedding planes. In such a situation, gravity-induced failures occurred frequently on the cut slope. In the present case, the impact of rainfall on cut slope was not dominant for the repeated failures, although the 4th failure was triggered by rainfall, followed by local instability of the cut slope.

Numerical modeling

To better understand the repeated failures, numerical modeling of the cut slope is conducted using FLAC3D code (Itasca, 1997). Numerical modeling simulates the gravity-driven deformation of a three-dimensional model, taking into account the main excavation process due to its predominant role in the repeated failures. Geological model is constructed based on the geological and geomorphic conditions of the slope (Fig. 9). The initial state of the model is represented, in which the state variables are defined and forces within the mesh are in equilibrium. Nodes at the bottom boundary of the model are constrained in both horizontal and vertical directions. Both side boundaries of the model can move freely but do undergo end stresses. The pore water pressure is ignored, because the effect of water on the failure of cut slope is negligible. However, the model only simulates the main four excavations, and it cannot consider the local destruction and deformation. Slope materials in the constitutive model are considered according to the geological features. The elasto-plastic Mohr-Coulomb yield criterion is adopted for the materials of the cut slope. The yield criterion is the classic and widely used one in the geotechnical engineering. Thus, the detailed descriptions of the Mohr-Coulomb criterion can refer to FLAC3D manual (Itasca, 1997). Some parameters of the materials used in the model are listed in Table 1, and they are determined jointly by laboratory tests and back calculation (Liu et al., 2005).

Fig. 9. Numerical model of the slope.
Table 1. Physical and mechanical properties of the materials used in the model.

Fig. 10, Fig. 11 illustrate the changes in shear stress and normal stress on cut slope induced by excavation. Simulation results show that each excavation induces reduction in normal stress on cut slope and the increase in shear stress along bedding planes. Meanwhile, large-scale excavation will lead to large range of the stress redistribution. Larger shear stresses are mainly distributed on the weak bedding planes, the interface between bedrock and loess, and the toe of cut slope (Fig. 10). This implies that the concentration of shear stress exists on cut slope. However, the normal stress reduction zones, i.e. the unloading zones, occur on displaced mass, which locates above the shear stress concentration zone on bedding planes (Fig. 11). This implies that the unloading occurs only on the disturbed slope.

Fig. 10. Distribution of shear stress (τy)
Fig. 11. Distribution of normal stress (σy).

The simulations reveal that the stress release occurs on the cut slope, followed by the gravitational instabilities along the bedding planes of mudstone.


The repeated failures on a high cut slope are analyzed based on the field investigations after each failure. Numerical modeling is conducted to represent and better understand the failure histories. The following conclusions can be drawn:
Geological investigations based on geomorphic and sliding surface features after each failure show that the repeated failures are related to the exposure of the weak bedding planes and the toe unloading on the cut slope due to excavation.

Numerical modeling reveals that the excavation-induced stress release leads to the repeated gravitational instabilities of cut slope due to the reduction in normal stress and the increase in shear stress along the bedding planes of mudstone.

The combination method of geological investigation and numerical modeling is very helpful in avoiding unreasonable decision prior to design and construction.

The limit in this study is observed in the presence of the actual failure process of cut slope induced by multi-excavation, for lack of sufficient in-situ monitoring. Thus, for further study on similar cases, it is strongly recommended to conduct field monitoring of the deformation of ground surface and shear zone as well as the water table.

Source: Human-induced landslide on a high cut slope: a case of repeated failures due to multi-excavation

Authors: Fanyu Zhang, Gao Liu, Wenwu Chen, Shouyun Liang, Ransheng Chen, Wenfeng Han

Leave a Reply

%d bloggers like this: