Introduction

Red-bed landslides account for a significant proportion of the geohazards that occur in China, potentially causing severe damage to both buildings and city infrastructure (Wang et al. 2018), especially if many villages are located on the landslides. The occurrence and development of many red-bed landslides are a slow, progressive, and gradual failure process (Miao et al. 2016). After the first slip, the shear zone soil reached the residual state, and the subsequent creeping movement will be governed by the residual strength. Then the landslides can enter a dormant phase and strength recovery occurs, until instability conditions such as heavy rain or seismic events produce temporary attainment of the available strength along the preexisting shear plane inside the landslide, and the landslides will slide again.

As early as 1977, Chandler (Chandler 1977) investigated the Barnsdale landslide and indicated that the shear strength of slip-surface zone was recovered during such a stable period prior to re-sliding, which is reflected in the increase of shear strength (Timothy et al. 2005). The dormant phases can significantly increase the creep threshold stress (Wang et al. 2018), which also means that reactivated shear strength will be higher than the residual value, depending on the holding time and normal stress (Carrubba and Fabbro 2008). This phenomenon of friction strengthening or healing has also been found occurring in fault gouge (McLaskey et al. 2012; Carpenter et al. 2015). In recent years, there have been many studies on a variety of special soils or fault gouge, including experimental study of strength recovery from residual strength on kaolin clay by torsional ring shear test (Bhat et al. 2013), ring shear tests of Italian Cormons flysch (Carrubba and Fabbro 2008) and phyllite colluvium (Gibo et al. 2002), and using microphysical model to explain healing behavior of carbonate fault gouge (Chen and Spiers 2016; Chen et al. 2020). These analyses are important to determine whether or not strength gain occurs and the practical significance of the strength gain in design of the remedial measures.

In this paper, we investigate the strength regain of a red-bed landslide during dormant phases. First, we conduct SHS experiments on the soil to observe its healing and summarize the basic characteristics. Then, the impact of different normal stresses, holding time, and drainage conditions on strength recovery is investigated. Finally, ultrasonic pulse velocity instrument is used to test the wave velocity changes of soil before and after holding to review the effects of healing.

Study area and soil properties

The experimental soil samples for this study were taken from the Xiejiawan landslide, which is a typical red-bed landslide located in Chengdu, Sichuan province (see Fig. 1). The red-bed is composed of red continental clastic rocks as the dominant (Cheng et al. 2004). Because of unique engineering properties, the red-bed is sliding-prone strata in which geological hazards are common (Wu et al. 2018). The Xiejiawan landslide was initiated by the heavy rainfall in the 2018 rainy season, then it stopped sliding and turned into dormant phase until the 2019 monsoon the landslide deformed again. The landslide presently is in creeping condition.

Fig. 1
figure 1

Location of study region (1 Rad-bed. 2 Stratigraphic boundary. 3 Glide direction. 4 Landslide-mass. 5 Shear zone. 6 Sampling point.)

Full size image

The landslide is located in the low mountain terrain of the Longquan Mountains situated on the edge of the Chengdu Plain. The Mesozoic Jurassic (J), Cretaceous (K) and Cenozoic Quaternary (Q) are exposed within the landslide range. The deformed range of the landslide covers a total area of 6.7 × 104 m2 and the volume is evaluated to be 6.7 × 105 m3. The dip angle of the landslide slope is 20°–30°. The bedrock of the landslide area is the purplish red mudstone of the Jurassic Penglai formation. The dipping of the rock strata is consistent with the landslide slope, and the rock mass has weak weather resistance with fissure developed and easy to form a muddy interlayer.

The sampling point is located at the exposed sliding zone in the landslide rear edge. The soil is reddish-brown, with gravel, no bedding, no special smell, and soil mass sections are smooth. The material properties of sample soil are shown in Table 1. Fig. 2 presents the grading curve of the soil.

Table 1 The material properties of the shear zone soil of Xiejiawan landslide
Full size table
Fig. 2
figure 2

Grading curve of sampled soil

Full size image

SHS experiments

Experimental procedure

The reversal direct shear experiments were used to measure the residual strength and the strength recovery of the soil. The sampled soil was remolded in laboratory to simulate natural water content and saturated conditions. We selected 3 groups with 4 soil samples in each group for simultaneous testing. The specimens used in the SHS experiments have a height of 2 cm, effective shear area of 30 cm2. The shear rate was controlled at 0.06 mm/min. Before the SHS tests, the residual strength of the soil would be tested first. We performed contrast experiments on samples with natural moisture content and saturated samples. The samples were sheared repeatedly under different normal stresses of 50, 100, 150, and 200 kPa, and the residual strengths obtained were 38.7, 67.6, 87.5, and 112.1 kPa, respectively, for soils with natural moisture content. The residual strengths of saturated samples obtained at the same normal stress were 29.4, 58.3, 76.4, and 103.4 kPa, respectively. In the SHS experiments, the soil sample was sheared to the residual state first and then holding for a period of time. Because the sliding mass in the field remains subjected to a shear stress after movement, the shear force applied at the end of residual strength experiments was maintained on the specimen throughout the dormant period to simulate field conditions (Stark and Hussai 2010). After the holding period, the shear takes place in the same direction as before stopping until the residual state is reached again.

Comparative experiments of soil samples with natural moisture content and saturated

Figure 3 is the result of a set of SHS tests showing details of the creep, stress buildup, and post-peak frictional weakening. Under the same normal stress and holding time, both the residual strength and the recovery value of the saturated samples are lower than the natural moisture content samples. Obvious peak values appear when the shear is started again, and the shear strength value is higher than the residual strength value.

Fig. 3
figure 3

Data curve of a representative experiment (Stress relaxation and recovery at 100 kPa for 12 h)

Full size image

Figure 4 shows the experimental results of a strength recovery under different normal stresses and holding time. It can be seen that whether the soil is in natural moisture or saturated condition, the degree of self-healing continues to increase with the increase of the holding time and normal stress (see Table 2). In the slow-speed deformation sliding, the steady-state shear strength is weakly dependent on normal stress (Ujiie and Tsutsumi 2010), and taking the holding time of 12 h as an example, the recovery value is 7.3, 8.3, 10.1, and 12.8 kPa at 50, 100, 150, and 200 kPa normal stresses, respectively. It can be seen that the healing of the shear plane is more dependent on the holding time. But the normal stress plays a decisive role in the residual strength of the soil. When the holding time reaches 3000 s, the measured shear strength exceeds the initial peak shear strength of the soil samples. This is because the specimens are made by disturbed soil, and the original structures of the soil were destroyed in remolding processes. The effects of holding time and normal stress on strength recovery are discussed in detail in the following part of the paper.

Fig. 4
figure 4

Schematic diagram showing results of a strength recovery experiments under different normal stresses and holding time

Full size image
Table 2 Comparison of strength recovery values for soil samples with different moisture content (1: samples of natural moisture content; 2: saturated soil samples)
Full size table

Effects of holding time

It can be seen from Fig. 4 that the strength recovery is correlated with holding time. We summarized the effect of holding time on strength recovery in this section. From Fig. 4, when holding time is short (less than 3000 s), the differences of recovery values are small, and there is no obvious regularity. This phenomenon was also discovered in previous studies (Ikari et al. 2016), so only the data for holding time over 3000 s are used in subsequent analysis. Figure 5 shows that the recovery value log-linear depends on the holding time for soil samples of natural moisture and saturated, respectively. And the saturated soil samples have the same correlation, but the recovery value is slightly smaller.

Fig. 5
figure 5

Variation of factor of shear strength with holding time under 100 kPa

Full size image

Mechanisms of the strengthening of landslide soil may be due to the dislocation motions of particles and the reconstruction of soil structure with compaction in the holding time. When the soil is in the residual state, the soil particles are redirected due to the shearing effect. It is difficult for the particles facing each other to establish a connection. After a certain pressure and holding time, dislocation motions occur and the interaction between the particles will increase, which will lead to an increase in shear strength when shearing again. Carpenter et al. (2016) suggested that strength recovery is related to the mineral characteristics and strong, angular minerals (such as quartz and feldspar) content of the soil. Samples that exhibit large compaction and dilatancy are generally composed of stronger, angular minerals and exhibit higher rates of friction strengthening (Carpenter et al. 2016). For the typical red-bed soil in this study, the phenomenon is more obvious due to the higher quartz and feldspar content. The difference in mineral composition will lead to different soil structure, and the soil structure directly determines the peak strength. The experiments of Bhat et al. (2013) proved that the soil with the smaller difference between the peak strength and the residual strength shows a lower value of recovered strength. This indicates that the soil has structural recovery during the holding period.

Regression analysis is then performed on the recovery rate with the holding time more than 3000 s. The definition of recovery rate is the ratio of the strength recovery value to the holding time. The strength recovery value is the difference between the peak strength obtained by shearing again and the residual strength. Figure 6 shows that the strength recovery rate conforms to the power law relationship with holding time in logarithmic coordinates. At the same time, it can be seen that the recovery rate slows down with the increase of holding time. This may be because, after a holding time, the soil samples are relatively dense, the pores have been reduced, the structure has been optimized, and the contact and connection of soil particles tend to be sufficient. In the study of fault healing, McLaskey et al. (2012) proposed that the creep mechanism caused the healing phenomenon, and these effects gradually become full with time, which led to the weakening of the healing rate. With the increase of the holding time, the contact between the soil particles is more sufficient, the degree of cementation is also increased, the particle distribution method is optimized, and the rough contact creep of the shear plane and the adhesion at the rough contact continue to strengthen over time (McLaskey et al. 2012).

Fig. 6
figure 6

Relationship curve between recovery rate and holding time under 100 kPa

Full size image

Effects of normal stress

Figure 7 shows the effects of normal stress on recovery rate values for 12 h holding time. Under the same holding time, the recovery rate increases almost linearly with the increase of normal stress for moisture condition of saturated and unsaturated. The recovery rate increases 1.3 × 10−4 kPa/s and 1.5 × 10−4 kPa/s when normal stress increased from 50 kPa to 200 kPa for moisture condition of natural and saturated, respectively. A number of previous studies have concluded that a higher normal stress is beneficial to the development of mechanical strength of porous media (Åhnberg 2007; Fahey et al. 2011). This is because larger normal stress makes the shear plane particles cement faster, and the soil becomes denser faster.

Fig. 7
figure 7

Recovery rate values of different normal stresses for 12 h holding time

Full size image

Effects of drainage conditions

In this section, the effects of drainage condition on soil healing are investigated. Filter paper is used to control the drainage of samples tested. By using different layers of filter papers, the drainage condition is changed in shearing. The medium-speed qualitative filter paper with a pore size of 30–50 μm was selected. The soil moisture content will be measured after the experiments to evaluate the degree of drainage under different drainage conditions. Figure 8 shows the shear strength recovery of the soil under different drainage conditions after being held for 12 h at normal pressure of 100 kPa. It should be noted that the results that the addition of filter paper makes the water absorption effect outstanding, thereby accelerating drainage. It can be seen that healing of this red-bed landslide soil is very sensitive to drainage condition, and a better drainage condition can enhance strength recovery of the landslide soil prominently. Therefore, the reasonable setting of drainage measures may greatly improve the stability of the red-bed landslide. Drainage conditions directly affect the change of soil moisture content, and the decrease of soil moisture content leads to the decrease of pore water pressure and the increase of effective soil stress. In addition, the decrease of soil moisture content will increase the matrix suction, and the increase in matrix suction also contributes to the shear strength acquisition (Fredlund et al. 1978; Khalil et al. 2004). In unsaturated soil, all elements inside the soil structure will adjust and change during the dormant phases in order to adapt to the process of the two-phase movement of water and gas in the pores of the soil. But for saturated soil, the water in the pores cannot be discharged in time, and there is no compressibility, the water will bear all the load inside the structure. At this circumstance, the soil particles are not very stressed, so the structural adjustment is limited.

Fig. 8
figure 8

The shear strength recovery of the soil under different drainage conditions after being holding for 12 h at 100 kPa (a: 5 layers filter paper drainage conditions; b: Complete drainage conditions; c: 3 layers filter paper drainage conditions; d: 1 layer filter paper drainage conditions)

Full size image

Ultrasonic experiments

The James Instruments V-Meter Mark IV is an ultrasonic pulse velocity instrument. It can generate low frequency ultrasonic pulses, and measure the time taken for ultrasound to travel from one transducer to the other through the material experimented. At the laboratory scale, ultrasonic observations may be a useful tool to infer porosity changes and fabric evolution, and at the field scale, elastic wave speed changes within damage zones may be a proxy for bulk porosity changes and could provide critical information on shear zone strength evolution (Kaproth and Marone 2014). The experiment of Kaproth and Marone (2014) once provided detailed observations of fault zone damage during fault slip and subsequent fault healing, which confirmed the availability of elastic wave test in material strength recovery test.

We performed P wave detection on the soil before and after the dormant phases, and compared the results of waveform and wave speed. The P wave passing through the soil samples will inevitably pass through the shear plane. During the entire experiment, only the soil material on the shear plane was destroyed and recovered. Therefore, the change of waveforms before and after holding can be used to clarify the healing of the shear zone. We still selected 3 groups of 4 soil samples in each group for parallel tests under different normal stress conditions. Table 3 shows the results of wave velocity differences and travelling time differences for soil samples with the natural moisture content with holding time to be 3000 s, 10,000 s, 12 h and 24 h. All velocities measured after the dormant period are greater than the initial values, and the propagation time is less than the time required before the dormant period.

Table 3 P wave velocity difference and time difference before and after dormant phases of soil samples
Full size table

Figure 9 shows typical waveforms of soil sample tested before and after 12 h of holding under normal stress of 50 kPa. It can be clearly seen that the waveforms have changed a lot after holding, the wave curve lasts longer at the peak after the dormant period, indicating that the energy of the P wave decays faster in the residual state before healing. This is the effect of the layered structure surface of the soil sample (the largest layered surface is the shear plane of the soil sample). The energy attenuation of the waveform after holding became weaker, and the peak duration is longer, indicating that the porosity of the soil sample decreases after holding, and the soil is more complete and homogeneous. Figure 10 shows the relationship of wave speed difference and holding time, and it is found that they are linearly positively correlated. We can intuitively see that with the increase of holding time, the wave speed difference (The wave velocity difference is the speed after holding minus the wave speed in the residual state, and the greater the difference, the greater the degree of recovery of the soil structure.) is gradually increasing. This means that the soil structure is recovered more completely after holding.

Fig. 9
figure 9

Configuration of ultrasonic experiments and waveform comparison of soil samples at 50 kpa before and after holding for 12 h (a: Waveform of residual soil sample; b: Waveform of holding 12h soil sample)

Full size image
Fig. 10
figure 10

Relationship curve between holding time and wave speed difference

Full size image

Conclusions

This paper discussed the self-healing effect of typical red-bed landslides soil though laboratory experiments. The following conclusions can be drawn:

1 SHS experiments show the existence of self-healing of the shear plane. Laboratory experiments results indicate that with the increase of the holding time or normal stress, the strength recovery value increases obviously. It is found that the recovery value log-linear depends on the holding time, and the healing rate shows a power-law reduction relationship with the holding time in the logarithmic coordinate. Both the recovery value and the recovery rate of saturated soil are lower than that of the soil with natural moisture content.

2 Drainage condition has obvious influence on the strength recovery of the soil. Under the same holding time and normal stress, better drainage condition (in which the final soil moisture content lower), can enhance strength recovery of the landslide soil prominently. Therefore, the installation of drainage measures is particularly important in the prevention and treatment of red-bed landslides triggered by rainfall.

3 The difference of the waveform before and after the dormant phases can clearly reflect the change of the shear plane of the soil sample. The ultrasonic tests can be used to evaluate the degree of shear plane healing of landslides in field.