Shear Strength in Preexisting Landslides

The selection of shear strength parameters for the design and repair of slopes in landslide areas is important and difficult (Stark et al. 2005). Skempton (1964) concluded that if a failure has already occurred in clayey soils, any subsequent movement along the existing slip surface will be controlled by the drained residual strength. Skempton (1985) suggested that the strength of a clay will be at or close to the residual value on slip surfaces in old landslides, soliflucted slopes, bedding shears in folded strata, sheared joints or faults, and after an embankment failure. Therefore, the drained residual shear strength has been and is still being used for analysis of slopes that contain a preexisting shear surface.

D’Appolonia et al. (1967), Ramiah et al. (1973), Angeli et al. (1996, 2004), Gibo et al. (2002), Stark et al. (2005), and Carrubba and Del Fabbro (2008) suggested that preexisting shear surfaces may exhibit a higher shear strength than the drained residual value after a period of time in which the slope remains stable, i.e., does not experience shear displacement. This assertion of strength recovery may impact landslide mitigation and remedial measures because the increased strength could result in savings to insurance companies and/or landslide mitigating agencies. Therefore, the objective of this study is to investigate the possibility of strength recovery along preexisting shear surfaces using a suitable laboratory shear device, test procedure that models field conditions, variety of natural soils with a range of plasticity liquid limit (LL) 37 to 112%, a range of effective normal stresses from 100 to 600 kPa, and rest periods of up to 300 days. This paper describes the laboratory ring shear and direct shear strength recovery test procedures and the observed strength recovery behavior for a range of soil plasticity and effective normal stresses.

D’Appolonia et al. (1967) concluded that the shear surface in a cohesive colluvial soil in an ancient landslide in West Virginia underwent “healing” because the back-calculated shear strength was found to be greater than the laboratory determined drained residual value. D’Appolonia et al. (1967) performed direct shear tests on undisturbed specimens containing the preexisting shear surface obtained from shallow portions of the slip surface, i.e., at the slope toe and top, and these tests show a peak strength greater than the laboratory drained residual strength at effective normal stresses of 100 kPa or less.

Ramiah et al. (1973) investigated the strength gain in remolded and normally consolidated clay minerals (kaolinite with $\text{LL}=66\mathrm{\%}$ and bentonite with $\text{LL}=400\mathrm{\%}$ ) in reversal direct shear tests under three different effective normal stresses, i.e., 29.4, 58.8, and 98.4 kPa. After reaching the residual condition, the specimens were resheared after rest periods of up to four days. Ramiah et al. (1973) showed a strength gain for high plasticity soil (bentonite) even with a short rest period whereas low plasticity soil (kaolinite) does not show a strength increase.

Chandler (1984) presented a study of reactivated landslides which shows that the back-calculated friction angles are greater than those measured using a ring shear device on slip surface specimens. Chandler (1984) did not discuss the possibility of strength recovery of the preexisting shear surfaces and attributed the difference in friction angle to the movement on a single shear plane in the ring shear device and possibly several different shear surfaces or a shear zone in the field.

Angeli et al. (1996) used direct shear tests whereas Angeli et al. (2004) used ring shear tests to study the strength gain mechanism in different clays including London clay. Tests were performed on normally consolidated specimens using rest periods of up to five days for direct shear and nine days for ring shear for effective normal stresses ranging from 71 to 220 kPa but detailed results for only effective normal stress equal to or less than 100 kPa are presented. Angeli et al. (1996, 2004) reported an increase in the recovered shear strength with time during these direct and ring shear tests.

Gibo et al. (2002) used ring shear tests to measure the strength recovery for two different reactivated landslides. Tests were performed on remolded normally consolidated slip surface specimens at effective normal stresses ranging from 30 to 300 kPa using rest periods of two days. Gibo et al. (2002) concluded that it is reasonable to consider the recovered strength in the stability analysis of a reactivated landslide dominated by silt and sand particles and effective normal stress less than 100 kPa.

Stark et al. (2005) presented laboratory ring shear test results on two soils of different plasticities, i.e., Duck Creek shale $\left(\text{LL}=37\mathrm{\%}\right)$ and Otay Bentonitic shale $\left(\text{LL}=112\mathrm{\%}\right)$ , under a single effective normal stress of 100 kPa. This study was motivated by a 1991 landslide project near Seattle and suggests that a preexisting failure surface may undergo strength recovery and exhibit a shear strength that is greater than the residual value for rest periods of up to 230 days at an effective normal stress of only 100 kPa. It was also observed that the magnitude of recovered shear strength increases with increasing soil plasticity but the recovered strength is lost with small shear displacement after the rest period.

Carrubba and Del Fabbro (2008) conducted similar torsional ring shear tests as Stark et al. (2005) on Rosazzo $\left(\text{LL}=45\mathrm{\%}\right)$ and Montona $\left(\text{LL}=51\mathrm{\%}\right)$ flyschs from northern Italy using normally consolidated specimens, aging times of up to 30 days, and effective normal stresses of 25, 50, and 100 kPa. Carrubba and Del Fabbro (2008) reported more strength gain in Montona flysch than in Rosazzo flysch but all of the testing occurred at an effective normal stress of 100 kPa or less.

Fig. 1 summarizes the test results presented by these prior studies using the ratio between the drained recovered shear strength $\left({\tau }_{\text{rec}}\right)$ and drained residual strength $\left({\tau }_r\right)$ as a function of rest time. Even though these researchers used different devices and test procedures, all soils show a strength gain above the residual strength for a range of test procedures and shear devices at effective normal stresses of 100 kPa or less.

The fully softened shear strength $\left({\tau }_{\text{fs}}\right)$ corresponds to the peak shear strength of a normally consolidated soil with random particle arrangement which is the upper bound strength used for the analysis of slopes that have not undergone previous sliding. Therefore, ${\tau }_{\text{rec}}$ is not expected to exceed ${\tau }_{\text{fs}}$ in the laboratory or field because of the presence of a preexisting shear surface and alignment of the clay particles along the shear surface parallel to the direction of shear (see Fig. 2 where ${\tau }_{\text{fs}}$ is 42.0 kPa which is well above ${\tau }_{\text{rec}}$ ). Furthermore, ${\tau }_r$ is the minimum or lower bound strength which is used in the analysis of ancient and reactivated landslides and is also used as a reference strength for strength recovery because the recovered strength is the increase above the residual value. Thus, ${\tau }_{\text{fs}}$ and ${\tau }_r$ are considered upper and lower bound strengths, respectively, for the recovered strength and provide a means for evaluating the significance, if any, of the measured strength gain. The recovered strength ratio (RSR) is the difference between ${\tau }_{\text{rec}}$ and ${\tau }_r$ divided by the difference between ${\tau }_{\text{fs}}$ and ${\tau }_r$ , as shown in

If no strength recovery occurs, ${\tau }_{\text{rec}}$ equals ${\tau }_r$ and the RSR equals zero. If the soil gains strength to ${\tau }_{\text{fs}}$ (which is not likely/possible), ${\tau }_{\text{rec}}$ equals ${\tau }_{\text{fs}}$ and the RSR equals unity. Thus, the range of values of the RSR is zero to less than unity.

The data in Table 2 are used to calculate the RSR for each soil and each rest period at an effective normal stress $\left({\sigma }_n^{\prime }\right)$ of 100 kPa. Table 2 also presents the values measured for drained fully softened friction angle $\left({\varphi }_{\text{fs}}^{\prime }\right)$ , drained residual friction angle $\left({\varphi }_r^{\prime }\right)$ , and drained recovered friction angle $\left({\varphi }_{\text{rec}}^{\prime }\right)$ for each soil at ${\sigma }_n^{\prime }=100\text{kPa}$ . Corresponding RSR values calculated using Eq. (1) are plotted in Fig. 3. The plots show an increase in RSR with increasing the rest periods for all four soils tested. Fig. 3 also shows that the RSR at an effective stress of 100 kPa is measurably greater than the drained residual strength but is negligible at effective normal stresses greater than 100 kPa. Furthermore, RSR increases for all of the soils tested with increasing rest time albeit at different magnitudes (see Fig. 3). These results are in agreement with prior testing, e.g., Ramiah et al. (1973), Angeli et al. (1996, 2004), Gibo et al. (2002), Stark et al. (2005), and Carrubba and Del Fabbro (2008), which show a strength gain at ${\sigma }_n^{\prime }\le 100\text{kPa}$ (see Fig. 1). The main differences in the current study and other studies are the investigation of strength recovery at ${\sigma }_n^{\prime }>100\text{kPa}$ (up to ${\sigma }_n^{\prime }=600\text{kPa}$ ) and rest periods of up to 300 days or almost one year. Fig. 3 shows that test results at ${\sigma }_n^{\prime }>100\text{kPa}$ are substantially lower than those at ${\sigma }_n^{\prime }=100\text{kPa}$ with values of RSR less than 0.2 for Esperanza Dam and less than 0.1 for Madisette clay.

Duck Creek shale with the lowest plasticity $\left(\text{LL}=37\mathrm{\%}\right)$ and the smallest difference between ${\tau }_{\text{fs}}$ and ${\tau }_r$ exhibits the highest RSR at 100 kPa (see Fig. 3). Fig. 3 indicates that the rate of increase in RSR is decreasing after 300 days and ${\sigma }_n^{\prime }=100\text{kPa}$ so the RSR for Duck Creek shale probably will not reach unity. Otay bentonitic shale with the highest plasticity $\left(\text{LL}=112\mathrm{\%}\right)$ and the largest difference between the ${\tau }_{\text{fs}}$ and ${\tau }_r$ exhibits the lowest RSR for the rest periods considered. Fig. 3 also shows that the increase in RSR for Otay bentonitic shale is starting to decrease with time and probably will not exceed an RSR of 0.4. Silty clay from Esperanza Dam and Madisette clay from Los Angeles exhibit RSRs in between Duck Creek and Otay bentonitic shales which are in agreement with the liquid limit/plasticity of these materials.

Fig. 4 presents the normalized strength ratio (NSR) given by Eq. (2) as a function of rest time on a semilogarithmic scaleFig. 4 shows that the largest increase in NSR occurs for the highest plasticity soil and increases quicker than for lower plasticity soils at an effective normal stress of 100 kPa. The data in Fig. 4 may be important for landslides because many slides occur along high plasticity soil but a significant strength gain only occurred at an effective normal stress of 100 kPa.

In summary, higher plasticity soils exhibit a higher strength recovery at 100 kPa which may be caused by secondary compression, van der Waals attractions, thixotropic hardening, and/or particle reorientation. The ring shear specimens tested herein were submerged in distilled and deionized water, and the tests were conducted at a constant temperature of $21°\text{C}$ ( $70°\text{F}$ ) so desiccation, water chemistry, and temperature probably did not play a major role in the strength increase at ${\sigma }_n^{\prime }=100\text{kPa}$ . During the ring shear tests the water content of the specimens could not be determined because the specimens would have to be unloaded and new soil added to repair the ring shear specimen to measure water content. Water contents measured at the end of the test for each specimen were almost equal to the plastic limit of the soil. Some vertical settlement was observed during each rest period even though the applied normal stress remained constant during the rest period. This vertical settlement is probably due to secondary compression of the shear zone material during the rest period.

It is anticipated that at an effective normal stress of 100 kPa or less, the clay particles have a greater ability to rebound from being oriented parallel to the direction of shear causing a small increase in strength than at high normal stresses. At higher effective normal stresses the clay particles cannot rebound or reorient because of the high applied normal stress; as a result, the clay particles remain essentially parallel to the direction of shear and thus the shear resistance after healing is at or near the previously attained residual strength.

The recovered strength may have an impact on the analysis and repair of shallow landslides in high plasticity material because strength gain was observed at ${\sigma }_n^{\prime }=100\text{kPa}$ . ${\tau }_{\text{Rec}}$ appears to have little, to no, practical impact for deep landslides (see Fig. 4). Remediation of deep landslides is more of a concern because of the higher repair costs involved than shallow landslides. The recovered strength may be applicable to the shallow portion of a deep landslide which is usually fairly small. Thus, it is anticipated that the recovered strength will not significantly impact the analysis or repair of deep landslides.

In addition, the test results herein show that even at low effective normal stress, most of the strength gain is lost if the specimen undergoes a small shear displacement. This indicates that ${\tau }_{\text{rec}}$ reflects a brittle and sensitive soil structure (see Fig. 2) and probably should not be relied upon in design even at effective normal stresses of 100 kPa or less. Thus, the use of a recovered strength in the design of landslide remedial measures is not recommended at this time. However, the strength recovery may be useful in explaining the behavior of shallow landslides, such as explaining the amount and rate of slope creep or slope stability prior to reactivation as suggested by D’Appolonia et al. (1967). In summary, this study confirms the suggestion of Skempton (1964, 1985) of using the drained residual shear strength for remediation of reactivated landslides and for comparison with back-calculated shear strength values.

This study expands on previous studies on strength gain along preexisting shear surfaces in terms of rest periods, range of effective normal stresses, particle size of the material tested, range of plasticity of soils, and stress condition during rest period. Based on the laboratory ring shear strength recovery test results for four natural soils that exhibit a range of plasticity described herein, the following conclusions are presented: