Slope Stability 2000

A dormant earth flow crossed by Forest Highway 78 in Montrose County, Colorado, reactivated as an earth slide — earth flow in June 1997. The earth slide — earth flow moved at rates of up to a few meters per day and accumulated more than 150 meters of displacement during the summer of 1997, causing considerable damage to the highway and temporary road closures. An irrigation ditch crossing the earth slide — earth flow was also damaged and required reconstruction several times during the summer. Movement slowed rapidly and has essentially stopped. This paper presents the investigation into the cause of the reactivation, the rapid cessation of movement, and the probability of a similar period of activity occurring in the future. The investigation includes study of subsurface conditions, sliding history in the area, climate, and manmade impacts to the slide, such as the road and irrigation ditch.

This paper focuses on the importance of three-dimensional (3D) slope stability analyses in practice. Commercially available 3D slope stability software does not consider the shear resistance along the two sides of the slide mass that parallel the direction of movement in calculating the 3D factor of safety (Stark and Eid 1998). Consequently, the 3D factor of safety may be underestimated and the back-calculated shear strengths may be overestimated or unconservative. A method for incorporating the shear resistance along the two sides of a slide mass in existing 3D software is presented. A parametric study is used to investigate the importance of 3D end effects by providing a comparison of two-dimensional (2D) and 3D analyses for various slide mass geometries and shear strengths along the failure surface. A field case history is used to illustrate the use of the parametric study results and the importance of conducting a 3D analysis in practice.

All limit equilibrium methods of slope stability analysis 1) calculate a factor of safety which is assumed to be the same at all points along the potential slip surface, 2) use only force and moment equilibrium and employ varying assumptions to make the problem statically determinate, 3) assume that the potential sliding mass is a rigid body, 4) need a direction of movement of a potential sliding mass for three-dimensional cases, and 5) calculate a yield acceleration to be used in the Newmark method. To facilitate the slope stability analysis and to avoid some of the above limitations, a numerical technique has been developed and a computer program written to carry out the analysis technique. A single analysis determines local factors of safety and a pattern of induced deformations of a potential sliding mass due to gravity, hydrostatic forces, and base motions. Results of analyses for two example slopes are given to demonstrate the comparison between this technique and the conventional methods.

The slope stability analysis for MSW landfills is usually performed using conventional method of slices or translational wedge method considering potential failure surfaces at limit equilibrium. The safety factor is then determined by comparing the sum of the resisting forces to the driving forces, mobilised along the potential failure surfaces. Usually a safety factor of 1.5 is considered acceptable, other specific values can also be mandated by regulation. It is however uncertain whether or not conventional limit equilibrium methods are applicable to MSW because of their ability to undergo large strains without reaching failure. This paper presents a case study concerning the stability analysis of the Warsaw old town Gnojna Grora hill. The problem of its stability has its roots in the 17^th century when Gnojna Grora was used as a municipal landfill. The paper describes its historical background, and the engineering analysis carried out to verify its stability. The paper shows that there is a high risk of failure of the old town hill if the hydraulic conditions change.

Evaluating sliding stability of rock slopes is probably one of the most overlooked and least understood factors in geotechnical engineering today. The literature has numerous case histories of rock sliding failures that have lead to catastrophic failure of critical structures, some of the most notable being St. Francis Dam, California and Malpasset Dam, France. These case histories serve as reminders that the inadequate or improper identification, characterization, and/or evaluation of the sliding of rock blocks formed by critically oriented discontinuities within a rock mass can lead to catastrophic loss of life and/or personal property. Inherent in any characterization and evaluation of subsurface features, whether it is for dams, tunnels, slopes, foundations, etc., is the uncertainty in identifying the critical features present in the rock mass. Of utmost importance in rock slope stability evaluations are the characteristics of the discontinuities and surfaces of weakness in the rock mass. These characteristics include: orientation and attitude, infilling type, seepage conditions, roughness, among others. In addition to single discontinuities, the intersection of critical discontinuities forming potential 3-dimensional blocks must also be considered.

A limitation of reliability approach to slope safety has been that reliability evaluation yields only notional probability of failure. It would be desirable if the assessed notional probability can be indeed calibrated to the real probability of failure. In this way, landslide risk could be managed with confidence and choices among alternative options could be made more defensible. This paper proposes a calibration procedure based on the Bayesian approach. It will first integrate the analytically assessed failure probabilities of a large set of soil slopes in Hong Kong with the corresponding observed performances of these soil slopes. Both failed and unfailed slopes were used in the calibration process. Through this exercise, the reliability assessed for a given slope in Hong Kong with similar soil material to those used in the calibration study can be calibrated to a realistic reliability measure.

A model of slope stability is presented using a non-homogeneous Bingham model. The constitutive parameters can be calibrated from in situ observations only (inclinometers). It is shown that the yield stress is varying nonlinearly with depth. An example is given. The case of possible slip at the bed rock is discussed.

The factor of safety of a slope can be computed with a finite element or finite difference code by reducing the soil shear strength in stages until the slope fails. The resulting factor of safety is the ratio of the soil's actual shear strength to the reduced shear strength at failure. This "shear-strength reduction technique" can also be applied to geotechnical stability problems such as cantilever walls, braced excavations and retaining walls. The strength reduction technique is demonstrated for several geotechnical stability problems, illustrating some of the advantages and limitations of the technique.

This paper presents the case history of a natural slope, about 10 meters in height, in a sensitive, low plasticity, glaciomarine clay near Portland, Maine. The area in the vicinity of the slope has a history of landslides. A field and laboratory testing program was conducted which included field and laboratory vane shear (VST), <emph type="32">CK_oUC triaxial, and consolidation tests. Based upon regional geotechnical practice, the low undrained shear strengths measured behind the slope using the VST result in a stability factor of safety of 0.64 indicating failure for undrained conditions. However, the observed stable slope performance indicates that the actual factor of safety must be greater than unity. Regional geotechnical practice generally does not use corrections for measured s_u(FV) values in sensitive glaciomarine soils. Application of Bjerrum's anisotropic correction to the s_u(FV) measured at the site and considerations of stress history result in a computed factor of safety which is more consistent with the observed field performance of the slope. The application of VST anisotropic correction factors is recommended for future undrained stability analyses in this material.

Two centrifuge experiments investigating the effects of sinkhole formations on the stability of earth dams are discussed in this paper. A special container and a trap door assembly were designed to simulate formation of sinkholes. Calibration tests on level ground sand models were also conducted. The tests on embankment models provided insights into the mechanisms involved in the cavity formation inside embankments, as a result of sinkhole development. The two embankment models of compacted Bonnie silt were identical except the presence of water reservoir in the second model. However, the water reservoir in this test was contained in an impermeable latex membrane bag, so no seepage was developed. The sinkhole was about 3.6 m deep and 11 m in diameter at the base of a 35.9 m tall embankment in terms of prototype dimensions. In the test with no water reservoir, the surface of the embankment did not suffer major deformations; although a large cavity was formed above the sinkhole. In the test with water reservoir, the shear zone reached the dam surface which created a 3.6 m deep and 11 m diameter depression in the dam surface in terms of prototype dimensions.

The use of drilled shaft concrete piles is generally not a widespread method of stabilizing landslides along the Alberta Highway Infrastructure as a result of perceived high costs, the uncertainty of performance of these structures in the long term and the ready availability of land for realignment, offloading and toe berm construction. This paper presents a case history of a roadway embankment slide which was stabilized using a tieback drilled shaft concrete pile retaining wall. Although the piles were not installed to the desirable depth, the wall has performed satisfactorily since its construction some 10 years ago with none or minimal maintenance to the surface of the roadway.

Stabilization of an unstable section of roadway along a major river valley experiencing deep seated movements was effected through the construction of a pile wall, simple drainage measures and a gabion wall. Since the construction of the pile wall in 1997, the effectiveness of these remedial measures has been evaluated through instrumentation monitoring and observation. So far the roadway is performing well.

With the advances in computer technology and the increased accessibility and flexibility of computer software developed for geo-engineering applications, numerical analysis has gained wide recognition as a qualitative tool to increase understanding and guide prediction of anticipated behavior during underground construction or mining operations. Successful practical problem solving in geo- engineering depends on formulation of a concise conceptual model that correctly incorporates the key factors for the situation at hand. Historically, this conceptual model was built on empirical knowledge and existed only in the mind of the experienced and well-informed engineer. Numerical analysis provides a way to present the conceptual model explicitly, such that it can be reviewed critically, and the effect of possible assumptions and design changes examined and discussed. This paper reviews four case examples to illustrate some of the many ways in which numerical analysis can and has been used to provide valuable assistance in improving practical geo-engineering design.

An investigation has been performed into the stability of an undrained clay slope having spatially randomly varying shear strength. The results of the study lead to a direct comparison between the probability of slope failure and the traditional Factor of Safety for a range of statistically defined input shear strength properties. The results highlight the influence of the spatial correlation length, a variable which is routinely omitted from conventional probabilistic studies in geotechnics.

Many deterministic and probabilistic models can be applied to reliability analysis of earth slopes. This paper investigates the effect of commonly used models by a set of comparative studies based on three prototype embankments. The deterministic models considered are the simplified Bishop method, the modified Swedish method, and Spencer's method. For Spencer's method, both circular and non-circular failure surfaces are considered. The probabilistic models considered are the mean-value first order second moment method (MFOSM), the point estimate method (PEM), and the advanced first order second moment method (AFOSM). The results indicate that the effect of these different models on the calculated slope reliability index varies with slope geometry and soil strength parameters. The effect of deterministic slope stability method is not significant; but the shape of the failure surface can significantly affect the slope reliability, especially for layered embankments. No practical difference was observed between the results of reliability analyses done by the mean-value First Order Second Moment method (MFOSM) and Point Estimate Method (PEM). For the cases studied the AFOSM gave a more conservative reliability index, but it requires more computational effort due to the iterative nature of slope stability problems and numerical problems may occur.

At present, there is little knowledge concerning the shear strength of clays containing floating rock particles with concentrations from 0 to 30%. In practice, the effect of rock particles is typically disregarded in shear strength analysis. The two primary reasons for this are: 1) a lack of agreement concerning the influence of rock particles on material strength, and 2) the expense and difficulty of testing clay-rock mixtures with existing laboratory procedures. These factors have impeded the development of successful slope remediation design techniques for colluvium derived from resistant sedimentary rocks or spoil produced from surface mining. This study reviewed 31 technical papers which contain analysis of shear strengths for clay and sands with varying mixtures of rock particles. These technical papers, published over the last 40 years, are comprised of field case studies, laboratory evaluations, and theoretical analysis. Evaluation of this body of knowledge has shown that the shear strength gradually increases with increasing percentages of floating particles in unsaturated clays.

In this paper, a new approach for modeling of three dimensional slope stability problems is presented. To this end, p-version of the finite element method together with large deformation hyperelastic-plastic formulation is utilized to model localized, continuous deformation that has been observed in failure mechanisms of slopes. In particular, it is shown how the new method can be used with a rather small number of finite elements to model sharp deformation gradients resulting from shear localization during slope failures. Method is well suited for both 2D and 3D computations. In 3D, we show clear advantages of our formulation over classical approach which rely on large number of elements or special elements with enhanced shape functions. In addition to that, solution advancement strategies that allow for accurate following of the equilibrium path in hardening, softening and snap-back regimes are presented. Proposed method is used to simulate behavior of three dimensional slopes. Reduction of safety factors (obtained for plain strain conditions) is observed as the out the of plane curvature of the slope is increased.

By using the upper bound theorem of limit analysis, and assuming the yielding of soil with its partial shear strength, stability calculations have been performed for determining the factor of safety of soil slopes. During failure, the slope material was assumed to rotate as a single rigid mass bounded with logarithmic spiral rupture surface. A detailed stepwise solution procedure has been provided. The results have been compared with those obtained using the Bishop simplified method. In all the cases except for very steep slopes, the upper bound theorem of limit analysis provides almost the same answers as predicted by the Bishop method. For very steep slopes, the Bishop method gives a conservative estimate of the factor of safety as compared to the upper bound limit analysis approach.

A significant advantage of the finite element method over traditional limit equilibrium methods for analyzing slope stability is its ability to monitor the progress of the failure zone up to and including overall shear failure. This advantage results from the finite element's ability to calculate stresses in the slope that result from the application of gravity, or other applied loads, and subsequently compare these stresses to some failure criterion. Within the method, various construction procedures (i.e. loading paths) such as the embankment or excavation can be modeled. In the present paper, finite element stability analysis of a slope with a relatively simple elasto-plastic, Mohr-Coulomb soil model is used to investigate the spread of yield within the slope under differing loading strategies. In particular, loading is accomplished in three ways: gravity applied instantaneously to an initially weightless slope, gravity applied incrementally to an initially weightless slope, and through building the slope up one row of elements at a time (i.e. embankment construction). The influence of loading strategies on the spread of failure is highlighted through the use of contour plots, which indicate the spread of yield within the slope and hence the location and shape of the potential failure surface. Furthermore, the affects of the flow rule used on the yielding zones is shown.

This paper is based on a critical state soil mechanics concept that liquefaction occurs when soil is on the dry side of critical states, near zero effective stress, and in the presence of high hydraulic gradients. In this view liquefaction is one of a group of phenomena; including piping, boiling, fluidization; with pipes and channels and hydraulic fractures, internal erosion and void migration. This paper will refer to some aspects of the failures of Fort Peck, Baldwin Hills, and Teton Dams in support of this view. Casagrande held an opposite view that liquefaction occurs by a chain reaction among sand grains on the wet side of critical states. Cam-clay provides a model for ductile stable yielding and deformation of an aggregate of grains wetter than critical states. A layer of such sediment can form folds during deformation. If a soil aggregate is denser than the critical state, it can fail with fault planes on which gouge material dilates and softens, or it can fracture and crack into a clastic debris, or develop pipes and channels. The critical state explanation of rapid failure is rapid transmission of pore water pressure through such opening cracks or channels. The Baldwin Hills and Teton dam failures were failures with cracks and pipes. In the case of the Fort Peck failure we suggest that high pore pressures from the core hydraulic fill were transmitted in the layer beneath the part of the dam that failed; Casagrande's view of the failure as evidence of a "chain reaction" is questioned. Selection and control of fills to ensure ductility and avoid over compaction and measures to ensure stability are discussed.

The difficulty in estimating engineering properties of peat has prompted the need to use combined field and laboratory tests coupled with an empirical procedure for stability analysis of embankments on peaty ground. Accordingly, this paper presents a case history of stability analysis for construction of an embankment on peaty ground for an expressway in multi-loading stages that failed at a height of 6.3 m. The method outlined here, which is the current state of the practice in Japan, employs cone penetration tests and direct shear tests using Mikasa's apparatus for the process of investigation, analysis, and quality control with a consistent approach.

Twelve homes were constructed in 1949 at the top of a sea cliff along Esplanade Drive in the City of Pacifica, located on the northern coast of San Mateo County, California. The rear yards of those properties were bounded by an approximately 20-meter (70-foot) high cliff that has retreated episodically at an average rate of 0.5 to 0.6 meter (1.5 to 2 feet) per year over the past 146 years. During the heavy storms of the 1997/1998 El Niño winter, a severe episode of cliff retreat undermined seven homes and threatened three others. All ten homes were condemned and demolished by the City of Pacifica. In this study we analyze geologic, tide, wave, rainfall and wind data in an attempt to determine the causes of this most recent erosion event. We identify the following possible contributory causes of the cliff retreat: 1) wave-induced undercutting of the cliff landward of an old revetment, 2) reduction in beach width over time, 3) reduction in cliff-face stability owing to infiltration from heavy rains, 4) erosion of the cliff face by groundwater piping, and 5) wind-induced erosion of loose dune sand at the top of the cliff. While these factors may explain the retreat of the cliff below the twelve homes along Esplanade Drive, the question remains as to why other geologically similar sites in the region were not severely eroded during the 1997/1998 El Niño winter.

The stability of many embankments on soft soils is typically assessed during construction by monitoring the behavior of key parameters through the installation of geotechnical instruments. The measurements from the instruments are typically presented in graphical form and then compared to the results from engineering analyses. This procedure generally works well if there are not a lot of instruments to be monitored, the stratigraphy and embankment configuration are relatively uniform, and construction occurs over a relatively short period of time thereby limiting the amount of analyses that need to be performed. However, for many large landfills over soft soils this procedure can be rather cumbersome because the subsurface conditions and side slopes are likely to change over time, many instruments are needed over a large area, and landfill construction may occur over many years. This paper presents a unique methodology that has been implemented at the Fresh Kills Landfill in New York City to rapidly assess the stability conditions of the landfill during construction based on the large number of installed instruments and computer engineering analyses.

Several large active and ancient landslides have failed along seams of bentonite located on the Palos Verdes peninsula, California. This is a desirable location for residential development due to its commanding ocean views. Numerous geotechnical reports regarding the residual shear strength of bentonites have been prepared over the years. The shear test procedure used has primarily been the direct shear test, but triaxial and ring shear testing have been performed. In-situ testing of block samples has also been performed on at least two occasions. Data indicates that Palos Verdes bentonite samples that have been tested for shear strength have similar mineralogical properties and similar Atterberg Limits. It would be expected that the residual shear strength values obtained would be similar. However, the actual residual shear strength values measured by various researchers vary widely. Direct shear and ring shear tests on samples of remolded bentonite performed for a study by the first author indicate that testing methods can be a significant source of variation in the measured residual shear strength. The residual shear strength determined by various methods was used in slope stability analyses of the Abalone Cove landslide, for which the slide geometry and groundwater conditions at the time of movement are known. The analyses indicate that the residual shear strength of the bentonite determined by ring shear testing resulted in the best fit as determined by a factor of safety of 1.0.

This paper presents a design method for slopes with a row of piles to enhance slope stability or to prevent slope failure. The basic idea of the method is to allow that two different slip surfaces can be assumed in upslope and downslope earth masses of the row of piles. Contrary to the conventional limit equilibrium procedure where a factor of safety is calculated for a given slip surface, in the present method forces acting on the stabilizing piles are estimated based on a given slip surface whose factor of safety is prescribed to ensure the slope stability. By assuming a slip surface in upper and lower earth masses respectively, the forces acting on the piles can be calculated from an existing limit equilibrium method with a prescribed value of the factor of safety. The Bishop method is employed, and a repeated trial procedure is used to find the most dangerous situation for the piles and the corresponding two critical slip surfaces in the upside and downside of the pile row. It may be understood that if the stabilizing piles are still sound under the most dangerous situation, then the slope will have, at least, the prescribed (target) factor of safety. In the proposed method, we clearly distinguish between forces which piles in a row can bear, i.e. the horizontal bearing capacity as passive piles, and forces which piles in a row should bear under the target factor of safety. The consideration in which the latter does not exceed the former makes possible a more rational design for stabilizing piles. The results of some examples illustrate the effectiveness of the proposed method.

Bluestone Dam is a concrete gravity structure founded on interbedded orthoquartzite and carbonaceous shale bedrock. Conventional analyses indicate the dam is unstable at pools approaching the probable maximum flood. Measurements of uplift pressures acting on the base of the dam at elevated pools are not available because 1) the dam has been operated at pool elevations significantly less than those used in design and 2) storms of record have not approached design levels. Numerical modeling was selected to aid in the extrapolation of uplift pressures operable at design pool levels. A preliminary modeling effort was undertaken using the distinct element program UDEC. The preliminary modeling effort had three primary objectives: 1) develop an approach for emulating drain performance and incorporating this within the numerical model, 2) determine the geometric and material parameters that have the greatest influence on uplift pressures for development of an efficient subsurface investigation program; and 3) assess whether numerical modeling can provide reliable information for use in the design process. The scope of this paper includes the parametric study and describes the developed exploration program. The reader is referred to a companion paper published in the 1999 ASDSO Dam Safety Proceedings, St. Louis, Missouri for descriptions of the developed drain model and related uplift pressure distributions.