d

2n International Conference on New Developments in Soil Mechanics and Geotechnical Engineering,

Seismic stability assessment of an existent countermeasure

Deniz Ülgen
Research Assistant, Civil Engineering Department, Middle East Technical University, Ankara, Turkey
e-mail: [email protected]

Selman Sağlam
Research Assistant, Civil Engineering Department, Middle East Technical University, Ankara, Turkey
e-mail: [email protected]

KEYWORDS: Slope stability, seismic effects, permanent displacement

ABSTRACT: The study includes seismic slope stability assessment of a residential site which was
experienced a prior landslide. The landslide induced by an uncontrolled excavation has been
controlled with a rock-fill bracing at slope toe. Pre-failure and post-failure (i.e. with and without
countermeasure) slope stability analyses were performed using limit equilibrium and finite element
methods. The results are utilized to assess effectiveness of the current countermeasure, and to
examine whether the safety factors of the most critical surfaces are high enough. Then, seismic
slope stability of the slope which is in an active earthquake region was analyzed with pseudo-static
and dynamic finite element analyses. Pseudo-static analyses were performed by using Slide V5. As
known, pseudo-static analysis estimates the factor of safety but gives no information about the
permanent displacements. So as to find permanent displacements dynamic analyses were
performed with Plaxis V8. Different scenario earthquakes were selected for the dynamic analyses.
The safety factors obtained by pseudo-static and estimated permanent displacements by dynamic
analyses are discussed in a comparative manner. Finally, the results are evaluated from the point of
the seismic effectiveness of current countermeasure.

1 INTRODUCTION

Landslides have caused devastating damage and loss of life throughout history. One of these
landslides, which was experienced in a residential site of Muğla, Turkey, is to be evaluated within
the context of this study. The landslide induced by an uncontrolled excavation was immediately
controlled with rock-fill bracing at slope toe. However, two residential buildings standing at top of
the slope were exposed to great amounts of damage.
Pre-failure and post-failure (i.e. with and without countermeasure) stability performances of the
slope were firstly analyzed under static conditions using limit equilibrium and finite element
approaches. The results are utilized to assess effectiveness of the current countermeasure, to
examine whether the safety factors of the most critical surfaces are high enough. Then, seismic
stability of the slope which is in an active earthquake region was analyzed with pseudo-static and
finite element methods. The predictive relationship proposed by Bray and Travasarou (2007) was
also utilized to estimate permanent displacements that are aimed to compare with the results of
finite element analyses. The seismic analyses were performed in accordance with real
representative earthquakes.

298
 d
2n International Conference on New Developments in Soil Mechanics and Geotechnical Engineering,
28-30 May 2009, Near East University, Nicosia, North Cyprus

2 BACKGROUND OF THE PROCEDURES USED IN THE ANALYSES

2.1 Static Slope Stability

The most commonly used methods of static slope stability analyses are the limit equilibrium
methods. These methods have been used extensively because of the reliable results corresponding
to careful selection of appropriate input parameters. As known that when shear resistance of a soil
is totally mobilized on a potential failure surface, slope becomes unstable in limit equilibrium
approach. The soil on the potential failure surface is assumed to be rigid-perfectly plastic and the
factor of safety against failure is assumed as constant over the entire failure surface. In real slopes,
however, the factor of safety for each element of soil on the failure surface is not constant. Thus, it
is possible to have a failure even if the factor of safety based on peak shear strength is above 1.
Kramer (1996) suggests that the stability of the materials exhibiting strain softening behavior can
be analyzed reliably by using residual shear strengths.
The finite element method is used to simulate the relationship between stress and deformation
of slope materials. It is a powerful tool which can predict stresses, deformations and pore water
pressures of a specified soil profile. It can also predict the most critical zones and corresponding
safety factors within a slope. Thus, the most likely mode of failure can be identified and
deformations up to and sometimes beyond the point of failure can be calculated.

2.2 Seismic Slope Stability

The seismic shear stresses imposed additionally upon the corresponding static stresses may cause
inertial instability of the slope if the shear resistance is mobilized along a potential failure surface.
The seismic performance of the slope stability is commonly evaluated either by a pseudo-static
analysis that produces a factor of safety against failure or by calculating seismic induced permanent
slope displacements.
The pseudo-static approach, in which seismically induced horizontal and vertical forces
additional to those of static forces are taken into account, intends to calculate the minimum factor
of safety against sliding. The earthquake effects on a potential sliding mass is represented by
horizontal and vertical seismic coefficients. The inertial forces produced by earthquake shaking are
expressed as product of seismic coefficients and the weight of potential failure mass. Accordingly,
the factor of safety is expressed as the ratio of resisting force to driving force. The vertical pseudo-
static force typically has less influence on the factor of safety since it reduces (or increases,
depending on its direction) both the driving force and resisting force and for this reason the vertical
seismic effects are frequently neglected. The results of pseudo-static analyses critically depend on
the value of horizontal seismic coefficient, kh. Therefore, selection of an appropriate seismic
coefficient is the most important aspect of a pseudo-static analysis. Terzaghi (1950) suggested the
use of kh values of 0.1, 0.2 and 0.5 depending on the earthquake severity. The suggestion relating
pseudo-static design criteria by Seed (1979) was based on a study of earth dams. It is indicated that
the safety factors become greater than 1 if the seismic coefficients are taken as between 0.10 and
0.15. Nevertheless, it should be noted that the soil materials used in construction of earth dams do
not show more than 15% strength loss upon cyclic loading. Marcuson (1981) recommended to use
one-third or one-half of the maximum acceleration as seismic coefficients for dams. Hynes-Griffith
and Franklin (1984) suggested that the use of seismic coefficient (kh) equal to 50% of the peak
ground acceleration would not develop large deformations in earth dams with pseudo-static factors
of safety greater than 1. Despite the various recommendations on selection of seismic coefficients,
it seems clear that the pseudo-static coefficient is fairly related with the anticipated or experienced
peak ground acceleration and engineering judgement.
Stress-deformation analyses of seismic slope stability are usually carried out using dynamic
finite element programs. The seismically induced permanent deformation of a slope is calculated as
integrating the seismically induced permanent strains of each element obtained after meshing
process. The permanent strains within individual elements can be estimated in different ways.

299
 Seismic stability assessment of an existent countermeasure against landslide in an urban area
Ülgen D., Sağlam S.

Strain potential approach, which was developed by Seed et al. (1973), is mainly based on
prediction of strain potential of an in situ soil element according to imposed cyclic stresses. The
strains to be developed in the field are assumed to be the same as those developed in a laboratory
test specimen loaded under similar conditions. Stiffness reduction approach was then developed to
calculate permanent strains using reduced stiffness values which are obtained by seismic strain
potential of a nodal point (Lee, 1974; Serff et al. 1976). The strain potential and the stiffness
approaches are very approximate procedures. Therefore, the finite element analyses employing
non-linear soil models (which are rather accurately represents real stress-strain relationship of soils)
may be more useful to find accurate results.
Seismically induced permanent deformations of slopes are commonly estimated by means of
the procedures based on the landmark works of Newmark (1965) and Makdisi and Seed (1978). In
the pioneering method, potential landslide is represented by a rigid block resting on an inclined
plane, and the expected displacement is calculated by integration of the equation of motion for the
rigid block. The limit acceleration triggering the sliding is a requirement in the integration of the
acceleration time history. Thus, the limit acceleration, which is termed as yield acceleration, is
determined as the value for which the factor of safety will drop below 1. There have been many
procedures proposed in the light of rigid block sliding method. Bray and Travasarou (2007)
developed a simplified semi-empirical predictive relationship for estimating permanent
displacements. A non-linear fully coupled stick-slip sliding block model was utilized to obtain a
data set that was used to develop a predictive model based on probability.

3 SITE EXAMINATION

The toe of the slope in evaluation was excavated for purpose of constructing a residential building.
Having reached an excavation height of 3 m, the stability of the slope was ruined. Then, the
landslide was immediately controlled with rock-fill bracing at slope toe. The cross section of the
slope is drawn in Figure 3, and the photos of the site are depicted in Figure 1 and Figure 2
respectively.
Cracks having horizontal gaps up to 50 cm and vertical gaps up to 30 cm were observed at top
of the slope. The slope inclination is about 35o. The fill supporting the slope at the toe consists of
rock which was well graded with moderate to low size pieces. The observations and investigations
conducted both in site and laboratory demonstrates that the predominant soil type of the slope is
slope debris. The great majority of the granules observed in silt matrix are limestone. The slope
debris is followed by limestone rock after a depth of 15-20 m measured within boring logs drilled
at top of the slope.

(a) (b)
Figure 1 a) Cutting at the toe of existing slope b) Rockfill bracing at slope toe for stabilizing slope failure

300
 d
2n International Conference on New Developments in Soil Mechanics and Geotechnical Engineering,
28-30 May 2009, Near East University, Nicosia, North Cyprus

Figure 2. Slope cracks occurred due to 3m excavation at toe

4 STATIC ANALYSES

The limit equilibrium analyses were performed by means of Slide V5 in which the Bishop
simplified method was selected to compute FS values. The finite element analyses were performed
with Plaxis V8. The FS values were computed with respect to the slope stability cases of pre-failure
stability (before the excavation), post-failure stability (just after the excavation), and stability with
rock-fill. The model representing the slope was constituted due to the investigations conducted both
in site and laboratory. The materials contained in the slope soil were idealized within the model as
two main soil types; one of which represents slope debris and clayey silt encountered above the
bedrock of limestone, and the other represents the limestone. The geometry of the model in plain
strain condition and the material properties used in the analyses are shown in Figure 3. The slope
soil was modeled with Material 1, the underlying limestone was modeled with Material 2, and the
Material 3 was used to model the rock-fill which was replaced with the excavated soil. The
materials’ stress strain behaviour was represented by Mohr-Coulomb soil model.

Material 3
 (kN/m3)= 20
4 5
Material 1
 (kN/m3)= 18 E (kN/m2)= 250000
E (kN/m2)= 50000 c (kN/m2)= 1
c (kN/m2)= 5, cresidual (kN/m2)= 5 = 40o
= 30o, residual= 27o
1 3
4 7

1 8 6 2
2 2

Material 2
 (kN/m3)= 20
E (kN/m2)= 200000
c (kN/m2)= 80
y
= 38o
0 x 3

Figure 3. The geometry of the model used in the analyses for rock-fill bracing.

The most critical failure surfaces determined by either Plaxis V8 or Slide V5 and the
corresponding FS values are indicated in Figure 4. The FS values and the critical failure surfaces

301
 Seismic stability assessment of an existent countermeasure against landslide in an urban area
Ülgen D., Sağlam S.

determined by limit equilibrium and finite element analyses are very similar. The FS values are in a
narrow range around the limit value of 1. In the analyses, it drops below 1 as the excavation height
reaches 2 meters. The FS values shown in Figure 4b are for an excavation of 3m height. The slope
material was modeled by the residual parameters of cohesion (c) of 5kN/m2 and an internal friction
angle () of 27o in the analyses of the case with rock-fill. The FS values for this case are observed
to be less than those of pre-failure case. This tendency can be explained by the residual response of
the slope soil.

FS = 1.1 FS = 0.98 FS = 1.01

(a)

FS = 1.15 FS = 0.99 FS = 1.08

(b)

Figure 4. The most critical failure surfaces and the corresponding FS values determined by (a) Plaxis V8 and
(b) Slide V5.

5 SITE SEISMICITY

Muğla is located in the southwest of Aegean Region which is well-known as a seismically active
region. Most of the earthquakes occurred in this area show normal faulting mechanism. Normal
faults form the graben system and indicate the deformation continuity occurring in the Aegean-
Western Anatolia Plate. Sezer (2003) studied the seismic activity and risk in Muğla seismotectonic
region located between 36.00°-37.50°N latitudes and 26.00°-30.00°E longitudes. A series of
statistical analyses were performed so as to determine the seismic risk. The analyses were based on
the earthquakes (Surface wave Magnitude, Ms ≥4) occurred in the region between the centuries of
20 and 21. Epicenter maps of the earthquakes having magnitudes greater than 4.0 are given in
Figure 5. As a result of the analyses, Sezer (2003) found that there is a 95 percent probability for a
4.1 to 6.5 magnitude earthquake in Muğla region. Moreover, he suggested that design earthquake
magnitude should be selected as 6.8 for residential buildings.
From the results of probabilistic studies and the seismotectonic features of the region, it is
concluded that a large magnitude event occurring in Ula-Oren fault would govern the design
earthquake ground motions at the site. For an assessment of design motion parameters, an
earthquake having a large magnitude of Mw=6.8 is considered. The distance to the causative fault is
assumed to be as 20km.

302
 d
2n International Conference on New Developments in Soil Mechanics and Geotechnical Engineering,
28-30 May 2009, Near East University, Nicosia, North Cyprus

Utilizing the attenuation relationships suggested by Abrahamson & Silva (1997) and Kalkan &
Gülkan (2004), and taking the shortest distance from the dam site to the causative fault as 20 km,
the calculated maximum accelerations at the site are given in the Table 1.

Table 1. Expected peak ground acceleration values at site

Since there is no earthquake recorded at or near vicinity of the site, 5 earthquake records are
selected as scenario events for the dynamic analyses. Selected earthquakes and their characteristics
are given in Table 2.
Table 2. Scenario earthquakes used in the analyses
Fault Magnitude Distance PGA Significant Period Ground
Name
mechanism (Mw) (km) (g) duration (sec) Tp(s) type
Reverse normal Northridge 6.7 23 0.172 7 0.24 Rock
Imperial
Strike Slip 6.9 16 0.134 20 0.1 Stiff Soil
Valley
San
Reverse normal 6.8 30 0.134 8 0.1 Rock
Fernando
Strike slip-
Kobe 6.9 16 0.212 12 0.58 Soft soil
reverse oblique
Reverse
Spitak 6.8 30 0.175 8 0.3 Rock
Oblique

(a) (b)

Figure 5. Earthquakes with Ms≥4 in Muğla seismotectonic region during the period of (a) 1900-1950 and (b)
1950-2000 (Sezer, 2003).

6 SEISMIC ANALYSES

Seismic stability of the slope (currently braced by rock-fill) which is in an active earthquake region
was evaluated through the analyses performed with pseudo-static and dynamic finite element
methods. Additionally, the method proposed by Bray and Travasarou (2007) was used to estimate

303
 Seismic stability assessment of an existent countermeasure against landslide in an urban area
Ülgen D., Sağlam S.

permanent displacements as well. In the analyses, the model representing the current situation of
the slope was used.
The pseudo-static analysis was performed with Slide V5. As depicted in Table 1, the expected
peak ground acceleration value at the site is given 0.14g. The seismic coefficient (kh) was selected
as 50% of peak ground acceleration, as recommended by Hynes-Griffith and Franklin (1984). The
corresponding FS value is computed as 0.947. Although the FS is less than 1, estimations of the
expected permanent displacements would provide more profound understanding of seismic
performance of the slope owing to the difficulty in the selection of seismic coefficient.
Plaxis V8 dynamic analyses were performed by means of the records of 5 different earthquakes
(Table 2) which are of the seismic characteristics convenient for the seismicity of the site. In the
analyses, elasticity modulus (E) of the slope soil set to obtain a shear wave velocity (Vs) of 100
m/sec. The dynamic loading was applied as acceleration history of relevant earthquake. The
dynamic stress strain behavior was represented again with the Mohr-Coulomb model. The potential
permanent displacements computed by Plaxis V8 are displayed in Table 3.
Table 3. Expected permanent displacements estimated by Plaxis V8 and the procedure proposed by Bray and
Travasarou (2007)
Earthquake Name Bray and Travasarou(2007) Plaxis V8
Northridge 36cm 35cm
Imperial Valley 23cm 50cm
San Fernando 13cm 20cm
Kobe 60cm 70cm
Spitak 88cm 40cm

Bray and Travasarou (2007) developed a simplified semi-empirical predictive relationship for
estimating permanent displacements. A probabilistic approximation was founded on a data set
obtained by the analyses performed with a non-linear sliding block model. The permanent
displacements probable to occur under the selected earthquake loadings were also calculated by the
proposed procedure. The results with respect to the median estimates of the procedure are
displayed in Table 3. As it is seen, the results of the Plaxis V8, except of the Spitak earthquake, are
higher than the median estimates of the procedure of Bray and Travasarou (2007). The
displacements are almost the same for the Northridge earthquake. The displacement estimated with
Plaxis V8 is greater than twice the median estimation for the Imperial Valley, whereas vice versa is
valid for the Spitak earthquake. The procedures estimate near permanent displacements for the
relevant earthquakes, besides the Imperial Valley and the Spitak earthquakes.

7 CONCLUSIONS

A slope which experienced a prior landslide induced by an uncontrolled excavation has been
evaluated throughout the study. Pre-failure (before excavation), post-failure (just after excavation
without rock-fill) slope stability cases were analyzed statically in addition to the slope with
countermeasure of rock-fill. The most critical failure surfaces and the static FS values computed by
limit equilibrium and finite element methods are almost the same for all the cases. The FS values
are in a narrow range around the limit value of 1. The observed result would not be unexpected for
the slopes composed of such a vulnerable material of slope debris, and having an inclination angle
of about 35o. It should be noted that once the stability is ruined, strength of the slope material
becomes in residual conditions. Therefore, the FS value computed for the case of slope with
countermeasure is less than that of pre-failure case although the toe of the slope is composed of
high strength material.
It would be fair enough to say that the seismic slope stability assessment is more complicated
than that of static slope stability. The cyclic fabric of earthquake loading makes the material have
consistently changing stress strain behavior during the seismic load imposed. Thus, the correctness

304
 d
2n International Conference on New Developments in Soil Mechanics and Geotechnical Engineering,
28-30 May 2009, Near East University, Nicosia, North Cyprus

of the model representing the relationship between stress and strain becomes more significant in
seismic analysis. Although the non-linear soil models are more appropriate to figure out the real
soil behavior, in this study Mohr-Coulomb model was used for the sake of consuming the time
effectively. In addition to difficulty in the selection of best soil model, designation of the probable
ground motion is another drawback of seismic analyses. It is expected to represent characteristics
of a future earthquake that may result great uncertainties.
The pseudo-static approach has been commonly used due to its simplicity. However, the
ambiguity in determination of the seismic coefficient (kh) is one of the major drawbacks of the
method. The FS value computed by pseudo-static approach is 0.947 with the seismic coefficient
selected as 50% of the peak ground acceleration (0.14g). Although FS is less than 1, the indication
of seismic failure needs to be improved due to the closeness of 0.947 to limit value of 1.
Potential permanent displacements are estimated by means of dynamic finite element analyses
and the probabilistic approximation proposed by Bray and Travasarou (2007). The procedures were
performed by using 5 different earthquakes representing seismic characteristics of the site. It is
seen that the results of the procedures are close for the earthquakes of Northridge, San Fernando
and Kobe. However, the displacements estimated by the finite element method are not consistent
with the ones estimated by probabilistic approximation for the earthquakes of Imperial Valley and
Spitak. The displacements computed by Plaxis V8 are ranging between 20-70cm, and the median
displacements calculated by probabilistic approximation are ranging between 13 and 88cm. Thus,
considering the permanent displacement estimations and the pseudo-static FS of 0.947, the current
slope with countermeasure may be expected to fail or experience great displacements during an
earthquake similar with the governing design earthquake.

REFERENCES

Abrahamson, N.A. and Silva, W.J. (1997). Empirical Response Spectral Attenuation Relations for Shallow
Crustal Earthquakes. Seismological Research Letters, Vol. 68, No. 1, pp. 94–127.
Bray, J.D. and T. Travasarou (2007). Simplified Procedure for Estimating Earthquake-Induced Deviatoric
Slope Displacements. Journal of Geotechnical and Geoenvironmental Engineering, ASCE, Vol. 133, No.
4, pp. 381-392.
Hyness-Griffin M.E., Franklin A.G. (1984). Rationalizing the Seismic Coefficient Method. Miscellaneous
Paper GL-84-13, US Army Corps of Engineers. 21pp.
Kalkan, E. and P. Gülkan (2004): “Site-Dependent Spectra Derived from Ground Motion Records in
Turkey,” Earthquake Spectra, 20(4):1111-1138, November.
Kramer, S. (1996). Geotecnical Earthquake Engineering. Prentice Hall, Upper Saddle River, New Jersey.
Lee,K.L. (1974). Seismic Permanent Deformation in Earth Dams. Report No.UCLAENG-7497, School of
Engineering and Applied Science, University of California at Los Angeles.
Makdisi, F. I., Seed, H. B., (1978). Simplified Procedure for Estimating Dam and Embankment Earthquake
Induced Deformations. J. of Geotech. Eng. Vol.104, pp.849-867.
Marcuson, W. F. (1981). Moderator’s report for session on Earth Dams and Stability of Slopes under
Dynamic Loads. Proceedings, International Conference on Recent Advances in Geotechnical Earthquake
Engineering and Soil Dynamics, St. Louis, Missouri, Vol. 3, p. 1175.
Newmark, N.M. (1965). Effects of Earthquakes on Dams and Embankments. Geotechnique, Vol.5, No.2.
Seed, H.B. (1979). Considerations in the Earthquake-Resistant Design of Earth and Rockfill Dams.
Geotechnique, Vol.29, No.3, pp. 215-263.
Seed, H.B., Lee, K.L., Idriss, I.M. and Makdisi, R. (1973). Analysis of the slides in the San Fernando dams
during the earthquake of Feb. 9, 1971. Report No. EERC 73-2, Earthquake Engineering Research Center,
University of California, Berkeley, 150 p.
Serff, N., Seed, H.B., Makdisi, F.I. and Chang, C.Y. (1976). Earthquake-induced deformations of earth dams.
Report EERC 76-4, Earthquake Engineering Research Center, University of California, Berkeley, 140 p.
Sezer, L.İ., (2003). Seismic activity and risk in Mugla seismotectonic region, Quaternary Workshop IV, ITU
Euro-asia Institute of Earth Sciences, pp. 111-120. (in Turkish).
Terzaghi, K., (1950). Mechanism of Landslides, The Geological Survey of America, Engineering Geology
(Berkeley) Volume.

305