You are on page 1of 6

XVI Danube - European Conference on Geotechnical Engineering

07-09 June 2018, Skopje, R. Macedonia


Paper No. 055

Stability analysis of “Kukovi” landslide


ŠPAGO Azra1, HADROVIĆ Armin2, HAJDUK Ismet3, JAZVIN Ena4
1
Phd., Civ.Eng.; University “Dzemal Bijedic” Mostar, Faculty of Civil Engineering, Sjeverni logor bb, Bosnia
and Herzegovina; azra.krvavac@unmo.ba
2
Phd., Civ.Eng.; the same as above; armin.hadrovic@unmo.ba
3
Phd., Civ.Eng.; the same as above; ismet.hajduk@unmo.ba
4
M.Sc., Civ.Eng.; Ministry of Transport and Communications of the Herzegovina-Neretva Canton, Maršala Tita
91, Mostar, Bosnia and Herzegovina; enajazvin@gmail.com

Abstract. This paper presents conducted stability analysis for landslide „Kukovi“, Bosnia and Herzegovina. As
all geological cross sections are formed of rock mass that lies under upper layers of soil cover, analysis have
been run by using Hoek Brown failure criterion and Mohr-Coulomb failure criterion respectively. Data on
geotechnical parameters were determined by statistical analysis of results from laboratory experiments, field
tests and literature data. Values of safety factor are obtained by deterministic slope stability analysis using
characteristic values of parameters from statistical distributions. Sensitivity analysis for safety factors has been
completed and Monte-Carlo probability analysis adopted distribution curves for parameters.

Keywords: Slope stability; Deterministic analysis; Sensitivity analysis; Monte-Carlo probability analysis

1 INTRODUCTION

Road tunnel “Crnaja” as well as three railway tunnels on track Sarajevo-Mostar-Ploče are located in
landslide area “Kukovi”. Area is characterized by highly complex geomorphological structure and
hydrogeological conditions. Activity of landslide was determined in 1956 and ever since several
detailed investigative studies were performed on landslide toe, in vicinity of infrastructure. The most
recent research works and monitoring conducted by Faculty of Civil Engineering at Džemal Bijedić
University of Mostar (2010-2012, 2015) indicated no landslide activity and included laboratory tests
on samples, water tightness test in four drilling boreholes, data on groundwater level as well as
investigation in parts of landslide area that were not treated in previous studies. Obtained results from
this study served as input data for slope stability analysis in five characteristic profiles along the
landslide. Analysis was carried out using Rocscience limit equilibrium software Slide.

2 GEOLOGICAL CROSS SECTIONS AND LOAD CASES

Stability of five characterisic cross sections through slope has been analysed and different litological
layers have been determined as following:
Soil cover: a multicolored sandy clay, an enriched fragment of marl, sandstone, shale,
clay schist and limestone
Flysch T11: rosy or rosy-greenish series of leafy marls, shale, sandstone and clay schist
Flysch T12: gray-yellow series sheet-marl and thin plate-like marly limestone,
Flysch T13: gray-yellow series of thin plate-like marl and limestone.

Two load cases have been analysed: basic load case combination LC I – dead load as self weight +
maximum groundwater level (14.2m from surface) and seismical combination of load LC II – dead
load as selfweight + most common measured value of groundwater level (24.0 m from surface) +
seismic load (EN 1998-5 Eurocode 8).

© Ernst & Sohn Verlag für Architektur und technische Wissenschaften GmbH & Co. KG, Berlin · ce/papers 2 (2018), Nos. 2 & 3
https://doi.org/10.1002/cepa.724 wileyonlinelibrary.com/journal/cepa 523
Špago, Hadrović, Hajduk, Jazvin / DECGE 2018

3 GEOMECHANICAL PARAMETERS OF BEDROCK AND SOIL COVER

3.1 Hoek-Brown failure criterion parameters for bedrock (flysch)

Uniaxial compressive strength of intact rock is determined from statistical analysis of specimen
laboratory results (257 samples) and literature studies (Chau and Wong 1996, Marinos and Hoek
2001, Hobbs et al. 2002, Rusnak and Mark 2006, Akram and Bakar 2007, Agustawijaya 2007).

Figure 1. Distribution curve ci (UCS), minci = 5 MPa; maxci = 138 MPa; modci = 29,22 MPa; aveci = 38,86
MPa; stdevci = 17,76 MPa; varci = 315,47 MPa2; mean,log ci = 3,57MPa; stdev,log ci = 0,44 MPa.

Comparison of values of statistical analysis results (minci, maxci, aveci, stdevci) taken from individual
locations on site has not shown any particular deviations. For this reason, as well as inability to define
fine line between different layers, it was decided that layers T11, T12 i T13 shall be treated as one single
layer of flysch.
To conserve space, details are omitted for other geotechinical parameters and instead presented are
only statistical calculation results. Based on these results, distribution curves are created and later used
for Monte-Carlo probability analysis.
GSI values determined from field investigation and studies of literature (Hoek et al. 1998, Hoek 2000,
Marinos and Hoek 2001): minGSI = 15, maxGSI = 27, aveGSI = 21, stdevGSI = (aveGSI – minGSI)/1,64 =
3,6.
Hoek-Brown parameter mi: minmi = 3, maxmi = 11, avemi = 7, stdevmi = (avemi – minmi)/1,64 = 2,4 (Hoek et
al. 1998, Hoek 2000, Marinos and Hoek 2001).
Statistical analysis of laboratory testing results (50 samples) give statistical characteristics of beta
distribution for unit weight of rock: min =13,27 kN/m³, max = 27,60 kN/m³, mod = 27,61 kN/m³, ave =
23,18 kN/m³, stdev = 4,25 kN/m³.
Hoek-Brown failure criterion presented through diagram of major and minor principal stresses 1-3
and diagram n-n and defined with characteristic values modci = 29,22 MPa, aveGSI = 21, avemi = 7 has
been applied in software package SLIDE. Hoek-Brown failure criterion parameters are: m = 0,416646;
s = 0,00015412; a = 0,540887 (RockLab, www.rocscience.com). Shear strength parameters derived
from criterion are: c = 0,722 MPa; φ= 18,3˚.

Monte-Carlo analysis is based on predicted distribution of Hoek-Brown parameters for intact rock mi
and GSI. Software then automatically computes distributions for Hoek-Brown parameters mb, s i a.

3.2 Defining Mohr-Coulomb failure criterion parameters for cover

Input data for sedimentary cover are taken from laboratory testing on samples taken from boreholes.
Values of these parameters are: φ = 26°, c =13 kPa and  = 20 kN/m³.

524
Špago, Hadrović, Hajduk, Jazvin / DECGE 2018

4 RESULTS OVERVIEW OF SLOPE STABILITY ANALYSIS

4.1 Results overview for model with input parameters on bedrock

Overview of safety factors, for critical slide plane on each cross section (determined by using software
package SLIDE) for both load combinations is shown in Table 1. Analysis is based on the
Morgenstern-Price method of slices.

Table 1. Overview of safety factors per cross section

COVER FLYSCH
CROSS SECTION

LC I LC I LC II
Depth of Fs Depth of Fs Depth of Fs Depth of
sliding EC7- sliding EC7 - sliding EC7 - sliding
Fs Fs
plane approach plane approach plane approach plane
(m) 1 (m) 2 (m) 3 (m)
1 1,094 197 1,600 196 1,445 182 1,403 184 1,317
2 1,189 160 1,324 160 1,194 127 1,130 204 1,109
3 - 215 1,326 219 1,192 213 1,145 200 1,104
4 1,008 212 1,281 213 1,157 200 1,104 200 1,081
5 - 229 1,227 229 1,106 212 1,044 212 1,025

For flysch layers from Table 1 it can be observed that for characteristic values of parameters, obtained
safety factors in critical slide planes at very large depths are Fs>1. Factors of safety is increased from
cross section 5 to 1. Hence, it can be concluded that for characteristic parameter values which define
Hoek-Brown failure criterion, examined slope is stable with safety factors exceeding requested
(FS>1,00). This calculation is computed using maximum measured groundwater level. However, due
to great depths of sliding planes decrease of water table will not cause significant increase of safety
factors in all considered cross sections. Figure 2. shows critical sliding plane through flysch on cross
section 5 for LC II.

Figure 2. Critical sliding plane through flysch on cross section 5 for LC II.

525
Špago, Hadrović, Hajduk, Jazvin / DECGE 2018

4.2 Stability analysis of sedimentary cover

For cover layer, reverse analysis resulted in new parameters with safety factors FS≈1,0. These newly
obtained parameters are φ = 26°; c = 19 kPa;  = 20 kN/m³. Safety factor values are justifiable as
landsliding occurs in cover layers along entire area (Figure 3). Figure 4 shows possible local sliding
through cover layer as computed.

Figure 3. Local sliding at location “Kukovi” landslide.

Figure 4. Critical sliding plane through cover layer at cross section 4 for LC I.

Thus, it can be concluded that locations with thicker cover layer along slope may be subjected to local
sliding as parameter values could be lower than obtained from reverse analysis.

4.3 Results overview of “Sensitivity analysis” and Monte-Carlo analysis for rock

As deterministic method delivered the lowest values of safety factors for cross section 5, only these
will be addressed in following text. Sensitivity analysis plot is presented in Figure 5.

Figure 5. Sensitivity analysis for change of parameters ci, GSI, mi and rock unit weight.

526
Špago, Hadrović, Hajduk, Jazvin / DECGE 2018

As expected, slope stability is most sensitive to changes of uniaxial strength of rock, while it has quite
low sensitivity to variations of other parameters, namely rock unit weight and GSI (Figure 5). As
processing of test results gave large dissipation in uniaxial strength of rock (Figure 1) which vastly
affects results of slope stability analysis (Figure 5), it was decided to perform Monte-Carlo probability
analysis with adopted distributions for GSI, mi, i ci. Changes in the groundwater level have not been
considered in the sensitivity analysis, as deterministic method creates a sliding plane at a great depth
so that a slight variation of the groundwater level from the surface of the ground will have small
impact on the stability of the slope. Therefore, it is only looked at groundwater level obtained for the
maximum measured level. Also, the unit weight of the rock mass is not varied in the calculations as,
according to Figure 5, it does not affect the results of the analysis of the slope stability.

This analysis was conducted for 1000 values of ci in range from minci = 5 MPa to maxci = 138 MPa
for each of 4851 potential slope planes in cross section 5. Attained is overall sliding probability Pf of
4,2% (Figure 6) with corresponding reliability index RI=1,769. Minimum required slope reliability
index is RI=3,0 (Slide v6.0 User's manual, Probabilistic Analysis 2010) and, with respect to
importance of stability of observed slope from aspect of consequences arising from its instability, this
analysis has not provided sufficient reliability index.

Figure 6. Results of Monte-Carlo analysis at cross section 5.

Figure 7. The normal distribution of the probability of occurrence safety factor FS at cross section 5.

527
Špago, Hadrović, Hajduk, Jazvin / DECGE 2018

5 CONCLUSIONS

As deterministic methods of safety analysis of slope Kukovi give satisfactory coefficients of safety,
and by probabilistic methods insufficient reliability index, it can be deduced that slope is conditionally
stable. However, in layers of sedimentary cover, particularly in deforested parts of slope there is a
chance of local landsliding. Size of these landslides depends on cover layer thickness and slope
surface areas that are not covered with high vegetation.

High frequency of result dissipation, particularly in uniaxial strength tests of intact rock puts a request
on further studies to increase the number of tested samples to minimize scattering of results and
establish representation of rocks of different strength properties in flysch. Regarding groundwater
regime at great depths of sliding planes, in this calculation it is assumed that they are always
submerged, so that the measurement of groundwater levels in relatively shallow boreholes does not
carry any weight in assessment of groundwater status at great depths that potential sliding planes pass
through. However, due to complexity of problem, all uncertainties can be overcome with additional,
future investigation of Kukovi landslide, preceded by new Ground Investigation Programme which
would all together enable definite and more accurate assessment of the site.

ACKNOWLEDGEMENTS

Authors would like to thank to company “Elektroprivreda BiH d.d. Sarajevo“ that financally supported
conducted research within the project „Technical monitoring of the Kukovi lanslide“.

REFERENCES

Agustawijaya, D. S. (2007). The Uniaxial Compressive Strength of Soft Rock. Civil Engineering Dimension 9:
9–14.
Akram, M. and Bakar, M. Z. A. (2007). Correlation between Uniaxial Compressive Strength and Point Load
Index for Salt Range Rocks. Pak. J. Engg. & Appl. Sci. 1:
Chau, K. T. and Wong, R. H. C. (1996). Uniaxial compressive strength and point load strength of rock.
International Journal of Rock Mechanics and Mining Science & Geomechanics Abstract. 33:183-188.
EN 1998-5 Eurocode 8. Design of structures for earthquake resistance - Part 5: Foundations, retaining
structures and geotechnical aspects, 4.1.3.3, (5)P.
Hobbs, P. R. N., Hallam, J. R., Forster, A. Entwisle, D.C., Jones, L.D., Cripps, A.C., Northmore, K.J., Self S.J.
(2002). Engineering geology of British rocks and soils, Mudstones of the Mercia Mudstone Group.
Keyworth, Nottingham British Geological Survey.
Hoek, E., Marinos, P. and Benissi, M. (1998). Applicability of the geological strength index (GSI) classification
for very weak and sheared rock masses. The case of the Athens Schist Formation. Bull Eng Geol Environ 57:
151-160.
Hoek E. (2000). "ROCK ENGINEERING", Course notes by Evert Hoek. A.A. Balkema Publishers.
Marinos, P. and Hoek, E. (2001). Estimating the geotechnical properties of heterogeneous rock masses such as
flysch. Bull Eng Geol Environ 60: 85-92.
Rusnak, J., Mark, C. (2001). Using The Point Load Test To Determine The Uniaxial Compressive Strength Of
Coal Measure Rock. National Institute for Occupational Safety and Health Pittsburgh, PA.
Slide v6.0 User's manual (2010). Probabilistic Analysis. www.rocscience.com.
"Technical monitoring project of Kukovi landslide“; Institute for design, materials and structures of Faculty of
Civil Engineering at Dzemal Bijedic University of Mostar; 2010-2012 (bosnian language) .
"Technical monitoring project of Kukovi landslide“; Technical Institute Ltd. Mostar; 2015 (bosnian language)
www.rocscience.com.

528

You might also like