Geotechnical Hazard Assessment of Road Cuts Stabilisation

GEOTECHNICAL HAZARD ASSESSMENT OF ROAD CUTS
STABILITY IN MOUNTAINOUS AREAS IN SAUDI ARABIA
By
Hassan Mabrouk Basahel
February 2019
Department of Mining and Materials Engineering

McGill University
A thesis submitted to McGill University in partial fulfillment of the

requirements of the degree of Doctorate of Philosophy
ABSTRACT
The stability of rock slopes is crucial to public safety in highways passing through rock
cuts. Slope instability and failures occur due to many factors such as adverse slop geometries,
geological discontinuities, weak or weathered slope materials as well as severe weather conditions.
External loads like heavy precipitation and earthquakes could play a significant role in slope
failure.
In this thesis, several rock mass classification systems for rock slope stability assessment
were critically evaluated against known rock slope conditions in the south western region of Saudi
Arabia. Selected empirical methods were applied to 22 rock cuts, and the stability conditions were
identified. The results obtained from these methods were compared to each other and served to
highlight the limitations of each empirical classification system.
A probabilistic stability analysis approach is then developed. The stability condition of a

jointed rock slope in one of the selected rock cuts taken from the previously chosen 22 rock cuts
was assessed using deterministic and probabilistic approaches, under both dry and wet conditions.
The stability analysis was carried out using FLAC3D to generate a 3-dimensional model, and
bedding is simulated with a ubiquitous joint model, to determine the influence of the dominant,
unfavorable bedding orientation with respect to the slope face. The deterministic analysis was
implemented first using the mean values of the selected random variables, namely the dip, dip
direction and friction angle of the dominant discontinuity set, and the stability condition was
assessed. A Box-Behnken design (BBD) approach is then adopted to create the surface response
function, as a second order polynomial for the factor of safety. To do so, fifteen FLAC3D models
were generated in accordance with BBD. Based on this, 10,000 simulations of different slope
realizations were carried out using Monte-Carlo simulation technique, and the probability of
unsatisfactory of performance of the rock slope was assessed. It is shown that the probabilistic
approach provides more insight and confidence in the stability condition of the rock slope, under
both dry and steady state heavy rainfall conditions.
A new factor, herein defined as the failure intensity factor is introduced to describe the
severity of slope failure with respect to the roadway width, to distinguish between rock slopes
having the same probability of failure, yet not the same magnitude of failed material. A maximum
i
shear strain criterion is adopted to delineate the volume of sliding material and its effect on the
roadway, and 3 levels of the intensity factors – low, moderate and high – are defined.
Finally, the intensity factor is used to derive the geotechnical hazard index matrix. A 5-
level geotechnical HI matrix is proposed, based on reported recommended levels of low, moderate
and high probabilities of failure. The proposed Hazard Index matrix considers both the probability
of unsatisfactory of performance of the rock slope and the severity of the slope failure expressed
as the intensity factor.
ii
RÉSUMÉ
La stabilité des pentes rocheuses est essentielle à la sécurité publique dans les autoroutes
flanquées de rochers. L’instabilité de la pente et les effondrements de talus sont attribuables à de
nombreux facteurs comme la géométrie défavorable des pentes, les discontinuités géologiques, les
matériaux de talus mous ou patinés et les conditions météorologiques extrêmes. Les pressions
externes comme les fortes précipitations et les tremblements de terre pourraient jouer un rôle
important dans les effondrements de talus.
Dans cette thèse, plusieurs systèmes de classification de masse rocheuse permettant

d’évaluer la stabilité des pentes rocheuses ont fait l’objet d’un examen critique et ont été comparés
aux pentes rocheuses connues dans le sud-ouest de l’Arabie saoudite. On a appliqué certaines
méthodes empiriques à 22 tranchées rocheuses et identifié les conditions de stabilité. Les résultats
obtenus à partir de ces méthodes ont été comparés les uns aux autres et ont permis de souligner les
limites de chaque système de classification empirique.
Une analyse probabiliste de la stabilité est ensuite mise au point. L’état de stabilité d’un
talus rocheux diaclasé dans l’une des tranchées rocheuses sélectionnée à partir des 22 tranchées
rocheuses choisies précédemment a été évalué à l’aide d’approches déterministe et probabiliste,
dans des conditions sèches et humides. L’analyse de la stabilité a été effectuée à l’aide du logiciel
FLAC3D pour produire un modèle en 3 dimensions, et l’assise est simulée à l’aide d’un modèle à
joint systématique, afin de déterminer l’influence de l’orientation dominante et défavorable de
l’assise sur la face de la pente. L’analyse déterministe a d’abord été effectuée à l’aide des valeurs
moyennes des variables aléatoires choisies, à savoir l’inclinaison, la direction de l’inclinaison et
l’angle de friction de l’ensemble de discontinuité dominant, et l’état de stabilité a été évalué. Une
approche de conception « Box-Behnken » (BBD) est ensuite adoptée pour créer la fonction de
réponse de surface, en tant que polynôme de deuxième degré pour le facteur de sécurité. Quinze
modèles FLAC3D ont donc été générés conformément à l’approche BBD. À partir de ces modèles,
on a effectué 10 000 simulations de différentes réalisations de pente grâce à la technique de
simulation Monte-Carlo, et on a évalué la probabilité de rendement insatisfaisant de la pente
rocheuse. Il est démontré que l’approche probabiliste procure une compréhension accrue et une
plus grande confiance dans l’état de stabilité de la pente rocheuse, dans des conditions sèches et
sous de fortes et constantes précipitations.
iii
On introduit un nouveau facteur, défini dans le présent document comme le facteur
d’intensité de l’effondrement, afin de décrire la gravité de l’effondrement du talus par rapport à la
largeur de la chaussée, et distinguer les pentes rocheuses ayant la même probabilité
d’effondrement, mais pas la même magnitude de matériaux défaillants. Un critère de cisaillement
maximum est adopté pour délimiter le volume de matériau de glissement et ses effets sur la
chaussée, et 3 niveaux de facteurs d’intensité – faibles, modérés et élevés – sont définis.
Enfin, le facteur d’intensité est utilisé pour calculer la matrice des indices de risque
géotechnique. On propose une matrice d’indices de risque géotechnique à 5 niveaux en fonction
des niveaux recommandés de probabilités d'effondrement faibles, modérées et élevées. La matrice
des indices de risque proposée tient compte à la fois de la probabilité de rendement insatisfaisant
de la pente rocheuse et de la gravité de l’effondrement du talus exprimée en facteur d’intensité.
iv
DEDICATION
This work is dedicated to …
My mother
My father
My wife
My kids
My brothers and sisters
May Allah bless them all.
v
ACKNOWLEDGEMENTS
All praise is due to Allah Almighty for giving me the guidance and the strength to complete
this work.
I would like to express my deep gratitude and appreciation to Prof. Hani Mitri for his close
supervision and continuous encouragement throughout the duration of my Ph.D. Special thanks to
him for all the valuable tips and the wisdom guidance, all of which have added a lot to my
personality and tremendously enriched my knowledge in the field of scientific research.
I would also like to thank all my colleagues in the MDL in McGill University for their
encouragement and motivation during my study, especially Dr. Shahe Shnorhokian for his help
and valuable suggestions.
I would also like to extend my thanks and appreciation to Saudi Cultural Bureau in Canada
and Saudi Geological Survey (SGS) in Saudi Arabia for their financial and technical support. I am
also grateful to Eng. Hussain Al Otaibi (The President of SGS), Dr. Zohair Nowab (previous
president of SGS), Dr. Abdullah Al Attas (previous Asst. president of SGS) and Eng. Saleh Al
Sefry (Asst. president of SGS) for their encouragement and continues support. In addition, many
thanks go to all my colleagues in the Geological Hazards department in SGS due to their unlimited
support during my work in the PhD. Moreover, special thanks to Dr. Ahmed Youssif, Ali Bulkhi,
Basem Daheri and Alaa Bahamel for their help in the field work.
Also, I present my thanks to Mr. Mazen Abu Abdullah, from the department of Applied
Geology in the SGS, for his help with the maps, and Mohammed Al Hussein, from Engineering
Geology Lab, for his help in lab tests.
I am grateful to all my family. To my mother and father for their absolute love and infinite
support since my childhood until this stage of my life. I am proud of both of you. To my wife for
her love, friendship and faith in me. To my kids for their absolute innocence and warm hugs every
day. To my sisters and brothers for their support. To my mother and brother in law for their
kindness and encouragement.
I could always not go without thanking my friend Ghassan Al Sulimani for his sincere
friendship and continuous support.
vi
CONTRIBUTIONS OF THE AUTHOR
Three papers have been published during the PhD work jointly with the thesis supervisor. They
include one refereed conference paper and two journal papers. A fourth paper is currently in
preparation for submission to the Journal of Engineering Geology. The list of publications is as
follows.
1. Basahel, H. and Mitri, H.S. (2016) Validation of empirical rock mass classification systems
for rock slopes. In Proceedings of 3rd International Symposium on Mine Safety Science
and Engineering, pp. 127-132.
2. Basahel, H. and Mitri, H.S. (2017) Application of rock mass classification systems to rock
slope stability assessment: A case study. Journal of Rock Mechanics and Geotechnical
Engineering, 9(6), pp. 993-1009.
3. Basahel, H. and Mitri, H.S. (2019) Probabilistic assessment of rock slopes stability using
the response surface approach–A case study. International Journal of Mining Science and
Technology (In press)
4. Basahel, H. and Mitri, H.S. (2019) Incorporation of rock slope failure severity into the
geotechnical hazard assessment. In preparation for submission to Engineering Geology.
In the above, manuscripts 2 and 3 are reported in Chapters 3 and 4, respectively of the
thesis. Chapter 5 constitutes the material of the paper in preparation for Engineering Geology
(paper No. 4 above).
vii
TABLE OF CONTENTS
ABSTRACT……………………………………………………………………………………… i
RESUME ……………………………………………………………………………………….. iii
DEDICATION ………………………………………………………………………………….. v
ACKNOWLEDGMENTS ……………………………………………………………………… vi
CONTRIBUTION OF AUTHORS …………………………………………………………….. vii
TABLE OF CONTENTS ……………………………………………………………………… viii
LIST OF TABLES ……………………………………………………………………………… xii
LIST OF FIGURES ……………………………………………………………………………. xiv
LIST OF ABBREVIATIONS AND SYMBOLS …………………………………………….. xviii
CHAPTER 1: INTRODUCTION ……………………………………………………………….. 1
1.1 General ………………………………………………………………………………………. 1
1.2 Study Problem ……………………………………………………………………………….. 4
1.3 Scope of Work ……………………………………………………………………………….. 4
1.4 Objectives of the Study ………………………………………………………………………. 6
1.5 Thesis Outline ………………………………………………………………………………... 7
CHAPTER 2: LITERATURE REVIEW ………………………………………………………… 9
2.1 Introduction ………………………………………………………………………………….. 9
2.2 Uncertainty in the Stability Analysis of the Slope …………………………………………. 11
2.3 Deterministic and Probabilistic Approaches ……………………………………………….. 12
2.3.1 Deterministic Approach …………………………………………………………………... 12
2.3.2 Probabilistic Approach ……………………………………………………………………. 13
2.4 Probabilistic methods in Geotechnical Engineering ………………………………………... 13
2.4.1 First Order Reliability Method (FORM) …………………………………………………. 13
2.4.1.1 First Order Second Moment (FOSM) …………………………………………………… 15
2.4.2 Monte-Carlo Simulation (MCS) ………………………………………………………….. 17
2.4.3 Point Estimate Method (PEM) ……………………………………………………………. 17
viii
2.4.4 Design of Experiments ……………………………………………………………………. 21
2.4.4.1 Factorial Design ………………………………………………………………………… 21
2.4.4.1.1 Two Levels Factorial Design ………………………………………………………… 22
2.4.4.1.2 Three Levels Factorial Design ……………………………………………………….. 25
2.4.5 Response Surface Designs ……………………………………………………………….. 26
2.4.5.1 Central Composite Design (CCD) ……………………………………………………... 27
2.4.5.2 Box-Behnken Design (BBD) …………………………………………………………... 29
2.5 Conventional methods in Geotechnical Engineering ……………………………………… 30
2.6 Numerical methods in Geotechnical Engineering …………………………………………. 31
2.7 Empirical Methods …………………………………………………………………………. 31
CHAPTER 3: APPLICATION OF ROCK MASS CLASSIFICATION SYSTEMS TO ROCK
SLOPE STABILITY ASSESSMENT– A CASE STUDY …………………………………….. 33
3.1 Abstract …………………………………………………………………………………….. 33
3.2 Introduction ………………………………………………………………………………… 34
3.3 Location of the Study Area ………………………………………………………………… 35
3.4 Classification Systems for Rock Slopes ……………………………………………………. 39
3.4.1 Rock Mass Rating (RMR) ………………………………………………………………… 39
3.4.2 Slope Mass Rating (SMR) ………………………………………………………………… 40
3.4.3 Chinese Slope Mass Rating (CSMR) ……………………………………………………... 44
3.4.4 Continuous Slope Mass Rating …………………………………………………………… 45
3.4.5 Graphical Slope Mass Rating …………………………………………………………….. 46
3.4.6 Hazard Index (HI) ………………………………………………………………………… 46
3.5 Case Study ………………………………………………………………………………….. 47
3.6 Results ……………………………………………………………………………………… 48
3.6.1 Rock Mass Rating System ………………………………………………………………… 48
3.6.2 Kinematic Analysis ………………………………………………………………………. 52
3.6.3 Rock Slope Stability Assessment ………………………………………………………… 57
3.6.3.1 Structurally Controlled Failure …………………………………………………………. 58
ix
3.6.3.1.1 Planar-Type Failure Mode ……………………………………………………………. 58
3.6.3.1.2 Wedge-Type Failure Mode …………………………………………………………… 59
3.6.3.1.3 Toppling-Type Failure Mode ………………………………………………………… 63
3.6.3.2 Non-Structurally Controlled Failure …………………………………………………… 65
3.7 Discussion ………………………………………………………………………………….. 65
3.8 Conclusion …………………………………………………………………………………. 69
CHAPTER 4: PROBABILISTIC ASSESSMENT OF ROCK SLOPES STABILITY USING THE
RESPONSE SURFACE APPROACH-A CASE STUDY …………………………………….. 71
4.1 Abstract …………………………………………………………………………………….. 71
4.2 Introduction ………………………………………………………………………………… 72
4.3 Location and Geological Sitting of the Study Area ………………………………………... 75
4.4 Methodology ……………………………………………………………………………….. 77
4.5 Field Measurements ………………………………………………………………………… 77
4.6 Random Variables …………………………………………………………………………. 79
4.7 Point Estimate Method …………………………………………………………………….. 79
4.8 Box-Behnken Design (BBD) ………………………………………………………………. 81
4.9 Monte-Carlo Simulation Technique (MCS) ……………………………………………….. 84
4.10 Shear Strength Reduction (SSR) …………………………………………………………. 85
4.11 Analysis of Variance (ANOVA) …………………………………………………………. 86
4.12 Model Setup ………………………………………………………………………………. 88
4.13 Results and Discussion ………………………………………………………………….... 93
4.13.1 Kinematic Analysis …………………………………………………………………….. 93
4.13.2 Point Estimates …………………………………………………………………………. 95
4.13.3 Numerical Modelling …………………………………………………………………… 97
4.13.4 Box-Behnken Design ………………………………………………………………….. 104
4.13.5 Probability of Unsatisfactory Performance ……………………………………………. 111
4.14 Conclusion ……………………………………………………………………………….. 111
CHAPTER 5: METHODOLOGY FOR THE DEVELOPMENT OF SLOPE FAILURE
INTENSITY FACTOR AND GEOTECHNICAL HAZARD INDEX ………………………. 114
x
5.1 Abstract …………………………………………………………………………………… 114
5.2 Introduction ……………………………………………………………………………….. 115
5.3 Methodology ………………………………………………………………………………. 117
5.4 Intensity Factor Concept …………………………………………………………………... 117
5.5 Geotechnical Hazard Index (GHI) ………………………………………………………… 120
5.6 Model Setup ……………………………………………………………………………….. 122
5.7 Results and Discussion ……………………………………………………………………. 125
5.7.1 Box-Behnken Design (BBD) for Model 2 ………………………………………………. 125
5.7.2 Probability of Unsatisfactory Performance ……………………………………………... 130
5.7.3 Intensity Factor ………………………………………………………………………….. 131
5.7.3.1 Method 1. Calculate Area of Potential Failure Zone ………………………………….. 131
5.7.3.2 Method 2. Calculate Length of Potential Failure Zone ……………………………….. 133
5.7.4 Geotechnical Hazard Index ……………………………………………………………… 135
5.8 Conclusion ………………………………………………………………………………… 136
CHAPTER 6: CONCLUSION …………………………………………………………………138
6.1 Research conclusions ……………………………………………………………………… 138
6.2 Recommendation for Future Work ………………………………………………………… 140
6.3 Statement of Contribution …………………………………………………………………. 141
REFERENCES ………………………………………………………………………………... 142
xi
LIST OF TABLES
Table 2.1: Basic design table for a 23 factorial design (Montgomery, 2013) …………………. 24
Table 3.1: Rock mass rating system (After Bieniawski, 1989) ……………………………….. 41
Table 3.2: RMR Rating Adjustment for Discontinuity Orientations (Bieniawski, 1989) …….. 42
Table 3.3: Rock Mass Classes determined from total rating (Bieniawski, 1989) ……………....42
Table 3.4: Adjustment ratings for F1, F2, and F3 (Romana 1985, modified by Anbalangan et al.,
1992) ……………………………………………………………………………………………43
Table 3.5: Adjustment factor F4 for the method of excavation (Romana, 1985) ………………43
Table 3.6: Classes of SMR according to Romana (1985) ………………………………………44
Table 3.7: Rating criteria and factors for the drainage sub-factors 𝑓𝐷,𝑛 (Pantelidis, 2010) …….49
Table 3.8: Summary of the basic features of selected case studies ……………………………. 50
Table 3.9: Kinematic analysis results for structurally controlled failure mechanism …………. 57
Table 3.10: Stability analysis results for planar failure using rock slope classification
methods………………………………………………………………………………………… 60
Table 3.11: Stability analysis results for wedge failure using rock slope classification
methods………………………………………………………………………………………… 62
Table 3.12: Stability analysis results for toppling failure using rock slope classification
methods………………………………………………………………………………………… 64
Table 3.13: Stability analysis results for toppling failure using rock slope classification
methods………………………………………………………………………………………… 67
Table 4.1: Data of dip, dip direction and friction angle for joint set # 1 ………………………. 81
Table 4.2: A typical Box-Behnken Design for three random variables ………………………. 83
Table 4.3: Rock mass and joint surface properties used in the numerical analysis ……………. 91
Table 4.4: Groundwater properties for steady state fluid-flow analysis in FLAC3D analysis … 93
Table 4.5: The best fitted distributions and their corresponding statistical moments for each set of
random variables ……………………………………………………………………………….. 96
Table 4.6: The calculated point estimates for each set of variables …………………………… 97
Table 4.7: shows results of the factor of safety of dry and saturated conditions for all treatments of
slope realizations ……………………………………………………………………………… 98
xii
Table 4.8: Results of ANOVA for the rock slope in dry condition …………………………… 106
Table 4.9: Results of ANOVA for the rock slope in saturated condition ……………………… 106
Table 4.10: Simulated and predicted SRFs for slope realization in dry condition ……………. 107
Table 4.11: Simulated and predicted SRFs for slope realizations in saturated condition …….. 107
Table 5.1: Proposed geotechnical hazard description matrix …………………………………. 122
Table 5.2: Shows results of the factor of safety for all treatments of slope realizations for model 2
…………………………………………………………………………………………………. 126
Table 5.3: Results of ANOVA for the rock slope in dry condition for model 2 ……………… 126
Table 5.4: Summary of 𝑃𝐹 and 𝐼𝐹 results and corresponding scores for model 1 and model 2
…………………………………………………………………………………………………. 136
xiii
LIST OF FIGURES
Figure 1.1: Examples of (a) wedge, (b) planar, and (c) toppling failures along road cuts (Maerz
and Kim, 2000) …………………………………………………………………………………... 2
Figure 1.2: Examples of (a) raveling, (b) undercutting, and (c) rolling failures along road cuts
(Maerz and Kim, 2000) …………………………………………………………………………... 3
Figure 1.3: Location of the study area ………………………………………………………….... 5
Figure 1.4: Rock cut face with steep slope angle at curved corner of the road ………………….. 6
Figure 2.1: Conventional procedure for slope stability analysis ………………………………... 10
Figure 2.2: Uncertainty sources in a rock slope stability analysis (Baecher & Christian, 2003) ..12
Figure 2.3: Limit state concept (Haldar and Mahadevan, 2000) ………………………………... 14
Figure 2.4: Probability distribution of performance function (M) ……………………………… 16
Figure 2.5: Rosenblueth’s pints and weights for two variables, correlated or uncorrelated (Baecher
and Christian, 2003) ……………………………………………………………………………. 18
Figure 2.6: Rosenblueth’s points and weights for three variables, correlated or uncorrelated (after:
Rosenblueth, 1975; Christian & Baecher, 1999) ……………………………………………….. 20
Figure 2.7: Geometric view of the 23 factorial design (Montgomery, 2013) ………………….. 24
Figure 2.8: Geometric view of the 33 factorial design (Montgomery, 2013) ………………….. 25
Figure 2.9: Diagram of rotatable CCD generation for two factors; factorial points (blue) + center
point (green) and axial points (red) …………………………………………………………….. 28
Figure 2.10: Diagram of face-center CCD generation for two factors; factorial points (blue) +
center point (green) and axial points (red) ……………………………………………………… 29
Figure 3.1: Annual precipitation map of Saudi Arabian from 1960 to 2014, showing a maximum
average of 500 mm ……………………………………………………………………………… 36
Figure 3.2 (a): Locations of road cuts in the study area ………………………………………… 37
Figure 3.2 (b): A three-dimensional (3D) topographic map showing the roadway through rugged
terrains ………………………………………………………………………………………….. 38
Figure 3.3: Example of structurally controlled failure (site#22 in Figure 3.1) …………………. 38
Figure 3.4: Example of non-structurally controlled failure due to highly weathering (site#21 in
Figure 3.1) ……………………………………………………………………………………… 39
Figure 3.5: Methodology used in the present study …………………………………………….. 51
xiv
Figure 3.6: Discrete and continuous RMR results for jointed rock cuts ……………………….. 52
Figure 3.7: (a) Planar failure at site 4, and (b) Stereographic projection showing the potential for
planar failure ……………………………………………………………………………………. 53
Figure 3.8: (a) Wedge failure at site 19, and (b) Stereographic projection showing the potential for
wedge failure …………………………………………………………………………………… 54
Figure 3.9: (a) Potential for flexural toppling failure at site 5, and (b) Stereographic projection
showing for flexural toppling failure …………………………………………………………… 55
Figure 3.10: (a) Potential for block toppling failure at site 5, and (b) Stereographic projection
showing for block toppling failure ……………………………………………………………… 56
Figure 3.11: Results of original SMR, discrete RMR, and strike difference of slope-joint for planar
failure …………………………………………………………………………………………… 58
Figure 3.12: Results of original SMR, discrete RMR, and the strike difference of slope-joint for
wedge failure …………………………………………………………………………………… 61
Figure 4.1: Location map ………………………………………………………………………. 76
Figure 4.2: Geological map of the study area ………………………………………………….. 76
Figure 4.3: Case study …………..……………………………………………………………… 78
Figure 4.4: Stereographic projection of structural data for discontinuity sets of the case study
…………………………………………………………………………………………………... 80
Figure 4.5: Geometric representation of a three-variable Box-Behnken design ………………. 82
Figure 4.6: Initial model of the rock slope before excavation …………………………………. 89
Figure 4.7: The rock slope model after excavation ……………………………………………. 90
Figure 4.8: Mesh size corresponding to the conversions of displacement …………………….. 92
Figure 4.9: Ground water flow model …………………………………………………………. 94
Figure 4.10: Sectional view shows the pore water pressure distribution within the rock slope
………………………………………………………………………………………………….. 94
Figure 4.11: Histogram and exponential fit distribution for dip values ……………………….. 95
Figure 4.12: Histogram and triangle fit distribution for dip direction values …………………. 96
Figure 4.13: Histogram and normal fit distribution for friction angle values …………………. 97
Figure 4.14: Result of displacement in the reference model before an excavation ……………. 99
Figure 4.15: Shows displacement contours and simulated factor of safety for the reference model
in dry condition ………………………………………………………………………………… 99
xv
in saturated condition …………………………………………………………………………. 100
Figure 4.17: Factor of safety for slope realization # 4 in dry condition ………………………. 100
Figure 4.18: Factor of safety for slope realization # 5 in dry condition ………………………. 101
Figure 4.19: Factor of safety for slope realization # 11 in dry condition ……………………... 103
Figure 4.20: Factor of safety for slope realization # 11 in saturated condition ……………….. 103
Figure 4.21: Plot of simulated by predicted SRFs for slope realizations in dry condition ……. 105
Figure 4.22: Plot of simulated by predicted SRFs for slope realizations in saturated condition
…………………………………………………………………………………………………. 105
Figure 4.23 (a): Response surface plot for dip angle and dip direction against SRF values in the
case of dry condition …………………………………………………………………………... 109
Figure 4.23 (b): Response surface plot for dip angle and friction angle against SRF values in the
case of dry condition …………………………………………………………………………... 109
Figure 4.23 (c): Response surface plot for dip direction and friction angle against SRF values in
the case of dry condition ………………………………………………………………………. 110
Figure 4.24: Response surface plot for dip angle and dip direction against SRF values in the case
of saturated condition ………………………………………………………………………….. 110
Figure 4.25: Histograms and distribution fit for the calculated factor of safety for the slope in the
dry condition …………………………………………………………………………………... 112
saturated condition …………………………………………………………………………….. 113
Figure 5.1: Example for rock slope failure causing obstruction to roadway (High intensity) … 119
Figure 5.2: Example for rock slope failure (Low intensity) …………………………………… 119
Figure 5.3: Sketch view of the road width below a rock slope ……………………………….. 121
Figure 5.4: Initial model 2 for the rock slope before excavation ……………………………... 124
Figure 5.5: Model 2 after excavation …………………………………………………………. 124
Figure 5.6: Plot of simulated by predicted SRFs for slope realizations in dry condition for model
2 ……………………………………………………………………………………………….. 127
Figure 5.7: Response surface plot for dip angle against SRF values in the case of dry condition
for model 2 …………………………………………………………………………………….. 128
Figure 5.8: Response surface plot for dip angle dip direction against SRF values in the case of dry
condition for model 2 ………………………………………………………………………….. 129
xvi
Figure 5.9: Response surface plot for dip angle and friction angle against SRF values in the case
of dry condition for model 2 …………………………………………………………………... 129
Figure 5.10: Response surface plot for dip direction and friction angle against SRF values in the
case of dry condition for model 2 ……………………………………………………………… 130
Figure 5.11: Histograms and distribution fit for the calculated factor of safety for the model 2
…………………………………………………………………………………………………. 131
Figure 5.12: Calculate the critical area (in red) using method 1 for model 1 …………………. 132
xvii
LIST OF ABBREVIATIONS AND SYMBOLS
LE: Limit Equilibrium

SRF: Shear Reduction Factor
𝑃𝐹: Probability of Unsatisfactory Performance
𝐼𝐹: Intensity Factor
FORM: First Order Reliability Method
PEM: Point Estimate Method
MCS: Monte Carlo Simulation
FOSM: First Order Second Moment
RSM: Response Surface Method
CCD: Central Composite Design
BBD: Box Behnken Design
FDM: Finite Difference Method
FEM: Finite Element Method
BEM: Boundary Element Method
RQD: Rock Quality Designation
RMR: Rock Mass Rating
SMR: Slope Mass Rating
HI: Hazard Index
GSI: Geological Strength Index
SSR: Shear Strength Reduction
FOS: Factor of safety
UCS: Unconfined Compressive Strength
DRMR: Discrete Rock Mass Rating
CRMR: Continuous Rock Mass Rating
ANOVA: Analysis of Variance
𝐶𝑆𝑀𝑅: Chinese Slope Mass Rating
xviii
𝑅𝑀𝑅𝑏𝑎𝑠𝑖𝑐 : Basic Rock Mass Rating
β: Reliability index
𝑋𝑖 : Variable
𝑍: Performance function
𝐺: Gravity
𝑃𝑓 : Probability of failure
M: Margin of safety
R: Resistance
Q: Loads
𝜇𝑀 : Mean value of margin of safety
𝜇𝑅 : Mean value of resistance
𝜇𝑄 : Mean of the load
𝜎𝑀 = the standard deviation of margin of safety
𝜎𝑅 2 : Variances of the resistance
𝜎𝑄 2: Variance of the load
𝜌𝑅𝑄 = Coefficient of correlation between both (R) and (Q) variables
𝜎𝑅 : Standard Deviations of the resistance
𝜎𝑄 : Standard Deviations of the load
Y: Response function
𝜇𝑥 : Mean of the variable
𝑣𝑥 : Skewness
𝜎𝑥 : Standard variation
𝜌𝑥1𝑥2 : Correlation coefficient between two variables
𝐶𝑂𝑉𝑥1𝑥2 : Covariance of two variables
𝐴𝑐𝑜𝑑𝑒𝑑(±) : Coded variable, low (-) and high (+)

I: Identity column
xix
𝛽𝑖𝑗 : Regression coefficient
∈: Random error
𝜂: Response surface
η0: Number of replications
2k: Axial Points in central composite design
k: Number of independent variables
𝐾: Number of factors in the design
𝐹: Number of factorial points

F1: An adjustment factor, which depends on parallelism between joints and slope face strikes
F2: An adjustment factor that refers to joint dip angle
F3: An adjustment factor that reflects the relationship between the slope face dip and joint dip
F4: An adjustment factor that depends on the method of excavation
ζ : Slope Height Factor
λ: Discontinuity Factor
αj: Joint direction

αs: Slope face direction
αi: Trend of the line of intersection of two joints
βj: Joint dip
βi: Plunge of the line of intersection of two joints
P: Plane failure
W: Wedge failure
T: Toppling failure
𝐴: Parallelism between discontinuities and slope dip direction
𝐵: Discontinuity dip
C: Dip relationship between slope face and joint
ψ: Two correction factors (F1 and F2) are grouped in one term
𝑓𝑁𝐶 : Normal condition
xx
𝑓𝑇𝑀 : Triggering mechanism
𝑓1 : factor of safety
𝑓2 : relative orientation of the dominant failure surface plane with respect to the cutting face
𝑓3 : Refers to the geological strength index (GSI) used for non-structurally controlled and
differential weathering (undercutting) failure modes
𝑓4 : Which is the differential weathering and refers to the volume of the exposed rock mass per one-
meter length of slope
Im: Mean annual precipitation height
Icr: Critical annual precipitation
𝑓𝐷 : Drainage Factor
∅𝑟 : Residual friction angle
∅𝑏 : Basic friction angle
𝑟 : Rebound on weathered rock surface
𝑅 : Rebound on un-weathered rock surface
𝐶0 : Number of center points
𝑁: Number of experiments
𝐶𝑗 : Cohesion for the joint surface
∅𝑗 : Friction angle for the joint surface
𝐹: Factor of safety
𝑦𝑖𝑗 : the 𝑖𝑗 𝑡ℎ observation
𝑎 and 𝑏: the treatments
𝜇 : The overall mean
𝜏𝑖 : The effect of the 𝑖 𝑡ℎ parameter level
𝜖𝑖𝑗 : Random error
𝐻0 : Null hypothesis
𝐻1 : Alternative hypothesis
𝑆𝑆: Sum of squares
xxi
𝜎 0 : In-situ stress
𝜎𝑣0 : Vertical stress
𝜎𝑥0 , 𝜎𝑦0 : horizontal stresses
𝛾: Rock density
𝑘ℎ : Hydraulic conductivity
𝑔 : Gravitational acceleration
𝑒 : Joint aperture
𝑣 : Coefficient of kinematic viscosity of the fluid
𝑘: Permeability coefficient
𝜌𝑓 : Fluid mass density
𝐺𝐻: Geotechnical hazard
xxii
CHAPTER 1
INTRODUCTION
1.1 General
During the last few decades, the Kingdom of Saudi Arabia has allocated significantly large
budgets for the development of highway networks to connect suburbs and villages to main cities
across the country. Often enough, road construction projects have been faced with challenges
and difficulties especially in areas characterized by rugged terrains and mountainous topography,
where it is important to determine safe rock slope designs along the road and assess the best rock
cutting method. This has often led to design choices involving curved rock slopes having
variable dip angles, thus requiring special considerations to avoid safety issues related to
instability and potential of failure of rock cuts.
Various forms of rock slope failures are observed in the mountainous areas including rock,
earth, or debris flows on slopes by gravity. Such failures can occur either in an abrupt manner or
slowly with gradual slides of earth layers. Slope failures are known to be triggered primarily by
heavy rainfall, although in general, rock slope failures can also be attributed to earthquakes and
other natural causes, or to man-made causes, such as adverse slope geometry, or poor terrain
cutting and filling methods. Because the factors affecting rock slope failures can be natural or
man-made, they can occur in developed or undeveloped areas, or any area where the terrain is
altered for roads, houses, utilities, and buildings.
In general, there are three main factors that control the type and rate of slope failure. They
are:
1. Slope gradient: The steeper the slope the more likely that failure will occur.
2. Slope material: Sediments and fractured or poorly cemented rocks and sediments are
weak and more prone to fail under little or no additional loading
3. Water: When the slope materials are saturated with water, they may lose cohesion and
flow easily.
The welfare and safety of highway motorists requires that road rock cuts be made safe as
practicable. Catastrophic failures of rock cuts can result in property damage, injury, and even
1
fatalities. Highways impeded by even small spills of rock material present enormous safety
hazards.
Ensuring the stability of highway rock cuts, whether new or old, requires a thorough
evaluation of the structure of the rock. Failures usually initiate and follow pre-existing
discontinuities rather than break through intact rock. The mapping of discontinuity orientations is
a requirement, before or after the cut has been exposed (Piteau, 1979a; Piteau, 1979b; Piteau
1979g). Rock determined to be loose with the potential for failure must be removed or restrained
in some way (Piteau, 1979e). Prescribed designs for remediation and/or mitigation are easy to
find (Brawner, 1994; Konya and Walter, 1991; Piteau, 1979f; Franklin and Senior, 1997a).
In many terrains, discontinuities are oriented in such a way that they combine to create
wedge, planar, or toppling failure (Figure 1.1). Simple block and wedge pull-out failures are easy
to analyze with limit equilibrium analyses and numerical modelling (Hoek and Bray, 1981;
Piteau, 1979c; Piteau, 1979d).
In terrain underlain by flat lying sedimentary rock with vertical orthogonal jointing, planar
and wedge slides are usually absent, the predominant failure mechanism being raveling (Figure
1.2). Raveling can be difficult to recognize ahead-of-the-fact because it involves time-dependent
processes (Franklin and Senior, 1987b).
Figure 1.1: Examples of (a) wedge, (b) planar, and (c) toppling failures along road cuts (Maerz
and Kim, 2000)
2
Figure 1.2: Examples of (a) raveling, (b) undercutting, and (c) rolling failures along road cuts
(Maerz and Kim, 2000)
Rock slope stability analysis requires good knowledge of physical and mechanical
properties of the rock mass conditions such as the state of weathering and decomposition of the
rock mass, strength parameters especially for the discontinuities surfaces and the geometric
parameters of the rock slope (angles and directions of slope face and joints) in case of
structurally controlled failures (planar, wedge, and toppling) and strength condition of the rock if
the failure mechanism is not controlled by the discontinuities (non-structurally controlled). For
example, in case of raveling, undercutting, and high weathering rock mass conditions, the failure
mechanisms are more complex and difficult to analyze by using conventional methods such as
limit equilibrium (Franklin and Senior, 1997b). Furthermore, another aspect that should be
considered when analyzing the rock slope stability is the identification of triggering factors. This
is a significant task that helps understand the main cause or loading condition leading to failure.
Triggering factors could be rainfall (pore water pressure), earthquake (seismic loading), or man-
made (e.g. steep cut or poor blasting).
Rock slope stability assessment may encounter some difficulties and complications when
the stability problems are found in the rock cuts of mountainous areas which are often
characterized by variable slope angles and curvatures, as is the case in west and south western
regions in Saudi Arabia. In this area, most road-cut failures occur during or after rainstorms and
which are often severe and dense in specific periods during the year.
3
The main goal of this research is to develop a Hazard Index based on detailed rock slope
stability assessment that is suitable for the climatic and environmental conditions of the south
western part of Saudi Arabia. The area is known for the unfavorable conditions of heavy rainfall.
This will be treated as the primary triggering factor along with the properties of the rock slope
rock-soil mass structure.
1.2 Study Problem
The long-term stability of highway road cuts is of crucial importance to public safety. Road-
cut failures are known to cause not only traffic disruptions, but they can also be responsible for
accidents and fatalities. In Saudi Arabia, most of the rock cuts in highways and mountainous
roads, in the west and south western parts, are exposed to the hazard of slope failure due to heavy
rainfall conditions.
Most previous studies done in this area used simple, limit equilibrium, deterministic analysis
to calculate the safety factor that reflects the stability condition. Also, previous studies adopted
qualitative empirical methods to assess the severity level of potential slope failure without taking
into consideration the effect of rainfall on the stability of the rock slope.
Thus, the determination of the stability condition of highway rock slopes made up of variable
rock mass structures and subjected to heavy rainfall conditions is of crucial importance. This will
be the main focus of this research.
1.3 Scope of Work
The scope of this research is to analyze the stability condition of the rock cuts under the
effect of heavy rainfall in the mountainous roads in the south-west of Kingdom of Saudi Arabia.
Also, the estimation of the most significant factors in the rock slope whether, geometric or
strength factors, which could play an important role in instability of the slope, is of primary
concern. Finally, a new, geotechnical hazard index, that can be used in both systematic design
and analysis of rock slopes along with road cuts in Saudi Arabia will be developed.
The study area is located in south-western part of the Kingdom of Saudi Arabia (Figure 1.3).
The rock cuts in this area are in most cases prone to slope failures especially during the rainfall
season. The roads are characterized by being 2-way, relatively narrow (≈ 10 m in total) with 1 m
4
shoulders in both sides of the road, and are often curved resulting in sharp changes in the slope
face strike, and steep, variable slope face angles. Figure 1.4 depicts an example of a sharply
curved roadway.
Figure 1.3: Location of the study area
5
Figure 1.4: Rock cut face with steep slope angle at curved corner of the road
1.4 Objectives of the Study
The main goal of this research is to develop a robust methodology for the determination of
level of geotechnical hazard of rock slope failure under the effect of heavy precipitation with
application to roadway cuts in southern west Saudi Arabia where heavy rainfall conditions in
mountainous roads prevail. More specifically, the goals of this study are the following:
1. Review and examine the validity of existing empirical methods to assess rock slope
stability under rainfall conditions as is the case in the area of study in southern west Saudi
Arabia.
2. Develop an appropriate numerical modelling technique for rock slope stability analysis
taking into consideration the rock mass structure (dominant joint) and the influence of
pore water pressure. Derive suitable failure condition, e.g. using Shear Strength
Reduction Factor (SRF).
6
3. Develop a probabilistic analysis approach to help determine the likelihood of rock slope
failure while taking into consideration the random variation of the critical joint set
surface properties, namely the dip, dip direction and friction angle.
4. Develop a new, rock slope geotechnical hazard index for rock cuts in highways and
mountainous roads in southern west Saudi Arabia.
1.5 Thesis Outline
Chapter 1 presents background to the nature of the road slope cuts in the mountainous region
located in the south-west area of Saudi Arabia. The chapter then discusses the common problems
that engineers, and designers are faced with when designing roads in difficult terrain and the
associated risks which may threaten the public safety of roads and highways, especially in the
rainfall season where the severity level increases. Finally, the scope and objectives of this
research are presented.
Chapter 2 provides a detailed literature review of the kinematic and numerical modelling
approaches that are used in the stability assessment in rock slopes. Different probabilistic
methods are then addressed including the point estimate method and the design of experiment
approaches such as central composite and Box-Behnken design to establish a relationship
between independent variables and the response variable (safety factor). This relationship is used
with the help of Monte-Carlo simulation technique (MCS) to assess the probability of
unsatisfactory performance of rock slopes. Also, a brief introduction has been presented about
the empirical methods used in the study.
Chapter 3 is about the selected case study rock slopes. It discusses the application of selected
empirical methods to classify the rock mass quality of rock slopes and the assessment of the
stability condition for the selected case study rock slopes. The advantages and disadvantages of
these empirical methods are presented from the perspective of their applicability to rock slopes.
Chapter 4 presents the probabilistic approach to slope stability. It presents the assessment of
rock slope stability for a case study that was selected to represent a structurally-controlled failure
mechanism. The effect of dominant joint set where the dip, dip direction and friction angle of the
joint surface are selected as probabilistic variables. A response surface methodology represented
by the Box-Behnken design is used as the probabilistic method to identify the probability of
7
unsatisfactory performance of the selected case study. The stability of the rock slope is
conducted in dry and saturated conditions to reach the appropriate design value for the stability
factor of a rock slope in seasonal heavy rainfall areas.
Chapter 5 describes the development of a new geotechnical hazard index (HI) by introducing
a new factor, beside the factor of the probability of unsatisfactory performance (PF), defined as
the intensity factor (IF). This factor considers the severity of failure, and is derived on the basis
of anticipated volume of failed slope material using a maximum shear strain criterion.
Chapter 6 presents the summary, conclusion and the statement of contributions resulting
from the research undertaken. Suggestions and recommendations for the future work are also
discussed.
8
CHAPTER 2
LITERATURE REVIEW
2.1 Introduction
Rock slope stability assessment is one of the most important concerns in civil and mining
engineering projects, for instance, the rock cuts along highways, railways, rock cutting processes
in urban areas and assess the slopes in open pit mines. Also, in the mountainous areas the rock
cutting processes in these areas constitute the kind of difficulties and obstacles because of the
curvatures and different dip angles of the road paths, this requires a more realistic designs and
development needed considerations to avoid problems related to the stability and potential of
failures (Goodman, 1989; Hoek and Bray, 1981; Gurocak et al., 2008).
In the natural environments the rock slopes find in an inhomogeneous, and anisotropic rock
mass that exhibits complex and, in some cases, unpredictable behavioure (Nathanail et al., 1992;
Hajiabdolmajid and Kaiser, 2002). Therefore, understanding the failure mechanism, which
governed by the geological structures and the geometry of the slope is one of the most important
issues in rock slope stability assessment. The slope with inhomogeneous and complex structures
makes the failure mechanism more complicated. However, to create a realistic model, which
simulates a complex rock masses, the geometry of the slope and the properties of the slope are
required to be determined and presented in the numerical models. The inherent variability of these
geometrical and strength parameters should be investigated, and the impact of each input
parameters to the response values (factor of safety) required to be examined to find out the most
significant parameters that effect on the stability of the rock slopes (Duzgun et al., 2003; Grenon
and Hadjigeorgiou, 2008; Gravanis and Pantelidis, 2014).
In any slope stability assessment analysis, there are some steps should be followed, and
this procedure called “Conventional approach” (Figure 2.1), which is the first step for the stability
analysis is data collections by both field investigation and laboratory tests to produce a sufficient
data for the parameters of interest. And the qualification of these data depends on the investigation
techniques used, which vary from simple techniques such as, scan line method, to advanced
technique like remote sensing techniques, for example, LIDAR technique, this letter method is
suitable for unstable areas where the life of the people may be exposed to a risk (Jaboyedoff et al.,
9
2012). In addition to the field measurements, the laboratory tests should be conducted to estimate
the physical and strength properties for rock specimens to complete the required data for numerical
analysis.
Characterization of the obtained data to provide the required input parameters for a selected
numerical model, whether using deterministic or statistical methods. The numerical model depends
on the purpose of the analysis and the results desired and this is the right choice, according to the
analysis capabilities. In case of slope stability assessment, the numerical analysis often used to
evaluate a possible of failure surface, and it is usually expressed as “stability factor” as an output
that describes the stability of the slope. The stability factor could be determined by different ways
such as, reliability index (β), limit equilibrium (LE), Shear Strength factor (SRF), and probability
of failure (𝑃𝑓 ) (Arai and Tagyo, 1985; Chen, 1992; Hammah et al., 2004).
- Field Investigation or Lab test

Acquiring sufficient data - Selected modelling technique
- Desired level of reliability of the output
Characterizing the - Deterministic technique

obtained data - Statistical technique
Numerical modelling - Stability factor (possible failure surface)
Figure 2.1: Conventional procedure for slope stability analysis
10
2.2 Uncertainty in the Stability Analysis of the Slope
Identify the sources of uncertainty that affect the reliability of the output results are one of
the most important factors that often omitted in the conventional procedure (Duncan, 2000).
The uncertainty could be existed in the slope stability analysis stages, such as in the input
parameters, during the calculations, or in method that used to transfer the input parameters to
the numerical model (Li et al., 2011; Park et al., 2012).
A reliable design approach must be able to consider the variability of the input parameters
and uncertainties for evaluating the probability of failure in rock slope to identify the real
amount of the hazard and make the remedial and mitigations measurements to reduce the risk
to the acceptable level.
Uncertainty has a direct effect on the reliability of the analysis or the design, even though
the data acquisition techniques and modelling methods have been subjected to many
developments and improvements in order to reduce the level of uncertainty in the geometric
parameters and output results (Baecher & Christian, 2003). The sources of uncertainty in rock
slope stability analysis can be classified into three categories (Figure 2.2), natural variability,
knowledge uncertainty and other sources of uncertainty.
Natural variability, which consider most important source of uncertainty, especially when
dealing with rock mass and discontinuity network (Baecher & Christian, 2003). The inherent
variability could be different characteristics of the geological structures at different locations
(spatial), or different characteristics in one site but in different time (temporal). And the spatial
variability is a common source of uncertainty and observed in the geometric parameters such
as, the orientation and size of the discontinuities (Cai, 2011; Nadim et al., 2005; Shamekhi,
2015).
Knowledge uncertainty, this source can be existed if there is lack of knowledge of the
controlling geometric parameters, and human errors in data collection, or adding some
simplified assumptions in the modelling methods (model uncertainty), or the low level of
accuracy in the acquired data (Baecher & Christian, 2003). Also, one of the sources of the
knowledge of uncertainty is the data collection through the human errors could be happened
during the field measurements or laboratory tests.
11
Determination of uncertainty sources should be conducted by investigating the effect of
each input parameters on the responses (outputs), through all expected possibilities, and
identify the variability of the strength and geometric parameters into the numerical modeling,
then come up with a complete picture about the most possible response distribution (stability
factor) that determine the probability of failure for the rock slope. Therefore, a statistical
simulation offers a means for dealing with uncertainty. It can quantify uncertainty and estimate
the probabilities of occurrence of different outcomes. It can thus help engineers to develop
more robust and economic designs and solutions (El-Ramly and Morgenstern, 2002; Hammah
et al., 2009; Gokceoglu et al., 2000; Baecher et al., 1977).
- Temporal variability
Natural Variability
- Spatial variability
- Lack of knowledge
Knowledge uncertainty - Modeling uncertainty
- Measurement accuracy
Other sources of - Operational uncertainty

uncertainty - Decisional uncertainty
Figure 2.2: Uncertainty sources in a rock slope stability analysis (Baecher & Christian, 2003)
2.3 Deterministic and Probabilistic Approaches

2.3.1 Deterministic Approach
Many of the slope stability have always conducted using deterministic approaches for
calculating the factor of safety for design analyses. a determinisic analysis ignords many sources
of uncertainty, as it uses one set of values (mean values) for the input parameters to get a single
value for the stability factor (single output value).
12
The probability of failure is very sensitive to the degree of uncertanity and the inherent
variability in the input parameters (Tabba, 1984). In addition, different modes of failure with same
factor of safety may have different probabilities of failures (Einstein and Baecher 1982). Therefore,
a deterministic approach when used with the numerical methods used only for a general evaluation
of the slope stability but not used to make decision in practical designs.
2.3.2 Probabilistic Approach
To determine the sources of uncertainty in slope stability analysis which can not
implemented through the deterministic approache, the probabilistic (stochastic) analyses are
substituted (Nadim et al., 2005; Hammah et al., 2008). A stochastic approach considers one of the
most techniques to evaluate the stability of the slopes in terms of the probabilitiy of failure. To
investigate the stability behaviour of a rock slope considering the uncertainty due to the variation
of the geometrical parameters, different realizations of the slope should be constructed. To
generate representative realizations of the slope, statistical techniques are engaged in the process
of the rock slope stability assessment (Shamekhi, 2015).
Different probabilistic methods that address the applications of reliability methods in the
field of geotechnical engineering. The most common methods include: (I) First Order Reliability
Method (FORM). (II) Point Estimate Method (PEM). (III) Monte Carlo Simulation (MCS).
2.4 Probabilistic methods in Geotechnical Engineering

Many of probabilistic methods have been adopted for dealing with uncertainty in field of
geotechnical engineering. An overview of these probabilistic methods and their applications will
be discussed as follows:
2.4.1 First Order Reliability Method (FORM)

This method considers one of the several methods that identify problems associated with
uncertainty. The first step in evaluating the probability of failure of any structure is to identify the
specific performance criteria and the relevant input parameters (Christian and Baecher, 2003).
these input parameters called the basic variables Xi. the performance function (margin of safety)
can be describe as the following (Haldar and Mahadevan, 2000).
𝑍 = 𝐺(𝑋1 , 𝑋2 , … … … . . , 𝑋𝑛 ) (2.1)
13
The limit state (failure surface) between the safe and unsafe regions in the design parameter
space can be defined as Z=0 (figure 2.3). And it represents a state beyond which a structure can
no longer fulfill the purpose for which it was designed for. Figure (2.3) shows the limit state in
equation (2.1) of a two parameters input problem, that the failure can take place when Z<0, and
the probability of failure can be calculated by the integral (Haldar and Mahadevan, 2000) as:
𝑃𝑓 = ∫ … ∫𝑔( )<0 𝑓𝑥 (𝑥1 , 𝑥2 , … . . , 𝑥𝑛 𝑋𝑛 )𝑑𝑥1 𝑑𝑥2 … 𝑑𝑥𝑛 𝑑𝑥𝑛 (2.2)
Which 𝑓𝑥 (𝑥1 , 𝑥2 , … . . , 𝑥𝑛 𝑋𝑛 ) is the joint probability density function for the basic random
variables (X1, X2,…., Xn). The computation of the probability of failure (unsatisfactory)
performance (𝑃𝐹) from the equation (2.2), is called the full distribution approach and can be
considered to be the fundamental equation of the reliability analysis.
One of the applications of the FORM is to study the influence of spatial variability of soil
properties on slope stability (Christian et al, 1994; Duncan, 2000). One of the methods which used
the concept of the FORM is the First Order Second Moment (FOSM), this method is represented
by a first- order (linear) approximation with equivalent normal variables, and the information on
the distribution of random variables is neglected.
Figure 2.3: Limit state concept (Haldar and Mahadevan, 2000)
14
2.4.1.1 First Order Second Moment (FOSM)
The FOSM is common to express the system’s reliability in terms of a reliability index
(β). And it is based on a first – order Taylor series approximation of the performance function
linearized at the mean values of the random variables (Duzgun et al., 2003; Low, 2007).
The reliability index is a relation between the loads (Q) of the system and its ability to
carry those forces (resistance R). The margin of safety (M) defines as the difference between the
resisting (R) and the loads (Q), with the slope being unstable if the margin of safety is negative.
M= R – Q (2.3)
In order to bring the picture in terms of the application of this concept on the case of
rock slope, the Q could represent as the displacing force (d) and the R as the resisting force (r).
The shaded area in Figure 2.4 being proportional to the area of the shaded zone. The method of
calculating the area of the shaded zone is to calculate the probability density function of the margin
of safety, the area of the negative portion of this function is the probability of failure (Figure 2.4).
If the resisting and displacing forces are defined by normal distributions, the margin of safety is
also a normal distribution, the mean and standard deviation of which are calculated as follows:
Mean, for the margin of safety given by

𝜇𝑀 = 𝜇𝑅 − 𝜇𝑄 (2.4)
Which,
𝜇𝑀 =the mean value of margin of safety.
𝜇𝑅 = is the mean value of resistance.
𝜇𝑄 = is the mean of the load.
Standard deviation for the margin of safety given by
𝜎𝑀 = √(𝜎𝑅2 + 𝜎𝑄2 ) − 2𝜌𝑅𝑄 𝜎𝑅 𝜎𝑄 (2.5)
Which,
𝜎𝑀 = the standard deviation of margin of safety.
15
𝜎𝑅 2 , 𝜎𝑄 2 = are the variances of the resistance and the load respectively.
𝜌𝑅𝑄 = is the coefficient of correlation between both (R) and (Q) variables.
𝜎𝑅 , 𝜎𝑄 = are the standard deviations of the resistance and the load respectively.
Figure 2.4: Probability distribution of performance function (M)
In case the R and Q are not correlated to each other the coefficient of correlation will be
equal to zero (𝜌𝑅𝑄 = 0).
The reliability index (β) should be determined first in order to obtain the probability of
failure (𝑃𝑓 ). The reliability index can be given as follows:
In case there are a coloration between the R and Q, then the reliability index can be calculated
as:
𝜇𝑀 𝜇𝑅 −𝜇𝑄
𝛽= = (2.6)
𝜇𝑅 2 +𝜎 2 )−2𝜌
√(𝜎𝑅 𝑄 𝑅𝑄 𝜎𝑅 𝜎𝑄
In case of R and Q being uncorrelated, the reliability of index can be calculated as:
𝜇𝑀 𝜇𝑅 −𝜇𝑄
𝛽= = (2.7)
𝜇𝑅 2 +𝜎 2 )
√(𝜎𝑅 𝑄
16
The procedure to estimate the probability of unsatisfactory performance through the reliability
index is as follows (Abdlellah, 2014):
1. Estimate the statistical description of the model parameters which are usually described by
their means, variance, and covariance.
2. Establish an analytical or numerical model to compute the margin of safety (M).
3. Compute the statistical moment of the performance function (margin of safety) in terms of
means and variances.
𝜇 (𝑝𝑒𝑟𝑓𝑜𝑟𝑚𝑎𝑛𝑐𝑒 𝑓𝑢𝑛𝑐𝑡𝑖𝑜𝑛)
4. Calculate the reliability index, 𝛽 = (2.8)
𝜎(𝑝𝑒𝑟𝑓𝑜𝑟𝑚𝑎𝑛𝑐𝑒 𝑓𝑢𝑛𝑐𝑡𝑖𝑜𝑛)
5. Compute the probability of failure (𝑃𝑓 ).
2.4.2 Monte-Carlo Simulation (MCS)
The Monte Carlo simulation is a powerful technique used in the field of engineering and
geotechnical engineering projects, which dealing with uncertainty. Basically, the MCS depends on
generating a random number for each of the random variables in the problem, and making a
combination among all these random variables to perform a number of deterministic computations
(Hammah et al, 2008). More information about the MCS and its components can be found in
chapter 4.
2.4.3 Point Estimate Method (PEM)
Point estimate method is one of the popularly methods used in field of the geotechnical
engineering because its simplicity, accuracy and can apply it in a wide range of geotechnical
problems (Harr 1987, Wolff 1996, Duncan 1999, Miller et al. 2004, Chiwaye and Stacey 2010).
The PEM was originally proposed by Rosenbluth (1975), this method uses a series of point
estimating- point-by-point evaluations of the response function at selected values (known as
weighting points) of the input random variables to compute the statistical moments of the response
variable (Hammah, 2009).
This approach uses 2n calculations to estimate the statistical moments of the response
function (Y). In case of two variables x1 and x2 the PEM evaluation of (Y) moments involves 22
=4-point estimates. So the value of each variable will be a weighting values of (mean + one
standard deviation) and (mean – one standard deviation) which are denoted with “+” and “-
17
“respectively (Baecher and Christian, 2003). Then the four points will be (𝜇𝑥1 + 𝜎𝑥1 , 𝜇𝑥2 +
𝜎𝑥2 ), (𝜇𝑥1 + 𝜎𝑥1 , 𝜇𝑥2 − 𝜎𝑥2 ), (𝜇𝑥1 − 𝜎𝑥1 , 𝜇𝑥2 + 𝜎𝑥2 ), (𝜇𝑥1 − 𝜎𝑥1 , 𝜇𝑥2 − 𝜎𝑥2 ) .
In the event that the variables are not correlated the weight for each point is 1⁄4 . While
in case of the variables are correlated by a correlation coefficient ρ the weight points will be
integrated with the correlation coefficient as illustrated in Figure 2.5 (Baecher and Christian,
2003).
The correlation coefficient ρ between two variables (x1, x2) is defined as:
𝑐𝑜𝑣𝑎𝑟𝑖𝑎𝑛𝑐𝑒 𝑜𝑓 (𝑥1 𝑥2 )
𝜌𝑥1𝑥2 = (2.9)
𝜎𝑥1 𝜎𝑥2
Where,
𝜌𝑥1𝑥2 : Correlation coefficient between the two variables (x1, x2), and this correlation will be
positive linear relationship between x1 and x2 (e.g. if x1 increases then x2 increases too) when the
𝜌𝑥1𝑥2 = +1. If 𝜌𝑥1𝑥2 = −1 ; the relationship between x1 and x2 will be negative linear (e.g. if x1
increases, then x2 decreases).
X2
( 1 + ρ ) /4
( 1 - ρ ) /4
ϬX2
μX2
ϬX2
(1 + ρ )/4 (1 -ρ )/4
ϬX1 Ϭ X1
X1
μX1
Figure 2.5: Rosenblueth’s pints and weights for two variables, correlated or uncorrelated
(Baecher and Christian, 2003)
18
Covariance (x1, x2) measures how much the two variables vary together, and can be
calculated by the following equation:
∑𝑛
𝑖=1 𝑥1 𝑥2
𝐶𝑂𝑉𝑥1𝑥2 = − (𝜇𝑥1 𝜇𝑥2 ) (2.10)
𝑛
When Y is a function of three variables, x1, x2, and x3, then there are eight point in total
which are located at each combination one standard deviation above or low the mean for all the
variables. As such Rosenblueth defined a convention for the weight’s nomenclature where the first
sign refers to x1 and the second to x2 and so on and so forth; also, if the point is 𝜇𝑥𝑖 + 𝜎𝑥𝑖 then
the sign is positive, otherwise, it is negative; and finally, ρ12 represents the correlation coefficient
between x1 and x2 and so on.
The convention is presented in the following set of equations, as well as shown in Figure (2.6):
1
𝑃+++ = 𝑃−−− = 8 (1 + 𝜌12 + 𝜌23 + 𝜌31 ) (2.11)
1
𝑃++− = 𝑃−−+ = (1 + 𝜌12 − 𝜌23 − 𝜌31 ) (2.12)
8
1
𝑃+−+ = 𝑃−+− = (1 − 𝜌12 − 𝜌23 + 𝜌31 ) (2.13)
8
1
𝑃+−− = 𝑃−++ = (1 − 𝜌12 + 𝜌23 − 𝜌31 ) (2.14)
8
19
Figure 2.6: Rosenblueth’s points and weights for three variables, correlated or uncorrelated
(after: Rosenblueth, 1975; Christian & Baecher, 1999)
Some modifications have been made on the first scenario of Rosenbluth method by
Hongs (1998) to be more sufficient for complicated engineering problems. In this case the response
function (Y) became a function of multiple, skewed and uncorrelated variables. That means the
point estimates of each parameter are functions of the first three statistical moments (mean,
variance, and skewness) and number of input variables.
Therefore, this modification increases the flexibility of the PEM for a range of stochastic
distributions of the input parameters (Baecher and Christian 2003).
Despite all the advantages and capabilities of the point estimate method, it suffers from
a main disadvantage which is the curse of dimensionality that effects on the accuracy of the method
when the number of skewed input parameters exceeds a certain value (Baecher and Christian,
2003). This shortcoming is noticeably influence when PEM is used to predict the statistical
moments of the output parameters. Due to this reason the outcome from PEM cannot be used for
determining the probability of failure or for any prediction purposes as the point estimates are a
function of skewness and the number of variables especially, when the number of variables
20
exceeds six or the input variables are highly skewed, that resulting the calculated point estimates
scatter more from the mean values to the tails of the distribution.
Therefore, an alternative method is needed to overcome the effect of the curse of

dimensionality by reducing the number of the input variables to only those that significantly
control the behaviour of the output parameters.
A different procedure may be followed in this case by creating a prediction function.

Having this function, the stability factor can be directly estimated for any arbitrary slope realization
without the need to carry out numerical simulations.
To do this step, the Response surface Method is recommended to create a response

surfaces which represent a mathematical relationship between the independent and dependent
variables (Hammah et al., 2008). The outcome design should provide enough information to fir
first-order response surface that described by a linear relationship between the independent and
the dependent variables. Also, a second- order response surface described by quadratic terms.
Constructing the prediction model, an accurate design is required by conducting some

experiments and analyses to consider all possible combinations of the input variables (available
from point estimates) and generate efficient realizations of the slope, then come up with a realistic
design appropriate for predicting purposes. These experiments are the factorial design (full or
fractional) whether two levels or three levels, Central Composed design and Box-Behnken design.
2.4.4 Design of Experiments (DOE)
2.4.4.1 Factorial Design
Factorial designs are most efficient for type of experiments and the experimental designs
that commonly used by engineers and manufacturing companies are full and fractional factorial
designs at two levels and three levels (Antony, 2003). By a factorial design, that mean in each
complete trial or replication of the experiment all possible combinations of the levels of the factors
are investigated (Montgomery, 2013), then the effect of one or more input factors and their
interactions on the overall behaviour of the outputs (responses) can be evaluated.
21
2.4.4.1.1 Two Levels Factorial Design
A full factorial designed experiment consists of all possible combinations of levels for
all factors. The total number of experiments for studying “k” factors at two levels is 2k . For
example, when we have two factors, say A and B, each run at two levels. This design is called a
22 factorial design.
The 2k full factorial design is particularly useful in the early stages of experimental
work, especially when the number of design factors is less than or equal to 4.
Two kinds of effects are estimated in a factorial design to evaluate the influence of the input factors
(parameters):
1. Maine effects that describe the influence of the factors on the behaviour of the
responses.
2. Interaction effects, which describe the influence of the changes of a factor, at a

specific level of the other factors on the behaviour of the response values.
As mentioned above regarding the number of factors in a 2k factorial design can

implemented in form of full factorial design if these factors are less than 4 but in case these factors
increase, the number of runs will consume much time and do higher-order interactions (third-order
or higher) which are not important (Antony, 2003). Therefore, the information on the main effects
and low-order interactions may be obtained by running only a fraction of the complete factorial
experiment. The fractional factorial design is among the most widely used types of designs for
product and process design and for process improvement (Montgomery, 2013).
In this design, the levels of the factors can be defined qualitatively such as “low” and
“high” and denoted as coded variables “- “and “+” respectively. Figure (7) shows a geometric view
of the 23 factorial design. Having the natural variables (levels of the factors) substituted by the
coded variables by using the following equation:
22
(𝐴𝑙𝑜𝑤 +𝐴ℎ𝑖𝑔ℎ )
𝐴−
𝑐𝑜𝑑𝑒𝑑(±) 2
𝐴 = (𝐴𝑙𝑜𝑤 +𝐴ℎ𝑖𝑔ℎ )⁄
(2.15)
2
The natural variables in this design can be substituted by the high point estimate and the
low point estimate are then converted by using equation (2.15) to the coded variables.
A basic design with three factors 23 factorial design with coded variables is presented in Table
(2.1).
Column “I” is called the identity column and its values at all treatments are equal to +1.
Treatment combination (1) refers to a treatment in which all the three factors (A, B, C) are at their
low levels. Small letter “a” represents a treatment in which factor A is at its high level while factors
B and C are their low values. Respectively, “ab” represents the treatment in which factor A and B
are their high values while factor C is at its low level. The rest of the notation are interpreted with
the same logic.
Being there are only two levels for each factor, the response is assumed to be
approximately linear over the range of the factor levels chosen. Which induce a first-order response
surface and can be defined by using the following relation:
𝑦 = 𝛽0 + 𝛽1 𝑥1 + 𝛽2 𝑥2 + ⋯ + 𝛽𝑘 𝑥𝑘 +∈ (2.16)
Where,
𝑥, represents independent variables, 𝛽𝑖𝑗 represents regression coefficient, and ∈ is

a random error.
23
Table 2.1: Basic design table for a 23 factorial design (Montgomery, 2013)
Factorial effects
Treatments
I A B AB C AC BC ABC
1 + - - + - + + -
a + + - - - - + +
b + - + - - + - +
ab + + + + - - - -
c + - - + + - - +
ac + + - - + + - -
bc + - + - + - + -
abc + + + + + + + +
Figure 2.7: Geometric view of the 23 factorial design (Montgomery, 2013)
24
2.4.4.1.2 Three Levels Factorial Design
There are some extensions and variations of these design which are useful in some cases,
the factors in these designs are present at three levels. The three levels design is written as 3 k
factorial design. A factorial arrangement with k factors each at three levels. These levels of factors
are referred to as low, intermediate, and high. The notations used to represent these factors levels
by the digits 0 (low), 1 (intermediate), and 2 (high). Each treatment combination in the 3 k design
will be denoted by k digits. Where the first digit indicates the level of factor A, the second digit
indicates the level of factor B, and the kth digit indicates the level of factor K. Figure (2.8) shows
the treatment combinations in a 33 design (Montgomery, 2013).
Figure 2.8: Geometric view of the 33 factorial design (Montgomery, 2013)

The design system in 2k design presented previously, the notation that used in this design
was ±1 because it facilitates the geometric view of the design and because it is directly applicable
to regression modeling, and construction of fractional factorial.
In 3k designs, when the factors are quantitative, the system is often denoting the low,
intermediate, and high levels by -1, 0, and +1, respectively. This facilitates fitting a regression
model relating the response to the levels.
The reason that the three-level designs were proposed is to model possible curvature in the
response function and to handle the case of nominal factors at 3 levels. A third level for a
continuous factor facilitates investigation of a quadratic relationship between the response and
each of the factors, this relationship is described as following,
25
𝑦 = 𝛽0 + 𝛽1 𝑥1 + 𝛽2 𝑥2 + 𝛽12 𝑥1 𝑥2 + 𝛽11 𝑥12 + 𝛽22 𝑥22 +∈ (2.17)
Where,
𝑥, represents independent variables, 𝛽𝑖𝑗 represents regression coefficient, and ∈ is a random
error.
A third factor level allows the relationship between the response and the design factors to
be modeled as a quadratic.
Unfortunately, the three-level design is prohibitive in terms of the number of runs, and thus in
terms of cost and effort. For example, a two-level design with center points is much less expensive
while it still is a very good (and simple) way to establish the presence or absence of curvature.
Then, if the curvature is important, two-level design can be augmented with axial runs to obtain a
central composite design, this sequential strategy of experimentation is far more efficient than
running a 3k factorial design with quantitative factors (Montgomery, 2013).
2.4.5 Response Surface Designs

Response Surface Methods are statistical techniques that are useful for the modeling and
analysis of problems in which a response of interest is affected by many variables and the main
goal of this analysis is to optimize this response (Montgomery, 2013).
The response (y) is a function of the several independent variables (x1,x2…, xn ) as shows
in the following relation:
𝑦 = 𝑓(𝑥1 , 𝑥2 … . , 𝑥𝑛 )+∈ (2.18)
Where ∈ represents the error observed in the response (y). If the expected response is denoted by
𝐸(𝑦) = 𝑓(𝑥1 , 𝑥2 , … , 𝑥𝑛 ) = 𝜂 (2.19)
Then the surface represented by
𝜂 = 𝑓(𝑥1 , 𝑥2 … . , 𝑥𝑛 ) (2.20)
Where 𝜂 called a response surface.
26
The first step in the RSM is to find a suitable approximation for functional relationship
between the response (y) and the independent variables. If the response is well modeled by a
linear function of the independent variables, the approximating function is the first-order model
(Montgomery, 2013).
𝑦 = 𝛽0 + 𝛽1 𝑥1 + 𝛽2 𝑥2 + ⋯ + 𝛽𝑘 𝑥𝑘 +∈ (2.21)
If there is a curvature in the system, the approximation function will be in a higher degree, such
as the second-order model (Montgomery, 2013).
𝑘 𝑘 𝑘
𝑦 = 𝛽0 + ∑ 𝛽𝑖 𝑥𝑖 + ∑ 𝛽𝑖𝑖 𝑥𝑖2 + ∑ ∑ 𝛽𝑖𝑗 𝑥𝑖 𝑥𝑗 +∈ (2.22)

𝑖=1 𝑖=1 𝑖<𝑗=2
The response surface designs are available only for continuous factors, and the response
surface includes two main kind of designs, Central Composite Design (CCD) and Box-Behnken
Design (BBD).
In the event of interest about curvature in the response function, the response surface designs
are more efficient than the 3k design which is mostly not sufficient way to model a quadratic
relationship. And a 2k design augmented with center points, is an excellent way to obtain an
indication of curvature. Then if the curvature is important, the two-level design can be augmented
with axial runs to obtain a CCD, this sequential strategy of experimentation is far more efficient
than running a 3k factorial design with quantitative factors (Montgomery, 2013).
2.4.5.1 Central Composite Design (CCD)

Central Composite Design (CCD) is one of the commonly design used to fit second-order
models in an analysis. The CCD contains of a factorial or fractional factorial design (2k) with
center points (η0) and added to a group of axial points (2k), which is allow estimation of
curvature. Here, k is the number of independent variables and η0 is the number of replications for
a combination consists of the variables mean.
27
If distance from the center of design space to a factorial point is ±1 unit for each factor, the
distance from the center design to the axial point represented by α (axial distance). Figure (2.9)
shows the generation of rotatable CCD for Two factors.
Figure 2.9: Diagram of rotatable CCD generation for two factors; factorial points (blue) + center
point (green) and axial points (red)
The factorial points are mostly involved in the estimation of the linear terms and the
interaction terms in a second-order fit, while the axial points and the center points are used to
estimate the quadratic terms and the curvature of the response surfaces (Myers and Montgomery,
2002).
There are many types of the CCD in which the region of interest is better described either
by a rotatable design (sphere, see Figure 2.9), or face-center CCD (cuboid, see Figure 2.10). A
rotatable design holds similar scaled prediction variance for any pairs of design points which are
equally distanced from the design center. This will give a similar prediction quality for the second-
order model within the region of interest. To achieve this situation two equations will be used;
equation (2.23) by (Khuri and Cornell, 1987), and equation (2.24) by (Box and Draper 1987,
Mason et al. 1989).
28
1⁄
𝜂0 = 4 × 𝐹 2 + 4 − 2𝐾 (2.23)
4
∝= √𝐹 (2.24)
Where,
𝐾 is the number of factors in the design.
𝐹 is the number of factorial points.
Figure 2.10: Diagram of face-center CCD generation for two factors; factorial points (blue) +
center point (green) and axial points (red)
2.4.5.2 Box-Behnken Design (BBD)
Box-Behnken Designs are a class of rotatable second order designs based on three level of
factorial designs consists of the central point and the middle points of the edges (Box and Behnken,
1960). BBD is considered a highly efficient design for estimating the second-order response
surface that is able to capture the curvature in the system for a given independent variable
(Montgomery, 2013), the design is shown in Figure (4.5) in chapter 4. More details about the BBD
approach can be found in chapter 4.
29
In this study the BBD is applied as a probabilistic method, each combination of the design
points will generate different realization of the slope, and this realization could be investigated in
rotatable. Hence, a response surface is constructed for these designs and then can get a quantified
prediction of the second-order models from each analysis. According to each design prediction
model the probability of failure can be estimated.
2.5 Conventional Methods in Geotechnical Engineering

Rock slope stability has been addressed using different techniques. Two kinds of these
techniques are widely used in field of geotechnical engineering, Kinematic and kinetic methods.
Kinematic analysis is commonly used to predict the simple failure mechanism (planar, wedge,
and toppling) using stereo projection technique to project the orientation data on a 2D stereonet,
each plane of discontinuity is represented by a pole that contains information about the dip and dip
direction. Discontinuities with the same orientation form a cluster of poles on a stereonet and
considered as a discontinuity set (Price and Cosgrove, 1990).
Three simple rock failure mechanisms can be identified by the kinematic analysis, Plane,
wedge and toppling failures.
Plane failure involves the sliding of a rock mass along a single plane surface. In order that
sliding should occur on a single plane, the following potential conditions must be satisfied (Hoek
and Bray, 1981).
1. The plane on which sliding occurs must strike parallel or nearly parallel (within
approximately ±20°) to the slope face.
2. The failure plane must “daylight” in the slope face. This means that its dip must be
smaller than the dip of the slope face.
3. The dip of the failure plane must be greater than the angle of friction of this plane.
4. Release surfaces which provide negligible resistance to sliding must be present in the
rock mass to define the lateral boundaries of the slide.
Wedge failure involves translational sliding between two intersection sets of discontinuities.
As in the case of plane failure, a condition of sliding is defined by the dip of the line of the
intersection of two planes must be less than the inclination of the slope angle, and bigger that the
friction angle.
30
Toppling failure involves a non-sliding element in the mechanism in which individual blocks
or sets of blocks rotate about some fixed axes.
Kinetic analysis technique compares the magnitude of the driving and resisting forces that act
along the planes of the discontinuities to estimate the factor of safety (Coggan et al. 1998). If the
failure mechanism is more complicated this method become incapable of representing the true
behaviour of the slope, and the numerical modeling should be used in this case.
2.6 Numerical Methods in Geotechnical Engineering

Another technique is numerical modelling which is often used when the failure mechanism is
more complex and cannot be analyzed by kinematic methods. Numerical methods in general are
divided into three types:
1) Continuum methods: These are suitable for the slopes in which the behaviour is mainly
governed by the intact rock or rock masses with a few discontinuities that are not frequent enough
to create a block system (Jing, 2003). The most commonly used methods are Finite Difference
Method (FDM), Finite Element Method (FEM), and Boundary Element Method (BEM).
2) Discrete methods: These are often used when the rock mass is highly fractured or the length of
discontinuities is long enough to create single blocks, and the rock domain can be represented by
an assemblage of separated blocks that can either be rigid or deformable (Shi, 1988; Goodman,
1995).
3) Hybrid Methods: These methods are good for inhomogeneous and anisotropic nature of rocks.
In some cases, it is necessary to use a combination of numerical methods either to construct a more
realistic model or to reduce the cost of computational effort for example, DEM/BEM or BEM/FEM
(Jing, 2003).
2.7 Empirical Methods
Rock mass classification systems or empirical methods represent an important tool to

assess the engineering behaviour of a rock mass. Empirical relations between rock mass properties
and the behaviour of the rock mass in relation to a particular engineering application are combined
to give a method of designing structures in or on a rock mass. In a classification system, an
empirical relation between the rock mass properties and the behaviour of the rock mass in relation
31
to a particular engineering application are combined to give a method for the engineering structure
design.
Many of the classification systems have been proposed in order to identify the rock
cuttings that have a high risk of failure (Pantelidis, 2009). Most of these rock mass classification
systems involve factors relevant to general condition of rock mass (intact rock strength, RQD),
condition of discontinuities, groundwater condition. Most of these classifications involve most of
these factors, since these some of these factors represent the inherent properties and characteristics
of the rock mass.
Since rock mass classification methods have been initially designed for tunneling and
underground excavation application, these classification methods have undergone developments
and modifications for rock slope stability applications. Thus, in this thesis the empirical methods
that are used for the analysis of slope stability will be considered.
Rock mass classification procedures were developed as an alternative to analytical

methods of discontinuous rock mass. The classification systems were developed primarily
empirically by establishing the parameters of importance, giving each parameter a numerical value
and a weighting factor. This led, via empirical formulae, to a final rating for a rock mass. The final
rating is related to the stability of the underground excavation used for the development of the
classification system (Hack el al., 2003).
Five empirical methods have been adopted in this study for assessing the rock mass
condition and evaluating the stability condition of rock slopes. These empirical methods are Slope
Mass Rating (Romana, 1985), Chinese Slope Mass Rating (Chen, 1995), Continuous Slope Mass
Rating (Tomás et al., 2007), Graphical Slope Mass Rating (Tomás et al., 2012) and Hazard Index
(Pantelidis, 2010), all these empirical methods are described and addressed in chapter 3.
32
CHAPTER 3
APPLICATION OF ROCK MASS CLASSIFICATION SYSTEMS TO ROCK
SLOPE STABILITY ASSESSMENT– A CASE STUDY
3.1 Abstract
The stability of rock slopes is considered crucial to public safety in highways passing
through rock cuts, as well as to personnel and equipment safety in open pit mines. Slope instability
and failures occur due to many factors such as adverse slope geometries, geological discontinuities,
weak or weathered slope materials as well as severe weather conditions. External loads like heavy
precipitation and seismicity could play a significant role in slope failure. In this paper, several rock
mass classification systems developed for rock slope stability assessment are evaluated against
known rock slope conditions in a region of Saudi Arabia, where slopes located in rugged terrains
with complex geometry serve as highway road cuts. Selected empirical methods have been applied
to twenty-two rock cuts that are selected based on their failure mechanisms and slope materials.
The stability conditions are identified, and the results of each rock slope classification system are
compared. The paper also highlights the limitations of the empirical classification methods used
in the study and proposes future research directions.
Keywords: rock mass classification; graphical slope mass rating; continuous SMR; rock slope
stability.
33
3.2 Introduction
Rock slopes in most road cuts, especially in mountainous areas are prone to instability
problems due to variation in the rock mass conditions and external factors induced by the
environments such as seismic activities and water in the slope (Pantelidis, 2009). The material
characteristics of a rock slope, the height, the face angle, and the rock joint orientations play a
significant role in the instability problem of road cuts and slopes. In addition, the stability of rock
slopes may also be influenced by the road curvature, particularly in rugged terrains (Hoek and
Bray, 1981).
Slope stability problems attract major concerns from researchers, and consequently, several
techniques and methods for slope stability evaluation have been proposed. These methods can be
basically grouped into four categories, i.e. kinematic analysis, limit equilibrium, numerical
modelling, and empirical methods. Kinematic analysis is commonly used to predict potential
structural failure mechanisms (planar, wedge, and toppling) using stereonet projection technique.
This technique is used to project the orientation of discontinuities by pole, containing information
about the dip and dip direction of a joint on a two-dimensional (2D) stereonet (Price and Cosgrove,
1990). Limit equilibrium method compares the magnitudes of the driving and resisting forces that
act along the planes of slide to estimate the factor of safety (Coggan et al., 1998), and it is also
widely used to examine slope structural stability.
Numerical modelling is used in more complex slope geometries and failure mechanisms.
It is particularly useful when the above-mentioned methods cannot represent the behaviours of the
slope. Numerical modelling provides insight into the effect of stress distribution in the slope and
displacements on its behaviour (Wyllie and Mah, 2004). Rock mass classification systems or
empirical methods represent an important tool that is often used for preliminary assessment of the
engineering behaviours of the rock mass (Duran and Douglas, 2000). In this chapter, the focus is
put on empirical methods, and the goal is to assess their efficiency for determination of the rock
slope stability.
Rock mass classification systems have been commonly utilized in the field of geotechnical
engineering, especially for design purpose. They are widely used due to their simplicity and the
limited need for detailed information (Duran and Douglas, 2000). However, these classification
systems are initially established for underground excavations (Hoek, 2007). They are developed
34
based on the importance of parameters; each is assigned with a numerical value and a weighting
factor. By substituting these weighting values into an empirical formula, one can obtain the final
rating for a rock mass (Hack et al., 2003). Such rating is of value in the sense that it can help the
engineering judgment of underground projects for design purpose (Bieniawski, 1993).
Even though empirical rock mass systems are established for underground stability
assessment, some of classification systems have been applied to the evaluation of rock slopes by
calibrating some parameters to make it applicable to surface excavations in rock (Pantelidis, 2009).
Rock mass characterization and discontinuity conditions represent the backbone of most empirical
methods. These conditions can be summarised into five categories, i.e. unconfined compressive
strength (UCS) of intact rock, rock quality designation (RQD), spacing between discontinuities,
discontinuity condition, and groundwater condition. These five factors also refer to as the well-
known Bieniawski’s rock mass rating (RMR) (Bieniawski, 1973, 1976, 1979, 1989).
In this chapter, five rock mass classification methods are chosen. The selected methods are
original slope mass rating (SMR) (Romana, 1985), Chinese SMR (Chen, 1995), continuous SMR
(Tomás et al., 2007), graphical SMR (Tomás et al., 2012) and an alternative rock mass
classification for rock slopes (Pantelidis, 2010). The stability assessment for rock cuts using the
selected five empirical methods has been conducted and their validity was examined. All results
for rock slope description and stability categories have also been compared.
The rock slopes of the selected case study are in rugged mountainous terrain with complex
geological features. They have steep slope face angles and sharp curvatures along the road. They
are subjected to rain storms in different periods of the year, mostly in summer and winter seasons.
The average annual precipitation of rain falls is 500 mm (see Figure 3.1). Because of heavy rainfall,
most rock cuts, even slightly and moderately weathered, suffer from the stability problems, thus
different failure mechanisms may be encountered. In addition, the excavation method used to
create the rock cuts could play a role in the deterioration of the rock slopes. Poor blasting methods
were used prevalently in most road cuts, without indispensable rock support.
3.3 Location of the Study Area
The study area is located in the far south-western part of Saudi Arabia known as Jazan
region, between latitudes 16 and 18 north, and longitudes 42 and 44 east as shown in Figure
35
3.2(a). Twenty-two road cuts are selected as case studies along five main roads, which are
distributed as follows. Five sites along road No. 12, seven sites along road No. 8, four sites along
Al-Hasher road, five sites along Ar-Raith road, and one site along Al-Aydabi road (see Figure
3.2(a)). The selected case studies are in an area with rugged terrains (Figure 3.2 (b)) and are
selected based on the type of failure mechanism.
(1) Structurally controlled failures: different discontinuity orientations play a significant

role in the instability condition of a structurally controlled slope (Figure 3.3). The selected sites
are: 1, 2, 3, 4, 5, 6, 7, 10, 11, 12, 13, 14, 16, 7, 18, 19 and 22.
(2) Non-structurally controlled failure (stress-controlled): the stability condition of the rock
slope is influenced mainly by external forces, i.e. groundwater. Also, the degree of weathering is
high. In non-structurally controlled slopes, the traces of rock joints are not obvious and have no or
little effect on the stability behaviour of the slope (Figure 3.4). The selected sites are: 8, 9, 15, 20
and 21.
Figure 3.1: Annual precipitation map of Saudi Arabian from 1960 to 2014, showing a
maximum average of 500 mm.
36
Figure 3.2 (a): Locations of road cuts in the study area
37
Figure 3.2 (b): A three-dimensional (3D) topographic map showing the roadway through rugged
terrains.
Figure 3.3: Example of structurally controlled failure (site#22 in Figure 3.2 a).
38
Figure 3.4: Example of non-structurally controlled failure due to highly weathering (site#21 in
Figure 3.2 a).
3.4 Classification Systems for Rock Slopes
As previously mentioned, five methods are used and discussed in this paper. Since all SMR
methods are based on the scores of the basic RMR system, all the case studies are first analyzed
with the basic RMR value.
3.4.1 Rock Mass Rating (RMR)
RMR is established by Bieniawski (1973-1989) to evaluate the quality of rock masses for
underground projects. The RMR system consists of five basic parameters that represent different
conditions of the rock and the discontinuities. These parameters are: (1) UCS of intact rock, (2)
RQD, (3) spacing between discontinuities, (4) condition of discontinuities, and (5) groundwater.
This RMR system is known as “the basic RMR” and it gives a value that ranges between 0 and
100 (Bieniawski, 1973).
An additional parameter was proposed by Bieniawski (1976) to account for the influence
of the discontinuity orientation on the stability condition (correction factor). However, this
39
parameter is introduced for tunnel and dam foundations but not for slopes (Aksoy, 2008).
Therefore, Bieniawski (1989) implemented more descriptive details in the fourth parameter of the
basic RMR (the condition of discontinuities). Tables 3.1-3.3 show the classification criteria of
RMR and their different rock mass classes (Bieniawski, 1989). Also, in the case of considering the
effect of discontinuity orientation on the slope stability of a rock slope, he recommended the use
of the SMR system proposed by Romana (1985).
Bieniawski (1989) applied a set of discrete functions to the RMR system, leading to the so-
called discrete RMR (DRMR). This could result in different users deriving different RMR scores,
depending on their own experience and selection of the discrete values. To address this problem,
Sen and Sadagah (2003) modified the RMR system by converting the classical discrete functions
into continuous rating functions, leading to the continuous RMR (CRMR). In this modification,
the difference between RMR values estimated by different users can be controlled within 10%
(Sen and Sadagah, 2003).
In this chapter, both the DRMR and the CRMR were applied to structurally controlled road
cuts, and the resulting scores were compared. It should be noted that, in both DRMR and CRMR,
only the five parameters (UCS, RQD, spacing of discontinuities, condition of discontinuities, and
groundwater), which are presented by the basic RMR, are used for rock mass classification of road
cuts in the study area.
3.4.2 Slope Mass Rating (SMR)
SMR is a classical lump-rating classification system for rock slopes developed by Romana
(1985). The SMR system is originally derived from RMR system, where adjustment parameters
that represent the discontinuity orientations in relation to the slope attitude are added to the basic
RMR, as well as the effect of the excavation method. The SMR score is obtained by subtracting a
factor from RMR depending on the joint-slope relationship, and adding a factor depending on the
method of excavation, as expressed in the following equation:
𝑆𝑀𝑅 = 𝑅𝑀𝑅 + (𝐹1 × 𝐹2 × 𝐹3 ) + 𝐹4 (3.1)
40
Table 3.1: Rock mass rating system (After Bieniawski, 1989)
Parameter Range of values

Point-load For the low range,
Strength of
strength >10 4-10 2-4 1-2 uniaxial compressive
intact rock
1 index (MPa) test is preferred
mineral
UCS (MPa) >250 100-250 50-100 25-50 5-25 1-5 <1
Rating 15 12 7 4 2 1 0
Drill core quality RQD (%) 90-100 75-90 50-75 25-50 <25
2
Rating 20 17 13 8 3
Spacing of discontinuities >2 m 0.6-2 m 200-600 mm 60-200 mm <60 mm
3
Rating 20 15 10 8 5
Slightly rough
Very rough Slightly rough Slickensided
surfaces
surfaces surfaces surfaces, or Soft gouge >5 mm
Separation <1
Condition of discontinuities Not continuous Separation <1 Gouge < 5 mm thick, or
mm
4 (See Table 2) No separation mm thick, or Separation > 5 mm
Slightly
Unweathered wall Highly Separation 1-5 mm (Continuous)
weathered walls
rock weathered walls (Continuous)
Rating 30 25 20 10 0
Inflow per 10 m
tunnel length None < 10 10-25 25-125 >125
(L/min)
Ratio of joint
Groundwater
5 water pressure to
0 < 0.1 0.1-0.2 0.2-0.5 >0.5
major principal
stress
General condition Completely dry Damp Wet Dripping Flowing
Rating 15 10 7 4 0
41
Table 3.2: RMR Rating Adjustment for Discontinuity Orientations (Bieniawski, 1989)
Strike and dip
Vary Very
orientations of Favorable Fair Unfavorable
favorable Unfavorable
Discontinuities
Tunnels and
0 -2 -5 -10 -12
mines
Ratings
Foundations 0 -2 -7 -15 -25
Slopes 0 -5 -25 -50 -60
Table 3.3: Rock Mass Classes determined from total rating (Bieniawski, 1989)
Rating 100-81 80-61 60-41 40-21 <20

Class number I II III IV V
Very good Very poor
Description Good rock Fair rock Poor rock
rock rock
where F1 is an adjustment factor, which depends on the parallelism between the joint strike
(𝛼𝑗 ) (or the plunge direction of the intersection line of two planes (𝛼𝑖 )) and the slope face strike
(𝛼s ). It ranges from 1, when the joint and the slope face strikes are near parallel, to 0.15, when the
angle between strikes is 30 (see Table 3.4).
F2 refers to joint dip angle (𝛽𝑗 ) in the planar failure or the plunge of the line of intersection
of two planes (𝛽𝑖 ) in the wedge-type failure mode. It varies from 1 for joints dipping more than
45 to 0.15 for joints dipping less than 20° (see Table 3.4).
F3 reflects the effect of the angle between the slope face dip (𝛽s ) and the joint dip (𝛽𝑗 ) or
the plunge of the intersection line of two planes (𝛽𝑖 ). It ranges from 0 (very favorable) when 𝛽𝑗 −
𝛽𝑠 or 𝛽𝑖 − 𝛽𝑠 is more than +10°, to -60 (very unfavorable) when 𝛽𝑗 − 𝛽𝑠 or 𝛽𝑖 − 𝛽𝑠 is less than -
10° (see Table 3.4)
F4 is an adjustment factor that depends on the excavation method. The values are selected
empirically as shown in Table 3.5.
42
Table 3.4: Adjustment ratings for F1, F2, and F3 (Romana 1985, modified by Anbalangan et al.,
1992).
Very Very
Case Favorable Fair Unfavorable
favorable Unfavorable
P │αj – αs│
T │ (αj – αs) -
> 30º 30 – 20º 20 – 10º 10 – 5º < 5º
180º│
W │αi – αs│
P/T F1 0.15 0.40 0.70 0.85 1.00
P │ βj│
< 20º 20 – 30º 30 – 35º 35 – 45º > 45º
W │ βi│
P F2 0.15 0.40 0.70 0.85 1.00
T F2 1 1 1 1 1
P βj – βs
> 10º 10 – 0º 0º 0 to – 10º < -10º
W βi – βs
T βj + βs < 110º 110 – 120º > 120º – –
P/T F3 0 -6 -25 -50 -60
Table 3.5: Adjustment factor F4 for the method of excavation (Romana, 1985).
Blasting or Deficient
Method Natural slope Presplitting Smooth blasting
mechanical blasting
F4 +15 +10 +8 0 -8
Planar-type and toppling-type failure modes are considered in the SMR system. Anbalagan
et al. (1992) added a wedge failure to Romana’s classification. Wedge-type and planar-type failure
modes will be addressed separately, where the inclination and the direction of the line of
intersection of two planes are considered for wedge failure analysis. This modified SMR
(Anbalagan et al., 1992) is used in this chapter, and three types of the structurally controlled
failures, i.e. planar, wedge, and toppling, are used. These are identified in Table 3.4 as ‘P’ for
planar, ‘W’ for wedge and ‘T’ for toppling. Different classes of SMR scores that describe the slope
and its stability condition are listed in table 3.6.
43
Table 3.6: Classes of SMR according to Romana (1985)
Class SMR Description Stability Failures Support

I 81 – 100 Very Good Completely stable None None
II 61 – 80 Good Stable Some Blocks Occasional
Some Joints or
III 41 – 60 Normal Partially stable Systematic
Many Wedges
Planar or Big Important /
IV 21 – 40 Bad Unstable
Wedges Corrective
Completely Big Planar or
V 0 – 20 Very Bad Re-excavation
unstable Soil – Like
3.4.3 Chinese Slope Mass Rating (CSMR)
CSMR rating was proposed by Chen (1995), where two coefficients were added to
Romana’s system (SMR). These are the slope height factor (𝜁), and the discontinuity factor (λ), as
shown in the following equation (Chen, 1995):
𝐶𝑆𝑀𝑅 = (ζ × 𝑅𝑀𝑅) + [𝜆 × 𝐹1 × 𝐹2 × 𝐹3 + 𝐹4 ] (3.2)
where 𝜁 is a non-dimensional parameter accounting for the effect of slope height, and is
given by
𝜁 = 0.57 + 0.43 × 80/𝐻 (3.3)
where H is the slope height in meters; λ is a parameter accounting for the effect of discontinuity
type, and is defined as follows:
(1) λ =1 for faults of long weak seams filled with clay;
(2) λ = 0.8 to 0.9 for bedding planes of large-scale joints with gouge; and
(3) λ = 0.7 for joints of tightly interlocked bedding planes.
Regarding the slope height, the equation (3.3) is applicable to slope height (H) > 80 m. For a
slope height (H) ≤ 80 m, the parameter 𝜁 = 1.
44
3.4.4 Continuous Slope Mass Rating
The continuous SMR was proposed by Tomás et al. (2007). In this system, continuous
functions were proposed to replace the lump-rating system adopted in Romana’s system (1985).
The proposed F1, F2, and F3 continuous functions are expressed as follows.
16 3 1
𝐹1 = 25 − 500 arctan[10 (|𝐴| − 17)] (3.4)
where |𝐴|= |𝛼j − 𝛼s | for planar failure, |𝐴|= |𝛼j − 𝛼s − 180| for toppling failure,
|𝐴|= |𝛼i − 𝛼s | for wedge failure. The parameters 𝛼j , 𝛼s and 𝛼𝑖 are the joint dip direction, slope
dip direction, and the trend of the line of intersection of two planes, respectively.
9 1 17
𝐹2 = 16 + 195 arctan (100 𝐵 − 5) (3.5)
where B is (in degrees) the dip angle of the joint for planar-type and toppling-type failure
modes (𝛽𝑗 ), and the plunge of the line of intersection of two planes for wedge-type failure
mode(𝛽𝑖 ).
1
𝐹3 = −30 + 3 arctan 𝐶 (3.6)
1
𝐹3 = −13 − 7 arctan (𝐶 − 120) (3.7)
where C is defined in degrees as follows:
(1) The difference in angle between the joint dip and slope dip, i.e. 𝛽𝑗 − 𝛽s , for planar-type
failure mode;
(2) The difference in angle between the plunge of the line of intersection of two planes and the
dip of the slope, i.e. 𝛽𝑖 − 𝛽s , for wedge-type failure mode;
(3) The sum of the two dip angles of the joint and slope, i.e. 𝛽𝑗 + 𝛽s , for toppling-type failure
mode.
Equation (3.6) is used for slopes with planar or wedge-type failure mode, whilst Equation
(3.7) is used for slopes with toppling-type failure mode.
45
3.4.5 Graphical Slope Mass Rating
A graphical method was proposed by Tomás et al. (2012) for determination of the
correction factors of SMR. The correction factors F1, F2, and F3 for the joint sets are determined
graphically by using stereo plots, which are designed for the failure modes (planar, wedge, and
toppling). The correction factors F1 and F2 are grouped into one term ψ. The stereo plots determine
the position of the discontinuity pole (for planar-type and toppling-type failure modes) or the
intersection of planes (for wedge-type failure mode). The proposed stereo plots for F3 depend on
the failure mode. Therefore, the modified SMR equation is expressed as follows:
𝑆𝑀𝑅 = 𝑅𝑀𝑅 + (𝜓 × 𝐹3 ) + 𝐹4 (3.8)
3.4.6 Hazard Index (HI)
Pantelidis (2010) proposed a quantitative system, in which the hazard associated with rock
slope failure is assessed by considering two separate functions - normal condition (𝑓NC ) and
triggering mechanism (𝑓TM ) for surface water and groundwater conditions. These two functions
contain sub-factors that describe the level of hazard by assigning a score to each level. By selecting
the appropriate score for each category of normal and triggering factors, a final score is obtained
to represent the level of hazard of a rock slope failure.
The quantitative attribution of the normal condition (𝑓NC ) is classified into four sub-factors
based on the failure mechanism. A structurally controlled rock slope can be assessed by the factor
of safety (Goodman and Bray, 1977; Hoek and Bray, 1981), and by the relationship between slope
and joint orientations. Non-structurally controlled slope is examined by using the geological
strength index (GSI) (Hoek and Marinos, 2000), and by calculating the volume of suspended rock
(hanging rock mass due to undercutting) if the rock slope is affected by weathering. The rating
scores for each sub-factor are 1, 3, 6 and 10, where “1” represents favorable condition of the rock
slope and “10” is unfavorable.
The quantitative attribution of the triggering mechanism (𝑓TM ) is assessed by the influence
of surface and underground water (drainage factor), and the ratio of the mean annual precipitation
to the critical annual precipitation against the stability behavior of a rock slope. The drainage factor
𝑓D can be examined by field inspection of the slope materials and structures of the rock cut or slope
(see Table 3.7); it is then assigned by a score (1, 3, 6 or 10). The score of drainage factor is
46
multiplied by the ratio of the mean annual precipitation to the critical annual precipitation to get
the final score of the triggering mechanism 𝑓𝑇𝑀 of a rock cut as follows:
𝐼
𝑓TM = 𝐼m 𝑓D (3.9)
cr
where 𝐼m is the mean annual precipitation, 𝐼cr is the critical annual precipitation, and 𝑓D is
the drainage factor.
The hazard index, HI, for a rock cut is calculated as
1⁄
𝐻𝐼 = (𝑓NC 𝑓TM ) 2 (3.10)
The HI is given on a scale of 1 to 10 defining 4 intervals, namely 1-4 (Good), 4-6 (fair), 6-
8 (poor) and 8-10 (very poor).
3.5 Case Study
In this study, twenty-two road cuts have been chosen in the study area to examine their
stability conditions using the five selected empirical methods. Table 3.8 summarizes the basic
features of the selected sites and field observations.
The methodology used in this study is demonstrated in Figure 3.5. As can be seen, the
methodology begins with the selection of road cuts (22 sites) from the region based on anticipated
failure mechanism. The second step is field investigation and data collection. In the current study,
all data and measurements related to rock mass and discontinuities characteristics were gathered.
This includes the state of weathering, RQD, joint spacing, joint surface conditions, effect of water,
and the different directions of joint sets, as well as the geometrical slope properties such as slope
height, face dip and dip direction. Also, a visual assessment of the rock mass was carried out using
GSI to characterize the rock mass condition. The GSI is an important factor especially in the
assessment of highly weathered rock slopes (non-structurally controlled).
The third step in the methodology (see Figure 3.5) is to perform laboratory testing. Rock
samples were collected from each site for rock testing to determine the physico-mechanical
properties. The bulk density was identified for all rock specimens. The point load test was carried
out for each rock sample to determine the UCS of the rock.
47
The fourth step of the methodology shown in Figure 3.5 is for structurally controlled
slopes. Two types of analyses are employed in this stage, the kinematic analysis and the RMR.
The kinematic analysis is performed with the DIPS (stereographic projection software) software
(Rocscience Inc., 2014). The likelihood of failure and its types (planar, wedge and toppling) are
determined. In addition, the most critical failure surface in the rock slope is identified. The RMR
scores were calculated for all sites. As mentioned previously, all rock mass classification systems
adopted in this study basically use the RMR system, except for the HI method (Pantelidis, 2010).
Two versions of RMR systems were used: the discrete RMR (Bieniawski, 1989) that is used in all
SMR methods, and the continuous RMR (Sen and Sadagah, 2003) that is used only for continuous
SMR classification system.
The fifth step is applying the empirical methods to the twenty-two sites based on the results
of the kinematic analysis represented by the type of failure and the joint-slope relationship and
RMR score. All SMR systems (Original SMR, Chinese SMR, continuous SMR, and Graphical
SMR) have been applied to structurally controlled rock slopes. The HI method (Pantelidis, 2010)
has been applied to both structurally and non-structurally controlled failure of rock cuts. All the
results of empirical methods have been compared and their pros and cons in terms of application
to stability assessment of road cuts are highlighted.
The stability of rock cuts for the selected rock slopes has been assessed through application
of the rock mass classification systems, especially the ones developed for rock slopes. To do this,
RMR should be determined, and the type of failure mechanism for jointed rock mass (structurally
controlled failure) needs to be identified.
3.6 Results
3.6.1 Rock Mass Rating System
The results of the discrete RMR system (Figure 3.6) indicate that the rock cuts sites (1, 2,
3, 4, 5, 6, 10, 14, 16, 17 and 18) have RMR values of 41-57 and are classified as “fair”. The four
locations (7, 11, 12 and 13) have RMR values of 63-79 and are classified as “good”. Site 19 gives
poor rock quality with RMR value of 37.
48
Table 3.7: Rating criteria and factors for the drainage sub-factors 𝑓𝐷,𝑛 (Pantelidis, 2010)
Rating criteria and score

Drainage sub-factors
1 3 6 10
Underground water
Case 1
a. Potential for Favorable conditions Two favorable and one One favorable and two Unfavorable conditions
water infiltration Mantel: Impermeable unfavorable condition unfavorable conditions Mantle: Permeable
from the upslope Upslope gradient: are fulfilled. are fulfilled. Upslope gradient:
area, 𝑓𝐷,1𝑎 Abrupt -or- -or- Gentle
Coverage: The upslope One favorable and two One unfavorable and Coverage: The upslope
area is covered by dense intermediate conditions two intermediate area is bare or almost
evergreen forest or are fulfilled. conditions are fulfilled. bare (e.g. sparse
urban development vegetation, deciduous
development).
b. Potential for Water circulation seems Free drained cutting (I)Fair drainage of (II)Poor drainage of
build-up of impossible (very tight or (wide unfilled underground water underground water
hydrostatic hard-filled discontinuities) through water through through narrow unfilled
pressure, 𝑓𝐷,1𝑏 discontinuities or intact unfilled or soft-filled or soft-filled
rock) discontinuities discontinuities.
-and- -and-
(III)No or short freezing (IV)Long freezing
periods. periods
Surface water
Case 2
c. Potential for Surface water flow is Minor inflow which can (I)Fairly drained rock (II)Poorly drained rock
water inflow unlikely to trigger the be drained through mass with regards to the mass with regards to the
through exposed type of failure studied narrow discontinuities. expected inflow expected inflow
discontinuities (water inflow seems -or- quantity. quantity.
and build-up of impossible: very tight or Free drained cutting -and- -and-
hydrostatic hard-filled (wide unfilled (III)No or short freezing (IV)Long freezing
pressure, 𝑓𝐷,2 discontinuities or intact discontinuities) periods. periods.
rock.
Case 3
d. Instabilities due Surface water flow is Minor instabilities are Instabilities are likely Major instabilities are
to surface water unlikely to trigger the likely due to wash-out due to wash-out of likely due to wash-out
flow, 𝑓𝐷,3 type of failure studied of infilling material. infilling material. of infilling material.
(no loose stones or -or- -or- -or-
blocks, insignificant Water flow from Water flow from A large amount of water
surface water flow etc.) upslope may cause the upslope may cause the flowing from upslope
instability of only some transportation of a few (e.g. from ravine) may
small loose stones. loose blocks or stones. cause the transportation
of several loose blocks
or stones.
However, the results of continuous RMR show observable changes on RMR ratings with
increased RMR values. The RMR classifications for sites 5, 10 and 14 have changed from “fair”
to “good”. The scores for sites 11 and 12 have also increased to 91 and 84, respectively, and were
classified as “very good”.
49
Table 3.8: Summary of the basic features of selected case studies.
Slope Slope face Joint sets (Dip ()/Dip direction ())
Site
height (Dip ()/Dip Potential of failure mechanism Field observations
No. Set #1 Set #2 Set #3 Set #4 Set #5
(m) direction ())
Partially stable (jointed rock mass
1 18 60/040 52/043 75/357 63/238 39/063 - Structurally controlled
in the upper portion)
Partially stable (dispersed blocks
2 19 69/027 73/242 83/134 45/342 - - Structurally controlled
in the upper portion)
3 15 77/055 17/016 54/220 80/210 41/095 83/095 Structurally controlled Stable
4 8 60/269 55/283 65/229 64/116 84/137 - Structurally controlled Unstable (failed)
5 19 78/020 78/252 25/078 84/163 - - Structurally controlled Stable
6 11 80/026 39/260 20/016 79/229 - - Structurally controlled Partially stable
7 18 80/010 57/244 62/195 70/050 85/343 - Structurally controlled Stable
Non-structurally controlled
8 13 60/010 - - - - - Unstable in rainy season
(Highly weathered slope)
88/285
10 30 74/236 44/274 85/339 - - Structurally controlled Partially stable
70/285
74/023
13 18 16/021 63/020 35/081 85/103 - Structurally controlled Stable
67/030
16 33 70/130 36/276 82/254 39/075 83/173 - Structurally controlled Stable
17 23 76/154 43/292 49/072 45/102 64/186 78/130 Structurally controlled Partially stable
18 15 74/008 11/095 83/030 73/038 - - Structurally controlled Partially stable (small blocks)
19 36 74/190 53/219 63/133 53/113 - - Structurally controlled Unstable (failed)
20 11 52/045 - - - - - Stable
22 29 65/140 44/323 82/072 48/149 - - Structurally controlled Partially stable
50
Site selection
Field investigation
Laboratory testing
Kinematic analysis RMR calculations

(Structurally controlled failure) (DRMR & CRMR)
1. Original SMR
Empirical methods 2. Chinese SMR
Structurally controlled failure
3. Continuous
SMR
4. Graphical SMR
Results Structurally and non-
5. Hazard Index
controlled failure
Conclusion
Figure 3.5: Methodology used in the present study.
51
100
90
80
RMR value % 70
60
50
40
30
20
10
0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22
Discrete RMR (1989) 57 48 51 44 55 54 64 55 79 74 63 54 52 52 41 37 58
Continuous RMR (2003) 59 52 54 48 66 52 72 67 91 85 63 62 59 56 45 40 61
Figure 3.6: Discrete and continuous RMR results for jointed rock cuts.
3.6.2 Kinematic Analysis
The kinematic analysis of rock cuts was conducted for the structurally controlled failure
using DIPS 6.0 software (Rocscience Inc, 2014). The analysis results indicate that there is potential
for planar failure in 5 locations, one of which is site 4 that already failed as shown in the photo of
Figure 3.7. Nine locations show potential for wedge failure mechanism, where site 19 already
failed (see photo in Figure 3.8).
For toppling failure, five sites show potential for flexural toppling (Figure 3.9), with high
probability recorded for site 3, and five locations indicate potential for block toppling (Figure 3.10)
with the maximum probability of 25% at site 2. All results of the kinematic analysis are listed in
Table 3.9.
52
(a)
(b)
Figure 3.7: (a) Planar failure at site 4, and (b) Stereographic projection showing the potential for
planar failure.
53
(a)
(b)
Figure 3.8: (a) Wedge failure at site 19, and (b) Stereographic projection showing the potential
for wedge failure.
54
(a)
(b)
Figure 3.9: (a) Potential for flexural toppling failure at site 5, and (b) Stereographic projection
showing for flexural toppling failure.
55
(a)
(b)
Figure 3.10: (a) Potential for block toppling failure at site 5, and (b) Stereographic projection
showing for block toppling failure.
56
Table 3.9: Kinematic analysis results for structurally controlled failure mechanism.
Toppling failure
Planar Failure Wedge failure
Site No. Flexural
(%) * (%) * Block (%) *
(%) *
1 47.5 65.98 15 0.64
2 0 29.1 7.5 25
3 0 7.05 40 1.92
4 17.5 29.49 5 4.23
5 0 3.47 0 14.14
6 5 8.97 27.5 4.10
7 0 17.72 20 5.78
1 32.5 58.61 0 0.13
10**
2 32.5 82.39 0 0.13
11 0 2.71 30 17.68
12 0 2.57 30 4.24
1 22.5 34.53 0 0.77
13
2 30 40.82 0 1.28
14 0 15.64 0 18.08
16 0 13.08 0 6.41
17 0 39.72 0 8.35
18 0 22.21 0 0.51
19 5 62.64 0 0.26
22 35 40.38 0 17.69
*Percentage of joints mapped.
**Site has two slope faces with different dip and dip directions.
3.6.3 Rock Slope Stability Assessment
The stability of the rock cuts in the case study area was assessed with five rock mass
classification systems developed for rock slopes, namely original SMR (Romana, 1985), Chinese
SMR (Chen, 1995), continuous SMR (Tomás et al., 2007), graphical SMR (Tomás et al., 2012)
and the HI method (Pantelidis, 2010). Three main failure mechanisms were defined using
kinematic analysis for the structurally controlled rock cuts (planar, wedge and toppling), and their
results have been used in the selected rock classification systems. The rock cuts with non-
structurally controlled failure mechanisms were assessed only with the HI method, because this
system was designed to address both structurally and non-structurally controlled failures, unlike
the SMR method that is only applicable to structurally controlled failure.
57
3.6.3.1 Structurally Controlled Failure
3.6.3.1.1 Planar-Type Failure Mode
SMR (Romana, 1985) results for rock cuts indicate that all scores are below 40 and are
categorized from unstable to completely unstable. When the strike angle between the slope face
(𝛼s ) and dominant joint (𝛼𝑗 ) is 20° and less, SMR will yield significantly reduced scores. Thus,
the stability condition of the rock cut is decreased as can be seen from the bar chart in Figure 3.11.
However, the values of RMR scores are above 60, suggesting “good” quality rock. This will
improve the SMR values.
Figure 3.11: Results of original SMR, discrete RMR, and strike difference of slope-joint for
planar failure.
The Chinese SMR system is applied to the case studies without the slope height factor
because the heights of all sites are less than 80 m, thus only the discontinuity factor (λ) is
considered. The results of this method show that the SMR scores have been increased compared
to Romana’s system. The reason for this increase is the low values of the discontinuity factor (λ)
for most rock cuts with an average value of 0.7. This value implies tight joints with high cohesion,
which leads to increased SMR values and thus an increase in the degree of stability condition.
While this may not be the case in reality, it can be concluded that the Chinese SMR tends to be
less conservative than Romana’s SMR.
58
The continuous SMR (Tomás et al., 2007) results show that there is a noticeable increase
in the SMR scores, where most of the stability conditions have switched from unstable to partially
stable. However, when the continuous RMR results were used with the continuous SMR functions,
only a slight change in the SMR scores was observed, and the stability condition of only two sites,
7 and 10, changed from partially stable to stable. Sites 7 and 10 show stable and partially stable
conditions, respectively (see Table 10).
The graphical SMR method (Tomás et al., 2012) results show no difference with Romana’s
SMR results. This is not surprising as the underlying concepts used in the graphical SMR method
are the same as the ones in the original discrete SMR (Romana, 1985) for the correction factors
F1, F2 and F3. The difference is only in the method of application, which is based on using the
stereo plots to determine the correction factors in the graphical SMR method.
The results obtained from the HI indicate that sites 1 and 13 are categorized as “fair”, and
sites 4, 7, 10 and 22 are classified as “poor”. As can be seen from Table 8, the field observation
data show that site 1 is partially stable (especially the upper portion), while site 13 is in a stable
condition. Also, site 4 has already failed in reality, and site 7 shows a stable condition; however,
sites 10 and 22 are partially stable. All the analysis results for the potential of planar failure are
documented in Table 3.10.
3.6.3.1.2 Wedge-Type Failure Mode

The results of original SMR indicate partially stable condition at sites 3, 7, 10-1, 16 and
17. Rock cuts 1, 2, 10-2 and 13-1 are classified as unstable, and the results of wedge failure analysis
for sites 4, 13-2, 18 and 19 indicate completely unstable rock cuts. The analysis results for sites 18
and 19 have negative SMR values. This is because the RMR values are below 40 indicating poor
quality rocks, and the difference in the angle between the strikes of the slope and the line of
intersection of the two joints that form the wedge failure is less than 10 (see Figure 3.12).
In addition, by field observation, sites 18 and 19 show partially stable and unstable (failed)
rock cuts, respectively (see Table 3.8). The graphical SMR analysis has given the same results and
conclusion as the original SMR.
59
Table 3.10: Stability analysis results for planar failure using rock slope classification methods.
Continuous SMR Continuous SMR
Site Original SMR Chinese SMR Graphical SMR Hazard Index
using DRMR using CRMR
No.
Rating Stability Rating Stability Rating Stability Rating Stability Rating Stability HI Category
Completely Partially Partially Completely
1 7 22 Unstable 46 48 7 6 Fair
unstable stable stable unstable
Completely Completely Completely
4 9 13 34 Unstable 38 Unstable 9 6.5 Poor
unstable unstable unstable
Partially Completely
7 22 Unstable 34 Unstable 53 61 stable 22 7.4 Poor
stable unstable
Partially
1 34 Unstable 34 Unstable 59 71 stable 34 Unstable 6.5 Poor
stable
10
Partially
2 34 Unstable 34 Unstable 53 65 stable 34 Unstable 6.5 Poor
stable
Partially Partially Completely
1 21 Unstable 33 Unstable 52 52 20 6 Fair
stable stable unstable
13
2 3 21 Unstable 52 52 3 6 Fair
22 7 22 Unstable 42 45 8 7.4 Poor
60
Figure 3.12: Results of original SMR, discrete RMR, and the strike difference of slope-joint for
wedge failure.
The results of the Chinese SMR show a slight increase in SMR scores. For example, in site
2, the SMR classification is unstable, while the Chinese SMR gives the result of partially stable.
Also, at sites 10-2 and 22, the SMR scores are classified as completely unstable in the original
SMR, and they are unstable in the Chinese SMR.
Two kinds of results are calculated for continuous SMR by using discrete and continuous
RMR. Once again, the results of the continuous SMR with discrete RMR indicate an increase in
SMR ratings compared to the original SMR. Sites 10-2, 13, and 22, switch from completely
unstable and unstable to partially stable. After the continuous functions of RMR in combination
with the continuous SMR are applied, the SMR ratings of some locations change, such as site 2
where the stability condition goes from unstable to partially stable.
Also, at sites 7 and 10, the SMR predictions switch from partially stable to stable. Sites 10
and 22 are in partially stable conditions, while sites 7 and 13 are in stable conditions (see Table
3.11).
Using the HI method, the results for all rock cuts are classified “poor”, except for sites 2,
3 and 10 where the results show “fair” rock slopes. However, by field observation, sites 2 and 10
indicate partially stable and site 3 shows a stable condition. All the analysis results of potential of
wedge failure for the selected cuts sites are listed in Table 3.11.
61
Table 3.11: Stability analysis results for wedge failure using rock slope classification methods.
Continuous SMR using Continuous SMR using Hazard Index
Site Original SMR Chinese SMR Graphical SMR
DRMR CRMR (HI)
No.
Rating Stability Rating Stability Rating Stability Rating Stability Rating Stability Rating Stability
1 33(IV) Unstable 40(IV) Unstable 33(IV) Unstable 35(IV) Unstable 33(IV) Unstable 6 Fair
2 40(IV) Unstable 43(III) P. stable 37(IV) Unstable 41(III) P. stable 40(III) Unstable 6 Fair
3 43(II) P. stable 46(III) P. stable 39(IV) Unstable 42(III) P. stable 43(III) *P. stable 5.1 Fair
C. C. **C.
4 2(V) 6(V) 33(IV) Unstable 37(IV) Unstable 2(V) 6.5 Poor
7 55(III) P. stable 58(III) P. stable 53(III) P. stable 61(II) stable 55(III) P. stable 7.6 Poor
1 50(III) P. stable 50(III) P. stable 60(III) P. stable 71(II) stable 50(III) P. stable 6.5 Poor
10
2 28(IV) Unstable 28(IV) Unstable 59(III) P. stable 71(II) stable 28(IV) Unstable 6.5 Poor
1 21(IV) Unstable 33(IV) Unstable 52(III) P. stable 52(III) P. stable 21(IV) Unstable 6.3 Fair
13 C. C.
2 3(V) 21(IV) Unstable 52(III) P. stable 52(III) P. stable 3(V) 6.3 Fair
unstable unstable
16 44(III) P. stable 46(III) P. stable 41(III) P. stable 48(III) P. stable 44(III) P. stable 6.5 Poor
17 44(III) P. stable 46(III) P. stable 41(III) P. stable 45(III) P. stable 43(III) P. stable 7.4 Poor
C. C. C.
18 -9 6(V) 31(IV) Unstable 35(IV) Unstable -9 7.4 Poor
C. C. C.
19 -15 3(V) 29(IV) Unstable 32(IV) Unstable -15 8.5 Poor
C. C.
22 16(V) 29(IV) Unstable 48(III) P. stable 51(III) P. stable 16(V) 7.4 Poor
unstable unstable
62
3.6.3.1.3 Toppling-Type Failure Mode
The toppling failure analysis was conducted for two failure mechanisms, which are flexural
and block toppling. The flexural toppling can be treated in all rock slope classification systems
selected in this study. However, the HI method (Pantelidis, 2010) was basically developed to assess
all possibility of rock failures based on failure mechanism, thus this method can analyze both kinds
of toppling-type failure modes.
The results of SMR’s classification systems indicate that seven rock cut sites are prone to
flexural toppling failure. Based on the original and graphical SMR systems, the rock cuts are
classified as follows: sites 1 and 4 show unstable conditions, sites 3, 6 and 7 are partially stable,
and sites 11 and 12 are classified as stable. The Chinese SMR results show no change in the
stability categories, except for sites 1 and 12 where the stability conditions are partially stable. The
continuous SMR results show that the stability conditions for the rock cuts of sites 1, 3, 6, 7 and
12 are partially stable; for site 11, the stability condition is improved to good or stable, while site
4 remains unstable.
The SMR scores increase when using the continuous RMR with continuous SMR
functions. For example, for sites 4, 7 and 12, the stability conditions change to partially, good and
very good rock slopes, respectively.
The results of HI method for flexural toppling show that sites 1, 3, 4, 7.1 and 11 are fair
rock cuts, and sites 6 and 12 are good rock cuts, whilst site 7.2 is a poor rock cut. Regarding the
block toppling analysis carried out by the HI method, sites 10.1, 11, 12, 14, 17 and 22 are classified
as good rock cuts, and sites 10.2 and 2 are classified as poor; while in site 5, the rock cut is
classified as fair rock cut. In field observation, as shown in Table 3.8, sites 1, 2, 3, 4, 7, 10 and 11,
are partially stable rock cuts, while site 5 is potentially in a stable condition. All the results for
toppling failure analysis are described in Table 3.12.
63
Table 3.12: Stability analysis results for toppling failure using rock slope classification methods.
Continuous SMR using Continuous SMR using Hazard Index
Original SMR Chinese SMR Graphical SMR
Site No. DRMR CRMR (HI)
Flexural Block Flexural Block Flexural Block Flexural Block Flexural Block Flexural Block
1 40 (IV) NA 45 (III) NA 53 (III) NA 55 (III) NA 40 (IV) NA 5 -

2 - NA - NA - NA - NA - NA - 6.5
3 41 (III) NA 44 (III) NA 46 (III) NA 49 (III) NA 41 (III) NA 5.8-4.1 -
4 34 (IV) NA 35 (IV) NA 39 (IV) NA 43 (III) NA 34 (IV) NA 4.1 -
5 - NA - NA - NA - NA - NA - 4.6
6 44 (III) NA 47 (III) NA 49 (III) NA 47 (III) NA 44 (III) NA 3.9 -
1 54 (III) NA 57 (III) NA 58 (III) NA 66 (II) NA 54 (III) NA 4.8 -
7
2 43 (III) NA 49 (III) NA 54 (III) NA 62 (II) NA 43 (III) NA 7.6 -
1 - NA - NA - NA - NA - NA - 3.4
10
2 - NA - NA - NA - NA - NA - 6.2
11 66 (II) NA 87 (I) NA 82 (I) NA 94 (I) NA 66 (II) NA 4.6 2.5
12 72 (II) NA 75 (II) NA 77 (II) NA 88 (I) NA 72 (II) NA 2.9 1.5
14 - NA - NA - NA - NA - NA - 1
17 - NA - NA - NA - NA - NA - 1
22 - NA - NA - NA - NA - NA - 1.3
64
3.6.3.2 Non-Structurally Controlled Failure
As previously mentioned, this type of failure mechanism can only be analyzed by the HI
method. The results indicate that sites 8, 9, 15, 20 and 21 are classified as poor rock cuts as shown
in Table 3.13.
3.7 Discussion
The results of the study show that all SMR methods can be used for structurally controlled
failure of rock slopes (jointed rock mass), while the HI method proposed by Pantelidis (2010) is
applicable to both structurally and non-structurally controlled failure mechanisms. In addition, the
HI method is the only method that considers the effect of external factors, such as water, that can
trigger failure.
The analysis results have revealed a noticeable contrast between the discrete and
continuous functions, whether in the RMR or in the SMR’s methods. As noted in Figure 3.6, the
RMR scores increase when calculated with the continuous functions. This is because in the discrete
RMR, the evaluation process for basic RMR parameters is based on step functions, which can
increase the level of subjectivity. The advantage of the continuous RMR is that it gives a distinctive
score for each RMR parameter, thus it can reduce subjectivity.
The same can be said for the continuous SMR, where the scores increase compared to those
obtained from the discrete functions such as original SMR, Chinese SMR and graphical SMR. The
original and graphical SMR methods show similar results, and the classes of rock slope assessment
are almost the same. This similarity is not surprising as the underlying concepts used in the two
methods are the same for the correction factors F1, F2 and F3.
The Chinese SMR, a modified version of the original SMR, adds two factors, namely the
height factor (ζ) for slopes taller than 80 m and the discontinuity condition factor (λ). When
applying in this study, the height factor is eliminated because all road cuts in this study are less
than 80 m high. The Chinese RMR ratings are found to be higher than the original SMR. This is
due to the discontinuity condition factor (λ), especially for discontinuities with favorable
conditions. It can be concluded that the Chinese SMR tends to be less conservative than Romana’s
SMR.
65
The effect of the angle between the slope face and the critical failure surface in case of
plane failure (or the intersection line in wedge failure) is obvious in the values of discrete SMR
methods.
66
Table 3.13: Stability analysis results for toppling failure using rock slope classification methods.
Hazard Index
Slope face Surface Normal condition Triggering factor
Site# Structure GSI (2010)
(dip ()/dip direction ()) condition (𝑓𝑁 ) (𝑓𝑇𝑀 )
HI Category
8 60/010 Disintegrated Very poor 10 10 7.1 8.5 Poor
9 55/060 Very blocky Very poor 35 10 7.1 8.5 Poor
Disturbed
15 66/095 Very poor 17 10 7.1 8.5 Poor
/Seamy

21
56
67
When the angle is less 10, the SMR scores become very low, suggesting an
underestimation of the stability condition. This scenario can be seen in sites 1, 4, 13.2 and 22 for
planar failure (see Figure 3.11) and in sites 4 and 13.2 (see Figure 3.12) for wedge failure.
Furthermore, if the quality of the rock mass in discrete RMR is classified as “poor” or
DRMR ≤40, the discrete SMR score may be negative (DSMR < 0), as are the cases in sites 18 and
19 (see Figure 3.12). However, the continuous functions of SMR do not have this peculiarity and
the total score is always a positive quantity regardless of the value of the said angle. This may
suggest that the discrete SMR is not appropriate for slopes with weak rock mass, especially when
the angle between the slope face and the critical joint surface is less than 10. It also suggests that
the continuous SMR might be considered as a better method for stability assessment of rock slopes
amongst the SMR methods.
All SMR methods utilized in this study apply only to structurally controlled failure
mechanisms. The HI method (Pantelidis, 2010) is applicable to both structurally and non-
structurally controlled slopes, which considers the normal condition and the triggering mechanism.
Note that the triggering mechanism does not exist in all SMR methods. Also, the HI method
considers block toppling-type failure mode, which SMR methods do not account for. It is for these
reasons that the HI method can be considered as the preferred empirical method used in this study.
Although the HI method is superior to other empirical methods, it has some limitations.
Firstly, the notion of “hazard” suggests the use of statistical methods to estimate the likelihood of
occurrence based on the probability distribution function of random variables such as the joint
properties. In the HI method, the normal condition (𝑓NC ) function is calculated with a deterministic
approach, and hence it does not consider the inherent uncertainty of the basic input parameters like
joint dip and dip direction.
Furthermore, the HI method relies on the interpretation of field observations and linking
such observations to the descriptions reported in Table 3.7 to determine the triggering mechanism
function (𝑓TM ) due to water and slope drainage condition. The approach is subjective and could
lead to wide variation of results depending on the observer’s interpretation. A more robust
approach is to derive a range of (𝑓TM ) values based on characteristic cases with numerical
simulation while accounting for the ground water effect, especially in the case study area where
68
most failures are known to occur shortly after heavy rainfalls, i.e. under undrained condition. The
HI method describes the slope stability as good, fair and poor - a classification which is more often
suited to describe the rock mass quality. Rather, it would be more representative to adopt
descriptions such as high, moderate, and low hazard for slope stability as done by others
(Anbalagan, 1992; Pachauri and Pant, 1992).
Finally, it is worth mentioning that all empirical methods used in this study share another
shortcoming, namely they are all developed for linear structures. They are not designed for
mountainous roads which are often curved and result in rock cuts of concave and convex surface
shapes. The introduction of an adjustment factor to account for the slope face curvature would
make a valuable improvement to the existing empirical methods.
3.8 Conclusion
This chapter examines the rock mass classification systems, especially those developed for
the assessment of rock slope stability. Five empirical methods, original SMR, Chinese SMR,
continuous SMR, graphical SMR and HI method, are applied to 22 sites of rock cuts located in a
rugged area along mountainous roads in the southwestern part of the Kingdom of Saudi Arabia.
The following conclusions can be drawn from this study.
SMR methods consider the condition of the rock mass presented by the RMR system, and
the relationship between the dip and the dip direction of both the slope face and the joint. These
relationships are presented by the correction factors of F1, F2 and F3 as well as the effect of the
excavation method F4. SMR methods can only be applied to slopes that exhibit structurally
controlled failure mechanisms.
On the other hand, the HI method considers the influence of the normal rock cut condition
and the triggering factors (e.g. the presence of water). It can be applied to both the structurally and
non-structurally controlled failure mechanisms.
The original and graphical SMR methods show similar results and the resultant stability
conditions are almost the same because the underlying calculation formulae of F1, F2, F3 and F4
are the same. However, application of the technique is different.
69
The continuous functions, such as the continuous RMR and continuous SMR, are relatively
less subjective classification systems because they give specific scores for each input parameter
value. However, the effects of triggering mechanisms such as water and earthquakes are absent in
the SMR methods.
The Chinese SMR method is not suitable for rock cuts below 80 m, where the effect of
slope height is ignored. In the case of the rock slopes located in the study area, all rock cuts are
less than 80 m high.
The HI method is as good as continuous SMR, which does not underestimate nor
overestimate the stability evaluation. However, the HI method is superior as it takes into account
the effect of water, and this is observed in the comparison of analysis results of HI method with
SMR methods.
In conclusion, the continuous SMR and the HI could be the proper methods for the stability
assessment of the rock cuts. These methods could be further improved by adding the factor of
slope height below 80 m, and by analyzing the effect of pore water pressure on the stability
behavior of a rock slope through incorporation of the numerical modeling technique with the
analysis and derived adjustment factors. The influence of slope curvature should also be
incorporated especially in mountainous areas such as the Jazan region in the Kingdom of Saudi
Arabia where the case study is carried out.
70
CHAPTER 4
PROBABILISTIC ASSESSMENT OF ROCK SLOPES STABILITY USING THE

RESPONSE SURFACE APPROACH-A CASE STUDY
4.1 Abstract
This chapter examines the stability condition of a jointed rock slope in the south western
region of Saudi Arabia using deterministic and probabilistic approaches, under both dry and wet
conditions. The study area is characterized by complex geology in rugged terrains. The stability
analysis is carried out using the code FLAC3D to generate a 3-dimensional, ubiquitous joint
model, to determine the influence of the dominant, unfavourable discontinuity orientation with
respect to the slope face. The deterministic analysis is first implemented using the mean values
of the selected random variables, namely the dip, dip direction and friction angle of the dominant
discontinuity set, and the stability condition is assessed with a factor of safety based on the
classical frictional joint constitutive model. A Box Behnken Design (BBD) approach is then
adopted to create the surface response function as a second order polynomial for the factor of
safety. To do so, fifteen FLAC3D models are generated in accordance with the BBD. Based on
this, 10,000 simulations of different slope realizations are carried out using Monte-Carlo
simulation technique, and the probability of unsatisfactory of performance of the rock slope is
assessed. It is shown that the probabilistic approach provides more insight and confidence in the
stability condition of the rock slope, both under dry and steady state heavy rainfall conditions. A
discussion is presented on the significance of accepting lower safety factors when heavy rainfall
conditions are encountered.
Keywords: Rock slope stability, Box-Behnken Design, Probabilistic analysis of slopes, Response
surface.
71
4.2 Introduction
Discontinuities are a considerable source of uncertainty and variability in engineering

design of rock slopes. Rock slopes are often found in complex and rugged terrains such as
mountainous areas, where strength and geometric properties of the discontinuities are subjected
to change within a rock slope (Duzgun et al., 2003; Einstein, 2003; Fisher and Eberhardt, 2012;
Hadjigeorgiou and Harrison, 2011; Hammah et al., 2009; Jimenez-Rodriguez et al., 2006; Jaeger,
1971; Park et al., 2012).
Due to these variations in both the strength and geometric properties of discontinuities in
rock slopes, the field measurements of their geometric and mechanical characteristic are subject
to variation in terms of accuracy and in the error ratio (Hammah et al., 2009; Johari et al., 2013;
Koyama and Jing, 2007; Shen and Abbas, 2013). As a result, rock slope design studies involve a
lot of uncertainties. The origin of these uncertainties in the geotechnical data and analysis can be
from several sources, for instance, the methods used in the data acquisition, the analysis
techniques, the judgment that depends on the experience of the engineer, the methodology used
in the study and other sources, all of which could influence the output results (Baecher and
Christian, 2003; Morgenstern, 1995).
In the past few decades, many studies have considered the effect of the inherent
variability of geotechnical properties in rock masses onto the stability conditions of rock slopes
(Griffiths et al., 2009). This type of studies relies on determining a range of stability factors
(factor of safety) to describe the stability behaviour of a rock slope by incorporating different
combinations of controlling factors, such as cohesion and friction angle to be used as input
parameters (independent variables), with numerical modelling, and then identifying the most
influential factors that play a significant role on the output or factor of safety. This type of
method is known as the probabilistic approach. It is more appropriate than the deterministic
approach, which is based on the analysis of the mean value of the input variables to derive a
single value of the output or factor of safety.
72
Numerous probabilistic methods have been proposed in the field of geotechnical
engineering specifically in slope stability studies. Most of these methods have been developed to
address the influence of the inherent variability of strength properties for slope stability
assessment in a realistic manner (Park et al., 2012; Daftaribesheli et al., 2011; Duncan, 2000;
Davis and Keller, 1997; Griffiths and Fenton, 2004).
Some studies have also discussed the influence of geometric parameters, particularly, the
characteristics of joint networks, which have been successfully incorporated in the stability
assessment of rock slopes (Karami et al., 2007). In addition, the impact of different joint surface
properties on the overall stability condition of rock slopes has been captured and analyzed
(Dershowitz et al., 1985).
Hammah et al. (Hammah et al., 2009) discussed the application of probabilistic approach
with numerical modelling. Two probabilistic approaches have been adopted in the analysis, the
point estimate method (PEM) and Monte-Carlo technique, to evaluate the probability of failure
of a rock slope. This analysis takes into considerations the influence of joint network on different
stability conditions. The study concluded that the PEM gives good results for probability
assessment with reasonable computational cost, but the method suffers from curse of
dimensionality (increased number of variables). On the other hand, the Monte-Carlo method is a
powerful tool and more flexible to incorporate a wide range of probability distributions without
being affected by the curse of dimensionality.
Brideau. et al. (Brideau. et al., 2012) investigated the natural variability of the geometric
parameters in joint surfaces such as persistence, spacing and orientation on the stability
conditions of a rock slope. The authors used a 3D distinct element code for stability assessment.
The results show that the joint surfaces, with unfavorable persistence (fully extended) and both
fixed spacing and orientation values do not always show a conservative solution.
A response surface approach was also presented as an efficient tool to identify the
likelihood of the instability behaviour of a rock slope (Shamekhi and Tannant, 2015). This
approach has been applied using the Central Composite Design (CCD), which is one of the
statistical design methods used to implement experiments to examine the effects of the main
interactions of different levels of the independent variables on the resulting response (dependent
73
variable). As a result, an equation for the response, i.e. factor of safety, is established as a
function of the design variables (independent variables) of the response surface. The resulting
mathematical model or the response equation is used to estimate the probability of the
unsatisfactory performance in the rock slope (Hill and Hunter, 1966). This technique is efficient
because it can be used to generate a wide range of output values (dependent variables). i.e.,
factor of safety along with Monte Carlo technique, with low computational effort. This is
particularly useful when the input variables include geometric parameters, such as, the
orientation of discontinuities, where the technique allows for the change of the inputs without the
need to re-construct the model itself many times.
In general, the stability condition in the rock slopes with joint network is significantly
controlled by rock structures. This kind of failure mechanism is known as structurally controlled
failure. These rock structures are mainly represented by discontinuities, joints, bedding planes,
faults, etc. The geometric aspects, such as dip and dip direction, and the strength properties, such
as friction angle and cohesion of the discontinuous surfaces, are the critical parameters that play
the important role in the instability of rock slopes on road construction projects (Hoek and Bray,
1981; Kostić, 2017; Priest and Hudson, 1976; Park et al., 2005; Sturzenegger and Stead, 2009;
Wyllie and Mah, 2004).
In this chapter, the stability condition of an excavated rock slope in mountainous areas
using the response surface methodology is assessed. The analysis of a rock slope model with a
single joint set as a representative of the dominant joint is adopted. Three parameters, dip, dip
direction and friction angle for the rock joint surface are considered probabilistically, the extent
of the joint is assumed to be infinite (persistence =1), and the cohesion is assumed to be zero
with no infilling materials. The study area is exposed to very highly intense seasonal rainstorms
during certain time periods of the year and most of the rock failures occur in rain times (Basahel
and Mitri, 2017). Thus, we evaluate the probability of unsatisfactory of performance for the rock
slope in the case of dry condition and then consider the effect of pore water pressure. We analyze
both conditions of the rock slopes and compare their stability behaviours to identify the critical
situations. Consequently, the remedial measures and means of mitigations are suggested based
on the estimate of the critical factor of safety calculated in different weather conditions in the
study region. The Box Behnken Design (BBD) is selected as a response surface design to study
74
the effect of the variability of the three independent variables on the resulting output variable i.e.
factor of safety. The BBD is a three-level design used to establish an equation to predict the
factor of safety in the form of a second order polynomial. This approach is appropriate for the
case that involves three factors and usually gives accurate results like the CCD but with less
computational time (Box et al., 2005; Montgomery, 2013).
4.3 Location and Geological Sitting of the Study Area
The study area is in the south-western part of Saudi Arabia. It is situated between latitude
17˚ 12’ 00’’and 17˚ 15’ 00’’ N and between longitude 43˚ 02’ 00’’ and 43˚ 04’ 00’’ E (Figure
4.1). The selected case study is a road cut located on Road 12, which is one of the main and vital
roads linking the cities and villages in the region.
The study region is characterized by mountainous and difficult terrains. As a result, roads
in this region were constructed across very rugged terrain, where, most of the time, it resulted in
steep rock slopes with sharp slope angles and discontinuities that are almost parallel to the slope
face. These excavated road cuts experience serious instability problems, especially, during and
after rainstorms. Consequently, yearly rock slopes failures cause closure and damage to these
roads. Clearly, these situations threaten public safety of road users and result in injuries and even
deaths at times.
Geologically, the selected rock slope is in Fayfa Mountain (Jabal Fayfa) in Jazan region,
which is covered by fractured syenite of Precambrian age; Jabal Fayfa is part of the Arabian
shield that contains Proterozoic layers rocks. Proterozoic layers rocks are originally
metamorphosed to green-schist facies and locally metamorphosed to amphibolite or granulite
facies (Baamer et al., 2010). These are known as Sabya formation (Figure 4.2). The Sabya
formation houses the oldest rocks; they have been subjected to many tectonic movements that
resulted in deformation, metamorphism and alteration processes (Fairer, 1981). It consists of
many types of rocks such as, green-schist, quartz-biotite schist, quartz-sericite schist, amphibole
plagioclase hornfels, hornfels, phyllite, black slate quartzite and marble (Baamer et al., 2010).
Structurally, many faults have been detected and observed in the study area to tend north-
west (NW) and west (W).
75
Figure 4.1: Location map
Figure 4.2: Geological map of the study area
76
4.4 Methodology
The study involves several phases of analysis. It begins by defining the suitable case
study and identifying the parameters that control the instability condition. These parameters are
the dip, the dip direction and the friction angle of the discontinuity surface.
First, compute the three statistical moments (mean, standard deviation and skewness) for
each parameter from which can calculate the point estimates. Then, use these estimate points
with the mean value for each parameter to create realizations of the road cut by means of Box
Behnken design. Subsequently, compute the shear strength reduction factor (SRF) value for each
realization using FLAC3D code (Itasca, 2012). Identify the contribution and significance of each
parameter via the Analysis of Variance (ANOVA). Then, generate an adequate prediction model
for estimating the response (SRF). Finally, Monte-Carlo sampling technique is used to run
10,000 slope realizations and the probability of unsatisfactory of performance for the rock slope
is estimated for both dry and wet rock slopes.
4.5 Field Measurements
The selected case study, as mentioned earlier, is a rock slope located on Road 12 in Jazan region,
south-west of Saudi Arabia. Basically, the rock slope consists of Amphibolites rock with
moderate state of weathering. The rock slope contains three main joint sets and the potential
failure mechanism is structurally controlled. A detailed field investigation at the selected site has
included the following.
1. Determining the geometry of the rock slope, slope height, width and inclination angle of
the slope face.
2. Measuring the dips and dip directions of 40 joints.
3. Collecting the required measurements of rock mass parameters for the rock mass
classification (RMR system).
4. Collecting all measurements needed for determining the strength properties for different
joint set surfaces, such as, joint roughness coefficient and Joint wall compressive
strength.
5. Collecting rock samples for laboratory testing to examine the strength of intact rock using
the point load test.
77
The rock slope extends 50 m wide, approximately 20 m in height and the slope face tilts at an
angle of 60˚ with an azimuth angle of 040˚ (north-east). The roadway is 10 m wide and is
located directly below the rock slope with no ditch at the toe of the slope (Figure 4.3). Moreover,
the slope face is left without any support.
The rock mass of the selected rock slope is evaluated and categorized according to RMR system
developed by Bieniawski (Bieniawski, 1989). The RMR contains five parameters known as a
basic RMR; Intact rock strength (UCS), Rock quality designation (RQD), discontinuities
spacing, conditions of discontinuities and Groundwater.
Figure 4.3: Case study
78
4.6 Random Variables
Three parameters, namely dip, dip direction and friction angle, of the most vulnerable
surface, have been chosen as random variables. The selected parameters are considered
probabilistically in the analysis. Other data such as rock mass properties and slope dimensions
have been kept with their deterministic values.
The friction angle is estimated according to the relation of Barton and Choubey (Barton
and Choubey, 1977), as follows.
∅𝑟 = (∅𝑏 − 20° ) + 20(𝑟⁄𝑅 ) (4.1)
Where, ∅𝑟 is the residual friction angle, ∅𝑏 is the basic friction angle, 𝑟 is the rebound
on weathered rock surface and 𝑅 is the rebound on un-weathered rock surface. The values of 𝑟
and 𝑅 are estimated using the Schmidt Hammar (L-type) by measuring the hardness of the rock.
Forty readings of dip and dip direction of different joint sets surfaces were measured
manually using Brunton compass. These readings have been plotted using the stereographic
projection method using a software DIPS v. 6.0 (Rocscience Inc, 2014) for kinematic analysis.
The analysis has been done in a lower hemisphere projection. All poles and great circles that
represent the different orientations of discontinuities sets are shown in Figure 4.4.
As shown in Figure 4.4, fourteen values of dip and dip direction that belong to set #1 are
clustered within the critical zone of planar failure region (red area). Hence, these critical values
of dip, dip direction and friction angle, estimated from Equation (4.1), are considered as random
variables (independent variables or inputs) in the modelling and the probabilistic analysis. Table
4.1 shows the data for joint set # 1.
4.7 Point Estimate Method
The Point Estimate method (PEM), proposed by Rosenblueth (Rosenblueth, 1975), is one
of the most popular statistical methods used in the field of geotechnical engineering problems
because it is simple, accurate (Christian and Baecher, 2002; Christian and Baecher, 1999) and it
takes the numerical analysis uncertainty into account, (Harr, 1987; Miller et al., 2004; Morales
and Perez-Ruiz, 2007; Wolff, 1996). For a given set of data and for each input parameter, the best
fit distribution curve is determined.
79
Figure 4.4: Stereographic projection of structural data for discontinuity sets of the case study.
Then, the first two statistical moments, the mean value (μ) and the amount of dispersion
before and after the mean (standard deviation, σ), are extracted from this distribution. Sometimes
the third moment (skewness, s), which describes the shape of the distribution is considered, if
found. The resulting statistical moments for a known distribution are then used in a specified
mathematical function to calculate the point estimates (High “+” and low “-”) for each
parameter. Then, the point estimates are used in conjunction with another technique (i.e. Box-
Behnken Design) to establish a second order polynomial expression. Such polynomial is used to
calculate the response function (Y). In this study, Y represents the factor of safety (FOS) or,
alternatively, the shear reduction factor (SRF) and considered as the dependant variable or the
output in the modelling.
In this study, the PEM is applied in two scenarios based on the statistical moments of the
random variables 𝑥𝑖 , where 𝑖 is the variable index; the scenarios are described as follows.
a) If the response function (Y) is a function of a random variable 𝑥𝑖 (the input) and the three
statistical moments (mean “μ”, standard variation “σ” and skewness “s”) of 𝑥𝑖 are
known, the point estimates, that represent 𝑥𝑖 , are calculated using the following equation
(Hong, 1998).
80
𝑠 𝑠 2
𝑥𝑖± = 𝜇𝑥𝑖 + [ 2𝑥𝑖 ± √1 + ( 2𝑥𝑖 ) ] 𝜎𝑥𝑖 (4.2)
b) If the distribution of 𝑥𝑖 is symmetrical and approximately Gaussian, the point estimates

are calculated using the following equation (Rosenblueth, 1975):
𝑥𝑖± = 𝜇𝑥𝑖 ± 𝜎𝑥𝑖 (4.3)
Table 4.1: Data of dip, dip direction and friction angle for joint set # 1
No. Dip Dip Direction Friction angle

1 (°)
51 (°)
040 (°)
28
2 52 041 23
3 50 045 26
4 49 046 26
5 49 045 28
6 50 043 21
7 54 042 31
8 50 044 31
9 51 043 25
10 55 040 28
11 52 043 30
12 53 046 26
13 49 045 25
14 57 044 29
4.8 Box-Behnken Design (BBD)
BBD is one of the statistical experiments used for the analysis of the problems that are
influenced by several independent random variables. In this method, the effects of each variable
and their interactions on the resulting response are determined. As a result, a mathematical
function to estimate the response function (Y) is derived. For example, BBD is used for the
estimation of the factor of safety (FOS) for a wide range of independent variables and to estimate
the probability of unsatisfactory performance for the response (Hammah et al., 2009; Wong,
81
1985). Also, this kind of statistical experiments are used usually for the purposes of the
optimization of the response characterization to get an optimal design (Montgomery, 2013).
BBD is a rotatable design that involves three-level of design points 3k, where k is the
number of random variables. Unlike the two-level design, 2k, that produces a linear response
function and cannot capture the nonlinear effect of input variables on the resulted response Y,
BBD is considered a highly efficient design for estimating the second-order response surface that
is able to capture the curvature in the system for a given independent variable. The design is
shown geometrically in Figure 4.5 (Montgomery, 2013).
A typical Box-Behnken design table for three-random variables is shown in Table 4.2.
The number of runs is thirteen, which covers all variable level treatment combinations. Besides,
the center point is treated three times. This could be advantageous because the design is rotatable
by nature, and the repetition of the center point makes the variance of predicted response
constant throughout the region of interest and provides a good prediction (Box and Hunter,
1957).
Figure 4.5: Geometric representation of a three-variable Box-Behnken design.

BBD can be summarized as follows. A set of input variables is given into the design
frame, each variable is defined by three levels of weighting values, (+1) represents the maximum
82
level of the variable, (-1) represents the minimum level and (0) represents the mean level (Box
and Behnken, 1960). The number of experiments, that is required to develop a BBD, is
calculated by the following expression.
𝑁 = 2𝑘(𝑘 − 1) + 𝐶0 (4.4)
Where, 𝑘 is the number of random variables and 𝐶0 is the number of center points.
Table 4.2: A typical Box-Behnken Design for three random variables.
Run X1 X2 X3
1 -1 -1 0
2 -1 1 0
3 1 -1 0
4 1 1 0
5 -1 0 -1
6 -1 0 1
7 1 0 -1
8 1 0 1
9 0 -1 -1
10 0 -1 1
11 0 1 -1
12 0 1 1
13 0 0 0
14 0 0 0
15 0 0 0
As mentioned above, in the simplest case, which is the two-level design (2k) with each
variable treated in two levels of design points (represented by +1 and -1, high level and low
level, respectively), the response is approximated to be linear over the range of variable levels
chosen. This induces a first-order response surface that is defined using the following relation:
𝑌 = 𝛽0 + 𝛽1 𝑥1 + 𝛽2 𝑥2 + ⋯ + 𝛽𝑘 𝑥𝑘 + 𝜖 (4.5)
Where, 𝑌 is the response, 𝛽𝑖𝑗 is the regression coefficient, 𝑥𝑘 is the independent variable
and 𝜖 is a random error. On the other hand, the three-level design 3k that is like Box-Behnken
design, allows to model the possible curvature in the response function and to handle the case of
random variables at three levels. The third level in this design facilitates the investigation of a
83
quadratic relationship between the response and each of the random variable (Myers and
Montgomery, 2002); this relationship is described as following:
𝑌 = 𝛽0 + 𝛽1 𝑥1 + 𝛽2 𝑥2 + 𝛽12 𝑥1 𝑥2 + 𝛽11 𝑥12 + 𝛽22 𝑥22 + 𝜖 (4.6)
The input variables are treated randomly in the design, where the main and interaction
effects between different levels of all input variables and their influences on the resulting
response (Y) are examined and identified. At the end, a polynomial function for the response (Y)
is obtained. This function contains multiple terms of the random variables to estimate the
interested response, which in turn, used for the objective of prediction purpose of the probability
of the risk of failure.
4.9 Monte-Carlo Simulation Technique (MCS)
The Monte Carlo simulation technique is considered as a very powerful tool for the
engineer who has only a basic working knowledge of probability and statistics for evaluating the
risk or reliability of complex engineering systems (Haldar and Mahadevan, 2000). Monte Carlo
method is used in a wide range of engineering projects that deal with uncertainty. It generates a
random number for each of the random variables in the problem and it makes combinations
amongst all these random variables to perform several deterministic computations (Hammah et
al., 2008). The accuracy of the MCS depends on the number of simulations performed, and it
increases with the increase of the number of simulations.
The MCS consists of sampling a set of materials from their joint probability distributions
function (PDF) and introducing them into the model. As a result, a set of values represent the
response (i.e. factor of safety) is generated from MCS analysis and the information on the
distribution and moments of the response variable are then obtained (Mellah et al., 2000).
There are six essential elements that form the Monte Carlo simulation (MCS) technique,
which are described as follows (Haldar and Mahadevan, 2000).
1. Defining the problem in terms of all random variables.

2. Quantifying the probabilistic characteristics of all the random variables and the
corresponding parameters.
84
3. Generating the values of these random variables.
4. Evaluating the problem deterministically for each set of realization of all the random
variables.
5. Determining probabilistic information from such realizations.
6. Determining the accuracy and efficiency of the simulation.
The MCS has some advantages such as, flexibility in incorporating a wide variety of
probability distributions without much approximation, and the ability to readily model
correlations among variables. In contrast, as the accuracy of MCS increases by increasing the
number of the simulations this can be disadvantageous because it becomes computationally
expensive (Haldar and Mahadevan, 2000). However, in this study, the MCS is used only to
generate values for the random variables, while, the response has been found using BBD and the
response surface approach. This greatly reduces the computational cost and allows to increase
the number of observations and hence the accuracy of the results.
4.10 Shear Strength Reduction (SSR)
Factor of safety (FOS) considers the failure criterion most commonly used to assist the
stability behaviour of rock slopes. It can be determined for any slope parameter of interest, such
as water level, slope height and strength property, by taking the ratio of the parameter’s actual
value under a condition to its value under the critical condition. The critical condition is
identified as the initiation of a hazardous state such as a slope failure.
The acceptable FOS value depends on the nature of the problem being investigated.
Service life, importance and complexity of the project are some of the factors that dictate the
choice of an appropriate FOS. For example, acceptable FOS in civil engineering projects are
usually higher than those of mining engineering, and this is because the anticipated service life of
civil engineering works is usually longer than mining ones. For example, an inter-city roadway is
likely to last over 100 years, whereas an open pit mine lasts generally between 5 and 50 years.
The present study is of a rock slope found in complex terrains and is subjected to heavy seasonal
rainstorm. A slope failure along an inter-city roadway like Road 12 of the case study, would
create much concern to traffic safety. In light of these points, a threshold FOS of 1.5 was
adopted, below which the stability of the rock slope would be considered “unsatisfactory”.
85
FOS can be calculated numerically using a variety of methods; the most popular methods
are limit equilibrium (LEM), limit state and shear strength reduction (SSR). In this paper, the
shear strength reduction (SSR) is used because it is commonly used in conjunction with
numerical modelling in the field of geomechanics particularly in slope stability problems
(Dawson et al., 2000; Dawson et al., 1999; Griffiths and Lane, 1999 and Hammah rt al., 2005).
SSR calculates the SRF value, which represent the FOS value in the analysis by progressively
reducing the shear strength of the materials until the failure state is found.
The stability condition of the case study is represented by the structurally controlled
failure mechanism. Therefore, the geometric and strength properties for the potential failure
surface should be considered. Based on that, the ubiquitous – joint material is selected in this
study, where the strength values for intact material, C and Ø, and for the ubiquitous joint 𝐶𝑗 and
∅𝑗 are used in the analysis. The equation to calculate the reduction factor for the ubiquitous –
joint material will be as follows:
1
𝐶𝑗𝑡𝑟𝑖𝑎𝑙 = 𝐹𝑡𝑟𝑖𝑎𝑙 𝐶𝑗 (4.7)
1
∅𝑡𝑟𝑖𝑎𝑙
𝑗 = 𝑎𝑟𝑐𝑡𝑎𝑛 (𝐹𝑡𝑟𝑖𝑎𝑙 𝑡𝑎𝑛∅𝑗 ) (4.8)
Where 𝐹 is the factor of safety, 𝐶𝑗 is the cohesion for the joint surface and ∅𝑗 is the
friction angle for the joint surface.
In the ubiquitous joint model, additional factors can be controlled, by the user, besides
the strength parameters, (𝐶𝑗 , ∅𝑗 ), these are the geometric parameters of the potential failure
surface, the dip and dip direction. The change in these geometric parameters could play a
significant role on the results of the SRF values, especially, if the analysis implemented in three-
dimension numerical model code (i.e. FLAC3D).
4.11 Analysis of Variance (ANOVA)
When all numerical modelling simulations for rock slope realizations needed for BBD are
completed and their corresponding responses (SRF) are obtained, the treatment combinations for
all levels of random variables along with their responses are performed to establish the functional
relationship between the independent variables and the response. This has led to the approximate
86
proper response surface, which is used for predicting the probability of failure, as stated in
section 4.8.
Through the BBD analysis, the main effect of all random variables, their interactions and
their contributions on the variability of the obtained response is determined using a statistical
technique called the Analysis of Variance (ANOVA). This technique detects the differences
between the levels of the random variables and measures the amount of deviation of the
response, resulted from each treatment combination in the analysis, from the overall mean of the
responses. In other words, the Analysis of Variance is used in this study to identify the
considerable (or most influential) random variables that impact the response (i.e. SRF).
Each treatment combination between different levels of parameters in BBD is called an

observation. Any statistical experiment performed in the BBD involves several observations and
this set of observations is known as a model, and can be expressed using the equation below:
𝑖 = 1,2, … . . , 𝑎
𝑦𝑖𝑗 = 𝜇 + 𝜏𝑖 + 𝜖𝑖𝑗 { (4.9)
𝑗 = 1,2, … . . , 𝑏
𝑦𝑖𝑗 is the 𝑖𝑗 𝑡ℎ observation, 𝑎 and 𝑏 are the treatments, 𝜇 is the overall mean, 𝜏𝑖 is the
effect of the 𝑖 𝑡ℎ parameter level and 𝜖𝑖𝑗 is a random error; this model is also known as the effect
model (Montgomery, 2013).
The ANOVA technique works on testing the effect of each treatment combination of
parameter levels on the resulted response by means of the application of statistical hypotheses.
Two hypotheses are examined to test the treatment effect 𝜏𝑖 . The equations below (4.10-4.11)
show the two hypotheses tests for one factor:
𝐻0 = 𝜏1 = 𝜏2 = ⋯ = 𝜏𝑎 = 0 (4.10)
𝐻1 = 𝜏𝑖 ≠ 0 for at least one 𝑖 (4.11)
Equation (4.10) represents a null hypothesis that assumes the treatment effect or the
interaction with the other parameter at all levels is equal to zero, which means that this parameter
87
does not contribute to the variability of the response. These hypotheses are examined for all the
random variables (parameters) included in the design.
Equation (4.11) represents an alternative hypothesis, which assumes at least one

treatment effect at a certain level or interaction is not equal to zero. This means that this
parameter or its interaction has a significant effect on the variability of the response; in this case,
the null hypothesis is not true and should be rejected.
The percentage contribution for all random variables on the obtained response can also be
calculated. To do so, it is necessary to compute the values of the sum of squares (𝑆𝑆) of each
random variable and their interactions as follows:
𝐶𝑜𝑛𝑡𝑟𝑎𝑠𝑡𝐴𝐵𝐶…𝐾 = (𝑎 ± 1)(𝑏 ± 1) + ⋯ (𝑘 ± 1) (4.12)

1
𝐸𝑓𝑓𝑒𝑐𝑡 𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑠𝐴𝐵𝐶…𝐾 = 𝑛2𝑘−1 (4.13)
1
𝑆𝑆𝐴𝐵𝐶…𝐾 = 𝑛2𝑘 𝑐𝑜𝑛𝑡𝑟𝑎𝑠𝑡𝐴𝐵𝐶…𝐾 (4.14)
The percentage contribution for each random variable is calculated using the following
equation:
𝑆𝑆𝐴𝐵𝐶…𝐾
𝐶𝑜𝑛𝑡𝑟𝑖𝑏𝑢𝑡𝑖𝑜𝑛 𝑝𝑒𝑟𝑐𝑒𝑛𝑡𝑎𝑔𝑒 = × 100 (4.15),
𝑆𝑆𝑡𝑜𝑡𝑎𝑙
Where, 𝐴𝐵𝐶 … 𝐾 are the parameters, 𝑘 is the number of parameters, 𝑛 is the number of
replications.
4.12 Model Setup
An initial three-dimensional numerical model has been constructed using FLAC3D code
(Itasca, 2012). This model represents the rock mass model without an excavation, the height is
120 m from the datum in z direction, the width is 50 m in x direction and it extends deeply 120 m
in y direction, as shown in Figure 4.6.
88
Figure 4.6: Initial model of the rock slope before excavation
The state of in-situ stress (𝜎 0 ) is assumed as follows: the vertical stress (𝜎𝑣0 ) in the z
direction is given by the overburden pressure as described in the relationship:
𝜎𝑣0 = 𝛾𝑧 (4.16)
Where, 𝛾 is the unit weight of the rock mass and 𝑧 is the depth below the ground surface.
The horizontal stress, (𝜎𝑥0 , 𝜎𝑦0 ), for the numerical model is also assumed to be a function in 𝛾𝑧. It
increases below the ground surface towards the model datum, with gradient of 𝛾. Since the rock
material is homogenous, the horizontal stress in x and y is assumed to be equal:
𝜎𝑦0 = 𝜎𝑥0 = 𝛾𝑧 (4.17)
The boundary conditions for all the sides of the model are restricted horizontally and
allowed to move only along the z direction. The model base can move horizontally only, and the
ground surface is kept free without any restrictions. The model is excavated from the top left side
89
to construct a rock slope with specific dimensions, as show in Figure 4.7. The slope is at an angle
of 60˚, with a slope height of 20 m and road width of 20 m.
Figure 4.7: The rock slope model after excavation
As can be seen in Figure 4.7, the model size has been extended 5 times the slope height
below the level of the roadway in the negative direction of z axis, and 5 times the roadway width
behind the excavated slope in the positive direction of y axis to avoid the effect of boundary
conditions on the numerical analysis results.
The rock mass properties and the properties of the potential failure surface that are used
in the analysis are presented in Table 4.3. The mechanical properties of the rock mass are
estimated from the Hoek Brown classification investigation for the rock slope site. The results of
this investigation have shown that the rock strength is 133 MPa as obtained from the point load
test, the density of the rock sample is 2.9 g/cm3, the rock slope surface condition is described as
very blocky and the surface quality is fair, the Geological strength index (GSI) for the rock mass
is estimated to be 55, and the mi constant estimation, that is related to the amphibolite rock, is
90
26. In addition, the disturbance factor has been taken as 1 for poor blasting and the effect of
slope height has also been considered in the calculation of the rock mass properties, where the
height used is 20 m.
Table 4.3: Rock mass and joint surface properties used in the numerical analysis
Rock mass properties
Rock Bulk Shear Tensile
Friction Cohesion
density modulus modulus strength
angle (°) (Pa)
(kg/m3) (Pa) (Pa) (Pa)
2900 2.852e9 1.316e9 45 3.32e5 3.7e4
Rock joint surface properties
Tensile
Friction Cohesion Dilation
Dip (°) Dip dir. (°) strength
angle (°) (Pa) angle (°)
(Pa)
52 043 27 0 0 0
A sensitivity analysis for mesh quality of the model has been conducted. Some test runs
have been performed to refine the grid size of the mesh and obtain an optimum grid for the
analysis. These tests are performed by running the simulation and changing the geometric ratios
of the successive zone sizes in y direction. The default ratio is 1 for all zones, but this ratio is
changed to 1.1 from the slope face to the back of the slope, where the zones sizes are decreased
around the slope and are increased towards the back of the slope by this ratio (1.1). Then, the
ratio is changed to 1.15 for the second run and 1.2 for the third and so on. The best mesh was
found at grid ratio of 1.4, where the displacement and the SRF values appear to converge (Figure
4.8).
The numerical analysis is done in both dry and saturated conditions. For the groundwater
analysis, a steady state fluid-flow is adopted as the state of the groundwater within the numerical
model. Two properties related to the fluid-flow are important to conduct the analysis; the first
property is the hydraulic conductivity of the joint 𝑘ℎ , which can be calculated using the
following expression (Barton and Quadros, 1997).
91
Figure 4.8: Mesh size corresponding to the conversions of displacement.
The numerical analysis is done in both dry and saturated conditions. For the groundwater
analysis, a steady state fluid-flow is adopted as the state of the groundwater within the numerical
model. Two properties related to the fluid-flow are important to conduct the analysis; the first
property is the hydraulic conductivity of the joint 𝑘ℎ , which can be calculated using the
following expression (Barton and Quadros, 1997).
𝑔𝑒 2
𝑘ℎ = 12𝑣 (4.18)
𝑘ℎ is the hydraulic conductivity of the joint in m/s, 𝑔 is the gravitational acceleration (9.81 m/s2),
𝑒 is the joint aperture (m) and 𝑣 is coefficient of kinematic viscosity of the fluid (m2/s).
The second property is the permeability coefficient 𝑘 which can be obtained from (Itasca,
2012):
𝑘ℎ
𝑘=𝜌 (4.19)
𝑓 𝑔
𝑘 is the permeability coefficient (m2/(Pa/sec), 𝑘ℎ is the hydraulic conductivity (m/sec), 𝜌𝑓

is the fluid mass density and 𝑔 is the gravitational acceleration. In case of model with single joint
set, an anisotropic flow is assumed, where the three permeability values are specified as 𝑘1 , 𝑘2 ,
and 𝑘3 , and the flow orientation is assumed to follow the joint attitude (dip/dip dir.). The
92
directions 𝑘1 and 𝑘2 are defined at the same plane of the joint surface, however, 𝑘3 is
perpendicular to the joint plane and assumed to be zero. Table 4.4 shows the necessary properties
for the steady state fluid-flow analysis in FLAC3D numerical analysis.
The groundwater surface extends from the slope face towards the slope surface; this
extension has been chosen to be 2 times the slope height behind the toe of the slope (Wyllie and
Mah), as shown in Figure 4. 9. The pore water pressure is applied on the upper-stream and the
down-stream of the slope model with pressure gradient based on the concept of 𝛾𝑤 𝑧 , where 𝛾𝑤
is the unit weight of water (Figure 4.10).
Table 4.4: Groundwater properties for steady state fluid-flow analysis in FLAC3D
analysis
Groundwater properties for steady-state flow analysis
Permeability Water Water
Flow dip Flow dip Water bulk
coefficient Porosity density tension
(°) dir. (°) modulus
(k1=k2) (kg/m3) limit
1.2e-9 52 043 0.3 1e3 1e3 0
4.13 Results and Discussion
4.13.1 Kinematic Analysis
The results of the kinematic analysis of the rock discontinuities using DIPs software
(Rocscience, 2014) indicate that a cluster of fourteen poles are concentrated in an area that is
almost parallel to the strike of the slope face. This in turn allows a potential of plane failure to
occur (the red region), the mean attitude (dip/dip direction) of this cluster has been determined as
set # 1 and its value is 52/040. Hence, the joint set # 1 has been selected as the critical joint,
because a potential of plane failure is most likely to occur along this joint surface (Figure 4.4).
As shown earlier in section 4.1, most of the parameters that contribute to the rock slope
failures are related to the rock discontinuities. Thus, set # 1 has been selected for further analysis
and its parameters represented by dip, dip direction and friction angle have been chosen as
random variables (see Table 4.1) and as a major source of uncertainties in the stability analysis.
93
Figure 4.9: Ground water flow model
Figure 4.10: Sectional view shows the pore water pressure distribution within the rock slope.
94
4.13.2 Point Estimates
According to the order shown in the methodology, after defining the set of all random
variables, the next step is finding the location of the sampling points within the experimental
design of BBD by calculating the point estimates. To do so, the best fit distribution curve has
been assigned to each group of random variables and the first three statistical moments for each
distribution (mean ‘μ’, standard deviation ‘σ’ and skewness ‘s’) have been identified. Figures
4.11, 4.12 and 4.13 illustrate the best fitted distribution for each set of random variables with the
corresponding statistical moments; Figure 4.11 shows the histogram and the exponential fit
distribution for dip values, Figure 4.12 shows that the best fitted distribution curve is a triangle
and Figure 4.13 shows the Normal distribution fit for the given values of the friction angles. All
statistical moments that have been determined from the fitted distributions are shown in Table
4.5.
Figure 4.11: Histogram and exponential fit distribution for dip values.
After that, the point estimates for each set of variables are calculated using equation (4.2)
for the dip and dip direction variables, and equation (4.3) for the friction angle variable. The
choice of the appropriate equation to calculate the point estimates is according to the existence of
the third moment (the skewness) in the fitted distribution. Table 4.6 shows the point estimates
95
calculated based on the statistical moments of the fitted distribution for each set of variables.
These calculated point estimates are used as the sampling points for each random variable with
known statistical distribution for the further analysis in BBD to properly estimate the statistical
moments for the response (factor of safety) and construct the response surface that
approximately describes the relationship between the random variables (inputs) with the
response (output).
Figure 4.12: Histogram and triangle fit distribution for dip direction values.
Table 4.5: The best fitted distributions and their corresponding statistical moments for
each set of random variables.
Variable Best fit distribution Mean (μ) Standard deviation Skenwess (s)
(σ)
Dip Exponential 52 2.6 2
Dip Direction Triangle 43 1.8 - 0.6
Friction Angle Normal 27 2.9 0
96
Figure 4.13: Histogram and normal fit distribution for friction angle values.
Table 4.6: The calculated point estimates for each set of variables
Variable Mean Point estimate (𝒙+
𝒊 ) Point estimate (𝒙−
𝒊 )
Dip 52 57 47
Dip Direction 43 46 40
Friction Angle 27 30 24
4.13.3 Numerical Modelling
A three-dimensional numerical model using FLAC3D software (Itasca, 2012) is

constructed and the material properties for the rock and the single joint set (set#1) are assigned to
the model with the mean values; this is the reference model in this study. A sensitivity analysis
for the model grid is performed to identify the optimal mesh which is adopted for the analysis.
Several runs have been conducted and the mesh size of the grid was changed in each run based
on the procedure mentioned in section 11 to refine the mesh until the values of the maximum
Displacement and SRF are converged (convergence < 10%). At the end, the suitable mesh size
was found to be 1.4 in Y direction and the SRF value to be 1.64; these values are selected for the
analysis (see Figure 4.8).
97
Thirteen numerical realizations for rock slopes have been simulated. These different
realizations represent the variability of the different levels of the input variables involved in the
BBD experimental frame. The SRF values have been computed as dependant variables
(responses) for each run in both the dry and the saturated conditions and are presented in Table
4.7.
The reference model shows that the amount of displacement in the initial model before
executing the excavation is approximately zero in the entire model as expected (Figure 4.14).
When the model is excavated, the maximum displacement is computed; it is found that the
displacement is 15 cm in dry condition and is 21.1 cm in the saturated condition and the
corresponding factor of safety (FOS) are recorded as 1.64 and 1.31, respectively (Figures 4.15
and 4.16). This in turn reflects the role of groundwater pressure in increasing the amount of
displacement, which led to a decrease in the safety factor to below the threshold safety limit 1.5.
Table 4.7: shows results of the factor of safety of dry and saturated conditions for all
treatments of slope realizations.
Slope Dip Friction SRF

Patterns Dip SRF (dry)
realizations Direction Angle (saturated)
1 --0 47 40 27 1.30 1.12
2 -+0 47 46 27 1.37 1.08
3 +-0 57 40 27 2.64 2.20
4 ++0 57 46 27 2.95 2.49
5 -0- 47 43 24 0.91 0.71
6 -0+ 47 43 30 1.31 1.14
7 +0- 57 43 24 2.78 2.32
8 +0+ 57 43 30 2.85 2.35
9 0-- 52 40 24 1.57 1.30
10 0-+ 52 40 30 1.65 1.44
11 0+- 52 46 24 1.78 1.38
12 0++ 52 46 30 1.87 1.48
13 000 52 43 27 1.64 1.31
For dry analysis, the highest value of the factor of safety is recorded in realization # 4 and
it is 2.95 (Figure 4.17), as shown in Table 4.7. While, the lowest value of the factor of safety was
found in realization # 5 and it is 0.91 (Figure 4.18). From these results, it is obvious that the
factor of safety value is highly affected by the dip angle of the joint; FOS values drop down by
decreasing the angle of dip and vice versa, as noticed in the Table 4.7.
98
Figure 4.14: Result of displacement in the reference model before an excavation.
in dry condition.
99
Figure 4.16: Shows displacement contours and simulated factor of safety for the
reference model in saturated condition.
Figure 4.17: Factor of safety for slope realization # 4 in dry condition
100
Figure 4.18: Factor of safety for slope realization # 5 in dry condition.
Also, from Table 4.7, we note that the factor of safety takes high values (>2.5) when the
value of joint dip is in the high level (+), as presented in slope realizations # 3, 4, 7 and 8.
Whereas, the factor of safety drops down to below (< 1.4), which is below the threshold safety
limit, when the joint dip value is in the low level (-), as shown in realizations 1, 2, 5, and 6.
Moreover, the factor of safety values remain above the safety limit (> 1.5), when the dip angle of
the joint has values that are equal to the mean values (0), as shown in realizations from 9 to 13 in
Table 4.7, regardless of the levels of the combinations of the two other factors.
However, it is found that when the dip direction is in the high level (+) and the friction
angle is equal to the mean value (0), the factor of safety is below 1.5, as seen in realization # 2,
where the FOS value is 1.37. In addition, when the friction angle is in the high level (+) and the
dip direction of the joint is equal to the mean value (0), the amount of FOS remains below the
safety limit 1.5, as shown in realization # 6 (FOS = 1.31).
The results of the factor of safety in the low-level treatment combinations are presented
in realizations 1, 5 and 9. The FOS value is below 1 in treatment # 5, where the dip and the
friction angle values are in the low levels, while the dip direction is kept at the mean value.
101
However, in treatment # 1, when the friction angle is equal to the mean value and both the dip
and the dip direction values are in the low levels, the FOS value increased to 1.3. But, when the
dip is at mean level and the other variables (the friction angle and dip direction) values are in the
low levels, the computed factor of safety increased to 1.57. This confirms that the dip variable
plays an active role on the computed FOS and that the friction angle slightly influences the
resulted factor of safety, as seen in treatment # 1.
The numerical modelling results for the analysis of the influence of the groundwater
pressure on the stability conditions, for all slope realization treatments, have revealed that all
values of the factor of safety in general have decreased as compared to their counterpart values in
the dry state. The reference model shows a clear drop in the value of FOS from 1.64 to 1.31 (20
% drop) and the stability condition transformed from a satisfactory condition (> 1.5) in the dry
state to a critical condition (< 1.5) in the saturated state.
Furthermore, in slope realization #11, as another example, the factor of safety has
dropped by 22.5%, where the FOS changed from 1.78 in the dry state to 1.38 in the saturated
state, as shown in Figure 4.19 and Figure 4.20. Here the dip angle is equal to the mean value
(52˚), whereas, the friction angle value of the joint plane is in the low level (24˚). Also, in slope
realization # 12, although the friction angle and the dip direction values are in the high level, the
factor of safety decreased by 21 % and the FOS value changed from 1.87 to 1.48. This reduction
in the FOS happened because the dip angle is equal to the mean (52˚).
As seen in Table 4.7, the results of all numerical simulations for the groundwater analysis
give FOS values below the limit of safety ,1.5, in the range (0.71-1.48), except the cases with the
dip variable located in the high level (+), where the FOS values remain above 2, despite the
presence of the water pressure.
Therefore, this gives an indication that the dip angle of the joint is very influential on the
variability of the factor of safety values, and that the high value of the dip angle could enhance
the stability of the rock slope. For instance, the analysis shows that the high level of the dip angle
is 57˚ and it is almost located at the border of the daylight envelop towards the stereonet center
(see Figure 4. 4), this is in turn reduces the likelihood of the failure and increases the stability.
102
Figure 4.19: Factor of safety for slope realization # 11 in dry condition
Figure 4.20: Factor of safety for slope realization # 11 in saturated condition
103
4.13.4 Box-Behnken Design
In order to predict the probability of unsatisfactory performance for the rock slope of the
case study, a response function has been established through the design of experiments approach
using the Box-Behnken design. As explained earlier in section 4.8, the BBD is used to establish a
quadratic relation between the inputs (random variables) and the output (response variable).
To establish this relation, the regression coefficient for the linear, the quadratic and the
interactions should be established to identify the contribution of each random variable. To do so,
the Analysis of Variance has been used and the contribution of each random variable has been
identified.
Figure 4.21 and Figure 4. 22 show the plots of the simulated values of SRF against the
predicted SRF values for both the dry and the saturated conditions, respectively. Form these
plots, we see that all the SRF predicted points fall within the confidence curve (red-dotted lines)
and close to the perfect fit (the diagonal line) and the results of the goodness of fit, 𝑅 2 , are 0.99
and both models (dry and saturated) are significant.
Table 4.8 and Table 4.9 show the results of ANOVA, they indicate that the dip variable is
a significant source of variability on the resulted response (FOS), where the percentage of its
effect is 88.6 % in the case of the dry condition and 86.1 % in the saturated condition.
When we compare the numerically derived SRF, from FLAC3D, and the probabilistically
predicted SRF values from BBD, we can clearly see the consistency between the results in terms
of the significant effect of the dip variable on the stability behaviour of the rock slope, which is
represented by the variability of the factor of safety, see Table 4.10 and Table 4.11.
From the BBD experiment, mathematical expressions have been established to describe
the relations between the random variables and the response (SRF), as presented in equations
(4.20) and (4.21) for both dry and saturated conditions, respectively. These relationships are used
to construct the response surfaces that used to predict the values of SRF for any slope
realizations based on the variability of the random variables, these functions are as follows:
104
𝑆𝑅𝐹 = 46.49535 + (−1.25885) × 𝐷 + (−1.04175) × 𝐷𝐷 + (0.3757222222) × ∅ +
(0.0134) × 𝐷2 + (0.01) × 𝐷𝐷2 + (−0.001388889) × ∅2 + (0.004) × 𝐷 ×
𝐷𝐷 + (−0.0055) × 𝐷 × ∅ + (0.0002777778) × 𝐷𝐷 × ∅ (4.20),
𝑆𝑅𝐹 = 47.996538889 + (−1.26015) × 𝐷 + (−1.002527778) × 𝐷𝐷 + (0.4311111111) ×
∅ + (0.01285) × 𝐷2 + (0.0101388889) × 𝐷𝐷2 + (−0.000138889) × ∅2 +
(0.0055) × 𝐷 × 𝐷𝐷 + (−0.006666667) × 𝐷 × ∅ + (−0.001111111) × 𝐷𝐷 × ∅
(4.21)
Figure 4.21: Plot of simulated by predicted SRFs for slope realizations in dry condition
Figure 4.22: Plot of simulated by predicted SRFs for slope realizations in saturated
condition
105
Where, 𝐷 is the dip Angle, 𝐷𝐷 is the dip direction, ∅ is the friction angle and 𝑆𝑅𝐹 is the shear
strength reduction.
Table 4.8: Results of ANOVA for the rock slope in dry condition
Sum of
Source of variability P-value Contribution (%)
squares
Dip 5.0086125 < 88.5
0.0001
Dip Direction 0.0820125 0.0256 1.45
Friction Angle 0.0512000 0.0557 0.90
Dip*Dip Direction 0.0144000 0.2453 0.25
Dip*Friction Angle 0.0272250 0.1302 0.481
Dip Direction*Friction 0.0000250 0.9584 0.0004
Angle
Dip2 0.4143692 0.0009 7.32
Dip Direction2 0.0299077 0.1164 0.53
Friction Angle2 0.0005769 0.8027 0.01
Table 4.9: Results of ANOVA for the rock slope in saturated condition
Sum of Contribution
Source of variability P-value
squares (%)
Dip 3.5245125 < 0.0001 86.1
Dip Direction 0.0171125 0.1043 0.42
Friction Angle 0.0612500 0.0133 1.50
Dip Direction*Friction 0.0004000 0.7740 0.0098
Angle
Dip2 0.3810519 0.0002 9.31
Dip Direction2 0.0307442 0.0307442 0.75
Friction Angle2 0.0000058 0.0000058 0.00014
3D response surfaces have been constructed using the previous mathematical

expressions, where each response surface plot explains the relation between two random
variables and the SRF, while the third variable is fixed at its mean value.
The plots in Figure 4.23 illustrate the response surfaces of the slope realizations in the dry
condition. Plot (a) shows the response surface for two variables, the dip and the dip direction
against the SRF, while the third variable (friction angle) is fixed at its mean value of 27˚. The
106
SRF values are directly proportional to the dip when the dip angle is higher than 50˚, whereas the
dip direction does not have a clear impact on the SRF value, which takes the values in the range
(1.2-1.4).
Table 4.10: Simulated and predicted SRFs for slope realization in dry condition
Slope Dip Friction Simulated Predicted

Patterns Dip
realizations Direction Angle SRF SRF
1 --0 47 40 27 1.30 1.23
2 -+0 47 46 27 1.37 1.32
3 +-0 57 40 27 2.64 2.69
4 ++0 57 46 27 2.95 3.01
5 -0- 47 43 24 0.91 1.01
6 -0+ 47 43 30 1.31 1.33
7 +0- 57 43 24 2.78 2.76
8 +0+ 57 43 30 2.85 2.75
9 0-- 52 40 24 1.57 1.54
10 0-+ 52 40 30 1.65 1.69
11 0+- 52 46 24 1.78 1.74
12 0++ 52 46 30 1.87 1.90
13 000 52 43 27 1.64 1.64
Table 4.11: Simulated and predicted SRFs for slope realizations in saturated condition

Patterns Dip
1 --0 47 40 27 1.12 1.10
2 -+0 47 46 27 1.08 1.02
3 +-0 57 40 27 2.20 2.25
4 ++0 57 46 27 2.49 2.51
5 -0- 47 43 24 0.71 0.78
6 -0+ 47 43 30 1.14 1.15
7 +0- 57 43 24 2.32 2.31
8 +0+ 57 43 30 2.35 2.28
9 0-- 52 40 24 1.30 1.26
10 0-+ 52 40 30 1.44 1.45
11 0+- 52 46 24 1.38 1.37
12 0++ 52 46 30 1.48 1.52
13 000 52 43 27 1.31 1.31
Plot (b) in Fig. 23 shows the constructed response surface for the dip and the friction
angle variables against the SRF values, while the dip direction is fixed at its mean value of 043˚.
107
The dip variable strongly affects the variability of SRF values, where it changes from 0.9 at dip
angle of 47˚ to 3 at dip angle of 57˚, this means that a change of 10˚ in the dip angle causes a
dramatic variation in SRF. However, the friction angle affects the SRF only slightly, where the
SRF values positively increase with the increase in the angle of friction.
Plot (c) in Figure 4.23 shows the response surface for the dip direction and friction angle
variables against the SRF values, while the dip is fixed at its mean value of 52˚. This relation
shows that both variables have relatively equal influence on the response. The dip direction gives
low SRF values when the joint strike is almost parallel to the strike of the slope face and its value
enhances when the value of the dip direction increases; the SRF changes from 1.5 at dip
direction value of 043˚ to nearly 1.77 at dip direction value of 046˚. Similarly, the effect of the
friction angle is limited, where the SRF values changes from 1.50 to 1.65 when the angle of
friction increases from 24 to 30; the amount of change in SRF is approximately 0.15.
Figure 4.24 shows the response surface plot of the dip angle and friction angle variables
against the SRF values in case of the saturation condition. The dip direction is fixed at the mean
value (043˚). It is clearly seen that the SRF values are significantly affected by the angle of the
joint dip and the relation is linear. Whereas, the plot also shows a limited effect of the change in
the friction angle on the SRF values, where SRF value changes from 0.7 to 1.2 when the angle
changes in the range (24-30). When we investigated the dip and the dip direction variables
against the SRF values, while keeping the friction angle fixed at the mean value of 27˚, we found
that the effect of the dip on SRF values is pronounced, especially, when the dip angle exceeds
50˚, where the SRF increases sharply. On the other hand, the variability of the dip direction does
not make any influence on the SRF results as they are almost constant values. We also observe
that the variabilities of the dip direction and the friction angle do not have a pronounced effect on
the SRF values.
Through previous observations of the BBD results, we conclude that the dip angle of the
joint surface has a significant and clear role and control on the stability behaviour of the rock
slopes, in contrast to the role of the other variables, the dip direction and the friction angle, which
have a minor impact on the stability of slopes.
108
Figure 4.23 (a): Response surface plot for dip angle and dip direction against SRF values
in the case of dry condition
Figure 4.23 (b): Response surface plot for dip angle and friction angle against SRF values
in the case of dry condition
109
Figure 4.23 (c): Response surface plot for dip direction and friction angle against SRF
values in the case of dry condition
Figure 4.24: Response surface plot for dip angle and dip direction against SRF values in the case
of saturated condition
110
4.13.5 Probability of Unsatisfactory Performance
The probability of the unsatisfactory performance to evaluate the stability behaviour of

the case study has been done. This evaluation has been made in the two conditions, the dry and
the saturated slopes. The Monte-Carlo technique has been implemented to generate 10,000
random values for the input variables, the dip, the dip direction and the friction angle, for the
critical joint set (set #1) based on the statistical moments and the distribution that has been
determined for each variable. The SRF value is calculated for each combination of these three
variables using equation (4.20) for the dry slope and equation (4.21) for the saturated slope.
Thus, 10,000 values have been generated for the SRF and the best fit distribution is found for
both cases, the dry and the saturated, which was found to be an inverse Gaussian distribution. As
aforementioned, the limit of satisfactory performance for the stability is defined to be, SRF = 1.5.
Consequently, the probability of unsatisfactory performance for the dry and the saturated slopes
are identified. Figure 4.25 shows the histograms and the inverse Gaussian distribution fit for the
calculated factor of safety (SRF) for the slope in the dry condition, the probability of
unsatisfactory performance is found to be 36.4 % from the probability curve. Figure 4.26 shows
the histograms and the inverse Gaussian distribution fit for the calculated factor of safety; from
this distribution curve, the probability of unsatisfactory performance for slope in saturated
condition is found to be 64.2 %.
4.14 Conclusion
The stability of rock slopes in mountainous areas, which are subjected to heavy seasonal
rains, is examined in this study. Due to the complexity of these areas, the rock mass discontinuity
properties could be a source of uncertainties. Thus, slope stability studies will not give a clear
answer if the deterministic approach is used. Consequently, the probabilistic methods are
adopted in this work to find out the main sources of the uncertainty and to take into consideration
the variability of several parameters and the effect of the groundwater pressure on the slopes.
Three parameters namely dip angle, dip direction, and friction angle of the potential
failure surface were chosen as random variables. The factor of safety was selected as the
indicator for the instability condition (response). The required measurements were collected for
these variables, and the proper statistical moments and distributions were defined. Box-Behnken
Design (BBD) was used to incorporate the statistical and numerical methods together in the
111
analysis. A 3D numerical model was constructed; it includes one joint set, which represents the
potential failure surface (set # 1) that was determined through the stereographic projection
methods.
dry condition
Based on the BBD design, fifteen treatments are generated to examine the effect of
different level combinations amongst the three random variables on the factor of safety. Using
the BBD experiments, polynomial functions for two cases of dry and saturated slopes were
established for calculating the factor of safety as a function of the random variables. The results
of BBD indicate that the key factor, which significantly influences the factor of safety, is the
angle of dip of the failure surface. The other factors, dip direction and friction angle did not show
clear effect on the variability of the factor of safety values. The results of the numerical
modelling agree with the results obtained from the Box-Behnken Design, where both analyses
show that the effect of the dip angle of the failure plane is dominant on the instability of the
slope.
112
Figure 4.26: Histograms and distribution fit for the calculated factor of safety for the
slope in the saturated condition
Finally, the probability of unsatisfactory performance was determined for the dry and the
saturated slopes. The analysis was done through Monte-Carlo simulation and the polynomial
expression was established from the BBD analysis. The results indicate that the probability of
instability increases by 27.8% in the case of saturated slopes. It is concluded that the situation of
the saturated rock slope must be taken into account during the design or in the recommendations
for remedial measurements, especially in regions similar to the study area of this paper.
113
CHAPTER 5
METHODOLOGY FOR THE DEVELOPMENT OF SLOPE FAILURE

INTENSITY FACTOR AND GEOTECHNICAL HAZARD INDEX
5.1 Abstract
Stability of rock slopes has been assessed by various versions of rock slope classification
methods in literature. Many factors have been considered in the stability assessment process, such
as the materials of the rock slope (normal condition), the strength and the geometrical properties
of both the rock mass and the joint surfaces network. Afterwards, the classification system of rock
slopes process has been developed by adding the effect of water as a triggering mechanism beside
the factor of normal condition, in order to insert the external force within the classification frame
of the stability of rock slopes.
The evaluation process of these factors to identify the stability behaviour of the rock slope
has been conducted by adapting two approaches. A conventional approach, which computes the
mean values of the material and the geometrical properties to get a single value for the stability
factor (factor of safety), and a probabilistic approach, which computes a range of variables for the
material and the geometrical properties, to produce a range of factor of safety values through which
the probability of failure can be estimated.
In this study, an intensity factor, which represents the effect of potential failure of the rock
slope on the roadway is presented. The intensity factor has been assessed by estimating the area of
potential failure surface on the slope face and how much it affects the roadway.
Ultimately, A geotechnical hazard description matrix has been introduced, and it is a

result of the product of the probability of unsatisfactory performance (𝑃𝐹) estimated from the
probability of potential failure, and the intensity factor that represents the influence of potential
failure on the affected area of the roadway.
114
5.2 Introduction
Rock slope stability assessment is considered one of the most important and critical aspects
in the field of geotechnical engineering, because it provides the necessary safety for slope design
and hence improves public safety (Park and West, 2001; Giani, 1992; Pantelidis, 2009; Pantelidis,
2010 and Hoek and Bray, 1981). In the last several decades, many of rock slope stability
assessment techniques have been revealed. One of these techniques, which are widely employed
in the field of engineering design and support of rock slopes are the rock mass classification
systems. These systems were originally proposed for underground excavations such as, Rock
Mass Rating by Bieniawski (Bieniawski,73,76,79,89, 93). Afterwards, these classification systems
were subjected to some modifications to become appropriate for surface excavation applications,
such as rock slopes. Examples for these modified systems are, Slope Mass rating (Romana 1985
and 1993), Modified Slope Mass Rating (Anbalagan et al., 1992), Chinese Slope Mass rating
(Chen, 1995), Continuous Slope Mass Rating (Tomás et al., 2007) and Graphical Slope Mass
Rating (Tomás et al., 2012).
Further modifications were incorporated by Pantelidis (Pantelidis, 2010). He proposed a

quantitative system that is expressed in terms of a Hazard Index (HI), where the level of the hazard
is evaluated by two separate functions. The first function is the normal conditions that are
determined by a factor of safety based on the failure mechanism. The second function is the
triggering mechanism for surface water and groundwater conditions. Unlike other classification
systems, in Pantelidis’s system the effect of water is assessed separately. These modified
classification methods along with their pros and cons have been outlined in chapter 3.
All previous methods discussed the assessment of the rock slope failure based on
classifying the quality of its hazard level. We notice that the hazard level for a particular rock slope
is determined with only the consideration of its physical properties, the mechanical behaviour of
the rock mass, the discontinuities and the effect of an external force (i.e. water), which could trigger
the rock block to fail. Therefore, the analysis of the possibility of slope failure of a particular rock
slope, which is only based on normal conditions and triggering factors (such as climate conditions)
regardless of the volume of the failure zone, has led to a reduced or underestimated or inaccurate
hazard level.
115
The term of hazard has many definitions in literature. It can be expressed as the threats to
people’s lives and public property (van Staveren, 2009). Hazard can be defined as the probability
of an undesired event (Smallaman, 2000); or it could be defined as a condition that may produce
a certain degree of risk (Silva et al., 2008). In view of International Union of Geological Science
(IUGS), 1997, the hazard is defined as any situation that is likely to cause an undesirable
consequence. Here, it is worth it to point out the difference between the hazard and another term
that is often accompanied with it, which is risk. In the field of geotechnics, the standard definition
of risk is the product of the probability of a particular slope to fail and the negative effects on the
public safety, such as people’s lives, property and environment (IUGS, 1997), 𝑃𝑟𝑜𝑏𝑎𝑏𝑖𝑙𝑖𝑡𝑦 ×
𝐶𝑜𝑛𝑠𝑒𝑞𝑢𝑒𝑛𝑐𝑒𝑠. The consequences here mean the cost of economic losses caused by the rock
failures (Hoek, 2007; Mouratidis and Pantelidis, 2007; Silva et al., 2008 and Staveren and Martin,
2009).
To evaluate the consequences for the potential economic losses of any rock slope failure
properly, it is required to assess the potentially failure size that hits the roadway besides the
determination of the likelihood of the slope failure. So, in addition to the probability of
unsatisfactory of performance (PF), this term will be used here instead of probability of failure as
a reference to the potential collapsing part of the slope, we need to adjust the concept of Hazard to
include a new factor that delineate the volume of potential slide surface. We can name it, the
Intensity Factor (IF), thus, the hazard of a rock slope is developed in this chapter and consequently
defined as the product of the probability of unsatisfactory of performance (PF) and the Intensity
factor (IF). The intensity factor is assessed by identifying the area of potential failure surface as
defined by the maximum shear-strain. This is in line of previous research suggesting the use of
maximum shear strength as a criterion for the delineation of potential failure zones (Saad and Mitri,
2010; Hammah et al, 2008; Diederichs, 2007; Duncan, 1996).
Two rock slope models in dry condition are adopted in this study, the first model (model
1) is the rock slope that is presented in chapter 4, and the second model (model 2) has been created
with the same material properties of model 1, but the geometry of the model 2 has been subjected
to changes until it resulted in the same probability of unsatisfactory of performance as model 1.
Then, the results in terms of the potential failure volume are compared, and based on that, the
116
intensity factor is calculated. Ultimately, the hazard index has been identified and categorized as
very low, low, moderate, high or very high.
5.3 Methodology
The methodology used here is as follows:
1. Recall the case study model used in chapter 4, which is called here, model 1.
2. Adjust model 1 geometry, by changing the slope height and slope angle, but keeping the
material properties of the model. With each change being made in the model geometry, a
probability analysis is performed, and the result is checked. If the probability does not
match that in model 1, the geometric shape is changed, and the process is repeated. The
probability approach used in this analysis is Box-Behnken Design (BBD) and the same
procedure, that is used in chapter 4 to get the probability of unsatisfactory performance, is
applied here (see chapter 4).
3. After the required probability of unsatisfactory performance percentage (%) is achieved,
the final slope model is adopted as model 2, for further analysis along with model 1.
4. The potential failure surface is estimated using the maximum shear strain. Then, the
intensity of rock failure percentage (%) can be calculated based on two methods. The first
one, is the ratio of the area of the potential failure surface and the affected area on the
roadway. The second method, is the ratio of the length of the potential failure surface of
the slope and the roadway width.
5. After the probability of unsatisfactory performance (𝑃𝐹) and the intensity factor (𝐼𝐹) are
found, the geotechnical hazard index is calculated. As a result, the geotechnical hazard can
be categorized from the geotechnical hazard description matrix (Table 5.1).
5.4 Intensity Factor Concept

The intensity factor is a new factor added to the hazard index; it shows the impact of
possible area prone to collapse on the roadway.
As mentioned in section 5.2, the determination of the likelihood of a rock slope to fail can
be estimated by the factor of the probability of unsatisfactory performance. Then it is used to assess
the hazardous of the rock failure of the slope. As a result, the consequences of the economic losses
can be estimated based on that resulted probability of failure.
117
To estimate the amount of economic losses and the consequences of what the rock slope
failures will lag behind, such as destroying the roads and threatening the lives and properties, we
need to assess the magnitude/size of the rock failure and its impact on the public safety. This is of
particular interest, especially, in cases where the probability of failure of two geometrically distinct
rock slopes is the same, but they have different effects on the roadway. So, the concept of the
intensity factor for a rock slope failure is mainly to show the extent of the impact of rock failure
on the road (see Figures 5.1 and 5.2). Therefore, the hazard level of any rock slope should be
determined by two factors rather than one, the probability of unsatisfactory performance (𝑃𝐹) and
the new presented factor, which is the intensity factor (𝐼𝐹).
The intensity factor is estimated using two sub-factors: firstly, delineate the potential failure
zone on the slope. This step is done during the numerical modelling of the rock slope using
FLAC3D software (Itasca, 2012) by finding the area that has the maximum shear strain (critical
area). In this study, the maximum shear strain ratio that matches the global minimum stability state
will use to estimate the critical area in the slope that. The determination of the critical area is done
using two methods of estimation:
1. Method 1: estimate the area of critical zone on the slope and the affected area on the
roadway.
2. Method 2: estimate the vertical length of the critical zone along the slope face and the
affected road width.
Secondly, a sub-factor related to swelling is considered to the intensity factor calculation. This
factor will be called here as a swell factor. The swell factor is used to calculate the volume increase
of the fallen rock mass due to the breakage of the rock mass after crashing down on the road. In
general, since the rock mass in the area is fractured and weathered rock, the swell factor is adopted
to be 1.31 (31%) for weathered rock according to the geotechnical technical guidance manual
(2007) issued by US transportation.
The intensity factor (%) can be calculated using the following equation:
𝑃𝑜𝑡𝑒𝑛𝑡𝑖𝑎𝑙 𝑓𝑎𝑖𝑙𝑢𝑒𝑟 𝑠𝑢𝑟𝑓𝑎𝑐𝑒 𝑜𝑛 𝑡ℎ𝑒 𝑠𝑙𝑜𝑝𝑒 ×𝑠𝑤𝑒𝑙𝑙 𝑓𝑎𝑐𝑡𝑜𝑟

× 100 (5.1)
𝑎𝑓𝑓𝑒𝑐𝑡𝑒𝑑 𝑎𝑟𝑒𝑎 𝑜𝑓 𝑡ℎ𝑒 𝑟𝑜𝑎𝑑𝑤𝑎𝑦
118
Figure 5.1: Example for rock slope failure causing obstruction to roadway (High intensity)
Figure 5.2: Example for rock slope failure (Low intensity)
119
5.5 Geotechnical Hazard Index (GHI)
After we obtain the probability of unsatisfactory performance (𝑃𝐹) and the intensity factor
(IF), each factor is categorized into three different levels with their corresponding scores, low
(score 1), moderate (score 2) and high (score 3) by multiplying the scores that correspond to the
particular percentage of each factor by using the equation 5.2. As a result, a 3×3 matrix will be
produced including fife level of geotechnical hazard, 1 (very low), 2 (low), 3 – 4 (moderate), 6
(high) and 9 (very high). This geotechnical hazard matrix is illustrated in Table 5.1.
𝐻𝑎𝑧𝑎𝑟𝑑 𝐼𝑛𝑑𝑒𝑥 = 𝑃𝐹 × 𝐼𝐹 (5.2)

Where,
𝑃𝐹, the probability of unsatisfactory performance.
𝐼𝐹, Intensity factor
The geotechnical hazard description matrix is in a logarithmic scale and will classify as
follows, for the probability results, 0 - 10% (score 1), ˃10% - 30% (score 2) and ˃ 30% (score 3).
It is worth noting that, as the study location is located in rural area the 10 % of probability is chosen
here as an acceptable probability of failure for short terms projects and rural roads according to
Hoek (2000). For failure intensity results, 0 – 10% (score 1), 10% - 40% (score 2) and ˃ 40%
(score 3).
This intensity categories were established based on the area width and it is 12 m, which the
road width below the slope toe, paved road is 10 m (2 lanes, 5 m each), and a margin of 1m width
unpaved in both sides of the road (Figure 5.3). The different categories of failure intensity will be
classified as follows, score 1 represents the failure intensity, which could affect the first 2m of the
road below the slope toe, and score 2 for affected road width ˂ 6m and 3 for affected road width
˃ 6m. Table 5.1 illustrates the proposed geotechnical hazard index description matrix that resulted
from the product of 𝑃𝐹 × 𝐼𝐹 .
120
Figure 5.3: Sketch view of the road width below a rock slope.
121
Table 5.1: Proposed geotechnical hazard description matrix
PF
0% - 10% ˃ 10% - 30% ˃ 30%
1 2 3
1 2 3
0% -10% 1
(very low) (low) (moderate)
2 4 6
IF ˃ 10% - 40% 2
(Low) (moderate) (high)
3 6 9
˃ 40% 3
(moderate) (high) (very high)
5.6 Model Setup

As mentioned in section 5.2, two rock slope models have been adopted for numerical
modelling analysis. Model 1 represents the case study used in chapter 4. The initial geometry of
model 1 is in Figure 4.6 and the excavated model 1 is shown in Figure 4.7. Model 1 has been
chosen for the probabilistic analysis in chapter 4, and all the geometrical information of the model
before and after excavation can be found in section 4.2 in chapter 4.
Model 2 has been created using FLAC3D software (Itasca, 2012), by making some
modifications on the geometry of the initial model (model1), while keeping the same material
properties of model 1. The purpose of these modifications on the geometry of model 1 is to get a
numerical model that has the same results of the probability of unsatisfactory performance as
122
model 1, and then make further analysis on the impact of potential failure resulted from the
probability analysis of both models on the roadway.
Several modifications have been made on the initial geometry of model 1, and the final
geometry for the new initial model 2 is shown in Figure 5.4; its width is 50 m (x direction), the
height is 140 m (z direction) and the depth is 120 m (y direction). After excavation, the initial
model 2, has a slope with an angle of 56° and at height of 40 m (Figure 5.5). As can be seen the
slope angle has decreased by 4° with an increase of about 20 m in height as compared to the
excavated model 1.
As can be seen in Figure 5.5, the model has been extended by 3 times the slope height
below the level of the roadway, in the negative direction of z axis, and 3 times the roadway width
behind the excavated slope, in the positive direction of y axis, in order to avoid the boundary
condition’s influence on the numerical analysis results.
The rock mass properties and the properties of the potential failure surface, which are used
in the analysis are the same as the material properties in model 1 (see Table 4.3).
The state of the in-situ stress (𝜎 0 ) has been re-calculated according to the geometry of
initial model 2, using the equations presented in 4.16 for (𝜎𝑣0 ) and in 4.17 for (𝜎𝑥0 𝑎𝑛𝑑 𝜎𝑦0 ).
123
Figure 5.4: Initial model 2 for the rock slope before excavation
Figure 5.5: Model 2 after excavation
124
5.7 Results and Discussion
In this section, we will discuss the results of probability analysis for only model 2, where
all results of model 1 about the probability analysis have been addressed in chapter 4. Also, results
of the failure intensity on the roadway for models 1 and 2 is discussed here.
5.7.1 Box-Behnken Design (BBD) for Model 2
As mentioned in chapter 4, the following procedure for probability analysis was using Box-
Behnken Design (BBD). Thirteen numerical realizations for the rock slope required for the BBD
experimental frame are simulated. The SRF values have bee computed for all different level of the
input variables involved in the BBD design as dependant responses for each run and is presented
in Table 5.2.
As seen in Table 5.2, results of the numerical simulations for slope realizations, 1, 2,5 and
6 give factor of safety (FOS) values below than the limit of safety 1.5 in low level values of joint
dip, this confirms that the dip variable play a significant role on the stability condition of the rock
slope. While, if we compare the amount of change in the FOS values between slope realization #1
and slope realization #2, we can note that the value of FOS was increased slightly from 1.27 to
1.32, because the variable of joint dip direction changed from low level “- “ to high “+” and this
indicates the simple effect of the dip direction.
When comparing the slope realizations # 5 and # 6, the FOS values were increased from
0.89 to 1.36, this means that the variable of friction angle affect if the dip variable in a low level.
Whilst, in slope realization # 9 although the variables of dip direction and friction angle are in low
level “-“ the FOS is 1.67, above the limit of safety 1.5
The purpose of the BBD method is to establish a quadratic relation between the inputs
(random variables) and output (response variable) To do so, the linear, the quadratic and the
interactions have been established to identify the contribution of each random variable on the
resulted response. The different contribution of the random variables has been done using the
Analysis of Variance (ANOVA), Figure 5.6 shows the plots of the simulated values of SMR
against the predicted SRF values for model 2 in dry condition. As can be seen, all SRF points fall
within the confidence curve (red-dotted lines) and close to the perfect fit (diagonal line), and the
goodness of fit, 𝑅 2 , is 0.99 and the model is significant.
125
Table 5.2: Shows results of the factor of safety for all treatments of slope realizations for
model 2

Patterns Dip
1 --0 47 40 27 1.27 1.21
2 -+0 47 46 27 1.32 1.28
3 +-0 57 40 27 2.45 2.48
4 ++0 57 46 27 2.52 2.57
5 -0- 47 43 24 0.89 0.98
6 -0+ 47 43 30 1.36 1.36
7 +0- 57 43 24 2.48 2.47
8 +0+ 57 43 30 2.54 2.44
9 0-- 52 40 24 1.67 1.63
10 0-+ 52 40 30 1.75 1.80
11 0+- 52 46 24 1.77 1.71
12 0++ 52 46 30 1.85 1.88
13 000 52 43 27 1.63 1.63
Table 5.3 shows results of ANOVA, they indicate that the dip variable is a significant
source of variability on the resulted response (FOS), where the percentage of its effect is 92.36 %.
Table 5.3: Results of ANOVA for the rock slope in dry condition for model 2
Sum of Contribution
Source of variability P-value
squares (%)
Dip 3.3153125 < 0.0001 92.36
Dip Direction 0.0128000 0.2451 0.35

Friction Angle 0.0595125 0.0363 1.66
Dip Direction*Friction 0.0000000 1.0000 0
Angle
Dip2 0.0930519 0.0164 2.59
Dip Direction2 0.0378519 0.0730 1.05
Friction Angle2 0.0030519 0.5486 0.085
126
Figure 5.6: Plot of simulated by predicted SRFs for slope realizations in dry condition for
model 2
From the BBD experiment, mathematical expressions have been established to describe
the relations between the random variables and the response (SRF), as presented in equation (5.3).
Theis relationship is used to construct the response surfaces that used to predict the values of SRF
for any slope realizations based on the variability of the random variables, this function is as
follows:
𝑆𝑅𝐹 = 25.03715 + (−0.361483333) × 𝐷 + (−0.9715) × 𝐷𝐷 + (0.2115833333) ×

∅ + (0.00635) × 𝐷2 + (0.01125) × 𝐷𝐷2 + (0.0031944444) × ∅2 + (0.0003333333) × 𝐷 ×
𝐷𝐷 + (−0.006833333) × 𝐷 × ∅ (5.3)
Where, 𝐷 is the dip Angle, 𝐷𝐷 is the dip direction, ∅ is the friction angle and 𝑆𝑅𝐹 is the
shear strength reduction.
3D response surfaces have been constructed using the previous mathematical expressions,
where each response surface plot explains the relation between two random variables and the SRF,
while the third variable is fixed at its mean value.
Figure 5.7 shows the constructed response surface for the dip variable against the SRF
values, we observe the significant effect of the dip on the results of SRF values, and the SRF values
are directly proportional to the dip.
127
Figure 5.8 shows the response surface for dip and dip direction variables against the SRF
values, while the third variable (friction angle) is fixed at its mean value of 27˚. Note that the dip
effect is high on the results of SRF, while the dip direction has no obvious effect.
Figure 5.9 Shows the response surface for dip and dip direction variables against the SRF
values. It also shows the great influence of the dip variable on the SRF results, unlike the effect of
friction angle that appears simply.
Figure 5.10 shows the constructed response surface for dip direction and friction angle
variables against the SRF values. As observed, there is a little effect of both variables on the results
of SRF.
Figure 5.7: Response surface plot for dip angle against SRF values in the case of dry
condition for model 2
128
Figure 5.8: Response surface plot for dip angle dip and direction against SRF values in
the case of dry condition for model 2
Figure 5.9: Response surface plot for dip angle and friction angle against SRF values in
the case of dry condition for model 2
129
Figure 5.10: Response surface plot for dip direction and friction angle against SRF
values in the case of dry condition for model 2
5.7.2 Probability of Unsatisfactory Performance

The probability of the unsatisfactory performance (𝑃𝐹) to evaluate the stability of model 2 has
been done. The Monte-Carlo technique has been performed to generate 10,000 random values for
the input variables, the dip, the dip direction and the friction angle, based on the statistical moments
and the distribution that has been determined previously in section 4.13.2 in chapter 4 for each
variable. The SRF value is calculated for each combination of these three variables using equation
(5.3). Thus, 10,000 values have been generated for the SRF and the best fit distribution is found,
which was found to be a triangle distribution. As aforementioned in chapter 4, the limit of
satisfactory of performance for the stability is defined to be, SRF = 1.5. Consequently, the
probability of unsatisfactory performance is identified. Figure 5.11 shows the histograms and the
triangle distribution fit for the calculated factor of safety (SRF), the probability of unsatisfactory
performance is found to be 36.1 % from the probability curve.
130
Figure 5.11: Histograms and distribution fit for the calculated factor of safety for the model 2
5.7.3 Intensity Factor
In this section, the results of failure intensity on the roadway for model 1 and model 2 will
be presented. The potential failure surface was estimated by the maximum shear -strain criterion
in the rock slope models 1 and 2. Figure 5.12 and Figure 5.13 are numerical modelling analyses
show the maximum shear-strain contours that delineate the potential failure surface in model 1 and
model 2, respectively.
The of failure intensity was identify by two methods, as mentioned in section 5.4. The
results of These two methods are described as follows.
5.7.3.1 Method 1. Calculate Area of Potential Failure Zone
The area of potential failure surface that illustrated by the maximum shear-strain contours
is calculated for model 1 and model 2. Figure 5.12 shows the maximum shear-strain (in red)
analysis for model 1, and represents as potential critical area of failure. The critical area is
delineated, and it is 23 m2. The interested area of road is also delineated, which represents the road
width 12 m by 10 m in length, as shown in Figure 5.12.
131
Figure 5.12: Calculate the critical area (in red) using method 1 for model 1
Using the equation 5.1, the intensity factor for model 1, will be calculated as follows,
23 × 1.31
𝐼𝐹 = × 100
120
𝐼𝐹 = 25 %
So, the intensity of failure will cover 25 % of the interested area of road.
Figure 5.13 shows the maximum shear-strain (in red) analysis for model 2, and represents
as potential critical area of failure. The critical area is delineated, and it is 600 m2. The interested
area of road is also delineated, which represents the road width 12 m by 30 m in length, as shown
in Figure 5.13.
132
600 × 1.31
𝐼𝐹 = × 100
360
𝐼𝐹 = 218 %
So, the intensity of failure will cover more than 100 % of the interested area of road. This
means the potential rock failure will be disastrous.
5.7.3.2 Method 2. Calculate Length of Potential Failure Zone
The method of calculation here is different, here the length of the potential failure zone
along the slope will take into account instead of the area as shown in Figure 5.14. Model 1 in
Figure 5.14 shows the length of the potential critical zone (labelled in red) is 5 m. The interested
area of road is the road width 12 m.
133
5 × 1.31
𝐼𝐹 = × 100
12
𝐼𝐹 = 54.5 %
So, the intensity of failure will cover 54.5 % of the interested road width.
Figure 5.15 shows the length of critical area in model 2 is 30 m, and the affected area is
the road width 12 m.
134
30 × 1.31
𝐼𝐹 = × 100
12
𝐼𝐹 = 327.5 %
So, the intensity of failure will cover more than 100 % of the interested road width. This
means the potential rock failure will be disastrous.
5.7.4 Geotechnical Hazard Index
Results of the probability of unsatisfactory performance (𝑃𝐹) and the intensity factor (𝐼𝐹)
are in percentage (%) and can be converted to rating scores by using Table 5.1. Then, by
multiplying the scores of 𝑃𝐹 × 𝐼𝐹 the geotechnical hazard can be calculated, and the
corresponding category can be found from the geotechnical hazard description, which also
presented in Table 5.1.
Table 5.4 displays a summary of 𝑃𝐹 and 𝐼𝐹 (method 1 and method 2) results for model 1 and
model 2.
135
Table 5.4: Summary of 𝑃𝐹 and 𝐼𝐹 results and corresponding scores for model 1 and model 2
𝐼𝐹 (%)
𝑃𝐹 (%) Score
Method 1 Score Method 2 Score
Model 1 36.4 3 25 2 54.5 3
Model 2 36.1 3 218 (≥100) 3 327.5 (≥100) 3
So, the results for the geotechnical hazard (𝐺𝐻) assessment will for model1 and model 2 will be
as follows,
Model 1, using IF method 1
𝐺𝐻 = 3 × 2 = 6 , and categorized as high hazard
𝐺𝐻 = 3 × 3 = 9, and categorized as very high hazard
Model 2, using If method 1
5.8 Conclusion
The results of the probability for both rock slope models agree that the dip factor is the key
element, which significantly affects the factor of safety, this in turn affects the entire stability
behavior of the slope. Both Model 1 and Model 2 give almost the same probability of failure
despite the fact that the slope height and the slope angle are different, Model 1 gives (𝑃𝐹) = 36.4
and Model 2 gives (𝑃𝐹) = 36.1.
The max. shear strain indicates that the volume of the potential failure zones is different in
the two models. Although the probability of unsatisfactory performance is the same in both
models, Model 2 results show that the potential failure zone is larger than that of Model 1.
Two kinds of calculations have been used to calculate the Intensity Factor in the study. The
first method calculates the ratio of the total critical area of the potential failure zone, on the slope
face, and the total affected area, on the road. The second method calculates the ratio of the vertical
length of the potential failure zone, on the slope face and the distance of the road width that can
be affected by the fallen rocks.
136
It is noticeable that the slope height has a significant effect in terms of failure intensity,
where the potential failure surface in model 2 was larger than in model 1because the slope height
in Model 2 is double the slope height in Model 1.
A geotechnical hazard description matrix is proposed, and it is a result of the product of

the probability of unsatisfactory of performance (𝑃𝐹) and the Intensity Factor (𝐼𝐹) for both
models (Model 1 and Model 2). Hence, two kinds of Hazard Index have been obtained in this
study, the first one has been obtained using method 1 and the other has been obtained using method
2.
137
CHAPTER 6
CONCLUSION
6.1 Research Conclusions
In Chapter 1, the problem of slope in stability is defined and reasons of instability problems
especially, in mountainous areas located in south-west region of Saudi Arabia are highlighted,
such as, the difficulty of rock cuttings due to the rugged terrains. This led to design choices
involving curved slopes with sharp slope angles, particularly in turns. Furthermore, adverse slope
geometry and discontinuities as a result of poor blasting. One of the more significant roles of
instability in this region is climate effects represented by heavy rainfalls, which cause deterioration
of the rock masses and trigger rocks to collapse. The scope and objectives are presented that
mainly focus on developing a geotechnical hazard index by assessing the stability analysis using
a probabilistic approach by incorporating the statistical numerical techniques for more realistic
stability analysis and dealing with the uncertainties which usually accompany with rock mass and
discontinuities properties. In addition, a new factor that assesses the extent rock failure and its
effect on the roadway is introduced and incorporated in the geotechnical hazard assessment of the
rock slopes.
In Chapter 2, a literature review is presented on the kinematic and numerical modelling

approaches that are used in the stability assessment in rock slopes. Also, some probabilistic
approaches were addressed including point estimate methods, design of experiments such as,
Central Composite design and Box-Behnken design, which help establish a relationship between
input (independent) and output (response) variables. This established relationship accompanied
with Monte-Carlo technique assess the probability of unsatisfactory performance of rock slopes.
Additionally, a brief introduction has been presented about the empirical methods.
In Chapter 3, five empirical methods developed for rock slopes assessment are
implemented to identify their validity for rock cuts. Twenty-two sites were chosen as case studies
in the south-western region Saudi Arabia. These selected case studies located in a mountainous
area, which is characterized by complex and difficult terrains. The selected empirical methods are
the original SMR developed by Romana (1985), continuous SMR (Tomás, 2007), Chinese SMR
(Chen, 1995), graphical SMR (Tomás et al., 2012) and hazard index (Pantelidis, 2010). The SMR
methods are mainly based on the RMR (Bieniawski, 1989) and the relationship between the dip
138
and dip direction of both the slope face and the joint as well as the influence of excavation method,
and can only be applied on structurally controlled failure mechanisms. The hazard index method
considers the effect of normal triggering conditions (e.g. presence of water) and can be applied for
structurally and non-structurally controlled failure mechanisms. The results indicated that the
original SMR and continuous SMR almost give similar results and stability conditions because
they are derived from the same concept but the application of the techniques is different. All slope
heights of rock cuts in the studied area are less than 80 m. Therefore, the Chinese SMR is not
applicable in rock cuts in the study area because the factor of height is not critical. Hazard index
and continuous SMR are recommended as good methods because provide a good stability
evaluation, and the hazard index is the superior as it takes the into consideration the effect of water
as a triggering factor.
In Chapter 4, the stability analysis assessment for a selected case study was implemented
using a probabilistic method, which takes into account the variability of several parameters that
could be sources of uncertainty, and the influence of the groundwater pressure on the stability of
rock slopes. Three parameters were considered as random variables in the analysis, namely dip,
dip direction and friction angle. The factor of safety was chosen as an indicator for the instability
condition (response variable). The required measurements were gathered for the random variables,
and the appropriate statistical moments and distributions were determined. Box-Behnken (BBD)
design was adopted as a probabilistic method. Fifteen treatments are generated, and the effect of
different level combinations of the random variables are examined. As a result, two polynomial
functions for dry and saturated conditions are established. The results of BBD experiment show
that the most influential factor on the variability of values of factor of safety is the dip factor, other
factors such as, dip direction and friction angle did not show obvious effects on the variability of
the safety factor values. Both results of the numerical and BBD where compatible in terms of that
the dip angle of the potential failure surface is predominant on the instability of the slope. The
probability of unsatisfactory performance was determined through the polynomial equation and
the Monte-Carlo simulation for dry and saturated conditions. The results indicate that the
probability of instability increased by 27 % in saturated condition. And the scenario of saturation
condition is recommended for remedial measures, especially in regions subjected to heavy rainfalls
such as the study area.
139
In Chapter 5, a geotechnical hazard description matrix is proposed. This proposed matrix
is a result of the product of the probability of unsatisfactory performance and a new factor related
to effect of the potential failure on an interested area of roadway, this factor called intensity factor
(𝐼𝐹). The intensity factor was estimated by two methods of calculation. The first method calculates
the ratio of the total critical area estimated by the maximum shear strain of the potential failure
zone, on the slope face, and the total affected area on the road. The second method calculates the
ratio of the vertical height of the potential failure zone, on the slope face and the distance of the
road width that can be affected by the fallen rocks.
Two rock slope models were used here, Model 1 represents the case study presented in
Chapter 4, and Model 2 which is constructed by modifying the geometry; slope height and slope
angle of Model 1 while maintaining the same material properties. The probability of unsatisfactory
performance for Model 2 was estimated using the BBD and Mote-Carlo simulation, and it was
𝑃𝐹 = 36.1% . Although the geometry is different, the results of 𝑃𝐹 for Model 2 was almost the
same results of 𝑃𝐹 obtained for Model 1, it was 36.6% in Chapter 4. The intensity factor was
calculated twice for each model, one using method 1 and the other using method 2. Consequently,
two kinds of hazard index were obtained for each model based on the results of intensity factor.
6.2 Recommendation for Future Work
Despite the fact that this research has contributed to enhancing the performance of the
geotechnical hazard assessment of rock slopes and enriching the concept of hazard more
effectively, there is still room for further research. The following are some suggestions for future
work.
1. Look at the effect of different slope curvature (i.e. convex and concave) with different
angles of curvatures, gentle and sharp (i.e. in turns) curves on the stability condition of the
rock slopes. And incorporate this effect as factor into the geotechnical hazard index.
2. Incorporate more failure mechanisms such as, wedge and toppling failures into the analysis
of the hazard index by using a proper technique for modelling structurally-controlled
failure such as distinct-element method for blocky rock slopes.
3. Develop a quantitative geotechnical hazard analysis including the effect of seismicity as a
triggering factor on the stability condition, and compare it with the effect of water pressure.
140
6.3 Statement of Contribution
1. In this study, a comprehensive assessment of empirical methods of rock slope stability is

performed for the first time with application to a case study encompassing 22 sites of rock
cuts in the south west region of Saudi Arabia.
2. A combined numerical modelling – probabilistic analysis approach for the analysis of rock
slopes is developed using 3D finite difference modelling and BBD method, accounting for
the rock slope material properties as random variables.
3. The concept of the intensity factor is introduced for the first time to incorporate the extent
of the failure zone and its effect on the roadway safety.
4. A new geotechnical hazard index approach is developed to account for both the probability
of failure and the extent of failure and its impact on roadway safety, resulting in a 5-level
geotechnical hazard index matrix.
141
REFERENCES
Anbalagan, R., Sharma, S. and Raghuvanshi, T.K., 1992. Rock mass stability evaluation using
modified SMR approach. In: Jha, P.C. (Ed.), Rock Mechanics Proceedings of the Sixth
National Symposium on Rock Mechanics, pp. 258–268.
Anbalagan R., 1992. Landslide hazard evaluation and zonation mapping in mountainous terrain.
Engineering Geology, 32(4): 269-77.
Aksoy, C.O., 2008. Review of rock mass rating classification: Historical developments,
applications, and restrictions. Journal of Mining Science, 44(1), pp.51-63.
Arai, K. and Tagyo, K., 1985. Determination of noncircular slip surface giving the minimum factor
of safety in slope stability analysis. Soils and Foundations, 25(1), pp.43-51.
Antony, J., 2003. Design of Experiments for Engineers and Scientists.
Abdellah, W., 2013. Geotechnical Risk Assessment of Mine Haulage Drifts During the Life of a
Mine Paln. PHD Thesis, McGill University, p 327.
Baecher, G.B., Lanney, N.A. and Einstein, H.H., 1977, January. Statistical description of rock
properties and sampling. In The 18th US Symposium on Rock Mechanics (USRMS).
American Rock Mechanics Association.
Box, G.E. and Behnken, D.W., 1960. Some new three level designs for the study of quantitative
variables. Technometrics, 2(4), pp.455-475.
Baecher, G. B. and J. T. Christian., 2003., Reliability and Statistics in Geotechnical Engineering.

The Atrium, Souhern Gate, Chichester, West Sussex, England, John Wiley & Sons Ltd.
Box, G. E. P. and Draper, N. R., 1987. Empirical Model Building and Response Surfaces. Oxford,
England: John Wiley & Sons.
Bieniawski, ZT., 1973. Engineering classification of jointed rock masses. Trans the S Afr Inst
Civil Eng. 15:335–43.
Bieniawski, ZT., 1989. Engineering rock mass classifications. New York: Wiley, 251pp.
142
Bieniawski, ZT.,1976. Exploration for rock engineering: proceedings of the Symposium on
Exploration for Rock Engineering, Johannesburg, 1-5 November 1976. AA Balkema
Limited.
Bieniawski, ZT., 1979. The geomechanics classification in rock engineering applications. In:
Proceedings of the 4th ISRM Congress. International Society for Rock Mechanics.
Bieniawski, ZT., 1993. Classification of rock masses for engineering: The RMR system and future
trends. Rock Testing and Site Characterization, p. 553-73.
Brideau, MA., Chauvin, S., Andrieux P. and Stead D., 2012. Influence of 3D statistical
discontinuity variability on slope stability conditions. Lateral; 90:000.
Basahel, H. and Mitri, H.S., 2017. Application of rock mass classification systems to rock slope
stability assessment: A case study. Journal of Rock Mechanics and Geotechnical
Engineering. Dec 31;9(6):993-1009.
Box, GE., Hunter, J.S. and Hunter, W.G., 2005. Statistics for experimenters: design, innovation,
and discovery. New York: Wiley-Interscience; May 31 (Vol. 2).
Baamer, W.S., Youssef, A.M., Zabramawi, Y.A., Al-Katheeri, M.M., Baamer, M.F., Basahel, H.M
and Al-Ahmadi, H.J., 2010. Rock slope instability analysis along Jabal Fayfa main roads,
Jazan area, Kingdom of Saudi Arabia: Saudi Geological Survey Technical Report.SGS-
TR-2009-1. 2010; 65 p., 51 fig., 35 tables, 10 plates.
Barton, N. and Choubey, V., 1977. The shear strength of rock joints in theory and practice. Rock
mechanics; Dec 1;10(1-2):1-54.
Box, G.E. and Hunter, J.S., 1957. Multi-factor experimental designs for exploring response
surfaces. The Annals of Mathematical Statistics; 28(1):195-241.
Barton, N. and de Quadros, E.F., 1997. Joint aperture and roughness in the prediction of flow and
groutability of rock masses. International Journal of Rock Mechanics and Mining Sciences;
Apr 1;34(3-4):252-e1.
Brawner, C. O., 1994. Rockfall Hazard Mitigation Methods. Participant Workbook: FHWA Report
– FHWA-SA-93-085.
143
Chen, Z.Y., 1992. Random trials used in determining global minimum factors of safety of
slopes. Canadian Geotechnical Journal, 29(2), pp.225-233.
Cai, M., 2011. Rock mass characterization and rock property variability considerations for tunnel
and cavern design. Rock mechanics and rock engineering, 44(4), pp.379-399.
Christian, J.T. and Baecher, G.B., 1999. Point estimate method as numerical quadrature, Journal
of Geotechnical and Geoenvironmental Engineering, 125 (9): 779-786.
Christian, J.T., Ladd, C.C. and Baecher, G.B., 1994. Reliability applied to slope stability
analysis. Journal of Geotechnical Engineering, 120(12), pp.2180-2207.
Chiwaye, H. T. and Stacey, T. R., 2010. A comparison of limit equilibrium and numerical
modelling approaches to risk analysis for open pit mining. SIAMM . Journal of the South
African Institute of Mining and Metallurgy, 110, 571-580.
Coggan, J. S., Stead, D. and Eyre, J., 1998. Evaluation of techniques for quarry slope stability
assessment. Transactions of the Institutions of Mining and Metallurgy, Section B: Applied
Earth Science, 107, B139-B147.
Chen, Z., 1995. Recent developments in slope stability analysis. In: Fujii, T. (Ed.), Keynote
Lecture: Proc. 8th Int. Cong. Rock Mech, vol. 3, pp. 1041–1048.
Christian, J.T. and Baecher GB., 2002. The point‐estimate method with large numbers of variables.
International Journal for Numerical and analytical methods in geomechanics. Dec
25;26(15):1515-29.
Cornell, J.A. and Khuri, A.I., 1987. Response surfaces: designs and analyses. Marcel Dekker, Inc.
Duzgun, H.S.B., Yucemen, M.S. and Karpuz, C., 2003. A methodology for reliability-based
design of rock slopes. Rock Mechanics and Rock Engineering, 36(2), pp.95-120.
Duncan, J. M., 2000. Factors of safety and reliability in geotechnical engineering. Journal of
Geotechnical & Geoenvironmental Engineering, 126(4), 307-316.
Duncan, J. M., 1999. The use of back analysis to reduce slope failure risk. Journal of the Boston
Society of Civil Engineers, 14(1), 75-91.
144
Duncan, J.M., 1996. State of the art: limit equilibrium and finite-element analysis of
slopes. Journal of Geotechnical engineering, 122(7), pp.577-596.
Duran, A. and Douglas, K. J., 2000. Experience with empirical rock slope design. In Proc. Intl.
Conf. Geotechnical and Geolo.
Daftaribesheli, A., Ataei M. and Sereshki F., 2011. Assessment of rock slope stability using the
Fuzzy Slope Mass Rating (FSMR) system. Applied Soft Computing. Dec 1;11(8):4465-
73.
Davis, T.J. and Keller CP., 1997. Modelling uncertainty in natural resource analysis using fuzzy
sets and Monte Carlo simulation: slope stability prediction. International Journal of
Geographical Information Science. Jul 1;11(5):409-34.
Diederichs, M.S., 2007. The 2003 Canadian Geotechnical Colloquium: Mechanistic interpretation
and practical application of damage and spalling prediction criteria for deep
tunnelling. Canadian Geotechnical Journal, 44(9), pp.1082-1116.
Dershowitz, W.S., Gordon, B.M., Kafritsas, J.C. and Herda, H., 1985. A new three-dimensional
model for flow in fractured rock, 17th International Congress of the International
Association of Hydrogeologists. 1985; 441–448, Tucson, Arizona.
Dawson, E., You, K. and Park, Y., 2000. Strength-reduction stability analysis of rock slopes using
the Hoek-Brown failure criterion. InTrends in Rock Mechanics (pp. 65-77).
Dawson, EM., Roth, W.H. and Drescher A., 1999. "Slope stability analysis by strength reduction."
835-840.
Einstein, H. H. and Baecher, G. B., 1982. Probabilistic and statistical methods in engineering
geology, problem statement and introduction to solution. Rock Mechanics, Suppl. 12, 47-
61.
Einstein, H.H., 2003. "Uncertainty in rock mechanics and rock engineering-Then and now."
In 10th ISRM Congress. International Society for Rock Mechanics.
El-Ramly, H., Morgenstern, N.R. and Cruden, D.M., 2002. Probabilistic slope stability analysis
for practice. Canadian Geotechnical Journal, 39(3), pp.665-683.
145
Fisher, B.R. and Eberhardt, E., 2012. Assessment of parameter uncertainty associated with dip
slope stability analyses as a means to improve site investigations. Journal of Geotechnical
and Geoenvironmental Engineering. Jan 17;138(2):166-73.
Fairer, G.M., 1981. Reconnaissance geology of the Jabal Fayfa quadrangle, sheet 17/43C,
Kingdom of Saudi Arabia: Saudi Arabian Deputy Ministry for Mineral Resource Open-
File Report USGS-OF-02-87. no text scale 1:100,000.
Franklin, J. A. and Senior, S. A., 1987a. Outline of RHRON, the Ontario rockfall hazard
rating system: Proceedings International Symposium on Engineering Geology and The
Environment, Athens, Greece, pp. 647-656.
Franklin, J. A. and Senior, S. A., 1987b. Rockfall Hazards – Strategies for detection,
assessment, and remediation: Proceedings International Symposium on Engineering
Geology and The Environment, Athens, Greece, pp. 657-663.
Federal Lands Highway., 2007. Geotechnical technical guidance manual. US transportation.

https://flh.fhwa.dot.gov/resources/design/pddm/Geotechnical_TGM.pdf
Gurocak, Z., Alemdag, S. and Zaman, M.M., 2008. Rock slope stability and excavatability
assessment of rocks at the Kapikaya dam site, Turkey. Engineering Geology, 96(1-2), pp.17-
27.
Goodman, R.E., 1989. Introduction to Rock Mechanics, 2nd edition. Wiley, New York. 562 pp.
Grenon, M. and Hadjigeorgiou, J., 2008. A design methodology for rock slopes susceptible to
wedge failure using fracture system modelling. Engineering Geology, 96(1-2), pp.78-93.
Gravanis, E., Pantelidis, L. and Griffiths, D.V., 2014. An analytical solution in probabilistic rock
slope stability assessment based on random fields. International Journal of Rock Mechanics
and Mining Sciences, 71, pp.19-24.
Gokceoglu, C., Sonmez, H. and Ercanoglu, M., 2000. Discontinuity controlled probabilistic slope
failure risk maps of the Altindag (settlement) region in Turkey. Engineering Geology, 55(4),
pp.277-296.
Giani, G.P., 1992. Rock slope stability analysis. CRC Press.

146
Goodman, R.E., 1995. Block theory and its application. In Proc. of 35th US Symposium on Rock
Mechanics (edited by JJK Daemen & RA Schultz). AA Balkema. Rotterdam. 1995b. str (pp.
3-15).
Griffiths, D.V., Huang, J. and Fenton, G.A., 2009. Influence of spatial variability on slope
reliability using 2-D random fields. Journal of geotechnical and geoenvironmental
engineering. Feb 23;135(10):1367-78.
Griffiths, D.V. and Fenton, G.A., 2004. Probabilistic slope stability analysis by finite elements.
Journal of Geotechnical and Geoenvironmental Engineering. May;130(5):507-18.
Griffiths, D.V. and Lane, P.A., 1999. Slope stability analysis by finite elements. Geotechnique.
Jun 1;49(3):387-403.
Goodman, R.E., and Bray, J.W., 1977. Toppling of rock slopes. In: Proceedings, Specialty
Conference on Rock Engineering for Foundations and Slopes, 2. Boulder, CO: American
Society of Civil Engineers, pp. 201–234.
Hajiabdolmajid, V. and Kaiser, P., 2002, January. Slope stability assessment in strain sensitive
rocks. In ISRM International Symposium-EUROCK 2002. International Society for Rock
Mechanics.
Hammah, R.E., Curran, H., Yacoub, T. and Corkum, B., 2004. Stability analysis of rock slopes
using the finite element method.
Hammah, R.E., Yacoub, T.E. and Curran, J.H., 2008. Probabilistic slope analysis with the finite
element method. American Rock Mechanics Association (ARMA).
Hammah, R. E., Yacoub, T. E. and Curran, J. H., 2009. Numerical modelling of slope uncertainty
due to rock mass jointing. International Conference on Rock Joints and Jointed Rock Masses
Tucson, Arizona, USA.
Haldar, A. and S. Mahadevan., 2000. Probability, Reliability, and Statistical Methods in

Engineering Design. New York, NY, John Wiley & Sons, Inc.
Harr, M. E., 1987. Reliability-based Design in Civil Engineering. New York: McGraw-Hill.
147
Hong, H. P., 1998., An efficient point estimate method for probabilistic analysis. Reliability
Engineering and System Safety, 59(3), 261-267.
Hoek, E., and Bray, J., 1981. Rock Slope Engineering: The Institution of Mining and Metallurgy,
358 pp.
Hoek, E., 2000. Practical rock engineering.
Hammah, R.E., Yacoub, T.E. and Curran, J.H., 2009. “Probabilistic slope analysis with the finite
element method”. In 43rd US Rock Mechanics Symposium & 4th US-Canada Rock
Mechanics Symposium. American Rock Mechanics Association.
Hadjigeorgiou, J. and Harrison, J.P., 2011. Uncertainty and sources of error in rock engineering.
In12th ISRM Congress. International Society for Rock Mechanics.
Hill, W.J. and Hunter, W.G., 1966. A review of response surface methodology: a literature
survey. Technometrics. 8(4), pp.571-590.
Hammah, R.E., Yacoub, T.E., Brent, C. and Curran, J.H., 2005. "The shear strength reduction
method for the generalized Hoek-Brown criterion.
Hack, R., Price, D. and Rengers, N., 2003. A new approach to rock slope stability-a probability
classification (SSPC). Bulletin of Engineering Geology and the Environment. 62(2): 167-
84.
Hoek, E., 2007. Rock mass properties. Practical rock engineering. www. rocscience.
com/learning/hoek-s-corner.
Hoek, E. and Marinos, P., 2000. Predicting tunnel squeezing problems in weak heterogeneous rock
masses. Tunnels and Tunnelling International. 32(11): 45-51.
Itasca., 2012 .FLAC3D-Fast Lagrangian Analysis of Continua. 5th ed., USA: Itasca Consulting
Group Inc.
IUGS Working Group on Landslides, Committee on Risk Assessment., 1997. Quantitative risk
assessment for slopes and Highway slope instability risk assessment 399 123 landslides—
the state of the art. In: Cruden D, Fell R (eds) Landslide risk assessment. Balkema,
Rotterbam, pp 3–12.
148
Jaboyedoff, M., Oppikofer, T., Abellán, A., Derron, M.H., Loye, A., Metzger, R. and Pedrazzini,
A., 2012. Use of LIDAR in landslide investigations: a review. Natural hazards, 61(1), pp.5-
28.
Jing, L., 2003. A review of techniques, advances and outstanding issues in numerical modelling
for rock mechanics and rock engineering. International Journal of Rock Mechanics and
Mining Sciences, 40(3), pp.283-353.
Jimenez-Rodriguez, R., Sitar, N. and Chacon, J., 2006. System reliability approach to rock slope
stability. International Journal of Rock Mechanics and Mining Sciences. Sep 1;43(6):847-
59.
Jaeger, J.C., 1971. Friction of rocks and stability of rock slopes. Geotechnique. Jun;21(2):97-134.
Johari, A., Fazeli, A. and Javadi, A.A., 2013. An investigation into application of jointly
distributed random variables method in reliability assessment of rock slope stability.
Computers and Geotechnics. Jan 1; 47:42-7.
Koyama, T. and Jing, L., 2007. Effects of model scale and particle size on micro-mechanical
properties and failure processes of rocks—a particle mechanics approach. Engineering
analysis with boundary elements. May 1;31(5):458-72.
Karami, A., Greer, S. and Beddoes, R., 2007. Numerical assessment of step-path failure of
northwest wall of A154 Pit, Diavik Diamond Mines. InSlope Stability 2007: Proceedings of
the 2007 International symposium on rock slope stability in open pit mining and civil
engineering, Perth, Australia (pp. 293-305).
Kostić, S., 2017. Analytical Models for Estimation of Slope Stability in Homogeneous Intact and
Jointed Rock Masses with a Single Joint. International Journal of Geomechanics. Aug
2;17(10):04017089.
Konya, C.J. and Walter, E.J., 1991. Rock Blasting and Overbreak Control: FHWA Report –
FHWA-HI-92-001.
149
Li, D., Chen, Y., Lu, W. and Zhou, C., 2011. Stochastic response surface method for reliability
analysis of rock slopes involving correlated non-normal variables. Computers and
Geotechnics, 38(1), pp.58-68.
Low, B.K., 2007. Reliability analysis of rock slopes involving correlated

nonnormals. International Journal of Rock Mechanics and Mining Sciences, 44(6), pp.922-
935.
Mellah, R., Auvinet, G. and Masrouri, F., 2000. Stochastic finite element method applied to non-
linear analysis of embankments. Probabilistic Engineering Mechanics, 15(3), pp.251-259.
Miller, S. M., Whyatt, J. K. and McHugh, E. L., 2004. Applications of the point estimation method
for stochastic rock slope engineering. Rock Mechanics Across Borders and Disciplines, 6th
North American Rock Mechanics Conference, Gulf Rocks, USA.
Montgomery, D.C., 2013. Design and analysis of experiments. John wiley & sons.
Myers, R. H. and Montgomery, D. C., 2002. Response Surface Methodology Process and Product
Optimization Using Designed Experiments (2nd ed.). New York-USA: John Wiley & Sons.
Mason, R. L., Gunst, R. F. and Hess, J. L., 1989. Statistical Design and Analysis of Experiments.
New York-USA: John Wiley & Sons.
Morgenstern, N.R., 1995. Managing risk in geotechnical engineering. In10th Pan American
Conference on Soil Mechanics and Foundation Engineering.
Morales, J.M. and Perez-Ruiz, J., 2007. Point estimate schemes to solve the probabilistic power
flow. IEEE Transactions on power systems. Nov;22(4):1594-601.
Myers, R.H. and Montgomery DC., 2002. Response surface methodology. Willey, New York.
Maerz, N.H. and Kim, W., 2000. Potential use of ground penetrating radar in highway rock cut
stability. Geophysics, 2000.
Mouratidis, A. and Pantelidis, L., 2007. Rock failure risk assessment in highway maintenance
management.
Nathanail, C.P., Earle, D.A. and Hudson, J.A., 1992. Stability hazard indicator system for slope
failure in heterogeneous strata. In EUROCK (Vol. 92, pp. 111-116).
150
Nadim, F., Einstein, H. and Roberds, W., 2005, May. Probabilistic stability analysis for individual
slopes in soil and rock. In Proc. Int. Conf. on Landslide Risk Management (pp. 63-98).
Piteau, D. R., 1979a, Engineering geology considerations and basic approach to rock slope
stability analysis for highways. Part A., Rock Slope Engineering Reference Manual: FHWA
Report – FHWA-TS-79-208, 78 pp.
Piteau, D. R., 1979b, Methods of obtaining geological, structural, strength and related
engineering geology data. Part B, Rock Slope Engineering Reference Manual: FHWA
Report– FHWA-TS-79-208.,147 pp.
Piteau, D. R., 1979c, Approach and techniques in geological structural analysis. Part C, Rock
Slope Engineering Reference Manual: FHWA Report – FHWA-TS-79-208.
Piteau, D. R., 1979d, Slope stability analysis methods. Part D, Rock Slope Engineering
Reference Manual: FHWA Report – FHWA-TS-79-208.
Piteau, D. R., 1979e, Rock slope stabilization, protection and warning-instrumentation
measures and related construction considerations. Part E, Rock Slope Engineering Reference
Manual: FHWA Report – FHWA-TS-79-208.
Piteau, D. R., 1979f, Blasting for rock slopes and related excavation considerations. Part F,
Rock slope Engineering Reference Manual: FHWA Report – FHWA-TS-79-208.
Piteau, D. R., 1979g, Description of detail line engineering geology mapping method. Part G,
Rock slope Engineering Reference Manual: FHWA Report – FHWA-TS-79-208, 29 pp.
Park, H.J., Um, J.G., Woo, I. and Kim, J.W., 2012. Application of fuzzy set theory to evaluate the
probability of failure in rock slopes. Engineering Geology, 125, pp.92-101.
Park, H. and West, T.R., 2001. Development of a probabilistic approach for rock wedge
failure. Engineering Geology, 59(3-4), pp.233-251.
Price, N. J. and Cosgrove, J. W., 1990. Analysis of Geological Structures. Cambridge, England:
Cambridge University Press.
Pantelidis L., 2009. Rock slope stability assessment through rock mass classification systems. Int
J Rock Mech Min Sci 46(2):315–325. doi:10.1016/j.ijrmms.2008.06.003
Pantelidis L., 2010. An alternative rock mass classification system for rock slopes. Bull Eng Geol
Environ (online first). doi: 10.1007/s10064-009-0241-y
151
Priest, S.D. and Hudson JA., 1976. Discontinuity spacings in rock. In International Journal of Rock
Mechanics and Mining Sciences & Geomechanics Abstracts. (Vol. 13, No. 5, pp. 135-148).
Pergamon.
Park, H.J., West, T.R. and Woo I., 2005. Probabilistic analysis of rock slope stability and random
properties of discontinuity parameters, Interstate Highway 40, Western North Carolina,
USA. Engineering Geology. Jul 11;79(3-4):230-50.
Price, N.J. and Cosgrove JW., 1990. Analysis of Geological Structures. Cambridge, England:
Cambridge University Press.
Pachauri, A.K. and Pant, M., 1992. Landslide hazard mapping based on geological attributes.
Engineering Geology. 32(1-2): 81-100.
Rosenblueth, E., 1975. Point estimates for probability moments. National Academy of Sciences of
the United States of America, 72, 3812-3814.
Romana, M., 1985. New adjustment ratings for application of Bieniawski classification to slopes.
In Proc. Int. Symp. On the Role of Rock Mech., pp 49-53. Zacatecas.
Romana, M.R., 1993. A geomechanical classification for slopes: slope mass rating. In Rock testing
and site characterization (pp. 575-600). Pergamon.
Rocscience Inc., 2014. Dip v. 6.0 - graphical and statistical analysis of orientation data. Toronto,
Canada.
Shamekhi, E. and Tannant, D.D., 2015. Probabilistic assessment of rock slope stability using
response surfaces determined from finite element models of geometric
realizations. Computers and Geotechnics, 69, pp.70-81.
Shen, H. and Abbas, S.M., 2013. Rock slope reliability analysis based on distinct element method
and random set theory. International Journal of Rock Mechanics and Mining Sciences.
2013 Jul 1;61:15-22.
Sen, Z. and Sadagah, B.H., 2003. Modified rock mass classification system by continuous rating.
Engineering Geology. 67(3-4): 269–80.
152
Silva, F., Lambe, T.W. and Marr, W.A., 2008. Probability and risk of slope failure. Journal of
Geotechnical and Geoenvironmental Engineering, 134(12), pp.1691-1699.
Smallman, C., 2000. Challenging the orthodoxy in risk management. In Risk management and
society (pp. 53-79). Springer, Dordrecht.
Saad, B. and Mitri, H., 2010. Hydromechanical analysis of upstream tailings disposal
facilities. Journal of geotechnical and geoenvironmental engineering, 137(1), pp.27-42.
Sturzenegger, M. and Stead, D., 2009. Quantifying discontinuity orientation and persistence on
high mountain rock slopes and large landslides using terrestrial remote sensing techniques.
Natural Hazards and Earth System Sciences. Mar 3;9(2):267-87.
Tabba, M. M., 1984. Deterministic versus risk analysis of slope stability. 4th International
Symposium on Landslides. Toronto, Canada. 491-498.
Tomás, R., Delgado, J. and Serón, JB., 2007. Modification of slope mass rating (SMR) by
continuous functions. International Journal of Rock Mechanics and Mining Sciences. 44:
1062–9.
Tomás, R., Cuenca, A., Cano, M. and Garcίa-Barbra, J., 2012. A graphical approach for slope
mass rating (SMR). Engineering Geology. 124(4): 67-76.
van Staveren, M.T., 2009. Extending to geotechnical risk management. Georisk, 3(3), pp.174-183.
Wolff, T.F., 1996. Probabilistic slope stability in theory and practice, In Uncertainty in the
Geologic Environment, ASCE, New York, 419-433.
Wyllie, D.C. and Mah, C., 2004. "Rock slope engineering 4th Edition.
Wolff, T.F., 1996. Probabilistic slope stability in theory and practice. InUncertainty in the
geologic environment: From theory to practice (pp. 419-433). ASCE.
Wong, F.S., 1985. First-order, second-moment methods. Computers & structures. Jan
1;20(4):779-91.
153

Geotechnical Hazard Assessment of Road Cuts Stabilisation

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Geotechnical Hazard Assessment of Road Cuts Stabilisation

Uploaded by

Copyright:

Available Formats

GEOTECHNICAL HAZARD ASSESSMENT OF ROAD CUTS

STABILITY IN MOUNTAINOUS AREAS IN SAUDI ARABIA

Hassan Mabrouk Basahel

Department of Mining and Materials Engineering

A thesis submitted to McGill University in partial fulfillment of the

A probabilistic stability analysis approach is then developed. The stability condition of a

Dans cette thèse, plusieurs systèmes de classification de masse rocheuse permettant

This work is dedicated to …

My brothers and sisters

May Allah bless them all.

LE: Limit Equilibrium

𝐶𝑆𝑀𝑅: Chinese Slope Mass Rating

𝜇𝑀 : Mean value of margin of safety

𝜇𝑅 : Mean value of resistance

𝜇𝑄 : Mean of the load

𝜎𝑀 = the standard deviation of margin of safety

𝜎𝑅 2 : Variances of the resistance

𝜎𝑄 2: Variance of the load

𝜌𝑅𝑄 = Coefficient of correlation between both (R) and (Q) variables

𝜎𝑅 : Standard Deviations of the resistance

𝜎𝑄 : Standard Deviations of the load

𝜇𝑥 : Mean of the variable

𝜌𝑥1𝑥2 : Correlation coefficient between two variables

𝐶𝑂𝑉𝑥1𝑥2 : Covariance of two variables

𝐴𝑐𝑜𝑑𝑒𝑑(±) : Coded variable, low (-) and high (+)

𝐾: Number of factors in the design

𝐹: Number of factorial points

ζ : Slope Height Factor

αj: Joint direction

𝐴: Parallelism between discontinuities and slope dip direction

C: Dip relationship between slope face and joint

𝑓𝑁𝐶 : Normal condition

∅𝑟 : Residual friction angle

∅𝑏 : Basic friction angle

𝑟 : Rebound on weathered rock surface

𝑅 : Rebound on un-weathered rock surface

𝐶0 : Number of center points

𝐶𝑗 : Cohesion for the joint surface

∅𝑗 : Friction angle for the joint surface

𝑦𝑖𝑗 : the 𝑖𝑗 𝑡ℎ observation

𝑎 and 𝑏: the treatments

𝜇 : The overall mean

𝜏𝑖 : The effect of the 𝑖 𝑡ℎ parameter level

𝜖𝑖𝑗 : Random error

𝑆𝑆: Sum of squares

𝜎𝑣0 : Vertical stress

𝜎𝑥0 , 𝜎𝑦0 : horizontal stresses

𝑣 : Coefficient of kinematic viscosity of the fluid

𝜌𝑓 : Fluid mass density

𝐺𝐻: Geotechnical hazard

1.2 Study Problem

1.3 Scope of Work

Figure 1.3: Location of the study area

1.4 Objectives of the Study

1.5 Thesis Outline

- Field Investigation or Lab test

Characterizing the - Deterministic technique

Numerical modelling - Stability factor (possible failure surface)

Figure 2.1: Conventional procedure for slope stability analysis

Other sources of - Operational uncertainty

2.3 Deterministic and Probabilistic Approaches

2.3.2 Probabilistic Approach

2.4 Probabilistic methods in Geotechnical Engineering

2.4.1 First Order Reliability Method (FORM)