You are on page 1of 10

International Journal of Rock Mechanics & Mining Sciences 76 (2015) 182191

Contents lists available at ScienceDirect

International Journal of
Rock Mechanics & Mining Sciences
journal homepage: www.elsevier.com/locate/ijrmms

A new empirical system for rock slope stability analysis


in exploitation stage
Marisa Pinheiro a,n, Sara Sanches b, Tiago Miranda a, Adriana Neves b, Joaquim Tinoco a,
Alexandra Ferreira b, Antnio Gomes Correia a
a
b

ISISE, University of Minho, Guimares, Portugal


Ascendi, Matosinhos, Portugal

art ic l e i nf o

a b s t r a c t

Article history:
Received 11 September 2014
Received in revised form
4 March 2015
Accepted 8 March 2015

An empirical system was developed to obtain a quality index for rock slopes in road infrastructures,
named Slope Quality Index (SQI), and it was applied to a set of real slopes. The SQI is supported in nine
factors affecting slope stability that contemplate the evaluation of different parameters. Consequently,
each factor is classied by the degree of importance and inuence by assigned weights. These weights
were established through a statistical analysis of replies to a survey that was distributed to several
experienced professionals in the eld.
The proposed SQI varies between 1 and 5, corresponding to slopes in very good and very bad
condition state, respectively. Besides the advantage linked to a quantitative and qualitative evaluation of
slopes, the SQI also allows identifying the most critical factors on the slope stability, which is a
fundamental issue for an efcient management of the slope network in the road infrastructure, namely
in the planning of conservation and maintenance operations.
& 2015 Elsevier Ltd. All rights reserved.

Keywords:
Slope stability
Slope quality index (SQI)
Rock slopes
Hazard evaluation
Exploitation stage

1. Introduction
The existing problems in road infrastructures networks are, in
most countries, directly or indirectly related to the lack of quality assessment systems that can provide to the management structures, tools
to assist in the planning of conservation and maintenance operations
proactively. As such, there is the necessity to mitigate this problem
through a constant search of innovative and effective techniques,
allowing optimizing the long-life cycle of these infrastructures. Among
all the elements that compose the road network, slopes are the ones
that are subjected to less normative rules when compared with
bridges, road pavement and electronic equipment for instance.
Nowadays, the diversity of methods and techniques for slope stability evaluation during the design stage is huge (for instance using
limit equilibrium methods, FEM, DEM, probabilistic approaches, etc.).
However, methods and techniques for this evaluation during the
exploitation stage, i.e. using mainly information of what was really
built and its actual state reported from visual inspections, monitoring
systems and indirect information (like climatic and seismic zoning),
are scarce. In the past, a number of quality/classication systems
applied to slopes have been proposed, though they normally consider
a limited number of the factors involved in slope stability and usually

Corresponding author.
E-mail address: marisamotapinheiro@gmail.com (M. Pinheiro).

http://dx.doi.org/10.1016/j.ijrmms.2015.03.015
1365-1609/& 2015 Elsevier Ltd. All rights reserved.

are limited to the analysis of some failure modes. Despite the existence of these systems, a more complete one that is able to combine a
broader number of factors affecting the rock slope stability is still
lacking. Thereby, an innovative system named Slope Quality Index
(SQI) that integrates the evaluation of a broad range of internal and
external factors related to the slope quality and stability was developed. The SQI is based on the rock mass classication system
developed by Liu and Chen [1] that considers the rating of three
factors by their degree of importance (weights). To this system, a large
number of factors and parameters were added and several adaptations were made in order to improve it. Likewise, the methodology for
the weight assignment was the same used by Liu and Chen [1], which
is based in surveys made by experts in the eld of knowledge of slope
stability. Each one of the dened factors have a different weight resulting in SQI values ranging from 1 to 5, translating very good to very
bad slope quality conditions, respectively. Besides the classication of
the slopes and the identication of the ones in poorer conditions, the
SQI can support the development of hazard maps and aid in the
decision concerning the intervention plans. In other words, the SQI
system intends, not only to identify slopes in bad to very bad condition (extreme values), but also it is indented to be a tool to map the
overall condition of the slope network allowing a more rational
management of the network mainly in what concerns the planning
of short and long-term conservation and maintenance interventions
and also evaluate the evolution of the slopes in time (individually or at
more macro levels).

M. Pinheiro et al. / International Journal of Rock Mechanics & Mining Sciences 76 (2015) 182191

Considering the lack of information in this eld, and aiming to


increase the reliability of the SQI, some existing and validated systems
were used embedded in the SQI for the evaluation of some of the
factors. For instance the Rock Hazard Rating System modied
(RHRSm) [2], created as an improvement of the original RHRS system
[3], is used for the evaluation of rockfall potential. However, in the
present work the RHRSm was subjected to some additional adaptations and changes to better suit the purposes of the SQI, resulting in
an update, which was called RHRSm2. This updated system and the
SQI system description are presented in Section 2. In Section 3 the SQI
is applied to a set of slopes and in the last section the main conclusions of the work are sketched.

2.1. Concept
The main goal of the SQI system is the calculation of an index
based on nine factors directly or indirectly related to the stability of
the slope. As previously mentioned, the SQI contemplates other
evaluation subsystems to classify some of the nine factors. To obtain
the factors, a number of parameters have to be rstly evaluated in a
range between 1 and 5, the same as the SQI range for the sake of
consistency (Fig. 1). To obtain the nal value of the SQI the factors are
then weighted by their degree of importance. The computation of
each degree will be presented in Section 2.3. The range of SQI between
1 and 5 translate very good and very bad conditions, respectively.
Therefore, the calculation of the SQI is performed according to
n
X

Table 1
SQI system for rock slopes: qualitative, quantitative and risk level classication.
SQI

Slope state

Risk level

[1; 1.4]
[1.5; 2.4]
[2.5; 3.4]
[3.5; 4.2]
[4.3; 5]

Very good
Good
Medium
Bad
Very bad

Very low
Low
Medium
High
Very high

2.2. Rockfall hazard rating systems (RHRSm and RHRSm2)


Due to the importance of rockfalls in slope quality this issue is
treated in the SQI, in a rst stage separately and if the conditions
are poor some measures have to be immediately adopted, and in a
second stage this information is integrated in the overall system. A
number of systems dedicated exclusively to this problem already
exist. Some of the most well known systems are:

2. Slope quality index (SQI)

SQI

183

W i  Xi

(a) Rockfall Hazard Rating System (RHRS) proposed by Pierson


et al. [3] that later emerged as RHRSm modied and improved
by Budetta [2];
(b) Rockfall Hazard Ontario (RHON), proposed by Franklin and
Senior [4] that appeared as an alternative to the RHRS system
allowing the application to low height slopes;
(c) Missouri Rockfall Rating System (MORFH RS), proposed by
Youssef et al. [5];
(d) Rockfall Risk Assessment for Quarries (ROFRAQ), proposed by
Alejano et al. [6] more specic to open pit quarries;
(e) Slope Mass Rating (SMR), proposed by Romana [7].

i1

where W i is the weight of each one of the nine factors (varies


between 0 and 1) and X i is the value of the classication assigned
to each factor. Each factor is partially calculated through the
parameters sum using:
Xi

n
X

W 0i  X 0i

i1

where W 0i is the weight assigned to each parameter and X 0i is the


rating obtained for each parameter ranging from 1 to 5. In this sense,
to assign different importance levels to each factor and each parameter related to the factors, weighting is used at both levels (parameters and factors).
After the SQI estimation, a qualitative and quantitative scale can be
obtained providing a faster and more intuitive assess of the slope
condition. Furthermore, the ve dened classes can be linked to a risk
level for accidents that could occur in the slope (Table 1). For instance,
a slope with a SQI equal to 4, which is in a bad condition, shows a high
probability for accidents, meaning that an urgent action in the slope is
needed.

Level 1
SQI=[1;5]

Level 2
W=[0;1]

Level 3
X=[1;5]

Parameter 1
Factor A

Parameter 2
Parameter n

SQI
Factor N

Fig. 1. Scheme of the SQI methodology.

The original Rockfall Rating System (RHRS) comprise the evaluation of 10 different categories that are scored through an exponential
scale that ranges from 3 to 81, where higher values correspond to
higher risks. In [2] the authors presented some changes to this
method adding three categories and removing other three with the
aim of turning the system more tted to roadways and simplifying
their use (Table 2). The 10 categories of this modied version, the
RHRSm, are shown in the following paragraphs.
(a) Slope height: the risk is directly related to the slope height [2].
(b) Ditch effectiveness: this category evaluates the ability to
prevent a falling rock from reaching the road, so with a high
effectiveness this probability is reduced [8]. Some factors to
consider in this evaluation are: (a) height and slope angle;
(b) ditch sizes (height, width and depth).
(c) Average risk: the value for Average vehicle risk is obtained by
AVR

ADT  SL  100%
PSP

where AVR means Average Vehicle Risk and represents the spatial
probability of appearing a vehicle in the zone of rockfall hazard;
ADT means Average Daily Trafc, i.e., the average trafc per day
(vehicle/day); SL is the length of hazard zone (km) and PSP is the
speed limit of the roadway (km/h). An AVR higher than 100%
means that at any moment more than one vehicle is present
within the hazard zone [9].
(d) Decision sight distance (DSD): the weight of this parameter is
obtained according to the slope location. The risk of an accident
due to a falling rock is smaller when the visual distance of the driver concerning an obstacle increases [8,9].
(e) Roadway width: constant value measured along the road [8].
(f) Slope Mass Rating (SMR): the structural behavior of the rock
mass seems to have a high inuence in this global evaluation.
Hence, the presence of discontinuities, their orientation and

184

M. Pinheiro et al. / International Journal of Rock Mechanics & Mining Sciences 76 (2015) 182191

Table 2
RHRSm system (adapted from [2]).
Category

Slope height
Ditch effectiveness
Roadway width
Slope Mass Rating (SMR)
Block size
Boulder volume
Volume of rockfall per event
Rockfall history
AVR
Decision sight distance
Annual rainfall and freezing
periods

Rating criteria by score


3 Points

9 Points

27 Points

81 Points

7.5 m
Good catchment
21.5 m
80
0.3 m
26 dm3
2.3 m3
1 per 10 years
25%
Adequate
100%
h 300 mm or no freezing
periods

15 m
Moderate catchment
15.5 m
40
0.6 m
0.21 m3
4.6 m3
3 per year
50%
Moderate
80%
h 600 mm or short freezing
periods

22.5 m
Limited catchment
9.5 m
27
0.9 m
0.73 m3
6.9 m3
6 per year
75%
Limited
60%
h 900 mm or long freezing
periods

4 30 m
No catchment
3.5 m
20
1.2 m
1.74 m3
9.2 m3
9 per year
100%
Very limited
40%
h 1200 mm or long freezing
periods

the type of material should be evaluated according to this


empirical method [2,7].
(g) Volume of Rockfall per event/Block size: volume and size of
the blocks has signicant impact on risk and consequences of
rockfall.
(h) Rockfall history: slopes with track record of instabilities deserve more attention.
(i) Annual rainfall and freezing periods: unfavorable element for
the slope stability due to leaching and accumulation of water
in the rock fractures, such as the freezing cycles leading to a
faster deterioration of the rock mass [8].

Table 3
Resume of the RHRSm2 system and the corresponding scale on the SQI.

According to Budetta [2] and Budetta et al. [9], there is still a


need for the application of the RHRSm in different geological
environments to better validate the method. Pina [10] underlined
that this method shows yet some limitations, since it presents
subjectivity in the evaluation of some parameters while others
directly receive their score. The author also refers to the subjectivity and interpretation linked to the person who is giving
these scores, however the system shows good results and seems to
support well the required slope analysis in the exploitation stage.
In the scope of the development of the SQI it was decided to adopt
the RHRSm system to evaluate the quality of the rock slopes associated with the rockfall potential. However, some adaptations were
carried out and the new adapted system was called RHRSm2.
First of all, the nal classication of the RHRSm, i.e., the scale that
denes the risk of falling blocks was changed in order to obtain not
three levels of hazard, namely low, medium and high, but ve levels
(Table 3). This change allows a simpler integration of the overall
system into the SQI since it also complies ve categories.
Additionally, some new categories were added or changed to
the initial 10 in order to complement the RHRSm system. Those
categories are described in the following paragraphs.

category is linked to the block size, already integrated in the


initial categories of the RHRSm system.
(d) Climate: high rainfalls may create destabilizing phenomena on
the slopes and increase the probability of failures.
(e) Platform width: some adjustments were carried out in this
category to consider more realistic values in the authors
opinion.
(f) Bench width: the inclusion of this new category was justied
due to a higher ability in the blocks catchment with increasing
bench width.

(a) Slope inclination: the inuence of this parameter is truly


relevant in rockfall occurrences. Slopes with higher inclination
present higher risk of rockfalls. Likewise, higher will be the
energy resulting from the impact of the block in the platform
and higher the rebound and the probability of accidents.
(b) Inuence zone: the distance between the slope foot and the
nearest road considerably inuences the risk of the falling
blocks. So, if the inuence zone of a slope is classied as Far,
this means that the distance between the foot of the slope and
the closest road is high and the corresponding SQI rating
should be a low value (1) since the danger to the road is minor.
(c) Rock quantity: the number of loose blocks with potential of
falling is important. Usually, the higher the number of loose
blocks the greater the probability of accidents. However, this

General classication of the slope

SQI

RHRSm2 score

Hazard degree

Intervention

o 54
54162
163324
325486
4487

Very low
Low
Medium
High
Very high

Medium term
Short term
Urgent
Instant

1
2
3
4
5

Thereby, the RHRSm2 can be applied according to Table 4.


2.3. SQI factors and parameter denition
Terzaghi [11] presented a classication for the main causes
concerning slope instability, which can be grouped in internal, intermediate and external causes. In the internal causes are included all
the phenomena leading to an alteration of soil/rock parameters. The
externals causes cover geometric alterations, natural and articial
surrounding vibration, etc. The intermediate happen with the combination of the previous ones. In the SQI development a set of internal
and external causes were considered.
The factors and parameters of the SQI were set based on the
authors experience and opinions gathered with other experts in rock
slope stability. Moreover, important inputs were obtained in a set of
documents, namely: (i) the reports of Estradas de Portugal [12] and
Technological Research Institute [13] regarding the risk map for slopes
and riverbanks; and (ii) In [1417] for the necessary parameters to the
stability classication of rock slopes. The parameters were gathered in
9 groups named factors as shown in Table 5.
Inasmuch as this method contemplates a partial rating for each
factor, it provides a clear perception of which are the factors/parameters with the highest inuence in the quality condition. Therefore,
the SQI can provide a quantitative evaluation of the slope quality, the

M. Pinheiro et al. / International Journal of Rock Mechanics & Mining Sciences 76 (2015) 182191

185

Table 4
RHRSm2 system.
Category

Rating criteria by score

Slope height
Slope angle
Bench width
Ditch effectiveness
Inuence zone
Roadway width
Structural Condition
Weathering degree
Erosion
Block size
Volume of Rockfall per
event
Block quantity
Face irregularity
Water presence
Rockfall history
AVR
Decision sight distance
Climate

3 Points

9 Points

27 Points

81 Points

7.5 m
o 301
Z4 m
Good catchment
Far
21.5 m
Discontinuous joints, favorable
orientation
Small W1
Small
0.3 m
2.3 m3

7.522.5 m
301401
23 m
Moderate catchment
Medium
15.5 m
Discontinuous joints, random
orientation
Moderate W23
Moderate
0.6 m
4.6 m3

22.530 m
40501
r1 m
Limited catchment
Close
9.5 m
Discontinuous joints, adverse
orientation
Large W34
Large
0.9 m
6.9 m3

430 m
4501

No catchment
Very close
3.5 m
Continuous joints, adverse
orientation
Extreme W5
Extreme
1.2 m
9.3 m3

None
Low
Dry
Few falls
1/10 year
25%
Adequate
100%
Low rainfall
o 1000

Few
Medium
Humid
Occasional falls
3/year
50%
Moderate
80%
Moderate rainfall
10002000

Some
High
Dripping
Many falls
6/year
75%
Limited
60%
High rainfall
4 2000

Many
Very high
Fluid
Constant falls
9/year
100%
Very limited
40%

Table 5
Factors and parameters considered in the SQI system.
Factor

Parameter

Geometry
Geological
Drainage system
Inspections
Monitoring
Surroundings
Historical
Protection
Environmental/trafc

Height and inclination of the slope and width of the bench


Empirical classication systems (RMR, SMR or Q). Type of formation and risk of rockfall (RHRSm2)
Surface and deep drainage system (existence and conservation state)
Maintenance and conservation state evaluation
Results from monitoring systems including: inclinometers, topographic marks, piezometers, etc.
Existence of overloads (houses, etc.) and possible vibrations (works, etc.)
History of accidents on the slope and interventions
Surface protection (metallic mesh, bolts, etc.) and vegetal cover
Seismic zone, precipitation and trafc level

urgency of an intervention and in which parameters this intervention


should focus to provide the highest impact in the slope quality.
In order to obtain a value for each factor, a scale from 1 to 5 was
dened for each parameter. The denition of the intervals for each
parameter was carried out using existing references and adapting
some of the recommendations based on the authors experience, as
already referred [1418]. When these references were absent the
resource was the authors experience validated with discussions with
other experts. Nevertheless, these intervals are open for future
updating if accumulated experience points out to different values.
For example, to evaluate the geological factor one of the three
empirical classication systems for rock masses have to be applied,
namely: Rock Mass RatingRMR [19], Q [20], or SMR [7]. The RMR
system has ve classes: very poor, poor, fair, good and very good.
However, the Q system presents another additional class: Extremely
poor. So, in order to group the six classes the extremely poor and very
poor classes were merged in only one class. This option was made
since a rock with an extremely poor or poor quality class is rated with
a value of 5, which is the worst rating possible in the SQI system. This
factor also includes the rockfall hazard evaluation using the adapted
form of the RHRSm, the RHRSm2. As already stated, to simplify the
integration of the RHRSm2 system into the SQI, ve hazard categories
were dened, which are presented in Table 3.
In Appendix A all the factors, parameters, weights and ranges of
values necessary to calculate the SQI are presented.

The level of information existing in a slope network is variable and


the SQI system takes that into account through a factor and parameter
weight distribution. If some information concerning the nine factors
and the corresponding parameters is missing, the corresponding
weight should be distributed proportionally by the remaining.
2.4. Factor weight denition
Since the inuence of each factor in the slope stability is not the
same, each one was weighted by a coefcient that measures its
importance and inuence degree. To dene those weights a survey
was developed and distributed to a group of professionals composed
by practitioners that work in the company associated to the research
project in which the present work is based and academics of the
scientic area. All of them are Portuguese and work mainly in the
elds of rock mechanics and slope stability. In this survey the
professionals had to compare the relative importance of the factors
in a scale from 1 to 9, with the lowest value meaning the same
importance and 9 an extremely higher importance of the factor at
stake. The importance degrees were dened using the Analytic
Hierarchy Process method, proposed by Saaty [21].
The surveys were created in a digital platform to facilitate
distribution and the analysis of the results. The professionals were
grouped in three categories based on their level of knowledge in the
eld, namely: Expert, High knowledge and Regular knowledge. As

186

M. Pinheiro et al. / International Journal of Rock Mechanics & Mining Sciences 76 (2015) 182191

Regarding the parameter weight distribution, for instance if a slope


does not have the information concerning the type of formation, this
weight should be distributed by the remaining parameters of the
Geological factor. This weight distribution should be made in proportion to the assigned weight for each parameter (by degree of importance).

such, this division allowed, not only weigh the answers according to
the level of knowledge, but also analyze the existing differences in the
answers between each level.
A total of 31 answers were obtained from the survey showing a
great variety of answers from the three levels of knowledge. To analyze the results, the calculation methodology proposed by Liu and
Chen [1] was used. In Fig. 2, the weights of each factor are presented
divided by level of knowledge. It is possible to observe that there are
no signicant differences between the weights assigned by professionals with different levels of knowledge, aside from the monitoring
and visual inspections factors that are more valued by the Expert level.
To obtain the nal weights three different scenarios were considered: (1) same importance in the answers for the different knowledge levels; (2) increase in 20% the importance of the answers from
the Expert level and decrease in the same proportion the answers
from the professionals with Regular level of knowledge (Fig. 3); (3) the
same as in the previous case but considering a 30% variation (Fig. 3).
It was concluded that the differences between scenarios are rather
small. For practical purposes when applying the SQI users can adjust
these weights according to their experience and specic cases. In this
work the weights given in scenario 2 were adopted (Table 6).
Different weights were also assign for the different parameters
within the factors. The authors based on their experience and
discussion with other experts directly assigned these values. The
importance of all the factors and parameters regarding the SQI are
presented in Appendix A (the values placed below each name of
factor/parameter).
As already referred, the information concerning the factors and
parameters of the slope can be limited in some cases. As such, it is
recommended that in these cases, the weight of the missing factor/
parameter should be proportionality distributed by the remaining
ones in order that the outcome of the system still varies from 1 to 5.
Regular
18

17

16

3. SQI application
The SQI system was applied to a randomly chosen group of 17 rock
slopes located in the Beira Alta e Litoral highway Concession managed
by Ascendi and located in the west coast of Portugal. The overall result
for the 17 rock slopes are presented and for one specic slope the
calculation of SQI is presented in detail. For this set of slopes a score
equal or higher than 3.5 (Bad and Very Bad state) was dened to
activate the security alert that can result in monitoring systems
placement or even an emergency intervention for the worst cases.
In this group of slopes there were no monitoring data therefore the
weight of this factor (W0.11) was proportionally distributed by the
remaining eight factors. Furthermore the information concerning the
empirical systems (RMR, Q or SMR) was absent, so their weights were
distributed by the remaining parameters of the geological factor
proportionally. Given the existing information it was possible to apply
the RHRSm2 system to all the 17 slopes.
To validate the SQI results, an evaluation was performed through a
panel of experts. This panel of experts was composed by a part of
professionals involved in the surveys with a high level of knowledge
in slope evaluation. It was asked to this panel to assign a quality to the
slopes in a range from 1 to 5 after an in situ inspection. It is important
to point out that the goal of this system is allowing experts and nonexperts (with a basic knowledge in slope stability) the application of
High

15 15

15

14
14

14

13
12

Weight (%)

Expert

17

12

12

11

11

11

10

10

10

10

10

10

9
8

9
8

8
6
6
4
2
0
Geometry

Geological

Drainage system

Inspections

Monitoring

Surroundings

Historical

Protection

Environmental/
Traffic

Fig. 2. Factor weights distribution by the degree of knowledge.


Scenario 1

Scenario 2

Scenario 3

25

Weight (%)

20

20
17

16

16
14 14

15

10

10

11 11

12

13 13
11 11 11
8

9
7

10 10
7

5
0

Geometry

Geological

Drainage system

Inspections

Monitoring

Surroundings

Historical

Protection

Environmental/
Traffic

Fig. 3. Comparative analysis between scenario 1, 2 and 3: factor weights considering a difference of 20% and 30% between the 3 levels of knowledge.

M. Pinheiro et al. / International Journal of Rock Mechanics & Mining Sciences 76 (2015) 182191

evaluation for the 17 slopes. It can be stated that there are no


signicant differences between both evaluations.
For a quantitative comparison, the Root Mean Square Error (RMSE)
and the Mean Absolute Deviation (MAD) were calculated, as follows:
s
P
SQI  E2
RMSE
4
N

the system in a systematic way (since normally in this kind of infrastructures a very high number of slopes need to be assessed) and in a
less user-dependent mode achieving a slope quality evaluation similar
to the one that could result from an expert evaluation. This validation
allows concluding that the systematic application of the SQI will lead
to a classication not far from the one that an expert would provide.
Since the availability of experts may be scarce this is an important
asset of the system. The panel of experts assigned this value without
knowing the result of the SQI. Table 7 presents the SQI and the expert

MAD

Table 6
Factor weights adopted for the SQI system considering scenario 2.
Factor

Weight

Geometry
Geological
Drainage system
Inspections
Monitoring
Surroundings
History
Protection
Environmental/trafc

0.17
0.14
0.11
0.13
0.11
0.09
0.07
0.10
0.08

SQI

Expert evaluation

Value
2.54
2.99
2.89
2.85
2.11
2.76
2.92
2.93
2.70
2.60
2.85
2.35
3.27
2.25
2.84
3.03
2.91

Value

Medium
Medium
Medium
Medium
Good
Medium
Medium
Medium
Medium
Medium
Medium
Good
Medium
Good
Medium
Medium
Medium

State

3.00
3.00
3.50
2.50
2.50
3.00
3.80
3.00
2.50
3.00
2.80
2.50
3.50
2.00
3.00
2.90
2.80

Medium
Medium
Bad
Medium
Medium
Medium
Bad
Medium
Medium
Medium
Medium
Medium
Bad
Good
Medium
Medium
Medium

Table 8
Confusion matrix between the SQI values and the experts evaluation.
Expert evaluation
State
SQI value

S2
S3
S4

S2

S3

S4

1 (100%)
0
0

2 (15%)
11 (85%)
0

0
3 (100%)
0

1,00
0,90
0,80

Absolute difference

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17

State

N
1 X

jSQI  Ej
N i1

where N denotes the number of examples, SQI the obtained value for
the index and E the expert evaluation values. Values of 0.35 and 0.28
were obtained, respectively. These values corroborate the previous
conclusion of a good match between the SQI and expert evaluation.
Fig. 4 presents the absolute deviations between both assessments
for the 17 slopes. With the exception of two slopes, the absolute
difference is always lower than 0.5, which means that the SQI results
are in line with the expert evaluation and conrms its reliability. It
should also be noted that in all cases the error is lower than 1.
The calculation of the confusion matrix allows a better understanding of the SQI accuracy. This matrix compares the slopes in the
different states resulting from expert evaluation (columns) and the
SQI values (rows). For this set of slopes, only three states are
represented, the Good (S2), the Medium (S3) and the Bad (S4). As
such, the confusion matrix presents a 3  3 dimension (Table 8).
The overall accuracy of the SQI system in comparison with the
expert evaluation is higher than 70% in what concerns the
prediction of the slope state. In the cases the prediction is not
the same, the difference is of only one state (Good to Medium for
instance) with a tendency from the experts to be more cautious in
their evaluations.
In order to clarify the SQI application the overall process is
presented for slope 13 (Fig. 5). The score obtained by the application
of the RHRSm2 system for this slope was 432 translating a SQI

Table 7
SQI results and experts evaluation.
Slope

187

0,70
0,60
0,50
0,40
0,30
0,20
0,10
0,00

10

11

12

13

14

Slope number
Fig. 4. Absolute deviations between the SQI and the experts assessments for the 17 slopes.

15

16

17

188

M. Pinheiro et al. / International Journal of Rock Mechanics & Mining Sciences 76 (2015) 182191

Fig. 5. General view of the selected slope.


Table 9
SQI application for slope 13.
Factor

Weight

Parameter

Slope

Rating

Geometric

0.19

Geological

0.16

Drainage system

0.14

Visual inspections

0.12

History

0.08

Environmental/trafc

0.09

Protection

0.11

Surroundings

0.11

Height
Slope angles
Bench angle
Bench width
Type of formation
Formationweathering degree
Formationfaults
BlocksRHRSm2
Empirical systemQ, RMR or SMR
Surface drainagemaintenance state
Surface drainageconservation state
Surface drainagepresence
Deep drainagepresence
Bench drainage
Conservation state
Maintenance state
Accidents on sloperockfall
Accidents on slopeplane
Accidents on slopewedge
Accidents on slopecircular
Interventions
Seismic zonetype 1
Seismic zonetype 1
Annual rainfall (mm)
Trafcmaximum speed (km/h)
TrafcTMD
Surface protection
Vegetal cover
Overload
Surroundings vibrations

22 m
561
Correct
3m
Schist and greywacke
W3W4
Yes
432

2
1
Yes
No
None
4
2
Active
None
Active
None
Yes-Level 2
1.6
2.4
1700 mm
120 km/h
21,800 Vehicles
o 25%
None
None
None

3
4
1
2
4
4
5
4
?
2
1
1
5
5
4
2
4
1
4
1
3
1
2
4
5
4
5
5
1
1

parameter related to rockfalls of 4, meaning a slope with high risk of


rockfalls. This information by itself is important in the planning of
future interventions in this slope since this rating point out to the
necessity of urgent stabilization measures to prevent rockfalls. The
details of the SQI system are presented in Table 9.
The SQI for this slope was 3.27, meaning that it is in a Medium
state and requires, at the present moment, mainly maintenance
actions, despite the urgent need to apply measures to prevent rockfalls. Furthermore, a more detailed analysis of the SQI factors allows
the identication of the most critical factors, which in this case are the
geometric and the geological (sum of the two factors higher than 1.3).
Therefore, maintenance interventions focusing on these factors are
expected to have the highest impact on the slope quality. Ultimately, it
is important to emphasize that the obtained SQI value translates very
well the real condition of the slope.

4. Conclusion
In this paper a new system for rock slope stability analysis during
the exploitation stage called Slope Quality Index (SQI) system was

presented. It allows the calculation of an index that considers the


evaluation of nine different factors. Within these factors a number of
parameters have also to be assessed with scores ranging from 1 to 5.
To translate different levels of inuence of the factors and parameters
in rock slope stability weights were assigned to each one of them. The
parameters weights were quantied according to the authors experience where as for the factor weights a more complex methodology
was adopted, involving a survey that was distributed to professionals
that work in this eld of study.
The SQI scale varies from 1 to 5, meaning respectively very good
and very bad slope quality. Thereby, it was dened that for a score
equal or higher than 3.5, a security alert for the slope should be
activated.
The SQI was applied to a group of 17 rock slopes and the results
were compared to an evaluation made by a panel of experts. Only
small differences between them were observed showing that the SQI
system provides a realistic overview about the slope condition and can
surely be used as a quantitative and qualitative evaluation system for
slopes in exploitation stage. This validation allows concluding that the
systematic application of the SQI will lead to a classication not far
from the one that an expert would provide. Even if only there of the

M. Pinheiro et al. / International Journal of Rock Mechanics & Mining Sciences 76 (2015) 182191

ve slope classes are represented in the results, the authors belief that
considering a great amount of information as input and the very good
behavior for this set of 17 slopes, the SQI should perform at least
acceptably in the remaining classes. Moreover, the SQI also allows the
identication of the most critical and important factors contributing
for the overall condition of the slope, which provides an important aid
in the decision making process and planning of interventions.
In conclusion, the SQI system has a signicant interest for
companies and institutions that have to manage a great number of
slopes in the scope of transportation infrastructures since it
provides a realistic evaluation of slopes not only to identify the
slopes in bad and very bad condition (extreme values) but also to
allow mapping the overall condition of the slopes network. This
mapping can result in a more rational management in the
planning of short and long-term interventions and analyze the

189

condition evolution of the slopes (individually or in a more macro


level).

Acknowledgements
The authors would like to thank AdIInnovation Agency, for the
nancial support awarded through POFC program, for the R&D project
SustIMSSustainable Infrastructure Management Systems (FCOMP01-0202-FEDER-023113).

Appendix A
See Table A1.

Table A1
SQI table.
Factors

Parameters

Categories and ratings

Geometry 0.17

Slope Height (m)


0.5
Rating
Slope angle (1)
0.35
Rating
Bench angle
0.15
Rating
Bench width (m)
0.25
Rating
Formation type

o 10
Very low
1
o 30
Very gradual
1
Correct2

1020
Low
2
3040
Gradual
2
Incorrecta

2030
Medium
3
4150
Medium
3

3040
High
4
5160
Inclined
4

440
Very high
5
460
Too inclined
5

127
01

457
12

23

34

44

5
I

4
II

3
III

2
IV

1
V

1
Unweatered
(W 1)

2
Slightly
(W 2)

3
Moderately
(W 3)

4
Highly
(W 4)

5
Decomposed
(W 5)

1
Exist

457
o 51

51153

153333

333459

4459

1
401000

2
1040

3
410

4
14

5
0.0011

1
10081

2
8061

3
6041

4
4021

5
200

1
10081

2
8061

3
6041

4
4021

5
200

1
Very good

2
Good

3
Medium

4
Bad

5
Very bad

1
Good

2
Medium

3
Bad

1
Yes

2
No

127
Yes

457
No

127
Yes

457
No

127
Very good
1
Good
1

457
Good
2
Medium
2

Medium
3
Bad
3

Bad
4

Very bad
5

Geological 0.14

0.50

Blocks
0.20
Empirical systems

(Only one system: Q, RMR or


SMR)

0.30
Drainage system 0.11

Surface drainage

0.60

Deep drainage
0.20
Bench drainage
0.20
Rating
Visual inspections 0.13 Conservation state
0.60
Maintenance state
0.40

Typeb
0.40
Rating
Weathering degree
(W)c
0.30
Rating
Faultsd
0.30
Rating
RHRSm2
1.00
Rating
Q
0.33
Rating
RMR
0.34
Rating
SMR
0.33
Rating
Conservation state
0.35
Rating
Maintenance state
0.45
Rating
Presence
0.20
Rating
Presence
1.00
Rating

Classication
Rating
Classication
Rating

190

M. Pinheiro et al. / International Journal of Rock Mechanics & Mining Sciences 76 (2015) 182191

Table A1 (continued )
Factors

Parameters

Monitoringe 0.11

Load cells
0.25

Categories and ratings


Intervals
Rating
Intervals

Inclinometers
0.25

Rating
Intervals

Piezometers
0.25
Topographic marks
0.25
Historical 0.07

Accidents on the slope

Rating
Intervals
Rating
Rockfall
0.25
Rating
Plane
0.25
Rating
Wedge
0.25

1
[020%]
1
1
[0%20%]
1
1
[020%]
1
1
[020%]
1
None

2
[2040%]
2
2
[20%40%]
2
2
[2040%]
2
2
[2040%]
2
Inactive

3
[4060%]
3
3
[40%60%]
3
3
[4060%]
3
3
[4060%]
3
Some

4
[6080%]
4
4
[60%80%]
4
4
[6080%]
4
4
[6080%]
4
Active

5
[80100%]
5
5
[80%100%]
5
5
[80100%]
5
5
[80100%]
5
Very activef

1
None

2
Inactive

3
Some

4
Active

5
Very active

1
None

2
Inactive

3
Some

4
Active

5
Very active

1
None

2
Inactive

3
Some

4
Active

5
Very active

1
Level 3

2
Level 2

3
Level 1

127
2.1

3
2.2

457
2.3

2.4

2.5

5
1.1

4
1.2

3
1.3

2
1.4

1
1.51.6

5
o100

4
100500

3
5001000

2
10002000

1
42000

1
5060

2
6070

3
7090

4
90100

5
100120

1
o18,000

2
18001900

3
19002000

4
20002200

5
422,000

1
o25%

2
[2550%]

3
50%

4
[5075%]

5
[75100%]

5
Non exist

4
Punctual

3
Uniform

5
Yes

43i
No

127

5
Yes

1
No

0.70
Rating
Circular
0.25
Rating

Environmental/trafc
0.08

Interventionsg
0.30
Rating
Seismic zoneh
0.30

Type 2
0.60
Rating
Type 1
0.40
Rating

Annual rainfall (mm)


0.50
Rating
Trafc

0.20

Protection 0.10

Maximum speed
(km/h)
0.50
Rating
Average daily trafc
(TMD)
(Vehicles)
0.50
Rating

Surface protection
0.80
Rating
Vegetal cover
0.20
Rating
Overload
0.60
Rating
Surrounding vibrations
0.40
Rating

Surroundings 0.09

Type

II

III

IV

Formations
description

Metamorphic:
Gneiss, quartzite,
and Anbolite
Migmatite;
igneous:
Granite,
Granodiorite,
Diorite and
Gabbro

Metamorphic: Cornean;
Sedimentary: Conglomerate;
Igneous: Andesite, Norite, Obsidian
and Dolerite

Sedimentary: sandstone and


greywacke; Igneous: Basalt,
Tuff, Brechia, Dacite and
rhyolite

Metamorphic: Phyllite and


Metamorphic: Shale,
Slate; Sedimentary:
Milonite marble;
Sedimentary: Gypsum and Limestone, Siltstone and
mudstone
Anhydrite

M. Pinheiro et al. / International Journal of Rock Mechanics & Mining Sciences 76 (2015) 182191

191

Table A1 (continued )
Table A1 (continued)
Level

Group

Surface protection

Drainage system

Block stability

Support structures

Interventions
description

Use of natural or articial


materials to improve the
stability of the slope, like
vegetation cover, ditch
construction, etc.

Includes all the actions that


could be made in the drainage
system (deep and surface)

Removal of the blocks and/or


retention using metallic grids or
nailing

Retaining wall construction


that can be of concrete,
masonry, reinforced soil, etc.

For a Correct rating the angle of the bench should be opposed to the slope angle.
The rock types were set based on the groups dened by Naghadehi et al. [17] and can be modied and adapted according to the typical types of formations existing
in situ where the SQI would be implemented.
c
The weathering degree was established in Bieniawski [20].
d
This parameter should only be scored if faults present an unfavorable orientation for slope stability. If faults do not exist, this parameter should not be considered and
the corresponding weight should be distributed for the remaining parameters.
e
The monitoring factor aims to evaluate, using percentage ranges, the ratio between the measured value in each instrument and the threshold dened in the
design phase.
f
None: No registered accident; Inactive: small/medium scale accident in a 10 years time space; Some: small/medium scale accident in a 5 years time space; Active: small/
medium scale accident in a 3 years time space and large accidents in a 1 year time space; Very Active: small, medium and large scale accidents in a 1 year time space.
g
The following actions and corresponding levels were dened according to the company responsible for the highway concession in study.
h
The ranges of values were dened for each seismic action and zones using the information existing in Eurocode 8NA-3.2.1 (2).
I
The ratings are presented by range of values with the purpose of better reect the subjectivity inherent of this evaluation. The user should be able to decide if to a
specic parameter should be attributed the maximum rating (5), or if the same parameter do not have a high inuence on the SQI that justies a so high rating.
b

References
[1] Liu YC, Chen CS. A new approach for application of rock mass classication on
rock slope stability assessment. Eng Geol 2007;89:12943.
[2] Budetta P. Assessment of rockfall risk along roads. Nat Hazards Earth Syst Sci
2004;4:7181.
[3] Pierson LA, Davis SA, Van Vickle R. Rockfall hazard rating system implementation manual. In: Report FHWA-OR-EG-90-01. Federal Highway Administration,
US Department of Transportation; 1990. p. 80.
[4] Franklin JA, Senior SA. The Ontario Rockfall Hazard Rating System. In:
Proceedings of the international conference on engineering, geology and the
environment, Balkema, Rotterdam; 1997. p. 64756.
[5] Youssef A, Maerz NH, Fritz MA. A risk-consequence Rockfall hazard rating
system for Missouri highways. In: Proceedings 54th highway geology symposium, Burlington, Vermont; 2426 September 2003. p. 17595.
[6] Alejano LR, Stockhausen HW, Alonso E, Bastante FG, Ramirz Ovanguren P.
ROFRAQ: a statistics-based empirical method for assessing accident risk from
rockfall in quarries. Int J Rock Mech Min Sci 2008;45:125272.
[7] Romana M. (1985). New adjustment ratings for application of Bieniawski
classication to slopes. In: Proceedings of the international symposium on
the role of rock mechanics. ISRM, Zacatecas, Mexico; 1985. p. 4953.
[8] Nunes A, Guedes P, Cardoso A. Instability analysis in excavation rock slopes in
road environmental. Case study on EN22 road between Regua and Ponte de
Batedeiras. III Portuguese highway congress, Portugal; 2004. p.10. (in
Portuguese).
[9] Budetta P, Nappi M. Comparison between qualitative rockfall risk rating
systems for a road affected by high trafc intensity. Nat Hazards Earth Syst
Sci 2013;13:164353.

[10] Pina L. Quantitative risk assessment of rockfall hazard in the amal coastal
road. Barcelona, Spain: Politechnique University of Catalua; 2008. p. 90.
[11] Terzaghi K. Mechanism of landslides. Harvard University, Department of
Engineering; 1950. p. 41.
[12] Estradas de Portugal S.A. Contract document for work type, Volume VIII:
Chapter 15.06/07 Bridges construction Overpasses and intersections type
of works; 2009. (in Portuguese).
[13] Carvalho CS, Macedo ES, Ogura AT. Mapping hazards in slopes and riverbanks.
Brasilia, Brazil: Institute for Technological Research- IPT; 2007. p. 176 in
Portuguese.
[14] Gao Y, Zhang Y, Jiang L, Huang Z. Researches on index system of rock slope
safety evaluation for open pit mine. Procedia Eng 2011;26:16927.
[15] Lindsay P, Campbell RN, Fergusson DA, Gillard GR, Moore TA. Slope stability
probability classication, Waikato Coal Measures, New Zealand. Int J Coal Geol
2001;45:12745.
[16] Pantelidis L. Rock slope stability assessment through rock mass classication
systems. Int J Rock Mech Min Sci 2009;46:31525.
[17] Naghadehi MZ, Jimenez R, Khalokakaie R, Mohammad S, Jalali E. A new openpit mine slope instability index dened using the improved rock engineering
systems approach. Int J Rock Mech Min Sci 2013;61:114.
[18] McMillan P, Matheson GD. A two-stage system for highway rock slope risk
assessment. Int J Rock Mech Min Sci 1997;34(3-4):196.
[19] Bieniawski ZT. Engineering rock mass classications. New York, NY: Wiley;
1989. p. 1989.
[20] Barton N, Lien R, Lunde J. Engineering classication of rock masses for the
design of tunnel support. Rock Mech 1974;6:189236.
[21] Saaty TL. The analytic hierarchy process. New York, NY: McGraw-Hill; 1980.

You might also like