Vous êtes sur la page 1sur 16

A new general empirical approach for the prediction of rock mass strengths

of soft to hard rock masses


O.S. Dinc
a
, H. Sonmez
b,n
, C. Tunusluoglu
a
, K.E. Kasapoglu
b
a
Canakkale Onsekizmart Universitesi, Department of Geological Engineering, Applied Geology Division, C- anakkale, Turkey
b
Hacettepe University, Department of Geological Engineering, Applied Geology Division, 06800 Beytepe, Ankara, Turkey
a r t i c l e i n f o
Article history:
Received 28 June 2010
Received in revised form
3 January 2011
Accepted 4 March 2011
Available online 31 March 2011
Keywords:
Disturbance
Failure criterion
Hoek and Brown criterion
Rock mass strength
Soft rock mass
a b s t r a c t
It is almost impossible to prepare representative cores of rock masses including discontinuities patterns
for laboratory studies. To overcome these difculties, researchers have focused on developing empirical
equations for estimating of the stressstrain behavior of a rock mass, including measurements of the
discontinuity patterns. As can be seen in the literature, the uniaxial compressive strength value of rock
mass (UCS
RM
) can be estimated by reducing the uniaxial compressive strength of intact rock material
(UCS
i
) based on the quality of a rock mass, represented by variables such as Rock Mass Rating (RMR),
Geological Strength Index (GSI) and Q value. For this reason, a unique reducing curve form empirical
equation has limited application and generally, cannot be applied to all kind of rock masses from
particularly soft to hard rock masses. In this study, a new general empirical approach is constructed to
estimate the strength of rock masses of varying hardness. The new empirical equations have been
calibrated using data from ve slope failures and four sets of uniaxial compressive strength data of rock
masses. In the new empirical equations, the UCS
i
is considered not only to be a scale parameter used in
the strength reduction but also used to adjust the degree of strength reduction in conjunction with
elastic modulus of the rock material (E
i
). The disturbance factor on the rock mass is taken into
consideration by two separate reduction factors applied to the Structure Rating (SR) to capture
increasing joint density, and to the s and m
b
parameters of the HoekBrown criterion, to decrease
the degree of interlocking. Hence, non-interlocked (cohesionless under zero normal stress) rock masses
such as spoil piles can also be modeled in the new empirical approach.
& 2011 Elsevier Ltd. All rights reserved.
1. Introduction
As is well known, preparing of high quality core samples from
a rock mass containing discontinuities to perform laboratory tests
is very difcult due to the presence of discontinuities. To over-
come these difculties, researchers have focused on the studies
about developing empirical equations for estimating of the
stressstrain behavior of a rock mass, including measurements
of the discontinuity patterns. As a result of these studies,
numerous empirical equations have been proposed in the litera-
ture [111]. Most of these equations consider uniaxial compres-
sive strength of intact rock material (UCS
i
) as a scale parameter.
The UCS value of rock mass (UCS
RM
) can be estimated by reducing
the UCS
i
based on the quality of rock mass such as Rock Mass
Rating (RMR), Geological Strength Index (GSI), Q value, etc. [13].
At this time, the question of which empirical relation is the
best for predicting of the strength of a rock mass? cannot be
denitively answered. In fact each empirical relation has the
highest predictive capacity for those rock masses included in their
original databases created for development of the various empiri-
cal relations. For this reason, a unique reducing curve form
empirical equation has limited application to estimate UCS
RM
for
the vast range of rocks masses formed of the variety of rocks from
soft to hard.
Most of the empirical equations given in Table 1 consider just
RMR, GSI or the joint parameter (JP) as an input parameter. But
these characterization schemes may not explicitly include the
strength and deformability of intact rock material, which may
play an important role on the strength behavior of rock masses,
particularly for softer ones. For example, two rock masses having
similar discontinuity patterns may be composed of different rock
materials having different uniaxial compressive strengths and
deformabilities. Although the quality index of the two rock
masses (such as RMR or GSI) may be similar, the degree of
reducing ratio used to estimate UCS
RM
by reducing UCS
i
should
logically be different for the two cases. In fact, the degree of
reduction used in the case of the stronger strength and the less
deformable intact rock material, would be expected to be higher.
Contents lists available at ScienceDirect
journal homepage: www.elsevier.com/locate/ijrmms
International Journal of
Rock Mechanics & Mining Sciences
1365-1609/$ - see front matter & 2011 Elsevier Ltd. All rights reserved.
doi:10.1016/j.ijrmms.2011.03.001
n
Corresponding author. Fax:90 312 299 2034.
E-mail address: haruns@hacettepe.edu.tr (H. Sonmez).
International Journal of Rock Mechanics & Mining Sciences 48 (2011) 650665
However, almost all empirical equations in the literature produce
almost same UCS
RM
/UCS
i
ratios when the rock quality indexes of
the rock masses are similar (Fig. 1).
M uller [12,13] classied rock mass conditions from unfactured
rock (one piece system) to strongly fragmented rock mass (loose
rock mass-a lot of piece system) as illustrated in Fig. 2a.
M uller [13] stated that The stressstrain curve of a rock substance
does not end with the attainment of the maximum load (upper curve,
in a normal testing machine) but that, after the appearance of the
decisive rst fractures, the load bearing capacity of the rock drops
quickly or slowly with further deformation, depend on material and
rate of loading. Later, based on M ullers studies, nomenclature of
different rock mass conditions with stressstrain curve based on
system size was illustrated in detail as given in Fig. 2b [16]. As
discussed by M uller [13], the geological history of an intact rock
initially starts from an unfractured rock free from discontinuities
and ends as a mylonite composed of strongly fragmented rock
mass by tectonic forces, similarly post-failure behavior of rock
specimens in stiff testing machine. Evaluation of the stressstrain
behavior for post-failure part is much more difcult than pre-
failure part. In particular the behavior from the peak strength (for
intact rock material) to the residual strength (for completely
fragmented-crushed rock material) is almost impossible to obtain
by laboratory studies. However, the stressstrain curves of softer
rock mass and harder rock mass for the post-failure part should
be expected as different. The empirical equation proposed by
Vardar [6,14 from 16] considers types of rocks not only as a scale
parameter, but also in the degree of reduction applied on the UCS
i
.
However, the properties of discontinuities such as weathering,
inlling, roughness were not sufciently considered in this
empirical approach. As can be seen from Vardars equation given
in Table 1, while the number of Strength Reducing value (f)
depends on the rock type is considered, only the number of joints
(joint density) at the constructed area is used for dening the
quality of rock mass. According to the more recent number of
Strength Reducing values depending on the type of rock [15,16],
while the value of f for rock material such as obsidian (high
strength and low deformability) is 40, the value of f for softer
rocks such as claystone varies between 3 and 12. In other words,
the number of Strength Reducing values of rock mass decrease
when the strength of rock materials decreases and deformability
of the rock materials increases.
The number of Strength Reducing values in Vardars approach
represents the loss of strength by increasing of blockiness (or
joint density). However during geological history of rock materi-
als from intact (unfractured) rock to completely crushed material,
not only the joint density (or blockiness) of the rock mass
increases, but also the quality of the surface properties (such as
roughness, weathering and inlling) of the discontinuities
decreases. Therefore, the mechanical properties of intact rock
material and the surface condition of discontinuities should be
considered together to obtain more representative the number of
Strength Reducing values for the rock masses. Some recent
studies have focused on similar conclusions, particularly on the
overall strength of rock mass composed of softer rock materi-
als [1720]. These researchers dened a gradual transition from
interblock shear failure (which is well-modeled by the Hoek
Brown failure criterion (HB) and the GSI procedure) to rock mass
behavior mainly controlled by the intact parts of the rock mass
(Fig. 3). When the UCS
i
is greater than 15 MPa, the overall
strength of a rock mass (UCS
RM
) is dened by the conventional
unique curve relation of the HoekBrown failure criterion. The
UCS
i
of 0.5 MPa is selected as boundary between UCS of soft rock
material and UCS of the intact soil by the researchers. The impact
of the UCS
i
on the UCS
RM
gradually decreases from 15 to 0.5 MPa.
When the UCS
i
is equal to 0.5 MPa, the UCS
RM
is accepted as equal
to the UCS
i
for whole range of GSI. Although the lower and the
upper strength limits for softer intact rock being accepted as
0.5 and 15 MPa are reasonably well veried from numerical
modeling of [18], as indicated in [20] these limits are still open
to verication using eld observations. Hence some new con-
tributions were introduced to the HB empirical approach
by [1720], we suggest that some details still need to be explored
as indicated below.
As can be seen from Fig. 1, the empirical equations curves
occupy a zone of failure envelope instead of a unique curve
envelope. Consequently, in this study instead of a denite unique
curve form equation, we proposed a power type continuous
equation. The curves of equation become closer to the lower
bound (harder rock mass) given in Fig. 1 by increasing the UCS
i
and elastic modus of intact rock material (E
i
).
As given in Fig. 3, the normalized uniaxial compressive
strength (UCS
RM
/UCS
i
) becomes closer to 1 when the UCS
i
decreases to 0.5 MPa [20]. This implies that discontinuities such
Table 1
Some well-known selected empirical equations for estimation of the uniaxial
compressive strength of a rock mass.
Reference
Empirical equation
Yudhbir et al. [4]
UCSRM
UCSi
exp 7:65
RMR100
100
_ _
Ramamurthy [5]
UCSRM
UCSi
exp
RMR100
18:5
_ _
Vardar [6,14 from 16]
UCSRM
UCSi
a
1logNk
i=f
1logNk
i
Kalamaris and Bieniawski [7]
UCSRM
UCSi
exp
RMR100
24
_ _
Palmstr om [8]
UCSRM
UCSi
JP
Sheorey [9]
UCSRM
UCSi
exp
RMR100
20
_ _
Aydan and Dalgic [10]
UCSRM
UCSi

RMR
RMRb100RMR
_ _
b 6
Hoek et al. [11]
UCSRM
UCSi
exp
GSI100
9
_ _ _
a
UCS
RM
and UCS
i
: uniaxial compressive strength of rock mass and intact rock
material, respectively, RMR: Rock Mass Rating, GSI: Geological Strength Index, a:
curvature constant of Hoek and Brown criterion, a: ratio of reducing strength
of rock according to an-isotropic loading condition (0.2rar1), i: number of
inhomogenity, f: number of Strength Reducing value (3rf r40), N
k
: number of
rock pieces affected in the construction area (rock element number), JP: joint
parameter.
Fig. 1. Some well-known empirical relations plotted with normalized uniaxial
compressive strength (UCS
RM
/UCS
i
) and qualities of rock mass such as GSI or RMR.
O.S. Dinc et al. / International Journal of Rock Mechanics & Mining Sciences 48 (2011) 650665 651
as ssures in soils (which are clearly observed in some well
studied overconsolidated clay such as London clay and Ankara
clay) have little inuence on the overall strength of soil mass.
Many studies have been performed on the overall strength of
ssured clays which show crucial strength loss of intact over-
consolidated clay (ssure free samples) occur depending on the
properties of the ssures. Indeed, much can be learned from the
soil mechanics literature regarding the inuence of cracks. Some
important studies from the broad soil mechanics literature are
here summarized.
Terzaghi [21] rst described slope failures of gentle slopes in
ssured clays composed of stiff intact clay divided by ssures.
Bishop and Little [23 from 22] identied that the ratio of strength
of ssured soil mass to the strength of intact clay was obtained
about 45%. According to the experimental studies performed on
ssured Nanticoke clay by Lo [22], the overall undrained strength
of a soil mass decrease to approximately 25% of the intact clay
strength (Fig. 4). Based on the results of experimental studies,
decreasing on the strength was also found for Blue London clay by
Simons [24 from 22]. Marsland and Butler [25] also reported that
the mechanical behavior of ssured clays is very similar to jointed
rock masses, based on their studies performed on slope faces
excavated in ssured clays. Furthermore, a model proposed by
Silvestri [26] for ssured clays that as developed by Ladanyi and
Archambault [27] and Ladanyi [28] for dening jointed rock
masses. The relation between normal stress and shear strength
of ssured Champlain clay was evaluated to be as nonlinear by Lo
and Lee [29]. Feda et al. [30] indicated that although ssured clays
are treated as soils in many engineering applications, the actual
mechanical behavior of ssured clays (such as strength and
deformability) is very similar to the mechanical behavior of
jointed rock masses. Vallejo [31] reported that the use of the
linear MohrCoulomb failure criterion on ssured clays may lead
to error depending on the normal stress level. Sonmez [32]
evaluated the applicability of the Hoek and Brown failure criter-
ion for predicting the strength of Pliocene aged ssured Ankara
Fig. 2. (a) Illustrative classication of rock mass by M uller [12]. (b) Schematic illustration of the full scale stressstrength behavior of rock (from Vardar [16]).
Fig. 3. Normalized rock mass strength (UCS
RM
/USC
i
) as a function of Geological
Strength Index (GSI) showing the conventional HoekBrown relation and weak
and strong rock mass transition functions [20].
Fig. 4. The relation between sample size and strength for ssured Nanticoke
clay [22].
O.S. Dinc et al. / International Journal of Rock Mechanics & Mining Sciences 48 (2011) 650665 652
clay (Fig. 5). According to the results of the experimental studies
performed by Sonmez [32], the overall strength of macro-ssured
Ankara clay masses varies between the strength of the intact clay
and the residual strength of the intact clay depending on the
construction area and ssure properties such as spacing, orienta-
tion, inlling and roughness.
Hence, in summary; the geotechnical literature clearly indi-
cates that characterizations of the strength of ssured clays show
that they have very similar mechanical behaviors to jointed rock
mass. Based on the soil mechanics literature, the ndings suggest
that the normalized uniaxial compressive strength (UCS
RM
/UCS
i
)
can be expected to be considerably lower than 1 for jointed
(ssured) soil masses, and hence the upper bound of soft rock
masses should also be considered to be similar.
2. A new empirical approach: Compressive Strength Reducing
Ratio (SRR
c
)
In this study, the ratio of UCS
RM
to UCS
i
was named as
Compressive Strength Reducing Ratio (SRR
c
) similarly with Vardars
terminology [6,14 from 16], and an empirical approach was
introduced to estimate of SRR
c
.
More specically, it is written in literature that the UCS
i
is
taken into consideration as a scale parameter in the estimation of
strength of rock mass. In this study the UCS
i
and elastic modulus
of rock material (E
i
) were also taken into consideration for
contribution at adjusting the UCS
i
and estimation of UCS
RM
.
To overcome some limitations of the existing unique curve
form empirical equations given above, the features incorporated
in a new empirical approach to provide a better capability for
generalizing the overall strength of rock masses ranging from soft
to hard rock masses are: (1) the new empirical approach should
have applicability from intact rock to very weak rock masses
(such as heavily jointed/crushed and decomposed rock mass),
(2) the uniaxial compressive strength of rock material (UCS
i
) Fig. 5. Fissure sets in Ankara clay (a) no inlling and (b) carbonate inlling [32].
Fig. 6. The enhanced Structure Rating (SR
u
) (from Sonmez [32], Sonmez and Ulusay [34]) graph modied using the relations among J
v
, V
b
, S and RQD originally dened by
Palmstr om [38].
O.S. Dinc et al. / International Journal of Rock Mechanics & Mining Sciences 48 (2011) 650665 653
should be considered to be not only a scale parameter used in
strength reduction but also to be used in degree of reduction in
conjunction with the elastic modulus of the rock material (E
i
),
(3) the degree of blockiness of rock masses should be considered,
(4) the properties of discontinuity surfaces such as weathering,
inlling, roughness should be considered, (5) the inuence of
disturbance on the overall strength of rock masses should be
included, (6) the strength of non-interlocked (cohesionless under
zero normal stress) rock masses, such as spoil piles, should be
predictable.
Schematic illustration of the stressstrain curve of rock masses
from peak strength (for unfractured-intact rock) to residual
strength (for strongly fragmented-crushed material such as
mylonite) given in Fig. 2b was considered during the construction
of the new empirical approach as presented below.
2.1. Denition of rock mass structure in the new empirical approach
In this study, the Structure Rating (SR) and the Surface
Condition Rating (SCR) are taken into consideration to dene
quality of rock mass.
The SR proposed by Sonmez and Ulusay [33] and modied by
Sonmez [32] and Sonmez and Ulusay [34] was used in the new
empirical approach to dene the blockiness of rock masses. The SR
value of a rock mass can be determined by the Volumetric Joint
Count (J
v
) originally developed by Palmstr om [3537]. Although
some practical procedures were adopted by Sonmez and Ulusay [34]
particularly for heavily jointed rock masses, the relations among J
v
,
discontinuity space (S, or average block dimension), average block
volume (V
b
) and Rock Quality Designation (RQD) for the common
block shape by Palmstr om [38] is introduced to the SR concept to
increase its practical value. The enhanced form of the SR graph is
given in Fig. 6.
As shown in Fig. 6, the SR value varies from 100 for intact rock
material or massive rock mass to zero for very weak rock masses
such as heavily jointed/crushed rock masses. As discussed in
detail by Palmstr om [38], the RQD as a borehole data (one
dimensional) is actually limited in application for very weak
(heavily jointed) rock masses and rock masses having widely
spaced discontinuities (see bottom axis of Fig. 6). The values of
RQD have not been used alone to dene quality of rock mass, but
it was used as a parameter in some rock mass classication such
as RMR and Q [1,3]. Indeed any 1-D borehole data such as total
core recovery (TCR), intact core recovery (ICR) or the RQD should
be used in caution given their limitations for classication of
particularly heavily jointed rock masses. However, the value
of the SR determined by J
v
, V
b
and/or S is a more sensitive
measure of the blockiness of very weak rock mass such as heavily
jointed/crushed rock masses. Indeed, the main advantage of the
use of the SR is the full scale characterization of the various
structures of rock masses ranging from intact (unfractured) rock
materials to heavily jointed/crushed rock masses.
To dene the full range of stressstrain curve from intact rocks
to heavily jointed/crushed rock masses, the S-shaped power-type
function is preferred, although exponential type curves are
commonly seen in the literature [4,5,7,911] as given in Fig. 1.
The S-shaped equation was used by Sonmez et al. [39] and Hoek
and Diederichs [40] for estimation of deformation modulus of
rock masses. In this study, the S-shaped curve form equation used
by Sonmez et al. [39] was preferred because the S-shaped curve
has the capability to dene full scale stressstrain curve particu-
larly for the SR values close to 100 which encompass jointed one-
piece systems or intact-massive rock material. Otherwise a sharp
decrease is observed in exponential type curves particularly for
the SR values close to 100.
The number of Strength Reducing value for the softest rock
masses (such as heavily jointed claystone) is about 3 [15,16]. In
other words, the ratio of UCS
RM
to UCS
i
for heavily jointed/crashed
claystone rock mass was about 33%, according the Vardars
approach [6,14 from 16]. Furthermore, the ratio of the overall
strength of a ssured clay soil mass to the intact strength varies
between 25% and 45%, based on the studies performed by Lo [22],
Bishop and Little [23 from 22].
By considering the values of Strength Reducing Ratios for the
softest (claystone) rock mass and jointed soil mass provided in
the literature, the Compressive Strength Reducing Ratio of UCS
RM
to UCS
i
was set to approximately 25% when SR0 at the boundary
between heavily jointed softest rock masses and heavily jointed
(ssured) soil mass (Fig. 7). Hence, the upper bound of the
S-shaped curve representing the full scale stressstrain curve
for the softest rock mass, was calibrated by considering UCS
RM
/
UCS
i
25% when SR0 (Fig. 7). On the other hand the lower
bound of the S-shaped curve, which represent the full-scale
stressstrain curve for the hardest rock mass, was calibrated by
using the lower bound of exponential curves seen in Fig. 1.
The parameter C is introduced to the S-shaped empirical
equation to calibrate the curves between the upper and lower
bounds (Fig. 7), and depends on both the type of rock and the
condition of discontinuity surfaces in the rock mass. When the
value of C decreases to one, the S-shaped curve moves upward.
The lower bound (the hardest rock mass) and upper bound (the
softest rock mass) are xed considering the boundary conditions
discussed above by calibrating of C. The values of C were obtained
as 10 and 1.75 for lower bound and upper bound, respectively
after calibration.
Fig. 7. Graphical illustration of new empirical approach for estimation of the
overall uniaxial strength of rock mass.
O.S. Dinc et al. / International Journal of Rock Mechanics & Mining Sciences 48 (2011) 650665 654
2.2. The inuences of type of rock and properties of discontinuities
In the new empirical equation, the type of rock material is
described by its strength and deformability characteristics as
proposed by Deere and Miller [41]. The effect of strength and
deformability of intact rock material on the overall strength of a
rock mass was inuenced in the new empirical approach.
To dene the effect of strength and deformability of rock
material, a parameter A representing the hardness of intact rock
material, was introduced to the new empirical equations. The
value of A can be selected from lines on the modulus ratio graph
of Deere and Miller [41] (given in Fig. 8) by using E
i
and UCS
i
,
which was calibrated by back analyses of ve slope failures and
four uniaxial compressive strength data used by [8]. The value of
A varies from 1 for the softest rock material to 5 for the hardest
rock material. The back analyses are discussed in detail under
following sub-sections.
The surface conditions of discontinuities have crucial impact
on the overall strength of rock masses because the shear strength
of discontinuity surfaces is highly sensitive to the surface condi-
tion parameters such as weathering, roughness and inlling. For
this purpose, the Surface Condition Rating (SCR), which was
proposed by Sonmez and Ulusay [33] based on ratings in the
RMR scheme, was considered (Table 2).
The S-shaped curve reduces toward the lower bound for
harder rock masses when the value of A increases and SCR
decreases. Otherwise the S-shaped curve moves toward the upper
bound. Practically, the upper bound S-shaped curve in Fig. 7 is
obtained when Surface Condition Rating (for very good surface
properties) and A (for softer rock material) are 18 and 1,
respectively. On the other hand, when SCR0 and A5, the lower
bound S-shaped curve is obtained.
As discussed above, the S-shaped curve is directly controlled
by parameter C determined from SCR and A together in the new
empirical approach. Therefore, the relation of Cf(A,SCR) was
calibrated by back analyses of ve slope failures and four uniaxial
compressive strength data used in [8].
2.3. Disturbance and degree of interlocking in the new empirical
approach
The method of excavation used during construction in rock
mass, such as blasting, decreases the quality of the rock mass
condition, which is known as disturbance. The inuences of
disturbance have been discussed in [11,33,4245]. The qualitative
approach for dening the degree of disturbance on the overall
strength of rock mass to be used in the HB failure criterion was
rst introduced in [33]. A similar method was also adapted to the
HB criterion [11]. The inuences of the disturbance on the rock
mass are given schematically in Fig. 9.
As indicated in [33], one of two inuences of disturbance on
rock mass is observed on the spacing of discontinuities as a result
of new discontinuities occurred after disturbance. In other words,
the discontinuity density of a rock mass also increases when the
degree of disturbance increases (Fig. 9b). In the new empirical
approach, this effect is directly considered by a reduction para-
meter (r
SR
) on the value of Structure Rating (SR). Second inuence
of disturbance is the increase in the aperture of the discontinu-
ities (or decrease in the degree of interlocking of blocks), since a
rock mass will be became looser due to the disturbance (Fig. 9c).
A rock mass excavated and accumulated as a spoil pile is the
loosest form of the rock mass, a state for which the cohesion
reduces to almost zero under zero normal stress. In addition to
cohesion, internal friction angle of rock mass may be expected to
be less than original value due to the loosening (decreasing
degree of interlocking) of rock mass after disturbance. A mod-
ication of the criterion made in [46], to dene jointed rock mass
which can be evaluated as an attempt to model non-interlocked
(cohesionless under zero normal stress) rock mass by considering
s0 in the original criterion. In addition, in the 1997 version of
the criterion [2], the parameter s was xed to zero when the GSI is
lower than 25. But this xed limit of s0 was removed by [11] to
overcome some limitations discussed in [34]. The current form of
the HB equation [11] has no applicability for non-interlocked
(cohesionless under zero normal stress) rock masses.
As is well known, while the s parameter mostly controls the
cohesion intercept, the internal friction angle is highly sensitive to
Fig. 8. The graph for selection of hardness constant of intact rock (parameter A)
based on UCS
i
and E
i
(F: very low, E: low, D: medium, C: moderate, B: high, A: very
high strength, modied from Deere and Miller, [41]); back analyses results are
given on graph by points with the case number and the value of A in parenthesis.
Table 2
The Surface Condition Rating table proposed by Sonmez and Ulusay [33].
Roughness rating (R
r
) Very rough Rough Slightly rough Smooth Slickensided
6 5 3 1 0
Weathering rating (R
w
) Unweathered Slightly weathered Moderately weathered Highly weathered Decomposed
6 5 3 1 0
Inlling (gauge) rating (R
f
) None Hard lling o5 mm Hard lling 45 mm Soft lling o5 mm Soft lling 45 mm
6 4 2 2 0
SCRR
r
R
w
R
f
O.S. Dinc et al. / International Journal of Rock Mechanics & Mining Sciences 48 (2011) 650665 655
the m
b
parameter. By considering the roles of the s and the m
b
parameters, an improvement was introduced by using reduction
factors (r
s
and r
mb
) to be applied on s and m
b
depending on the
degree of disturbance when the new empirical approach is used
with the HB equations. As a result of this modication, the HB
criterion can be used for the range between intact rock to
completely crushed and non-interlocked (cohesionless under zero
normal stress) rock mass.
Consequently, in this study the disturbance of a rock mass is
taken into consideration by two separate reduction factors
applied to the SR to dene increasing joint density, and to s and
m
b
in the HB criterion to dene decreasing degree of interlock-
ing. Although the disturbance factor (D) proposed by Hoek
et al. [11] was considered in this study, the relation between D
and disturbance factor (d
f
) proposed by Sonmez and Ulusay [33]
was also considered, as illustrated in Fig. 10. The degree of
disturbance may be selected from the guidelines of [11] for D,
and the guidelines of [44] for d
f
. The relations of r
SR
f(D), r
s
f(D)
and r
mb
f(D) given in Fig. 11 was calibrated by back analysis.
Back analyses of ve well-documented slope instabilities and
four uniaxial compressive strength data of rock masses used by
Palmstr om [8] were considered for calibration of the reduction
factors based on the disturbance in the new empirical approach.
While the back analyses are given under following sub-sections,
the equations of the new empirical approach proposed in this
study are given below
SR
d
r
SR
SR
u
1
N
SCR

5:555SCR100
100
2
Fig. 9. (ad) The schematic illustration of inuences of the disturbance on undisturbed rock mass.
Fig. 10. The relation between disturbance factor proposed by Hoek et al. [11] and
Sonmez and Ulusay [33].
O.S. Dinc et al. / International Journal of Rock Mechanics & Mining Sciences 48 (2011) 650665 656
C 1:75A
0:65
AN
SCR
3
SRR
c

UCS
RM
UCS
i
C
o
, o
SR
d
100100SR
d

4000 2p
SR
d
=100
_ _ 4
where UCS
i
is the uniaxial compressive strength of intact rock
material, SRR
c
is the Compressive Strength Reducing Ratio on UCS
i
,
SCR is the Surface Condition Rating, N
SCR
is the normalized value
between zero to one for Surface Condition Rating, SR
u
is Structure
Rating of undisturbed rock mass, SR
d
is adjusted SR based on
disturbance, r
SR
is a reduction factor on SR based on disturbance,
A is hardness constant for intact rock material that can be selected
by using the E
i
and the UCS
i
from Fig. 8 and UCS
RM
is uniaxial
compressive strength of rock mass in which increasing joint
density by disturbance is considered on SR by r
SR
for dening
increase of joint density in rock mass, and decreasing of inter-
locking of rock blocks by disturbance is not considered, or in other
words, interlocking of rock mass is assumed as fully-interlocked.
The reduction factors for dening degree of interlocking and its
relation to the strength is discussed in detail under following
section.
The estimation graph for elastic modulus of intact rock (E
i
)
proposed in [39] based on an Articial Neural Network (ANN)
may be preferred when just USC
i
and unit weight (g) are available
and an experimental result for E
i
is not available (Fig. 12).
3. Relating the new empirical approach to the Hoek and
Brown criterion
The HoekBrown empirical equation has been a very popular
and attractive empirical failure criterion since it was proposed in
1980 because it has a high capability for dening the non-linear
strength envelopes of rock masses. Although its non-linear form
was preserved, many modications have been introduced in the
last 30 years. Hoek and Brown [47] studied on data belong to 14
different rock types, which the lowest uniaxial compressive
strength is 39.9 MPa for sandstone, for evaluation of m
i
. On the
other hand, 152 mm core samples of heavily jointed Panguna
andesite having uniaxial compressive strength of intact rock of
265 MPa from Bougainville in Papua New Guinea and re-com-
pacted graded specimens of the same andesite fragments were
considered as jointed rock mass in the development of the
criterion [47]. Therefore, it can be said that the criterion is more
representative for prediction of strength of medium to hard rock
masses in its current form.
As given in previous section, the Compressive Strength Reduc-
tion Ratio (SRR
c
) on the UCS
i
can be estimated considering both
blockiness of rock masses (with the SR) and intact rock properties
(both UCS
i
and E
i
). Because of the authority of the HB criterion in
rock engineering, the SRR
c
value of rock mass to be calculated by
the procedure given above, was related to the HB non-linear
Fig. 11. Relations between reduction factors and degree of disturbance.
Fig. 12. Estimation graph for E
i
from UCS
i
and g based on ANN [39].
O.S. Dinc et al. / International Journal of Rock Mechanics & Mining Sciences 48 (2011) 650665 657
form equation by the following equations similarly with [20]:
SRR
c

UCS
RM
UCS
i
5
s SRR
c
1=a
r
s
6
m
b
m
i
exp
9lnSRR
c
1=a

28r
mb
_ _
7
a 0:5
exp1009lnSRR
c
1=a
=15exp20=3
6
8
s
0
1
s
0
3
UCS
i
m
b
s
0
3
UCS
i
s
_ _
a
9
UCS
RM
UCS
i
s
a
10
where UCS
RM
is the uniaxial compressive strength of the rock
mass in which increasing the joint density by disturbance is
considered on SR by r
SR
for dening increase of joint density in
rock mass, and decreasing of interlocking of rock blocks by
disturbance is not considered, or in other words, interlocking of
rock mass is assumed as fully-interlocked, UCS
RM
and UCS
i
are
uniaxial compressive strength of rock mass (including both two
effects of disturbance on rock mass) and rock material, respec-
tively; SRR
c
is the Compressive Strength Reducing Ratio on UCS
i
;
m
i
, m
b
, s and a are the HB parameters; r
s
and r
mb
are reduction
factors for degree of interlocking of rock blocks depending on the
degree of disturbance.
The UCS
RM
determined from Fig. 7 is the uniaxial compressive
strength of rock mass which just consider degree of blockiness for
fully interlocked rock mass. Because in Fig. 7 while weathering,
inlling and roughness are used as properties of discontinuities,
degree of boundary alignment of blocks and aperture of disconti-
nuities which can be used for dening degree of interlocking of
rock mass were not considered. Therefore, it can be stated that
either in natural condition (means undisturbed) or after distur-
bance degree of interlocking may be different from fully inter-
locked rock mass. Hence, the UCS
RM
determined from Fig. 7,
should not be used for design purpose without considering
inuence of degree of interlocking given Fig. 11b. The denition
for the aperture was used from [1]. Degree of interlocking
can be easily selected from Fig. 11b, by using disturbance factor
(when any disturbance applied to the rock mass) and degree of
interlocking from the denition (without any disturbance-in
natural condition) given bottom axis of the graph.
As can be seen from the equation of the parameter a, an
iterative solution is necessary to evaluate for exact the value of a.
However, for practical proposes that a can be selected from Fig. 13
depending on the SRR
c
.
Although, the value of the a varies between 0.5 for intact rock
and 0.65 for very poor rock masses, according to [20] the value of
a may extend to one for softer poor quality rock masses which
may have close to linear envelope. However, the latest ranges of
the value of a used in [11] were preserved in this study. The use of
the new empirical approach is summarized as ow chart in
Fig. 14.
4. Calibration of the new empirical approach by back analyses
The new empirical approach for estimation of the strength of
rock mass from soft to hard rock mass was calibrated by ve well-
documented slope failures and four uniaxial compressive strength
data of rock masses used by [8]. In the new empirical approach,
the relations of Cf(A,SCR), Af(UCS
i
, E
i
) given in Fig. 8, r
SR
f(D)
and r
s
f(D) (and r
mb
0.5r
s
0.5) given in Fig. 11 were calibrated
for the limit equilibrium conditions of the ve slope failures and
the uniaxial compressive strength data sets of [8]. The summary
descriptions of the cases and the back analyses are summarized in
the following sub-sections.
4.1. Descriptions of slope failures
To calibrate of the new empirical approach, back analyses of ve
slope failures were performed. Four of these cases (Case 1 to Case 4)
were described in [33] and [48], and are summarized below.
Case 1: A slope failure occurred in heavily jointed schist rock
mass at Baskoyak barite open pit mine, in western Turkey was
studied in [49]. According to these authors, groundwater was not
encountered in the geotechnical boreholes drilled on the pit
benches. Hence, the pit slopes were considered to be drained
for the purpose of the back analysis. No blasting was performed
due to the heavily jointed nature of schist rock mass which was
excavated directly by excavators. Therefore an adjustment factor
for disturbance of D0.7 was selected based on the description
of [11]. The cross-section of failed slope is given in Fig. 15a, and
the rock mass properties are summarized in Table 3.
Case 2: A combined mode of slope failure was observed in
jointed marl rock mass at Goynuk, Turkey as documented in [50].
The properties of jointed marl rock mass, which include two joint
sets in addition to bedding planes, are summarized in Table 3, and
cross-section of the failure is given in Fig. 15b. The method of
excavation was similar to Case 1, therefore the disturbance factor
(D) was selected as 0.7 by Sonmez and Ulusay [34]. No ground-
water was observed in the failed slope [50]. While the circular
part of the failure surface was taken into consideration as mass
failure, the residual effective shear strength parameters of the
stratication plane used in back analysis were evaluated by direct
shear test to be taken as c
0
1.4 kPa and f
0
121 [50].
Case 3: A slope failure occurred at the Kisrakdere lignite open
pit mine, in western Turkey [33,48,51]. A rock mass sequence
consisting of compact and jointed marl and soft clay beds failed as
a result of steepening of the slope benches. The characteristics of
rock mass are tabulated in Table 3. The cross-section of the failure
is drawn in Fig. 15c. The excavation was performed by excavators
after a controlled blasting process. Therefore, the disturbance
factor (D) was selected as 0.93 in [33].
Case 4: Some spoil pile slope failures were recorded by Ulusay
et al. [52] at Eskihisar lignite open pit mine in western Turkey.
Fig. 13. The relation between the SRR
c
and the a parameter of the Hoek and
Brown criterion from iterative solution of the equation given in this study.
O.S. Dinc et al. / International Journal of Rock Mechanics & Mining Sciences 48 (2011) 650665 658
Sonmez and Ulusay [33] applied the HB failure criterion to the
failures by assuming the material to be disintegrated rock masses.
As discussed above, spoil piles are excavated rock material and
may be evaluated as the loosest form of in-situ rock mass. Hence,
the cohesion of spoil piles may be expected to be zero due to non-
interlocking of rock blocks at the time of accumulation. The
cohesionless (under zero normal stress) rock mass behavior can
be modeled in the new empirical approach by assuming r
s
0 for
D1. The cross-sections of the spoil pile failures are given
in Fig. 15d, and the characteristics of the spoil piles are summar-
ized in Table 3. In the new empirical approach, the inuence of
disturbance are considered separately, Therefore, the properties
tabulated in Table 3 are given by considering in-situ properties of
rock mass.
Case 5: A slope failure in andesite rock mass was observed
between Cayeli and Kaptanpasa in north-eastern Turkey. The
slope was excavated as a borrow pit to produce rock for con-
struction of roads. For this purpose the material was excavated
after the controlled blasting. Therefore, the disturbance factor (D)
was selected as 0.93 similar to Case 3 after reviewing Fig. 10. No
groundwater was observed on the excavated slope faces. The
general view and a cross section of the failure are provided as
Fig. 16, the properties of the rock mass are tabulated in Table 3.
4.2. Descriptions of the UCS
RM
data from [8]
Seven uniaxial compressive strength data were used during
the construction of RMi system by Palmstr om [8]. Sample 4 in [8]
was a large slide having 300,000 m
3
materials occurred in the
L angsele mine, Sweden in 1975. The sliding materials included
sericite-quarzite, dacite-tuff and gray schist and greenstone. This
case was not considered in the calibration of the new empirical
approach by considering highly anisotropic rock material compo-
sition of this case. Sample 5 composed of Caledonian schist having
high strength anisotropy parallel and normal to the schistosity
planes. Therefore, this case was not used in the calibration.
Cylindrical samples of 0.6 m diameter and 1.2 m length of Lower
Triassic sandstone with intercalations of silty claystone and tested
normal to the layering was also another anisotropic case used
in [8] as the sample 6 which was not used in the calibration of the
new empirical approach. Finally, four uniaxial compressive data
(Case 6 to Case 9) except the sample 4 to 6 [8] were evaluated in
the calibration of the new empirical approach. The summary
descriptions of these cases from [8] are given as follows.
Case 6 (Panguna andesite): A description of the test data of the
Panguna andesite was given by Jaeger [53]. These data were used
by [8] during calibration of the RMi system. While the following
Fig. 14. The ow chart for the use of the new empirical approach.
O.S. Dinc et al. / International Journal of Rock Mechanics & Mining Sciences 48 (2011) 650665 659
descriptions for the tests performed on Panguna andesite are
directly taken from [8], the parameters to be used in the calibra-
tion of the new empirical approach are given in Table 4.
It was difcult to obtain sufcient small 2550 mm
2
cylin-
ders of rock containing no planes of weakness for triaxial test on
intact rock. The test samples for triaxial tests were taken by very
careful drilling, using 150 mm diameter triple tube drilling
equipment. These samples were prepared and tested triaxially
by Jaeger [53]. The measured unconned compressive strength of
intact rock was 269 MPa. From the triaxial tests at low conning
stress carried out on 150 mm cores an unconned compressive
strength of 3.7 MPa was evaluated by Jaeger, who noticed that an
important effect during testing was the interlocking of adjacent
blocks. As very little has been published on large triaxial tests on
rock mass, Hoek and Brown [47] concluded from studies of test
results of the Panguna andesite that these may be taken as a
reasonable model for the in-situ strength of a heavily jointed hard
rock mass.
Case 7 (granitic rock from Stripa): The UCS
RM
data of granitic
(quartz monzonite) rock from the Stripa mine in Sweden was
obtained from laboratory tests on a 1 m diameter by 2 m long
specimen [8]. Properties of the Stripa Granitic rock mass were
reported by Thorpe et al. [54]. While there are two dominant joint
sets having similar properties, several small joints in the sample
were generally discontinuous [54]. The parameters to be used in
the calibration of the new empirical approach are given in Table 4.
Fig. 15. Cross-sections of the slope failures of (a) Case1: Beysehir, (b) Case 2: Goynuk, (c) Case 3: Kisrakdere and (d) Case 4: Eskihisar spoil piles (Sonmez and Ulusay
[3334]).
O.S. Dinc et al. / International Journal of Rock Mechanics & Mining Sciences 48 (2011) 650665 660
Case 8 (in-situ tests on mine pillars of sandstone in the Laisvall
mine): The tests were performed on 9 pillars to obtain design
values for the load bearing capacity of the roof in the Laisval Mine
in Northern Sweden. The test procedure was given by S oder and
Krauland [55 from 8] in which the pillars were subjected to
increasing stresses. The damage from blasting was reported as
minimal. The parameters to be used in the calibration of the new
empirical approach are given in Table 4.
Case 9 (Paleozoic siltstone from Germany): The UCS
RM
of Carbo-
niferous dark siltstone from Hagen in Germany was obtained on
the tests performed on cylindrical samples 0.6 m diameter and
1.2 m long. According to personal communication between
Palmstr om and Mutschler, no prominent joint set occurred in
this sample, and the mainly short joints were orientated in
various directions [8]. The parameters to be used in the calibra-
tion of the new empirical approach are given in Table 4.
4.3. Back analyses of the cases
In the new empirical equations, SR, SCR and A are major input
parameters to estimate SRR
c
( UCS
RM
=UCS
i
). In addition to SRR
c
,
the degree of disturbance is also taken into consideration on the
SR by r
SR
, and on the s and m
b
parameters of the HB failure
criterion by r
s
and r
mb
. Therefore the calibrations of the relations
of Af(UCS
i
,E
i
) given by Fig. 8, r
SR
f(D) and r
s
f(D) (and
r
mb
0.5r
s
0.5) given in Fig. 11 were performed by using trial
and error approach for the limit equilibrium condition of incipient
slope failures (FOS1.0). In addition to the data from the slope
failures, four UCS
RM
data sets from [8] were also used for
calibration of the relation of Af(UCS
i
,E
i
) given by Fig. 8 and
r
SR
f(D). The back analyses of the slope failures and four UCS
RM
data are summarized in Table 5. The USC
i
and the E
i
were taken
into consideration for the selection of A. Although the factor of
Table 3
The rock mass characteristics of back-analyzed slopes (Case 1 to Case 4 are taken from [33,34]; Case 5 was investigated in this study).
Parameters Case 1 Case 2 Case 3 Case 4 Case 5
Spacing
a
(m) S
average
0.04 (three
joint sets)
S
1
0.37, S
2
0.65,
S
b
0.11
S
1
0.75, S
2
1.07,
S
3
0.13, S
b
0.4
S
1
0.71, S
2
0.82,
S
3
1.26, S
b
0.65 (in
spoil pile
S
average
0.083
b
)
S
average
0.35 (three joint
sets)
Condition of
discontiniuties and
ratings
Smooth to slickensided
surfaces (01), highly
weathered (1), soft
coating o5 mm (2)
Slickensided surfaces
(0), moderately
weathered (3), soft
coating o5 mm (2)
Smooth surfaces (1),
slightly weathered (5),
soft coating o5 mm
(2)
Smooth surfaces (1),
slightly weathered (5),
soft coating o5 mm
(2)
Upper part: slightly rough
(3), moderately weathered
(3), soft coating o5 mm
(2)
Lower part: slightly rough
(3), slightly to moderately
weathered (4(53)/2),
soft coating o5 mm (2)
SCR 4 5 8 8 Upper part: 8
Lower part: 9
J
v
(joint/m
3
) 75 13.3 12.5 4.96 8.6
SR 4.2 34.5 35.6 51.8 42.2
D
c
0.7 (d
f
0.97) 0.7 (d
f
0.97) 0.93 (d
f
0.90) 1.0 (d
f
0.8) 0.93 (d
f
0.90)
m
i
7 9.87 9.04 9.87 24
g (kN/m
3
) 23.6 18.5 21.0 16.1 24
UCS
i
(MPa) 5.2 4.8 40.0 4.15 Upper part: 62.7
Lower part: 90.8
E
i
(GPa)
d
9.0 3.8 9.0 2.0 Upper part: 22
Lower part: 30
Case 1: Baskoyak barite pit; Case 2: Goynuk lignite open pit; Case 3: Kisrakdere (Soma) lignite open pit; Case 4: Spoil pile instabilities at Eskihisar open pit; Case 5:
CayeliKaptanpasa barrow pit.
a
True spacing (S
1
, S
2
, S
3
for joints, S
b
for bedding planes, S
average
for average spacing).
b
Estimated by method of photoanalysis along x, y and z axes [33].
c
Adjustment factor for disturbance effect based on Hoek et al. [11] and the values in parenthesis based on Sonmez and Ulusay [33].
d
Estimated from Fig. 12 based on UCS
i
and g.
Fig. 16. General view (a) and the cross-section of the slope failure (b) that
occurred between Cayeli and Kaptanpasa in north-eastern of Turkey (Case 5).
O.S. Dinc et al. / International Journal of Rock Mechanics & Mining Sciences 48 (2011) 650665 661
Table 4
Denition of the parameters for the UCS
RM
data from Palmstr om [8] used in the calibration of the new empirical approach.
Panguna andesite UCS
RM
data (C6) Stripa Granitic rock UCS
RM
data (C7)
Denition used by Hoek and Brown [47] and
given by Palmstr om [8]
Denition considered in the back analysis of the
proposed empirical approach
Denition given by Palmstr om [8] Denition considered in the back analysis of the
proposed empirical approach
Spacing (S)o60 mm (three joint sets)
Volume of blocks (Vb)26 cm
3
(Palmstr om [8]
assessed from description and photo published
by Jeager [53])
Spacing (S)o60 mm (three joint sets)
Volume of blocks (Vb)26 cm
3
SR 10 from Fig. 6 based on three joint sets
having spacing of 60 mm
Spacing (S) of joint set-10.251.5 m
Spacing (S) of joint set-20.150.5
Spacing (S) of joint set-3 was not given
Volume of blocks (V
b
) 515 dm
3
Spacing (S) of joint set-10.251.5 m
(average0.88 m)
Spacing (S) of joint set-20.150.5
(average0.3 m)
Spacing (S) of joint set-3 was not given (but
assumed as 0.2 m)
Volume of blocks (V
b
) 515 dm
3
J
v_average
9.5 joint/m
3
SR 40 from Fig. 6 based on Jv_average and Vb
Roughness: J
r
3 (rough and irregular undulating,
based on Q system)
Inlling and alteration: J
a
2 (slightly altered joint
walls, non-softening mineral coatings, sandy
particles, clay-free disintegrated rock, etc.).
Comments of Palmstr om [8] about J
a
is that with
a rather weak lling and open joints and
veins a rating of J
a
34 was chosen
Roughness: R
r
5 (rough, based on Table 2)
Filling:R
f
2 (soft coating o2 mm, based
on Table 2)
Weathering: R
w
5 (slightly weathered, based
on Table 2)
SCR52512
Roughness (for set-1, set-2 and set-3): J
s
2
(rough based on RMi system)
Inlling and alteration (for set-1 and set-2:
J
A
34 (hard chlorite coatings).
Inlling and alteration (for set-3): J
A
1
(fresh)
Roughness: R
r
5 (rough, based on Table 2)
Filling:R
f
5 (hard coating o2 mm (4), and fresh
(6) based on Table 2)
Weathering: R
w
3 (moderately weathered, based
on Table 2)
SCR53313
UCS
i
269 MPa UCS
i
200 MPa
E
i
60 GPa (estimated using UCS
i
269 and assuming g26 kN/m
3
from Fig. 12) E
i
55 GPa (estimated using UCS
i
200 and assuming g26 kN/m
3
from Fig. 12)
Almost no disturbance, due to the use of triple tube
drilling equipment
D0.1 (almost no disturbance) Almost no disturbance (the sample was cut by
a slot drilling technique)
D0.1 (almost no disturbance)
UCSRM3.7 MPa Estimated UCSRM3.64 MPa from the new empirical
approach
UCSRM7.55 MPa Estimated UCSRM7.41 MPa from the new
empirical approach
Laisvall mine sandstone UCS
RM
data (C8) Paleozoic siltstone from Germany UCS
RM
data (C9)
Denition given by Palmstr om [8] Denition considered in the back analysis of the
proposed empirical approach
Denition given by Palmstr om [8] Denition considered in the back analysis of the
proposed empirical approach
Spacing (S) of joint set-10.21.2 m
Spacing (S) of joint set-20.31.5 m
Volume of blocks (V
b
) 0.10.3 m
3
Spacing (S) of joint set-10.21.2 m (average
0.7 m)
Spacing (S) of joint set-20.31.5 m (average
0.9 m)
Volume of blocks (V
b
)0.10.3 m
3
J
v_average
2.5 joint/m
3
SR 60 from Fig. 6 based on J
v_average
and V
b
No prominent joint set occurred in this
sample where mainly short joints were
oriented in various direction (Mutscher
(1993) from [8])
Volume of blocks (V
b
)510 dm
3
based on
Mutscher (1993) from [8]
Volume of blocks (Vb)510 dm
3
Although the SR is obtained as30 from Fig. 6
based on V
b
, SR55 (may vary between 50 and
60) is selected by considering no prominent joint
set denition given by Mutscher (1993) from [8]
Roughness (for set-1): J
s
1 (smooth surfaces
based on RMi system)
Roughness (for set-2): J
s
1.5 (slightly rough
surfaces based on RMi system)
Inlling and alteration (for set-): J
A
2 (assumed
clay on joint wall)
Inlling and alteration (for set-2): JA1 (fresh)
Roughness: R
r
2 (smooth (1) and slightly rough,
(3) based on Table 2)
Filling:R
f
4 (soft coating o2 mm (2) and none
(6) based on Table 2)
Weathering: R
w
5.5 (fresh (6) and slightly
(5) weathered, based on Table 2)
SCR245.511.5
In general unaltered or without coating
slightly rough to rough surfaces
Roughness: R
r
3 (slightly rough (3) based
on Table 2)
Filling: R
f
6 (no lling based on Table 2)
Weathering: R
w
6 (unweathered, based
on Table 2)
SCR46615
UCS
i
210 MPa UCS
i
65 MPa
E
i
55 GPa (estimated using UCS
i
210 and assuming g26 kN/m
3
from Fig. 12) E
i
22 GPa (estimated using UCS
i
65 and assuming g24 kN/m
3
from Fig. 12)
Damage from blasting was reported as minimal, so
almost no disturbance
D0.15, (minimal or almost no disturbance) Almost no disturbance (sample preparation of
0.6 m diameter and 1.2 m height)
D0.1 (almost no disturbance)
UCS
RM
20 MPa Estimated UCS
RM
18.9 MPa from the new empirical
approach
UCS
RM
6.8 MPa Estimated UCS
RM
7.07 MPa from the new
empirical approach
O
.
S
.
D
i
n
c
e
t
a
l
.
/
I
n
t
e
r
n
a
t
i
o
n
a
l
J
o
u
r
n
a
l
o
f
R
o
c
k
M
e
c
h
a
n
i
c
s
&
M
i
n
i
n
g
S
c
i
e
n
c
e
s
4
8
(
2
0
1
1
)
6
5
0

6
6
5
6
6
2
Table 5
Summary of the back analyses of the cases used in the calibrations of A, r
SR
, r
S
.
(a) Slope failures
Case no.: name Input parameters Back analysis
D SR SCR m
i
UCS
i
(MPa) g (kN/m
3
) E
i
(GPa)
a
A r
SR
r
S
FOS
C1: Beysehir 0.7 4.2 4 7 5.2 23.6 9 2.3 0.55 0.06 1.06
2.5 0.99
2.4 1.02
C2: Goynuk 0.7 34.5 5 9.87 4.8 18.5 3.8 2 0.55 0.08 1.07
2.2 0.08 1.02
2.1 0.06 1.01
2.2 0.06 0.99
C3:Kisrakdere 0.93 35.6 8 9.04 40.0 21.0 9 2.8 0.50 0.05 0.69
2.7 0.50 0.05 0.70
2.7 0.30 0.03 0.66
C4: spoil piles
Sec (1-1) 1.0 51.8 8 9.87 4.15 16.1 2.5 1.9 0.25 0 1.20
1.95 0.15 1.14
2.0 0.10 1.10
2.1 0.10 1.06
Sec (2-2) 1.9 0.25 0 1.12
1.95 0.15 1.06
2.0 0.10 1.03
2.1 0.10 0.99
Sec (3-3) 1.9 0.25 0 1.15
1.95 0.15 1.09
2.0 0.10 1.05
2.1 0.10 1.01
Sec (4-4) 1.9 0.25 0 1.15
1.95 0.15 1.09
2.0 0.10 1.05
2.1 0.10 1.01
C5: C- ayeliKaptanpasa barrow pit 0.93 42.2 8 (for upper part) 25
a
62.7 (for upper part) 25 22 (for upper part) 3.3 (for upper part) 0.30 0.03 0.99
9 (for lower part) 90.8 (for lower part) 30 (for lower part) 3.4 (for lower part)
(b) UCSRM data from Palmstr om [8]
Case no.: name Input parameters Output Back analysis
D SR SCR UCS
i
(MPa) g (kN/m
3
)
b
E
i
(GPa)
a
UCS
RM _measured
(MPa) (1) A r
SR
UCS
RM_predicted
(MPa) (2) (2/1)
C6: Panguna andesite 0.1 10 12 269 26 60 3.7 4.1 0.95 3.64 0.98
C7: Stripa Granitic 0.1 40 13 200 26 55 7.55 4.0 0.95 7.41 0.98
C8: Laisvall mine sandstone 0.15 60 11.5 210 26 55 20 4.0 0.93 18.9 0.94
C9: Paleozoic siltstone from Germany 0.1 55 15 65 24 22 6.8 3.3 0.95 7.07 1.04
a
E
i
was estimated from Fig. 12.
b
No information (assumed).
O
.
S
.
D
i
n
c
e
t
a
l
.
/
I
n
t
e
r
n
a
t
i
o
n
a
l
J
o
u
r
n
a
l
o
f
R
o
c
k
M
e
c
h
a
n
i
c
s
&
M
i
n
i
n
g
S
c
i
e
n
c
e
s
4
8
(
2
0
1
1
)
6
5
0

6
6
5
6
6
3
safety (FOS) of 0.7 was obtained from the back analysis of the
slope failure at the Kisrakdere open lignite pit mine, a limit
equilibrium condition (FOS1.0) was obtained for the rest of
the cases. As indicated in the summarized conditions of the slope
failures at the Kisrakdere open lignite pit mine, a rock sequences
consisted of compact jointed marl rock mass and soft clay beds
was observed in failed slope face having a height of about 80 m.
The sequence of the compact jointed marl rock mass and soft clay
beds was obtained from the photo of the slope faces and some
drilling logs behind the slope. However, it is possible that the rock
sequence in the slope was not be able to accurately characterized
depending on the large and chaotic view of the slope failure
(particularly toe of the slope was covered by failed materials, and
could not be observed after failure), resulting in the unusually
slightly low FOS of 0.7 from limit equilibrium condition (FOS1).
Four UCS
RM
data are closely estimated by the new empirical
equation.
5. Conclusion
The founding President of International Society for Rock
Mechanics (ISRM), Leopold M uller, always emphasized the dis-
continuous nature of rock masses and said that the existing
methods of continuum mechanics and soil mechanics were
inadequate for analyzing rock engineering problems [56]. In the
light of this statement, since the 1960s, studies about estimation
of strength and deformation properties of rock masses have been
one of the main research topics in the rock mechanics commu-
nity, due to the difculties encountered during preparation of
representative cores from rock masses for laboratory tests.
Although numerous empirical methods proposed for determining
the strength of rock masses, a few well-known empirical equa-
tions were presented in this study. While each of them has a
crucial contribution to determination of strength of rock masses,
the HB failure criterion has been widely used around the world.
Since it was proposed in 1980, the criterion has been commonly
preferred in many engineering project, and nevertheless many
improvements were proposed by both the originators of the
criterion and some other researchers to overcome some decien-
cies and limitations of the criterion. Only heavily jointed Panguna
andesite, which have sound intact rock pieces having UCS
i
of
265 MPa [47], was considered as a rock mass data during devel-
opment of the criterion by the originators. Based on the three
hard rock mass cases that two of them are same cases used in this
study as Case 7 and Case 8, Edelbro et al. [57] indicated that N
(rock mass number, [58]), RMi [8], Q [2] and GSI [11] yield a
reasonable agreement with the measured strengths of hard rock
masses. Although the prediction capacity of HB failure criterion
for hard rock masses may be deemed to be acceptable, it may not
accurately estimate the strength of soft to medium strength rock
masses. This limitation was underlined by some recent stu-
dies [1720], and they proposed some new improvements to
overcome this limitation. However, some details of these recent
studies, which are discussed in the present study, particularly on
the estimation of strength of soft rock masses, still need to be
explored. Some important studies from the broad soil mechanics
literature include useful information for dening the overall
strength boundary between soft rock mass and jointed (ssured)
soil mass. These ndings on the overall strength of jointed clays
were also taken into consideration during calibration of the new
empirical approach for the boundary between jointed soft rock
mass and jointed soil mass. Then the Compressive Strength
Reducing Ratio (SRR
c
) is introduced to the most popular non-
linear envelope of Hoek and Brown criterion by preserving its
latest form [11].
The commonly preferred input parameters in well-known
empirical equations such as uniaxial compressive strength of
intact rock material (UCS
i
), elastic modulus of intact rock material
(E
i
), joint density based on Structure Rating (SR) concept of [33],
are considered as input parameters to determine the SRR
c
in the
new general empirical approach. In addition, the disturbance of a
rock mass can be taken into consideration by two separate
reduction factors applied to the SR to dene increasing joint
density, and to the s and m
b
parameters of the HB criterion to
dene decreasing degree of interlocking from fully interlocked to
non-interlocked or cohesionless rock masses under zero normal
stress such as spoil pile.
Rock mass conditions, where the new general empirical
equations could be used, are similar to that of the HB failure
criterion. The value of Structure Rating (SR) may be expected as
higher than 70 for the rock masses including widely spaced
discontinuities. The dimensions of engineering application to be
constructed in/on rock masses divided by widely spaced disconti-
nuities should be taken into consideration to decide whether the
new empirical equation will be used or not. In other words, the
new empirical equations could not be preferred when the
potential failure mechanism in rock mass is expected to be as
discontinuity controlled.
Finally, it should be remembered that the strength envelope of
rock mass obtained from the new general empirical approach
should not be evaluated as a residual strength of rock mass. The
Compressive Strength Reducing Ratio (SRR
c
) approach was con-
structed on some scientic mostly agreed concepts of rock
mechanics, and calibrated by nine cases of rock mass from soft
to hard. M uller [13] indicated that Many attempts have been made
to handy calculation methods and simple formulation for everyday
use to engineer may be provided with simple working tools. I see a
danger in this: complicated things do not become simpler through
simplication at all cost. Things in geomechanics are complicated by
their very nature. Therefore, although empirical approaches are
practical tool to obtain some of design parameters, it should be
underlined that none of them should be used alone for nal
decision particularly by limited-experienced practitioners. Hence
the new empirical approach is open to improvement and discus-
sion, based on new cases and ideas.
Acknowledgment
This study was prepared from the ndings of research project
supported by TUBITAK (Project no. 108Y002). The authors thank
to Robert Zimmerman and Ed Medley for their valuable com-
ments as technical advisors of the project. In addition, the authors
thank to Arild Palmstr om for his permission to use the UCS
RM
data
given in his Ph.D. thesis, and to anonymous reviewers for their
valuable comments.
References
[1] Bieniawski ZT. Engineering rock mass classifcations. John Wiley and Sons;
1989. 215 p.
[2] Hoek E, Brown ET. Practical estimates of rock mass strength. Int J Rock Mech
Min Sci 1997;34(8):116586.
[3] Barton N. Some new Q value correlations to assist in site characterization and
tunnel design. Int J Rock Mech Min Sci 2002;39:185216.
[4] Yudhbir, Lemanza W Prinzl F. An empirical failure criterion for rock masses.
In: Proceedings of the fth international congress society of rock mechanics,
Melbourne, vol. 1; 1983. p. B18.
[5] Ramamurthy T. Stability of rock mass. Indian Geotech J 1986:174.
[6] Vardar M. Die Auswirkung der ingenieurgeologisch-felsmechanischen
Verh altnisse auf die Planung und Verwirklichung der Istanbuler Abwasser-
stollen in Geologie Felsmechanik Felsbau 1989. In: Egger P, Fecker E, Reik G,
editors. Festkolloquium L. M uller-Salzburg 1988. Clausthal: Trans Tech Pub;
1988.
O.S. Dinc et al. / International Journal of Rock Mechanics & Mining Sciences 48 (2011) 650665 664
[7] Kalamaris GS, Bieniawski ZT. A rock mass strength concept for coal incorpor-
ating the effect of time. In: Proceedings of the eighth international congress
of the rock mechanics, Rotterdam: Balkema; 1995. p. 295302.
[8] Palmstr om A. RMi-a rock mass characterization system for rock engineering
purposes. Ph.D. thesis, University of Oslo; 1995.
[9] Sheorey PR. Empirical rock failure criteria. Rotterdam: Balkema; 1997.
[10] Aydan O., Dalgic S. Prediction of deformation behaviour of 3 lanes Bolu
tunnels through squeezing rocks of North Anotolian Fault Zone (NAFZ). In:
Proceedings of the regional symposium on sedimentary rock engineering,
Taipei; 1998. p. 22833.
[11] Hoek E, Carranza-Torres CT, Corkum B. HoekBrown failure criterion-2002
edition. In: Proceedings of the fth north american rock mechanics sympo-
sium, Toronto; 2002. p. 26773.
[12] M uller L. Der Felsbau Band I. Stuttgart: Ferdinand Enke Verlag; 1963. [in
German].
[13] M uller L. Introductory lecture. In: M uller L, editor. Rock mechanics, CISM
courses and lectures 16. Wien: Springer; 1970.
[14] Vardar M. Zeiteinuss auf das Bruchverhalten des Gebirges in der Umgebung
von Tunneln. Karlsruhe: von der Fakult at f ur Bauingenieur und Vermes-
sungswesen der Universit at Karlsruhe (TH); 1977.
[15] Y uzer E, Vardar M Kaya Mekani gi.
_
IT

U Vakf, 1986, No.: 11, s: 75.


_
Istanbul [in
Turkish].
[16] Vardar M. Time dependent stability problems in tunnels and time-dependent
behavior of the rock mass. ITA/AITES Training Course Tunnel Engineering,
_
Istanbul, /http://www.ita-aites.org/leadmin%20/lemounts%20/general/pdf/
ItaAssociation/ProductAndPublication/Training/TrainingCourses%20/T_C_Var
dar_2005.pdfS; 2005.
[17] Carter TG, Diederichs MS, Carvalho JL. A unied procedure for prediction of
strength and post yield behaviour for rock masses at the extreme ends of the
integrated GSI and UCS rock competence scale. In: Proceedings of the 11th
congress of the international society for rock mechanics, Lisbon, London:
Taylor & Francis; 2007. p. 1614.
[18] Carvalho JL, Carter TG, Diederichs MS. An approach for prediction of strength
and post yield behaviour for rock masses of low intact strength. In:
Proceedings of the rst CanadaUS rock symposium, Vancouver; 2007.
p. 24957.
[19] Diederichs MS, Carvalho JL, Carter TG. A modied approach for prediction of
strength and post yield behaviour for high GSI rockmasses in strong, brittle
ground. In: Proceedings of the frst CanadaUS rock symposium, Vancouver;
2007. p. 2778.
[20] Carter TG, Diederichs MS, Carvalho JL. Application of modied HoekBrown
transition relationships for assessing strength and post yield behaviour at
both ends of the rock competence scale. In: SAIMM symposium on ground
support in mining and civil engineering construction, 30 March3 April 2008.
[21] Terzaghi K. Stability of slopes in natural clays. In: Proceedings of the rst
international conference on soil mechanics foundation engineering, Cam-
bridge, vol. 1, Mass; 1936. p. 1615.
[22] Lo KY. The operational strength of ssured clays. Geotechnique 1970;20(1):5774.
[23] Bishop AW, Little AL. The infulence of the size and orientation of the sample
on the apparent strenmgth of London clay at Maldon, Essex. In: Proceedings
of the geotechnical conference, Oslo, vol. 1; 1967. p. 8996.
[24] Simons NE. Discussion on shear strength of stiff clay. In: Proceedings of the
geotechnical conference, Oslo, vol. 2; 1967. p 15960.
[25] Marsland A, Butler ME. Strength measurements on stiff ssured Barton Clays
from Fawley (Hampshire). In: Proceedings of the geotechnical conference,
Oslo, vol. 1; 1967. p. 13945.
[26] Silvestri V. The long-term stability of a cutting slope in an overconsolidated
sensitive clay. Can Geotech J 1980;17(3):33751.
[27] Ladanyi B, Archambault G. Simulation of shear behavior of a jointed rock
mass. In: Proceedings of the 11th US rock mechanics symposium, Berkeley;
1969, paper ARMA 69-0105.
[28] Ladanyi B. The mechanics of landslides in Leda clay: discussion. Can Geotech
J 1970;7:5067.
[29] Lo KY, Lee CF. An evaluation of the stability of natural slopes in plastic
Champlain Clays. Can Geotech J 1974;11:16581.
[30] Feda J, Bohac J, Herle I. Shear resistance of ssured neogene clays. Eng Geol
1995;39:17184.
[31] Vallejo LE. Application of fracture mechanics to soil an overview. In: Vallejo J,
Liang YR, editors. Fracture mechanics (geotechnical special publication no.
43); 1994. p. 120.
[32] Sonmez H. Investigation of the applicability of the HoekBrown criteria to
the failure behavior of the ssured clays. Ph.D. thesis, Hacettepe University,
Ankara; 2001 (in Turkish).
[33] Sonmez H, Ulusay R. Modications to the geological strength index (GSI) and their
applicability to stability of slopes. Int J Rock Mech Min Sci 1999;36:74360.
[34] Sonmez H, Ulusay R. A discussion on the HoekBrown failure criterion and
suggested modication to the criterion veried by slope stability case
studies. Yerbilimleri (Earthsciences) 2002;26:7799. /http://www.yerbilim
leri.hacettepe.edu.tr/toc26.shtmS.
[35] Palmstr om A. The volumetric joint count-a useful and simple measure of the
degree of jointing. In: Proceedings of the fourth international congress of
IAEG, New Delhi; 1982. p. V2218.
[36] Palmstr om A. Application of the volumetric joint count as a measure of rock
mass jointing. In: Proceedings of the international symposium on funda-
mentals of rock joints, Bjorkliden, Sweden; 1985. p. 103110.
[37] Palmstr om A. The volumetric joint count as a measure of rock mass jointing.
In: Proceedings of the conference on fracture, fragmentation and ow,
Jerusalem; 1986.
[38] Palmstr om A. Measurements of and correlations between block size and rock
quality designation (RQD). Tunell Undergr Space Technol 2005;20(4):36277.
[39] Sonmez H, Gokceoglu C, Kayabasi A, Nefeslioglu HA. Estimation of rock
modulus: for intact rocks with an articial neural network and for rock
masses with a new empirical equation. Int J Rock Mech Min Sci 2006;43(2):
22435.
[40] Hoek E, Diederichs MS. Empirical estimation of rock mass modulus. Int J Rock
Mech Min Sci 2006;43(2):20315.
[41] Deere DU, Miller RP. Engineering classication and index properties for
intact rocks. Tech rep AFNL-TR-65-116, Air Force Weapons Lab, New Mexico;
1966.
[42] Laubscher DH. A geomechanics classication system for the rating of rock
mass in mine design. J South Afr Inst Miner Metall 1990;90(10):25773.
[43] Romana MA. Geomechanical classication for slopes: slope mass rating. In:
Hudson JA, editor. Comprehensive Rock Engineering, vol. 3. London:
Pergamon; 1993. p. 57599.
[44] Kendorski FS, Cumming RA, Bieniawski ZT, Skinner EH. Rock mass classifca-
tion for block caving mine drift support. In: Proceedings of the fth
international congress of rock mechanics, Melbourne; 1983. p. B5163.
[45] Sonmez H, Gokceoglu C. Discussion of the paper by E. Hoek and M.S.
Diederichs, Empirical estimation of rock mass modulus. Int J Rock Mech
Min Sci 2006;43:6716.
[46] Hoek E, Wood D, Shah S. A modied HoekBrown criterion for jointed rock
masses. In: Proceedings of the Eurock92, London: Thomas Telford; 1992. p.
20913.
[47] Hoek E, Brown ET. Underground excavations in rock. London: Institute of
Mining and Metallurgy; 1980.
[48] Sonmez H, Ulusay R, Gokceoglu C. A practical procedure for back analysis of
slope failures in closely jointed rock masses. Int J Rock Mech Min Sci
1998;35(2):21933.
[49] Ulusay R, Yucel Z. An example for the stability of slopes excavated in weak
rocks: Baskoyak Barite Open Pit., Earthsciences. Bull Earth Sci Appl Res
Center Hacettepe Univ 1989;15(2):1527. [in Turkish].
[50] Ulusay R, Ekmekc-i M, G okc-eo glu C, Sonmez H, Tuncay E, Erdo gan S. Slope
stability investigation for Himmetoglu lignite open pit mine. Hacettepe Univ
Rep, 1998, Proj 97-0058. [in Turkish].
[51] Sonmez H. TKI-ELI Soma linyitleri ac-ik isletmelerinde eklemli kaya k utlesi ic-
indeki sevlerin durayliliginin degerlendirilmesi. M.Sc. thesis, Hacettepe Univ,
Ankara, 1996 [in Turkish].
[52] Ulusay R, Yoleri MF, C- a glan D, Arkan F. Design evaluations for spoil piles at a
strip coal mine considering safety of the haul road. Int J Surf Min Recl Environ
1995;9:13340.
[53] Jaeger JC. Behavior of closely jointed rock. In: Proceedings of the 11th
symposium on rock mechanics; 1969. p. 5768.
[54] Thorpe R, Watkins DJ, Ralph WE, Hsu R, Flexser S.. Strength and permeability
tests on ultra-large Stripa granite core. Tech Info Rep 31, Lawrence Berkeley
Lab, Berkeley; 1980.
[55] S oder PE, Krauland N. Determination of pillar strength by full scale pillar
tests in the Laisvall mine. In: Proceedings of the 11th world mining congress
strata control in deep mines; 1990. p. 3959.
[56] Brown ET. Estimating the mechanical properties of rock masses. In: Proceed-
ings of the rst southern hemisphere rock mechanics symposium, Perth:
Austral Centre Geomech; 2008. p. 322.
[57] Edelbro C, Sj oberg J, Nordlund E. A quantitative comparison of strength
criteria for hard rock masses. Tunnell Undergr Space Technol 2006;22:5768.
[58] Goel RK, Jethwa JL, Paithankar AG. Correlation between Bartons Q and Bieniaws-
kis RMRa new approach. Int J Rock Mech Min Sci 1995;33(2):17981.
O.S. Dinc et al. / International Journal of Rock Mechanics & Mining Sciences 48 (2011) 650665 665

Vous aimerez peut-être aussi