See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/266496092

The Geological Strength Index (GSI): A characterization tool for assessing

engineering properties of rock masses

Chapter · July 2007

DOI: 10.1201/NOE0415450287.ch2

CITATIONS READS

77 80,154

3 authors:

P. Marinos V. Marinos
National Technical University of Athens National Technical University of Athens
83 PUBLICATIONS   3,205 CITATIONS    112 PUBLICATIONS   1,539 CITATIONS   

SEE PROFILE SEE PROFILE

Evert Hoek
University of Toronto
114 PUBLICATIONS   20,452 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

A large scale landslide in a coal mine within marly formations. Evaluation, analysis and rehabilitation View project

Improvement of slopestability in the camping facilities of “Elpis” in Ano Poroia of Serres, in Greece View project

All content following this page was uploaded by V. Marinos on 12 January 2015.

The user has requested enhancement of the downloaded file.

 THE GEOLOGICAL STRENGTH INDEX (GSI): A CHARACTERIZATION TOOL
FOR ASSESSING ENGINEERING PROPERTIES FOR ROCK MASSES
By Paul G. Marinos, Ph.D.,1 Vassilis Marinos,2 and Evert Hoek, Ph.D.3

ABSTRACT development of the Hoek-Brown criterion is to be pub-

lished in the first issue of a new international journal
The Geological Strength Index (GSI) is a system of entitled Soils and Rocks [Hoek and Marinos, in press].
rock mass characterization that has been developed in The present paper is an update and extension of the
engineering rock mechanics to meet the need for reliable paper by Marinos et al. [2005].
input data related to rock mass properties required as input
for numerical analysis or closed-form solutions for design- THE GEOLOGICAL STRENGTH INDEX (GSI)
ing tunnels, slopes, or foundations in rocks. The geological
character of the rock material, together with the visual Hoek and Brown recognized that a rock mass failure
assessment of the mass it forms, is used as a direct input criterion would have no practical value unless it could be
for the selection of parameters for predicting rock mass related to geological observations that could be made
strength and deformability. This approach enables a rock quickly and easily by an engineering geologist or geologist
mass to be considered as a mechanical continuum without in the field. They considered developing a new classifica-
losing the influence that geology has on its mechanical tion system during the evolution of the criterion in the late
properties. It also provides a field method for charac- 1970s, but they soon gave up the idea and settled for the
terizing difficult-to-describe rock masses. Recommenda- already published RMR system. It was appreciated that the
tions on the use of GSI are given and, in addition, cases RMR system (and the Q-system) [Bieniawski 1973; Barton
where the GSI is not applicable are discussed. et al. 1974] were developed for the estimation of under-
ground excavation and support and that they included
INTRODUCTION parameters that are not required for estimating rock mass
properties. The groundwater and structural orientation
A few decades ago, the tools for designing tunnels parameters in RMR and the groundwater and stress param-
started to change. Numerical methods were being devel- eters in Q are dealt with explicitly in effective stress
oped that offered the promise for much more detailed numerical analyses, and the incorporation of these param-
analysis of difficult underground excavation problems. eters into the rock mass property estimate results is
Numerical tools available today allow the tunnel inappropriate. Thus, it was recommended that only the first
designer to analyze progressive failure processes and the four parameters of the RMR system (intact rock strength,
sequentially installed reinforcement and support necessary RQD rating, joint spacing, and joint conditions) should be
to maintain the stability of the advancing tunnel until the used for the estimation of rock mass properties if this
final reinforcing or supporting structure can be installed. system had to be used.
However, these numerical tools require reliable input After several years of use, it became obvious that the
information on the strength and deformation characteristics RMR system was difficult to apply to rock masses that are
of the rock mass surrounding the tunnel. As it is practically of very poor quality. The relationship between RMR and
impossible to determine this information by direct in situ the constants “m” and “s” of the Hoek-Brown failure cri-
testing (except for back analysis of already constructed terion begins to break down for severely fractured and
tunnels), there was an increased need for estimating the weak rock masses.
rock mass properties from the intact rock properties and Additionally, since RQD in most of the weak rock
the characteristics of the discontinuities in the rock mass. masses is essentially zero, it became necessary to consider
This resulted in the development of the rock mass failure an alternative classification system. The required system
criterion by Hoek and Brown [1980]. A brief history of the would place greater emphasis on basic geological observa-
tions of rock mass characteristics; reflect the material, its
structure, and its geological history; and would be devel-
oped specifically for the estimation of rock mass properties
1
Professor, National Technical University of Athens, School of rather than for tunnel reinforcement and support. This new
Civil Engineering, Athens, Greece. classification, now called GSI, started life in Toronto,
2
Research assistant, National Technical University of Athens, Canada, with engineering geology input from David Wood
School of Civil Engineering, Athens, Greece.
2 [Hoek et al. 1992]. The index and its use for the Hoek-
Consultant, Vancouver, British Columbia, Canada.
Brown failure criterion was further developed by Hoek

87
[1994] and presented in Hoek et al. [1995] and Hoek and with the intact rock strength and should never be used
Brown [1997], but it was still a hard-rock system roughly independently of this parameter.
equivalent to RMR. Since 1998, Evert Hoek and Paul This index is based on an assessment of the lithology,
Marinos, dealing with incredibly difficult materials structure, and condition of discontinuity surfaces in the
encountered in tunneling in Greece, developed the GSI rock mass, and it is estimated from visual examination of
system to its present form to include poor-quality rock the rock mass exposed in outcrops, in surface excavations
masses (Figure 1) [Hoek et al. 1998; Marinos and Hoek such as road cuts, and in tunnel faces and borehole cores.
2000, 2001]. Today, GSI continues to evolve as the princi- The GSI, by combining the two fundamental parameters of
pal vehicle for geological data input for the Hoek-Brown the geological process—the blockiness of the mass and
criterion. the conditions of discontinuities—respects the main
geological constraints that govern a formation. It is thus a
geologically sound index that is simple to apply in the
field.
Note that attempts to “quantify” the GSI classification
to satisfy the perception that “engineers are happier with
numbers” [Cai et al. 2004; Sonmez and Ulusay 1999] are
interesting, but have to be applied with caution in order not
to lose the geologic logic of the GSI system. The quan-
tification processes used are related to the frequency and
orientation of discontinuities and are limited to rock
masses in which these numbers can easily be measured.
These quantifications do not work well in tectonically dis-
turbed rock masses in which the structural fabric has been
destroyed. In such rock masses, the authors recommend
the use of the original qualitative approach based on care-
ful visual observations. Thus, the “quantification” system
is only valid in the range of, say, 35 < GSI < 75, when the
rock mass behavior depends on sliding and rotation of
intact rock pieces, and where the spacing and condition of
discontinuities that separate these pieces and not the intact
rock strength control the behavior. When the intact rock
pieces themselves can fail, then the quantification is no
longer valid.
Once a GSI “number” has been decided upon, this
number is entered into a set of empirically developed
equations to estimate the rock mass properties that can
then be used as input into some form of numerical analysis
or closed-form solution. The index is used in conjunction
with appropriate values for the unconfined compressive
strength (UCS) of the intact rock, σci, and the petrographic
constant, mi, to calculate the mechanical properties of a
Figure 1.—General chart for GSI estimates from geo-
logical observations. rock mass, in particular the compressive strength of the
rock mass (σcm) and its deformation modulus (E). Updated
values of mi can be found in Marinos and Hoek [2000] or
FUNCTIONS OF THE in the RocLab program [Rocscience, Inc. 2007]. Basic
GEOLOGICAL STRENGTH INDEX procedures are explained by Hoek and Brown [1997], but a
refinement of the empirical equations and the relationship
The heart of the GSI classification is a careful engi- between the Hoek-Brown and Mohr-Coulomb criteria have
neering geology description of the rock mass, which is been addressed by Hoek et al. [2002] for appropriate
essentially qualitative, because it was believed that num- ranges of stress encountered in tunnels and slopes. Hoek
bers on joints were largely meaningless for weak and com- and Diederichs [2006] recently presented new equations
plex rock masses. Note that the GSI system was never for estimating rock mass deformation modulus incor-
intended as a replacement for RMR or Q, as it has no rock porating measured or estimated intact modulus.
mass reinforcement or support design capability. GSI
alone is not a tunnel design tool; its only function is the
estimation of rock mass properties. It is intimately linked

88
 SUGGESTIONS FOR USING GSI estimating the GSI value of a rock mass. How should the
numbers estimated from these sources be projected or
After more than a dozen of years of application of the extrapolated into the rock mass behind a slope or ahead of
GSI and its variations for the characterization of the rock a tunnel?
mass, this paper attempts to answer questions that have Outcrops are an extremely valuable source of data in
been raised by users about the appropriate selection of the the initial stages of a project, but they suffer from the
index for various rock masses under various conditions. disadvantage that surface relaxation, weathering, and/or
alteration may have significantly influenced the appear-
When Not to Use GSI ance of the rock mass components. This disadvantage can
be overcome by trial trenches but, unless these are
The GSI classification system is based on the machine-excavated to considerable depth, there is no guar-
assumption that the rock mass contains a sufficient number antee that the effects of deep weathering will have been
of “randomly” oriented discontinuities such that it behaves eliminated. Judgment is therefore required in order to
as a homogeneous isotropic mass. In other words, the allow for these weathering and alteration effects in assess-
behavior of the rock mass is independent of the direction ing the most probable GSI value at the depth of the
of the applied loads. Therefore, it is clear that the GSI proposed excavation.
system should not be applied to those rock masses in Excavated slope and tunnel faces are probably the
which there is a clearly defined dominant structural orien- most reliable source of information for GSI estimates, pro-
tation or structurally dependent gravitational instability. vided that these faces are reasonably close to and in the
However, the Hoek-Brown criterion and the GSI chart can same rock mass as the excavation under investigation.
be applied with caution if the failure of such rock masses is Borehole cores are the best source of data at depth, but
not controlled by such anisotropy (e.g., in the case of a it must be recognized that it is necessary to extrapolate the
slope when the dominant structural discontinuity set dips one-dimensional information provided by the core to the
into the slope and failure occurs through the rock mass). three-dimensional in situ rock mass. However, this is a
For rock masses with a structure such as that shown in the problem common to all borehole investigations, and most
bottom row of the GSI chart (Figure 1), anisotropy is not a experienced engineering geologists are comfortable with
major issue, as the difference in the strength of the rock this extrapolation process.
and that of the discontinuities within it is often small. For stability analysis of a slope, the evaluation is based
Anisotropy in cases of stress-dependent instability is dis- on the rock mass through which it is anticipated that a
cussed later in this paper. potential failure plane could pass. The estimation of GSI
It is also inappropriate to assign GSI values to exca- values in these cases requires considerable judgment,
vated faces in strong hard rock with a few discontinuities particularly when the failure plane can pass through sev-
spaced at distances of similar magnitude to the dimensions eral zones of different quality. Mean values may not be
of the tunnel or slope under consideration. In such cases, appropriate in this case.
the stability of the tunnel or slope will be controlled by the For tunnels, the index should be assessed for the
three-dimensional geometry of the intersecting discon- volume of rock involved in carrying loads, e.g., for about
tinuities and the free faces created by the excavation. one diameter around the tunnel in the case of tunnel
Obviously, the GSI classification does not apply to such behavior or more locally in the case of a structure such as
cases. the elephant foot of a steel arch. In more general terms, the
numerical models may include the variability of GSI
Geological Description in the GSI Chart values over the tunnel in “layers.” Drs. Edmund Medley
and Dimitrios Zekkos are currently considering developing
In dealing with specific rock masses, it is suggested a function defining the variation of GSI with depth for a
that the selection of the appropriate case in the GSI chart specific case.
should not be limited to the visual similarity with the
sketches of the structure of the rock mass as they appear in Anisotropy
the charts. The associated descriptions must also be read
carefully, so that the most suitable structure is chosen. The As discussed above, the Hoek-Brown criterion (and
most appropriate case may well lie at some intermediate other similar criteria) assumes that the rock mass behaves
point between the limited number of sketches or descrip- isotropically and that failure does not follow a preferential
tions included in the charts. direction imposed by the orientation of a specific discon-
tinuity or a combination of two or three discontinuities. In
Projection of GSI Values Into the Ground these cases, the use of GSI to represent the whole rock
mass is meaningless, as the failure is governed by the shear
Outcrops, excavated slopes, tunnel faces, and borehole strength of these discontinuities and not of the rock mass.
cores are the most common sources of information for

89
However, cases where the criterion and the GSI chart can been subjected to very heavy production blasting. The fac-
reasonably be used have been discussed above. tor allows for the disruption of the interlocking of the
However, in a numerical analysis involving a single individual rock pieces as a result of opening of the discon-
well-defined discontinuity such as a shear zone or fault, it tinuities. The influence of this factor is of great signifi-
is sometimes appropriate to apply the Hoek-Brown cri- cance to the calculated factors of safety.
terion to the overall rock mass and to superimpose the At this stage, there is relatively little experience in the
discontinuity as a significantly weaker element. In this use of this factor, and it may be necessary to adjust its
case, the GSI value assigned to the rock mass should participation in the equations as more field evidence is
ignore the single major discontinuity. The properties of accumulated. However, the experience so far suggests that
this discontinuity may fit the lower portion of the GSI this factor does provide a reasonable estimate of the
chart or they may require a different approach, such as influence of damage due to stress relaxation or blasting of
laboratory shear testing of soft clay fillings. excavated rock faces. Note that this damage decreases with
In general terms, when confinement is present, the depth into the rock mass and, in numerical modeling, it is
stress-dependent regime is controlled by the anisotropy of generally appropriate to simulate this decrease by dividing
the rock masses (e.g., slates, phyllites, etc.). A discussion the rock mass into a number of zones with decreasing
of anisotropy rock mass behavior in tunneling beyond the values of D being applied to successive zones as the dis-
commonly used classification systems is presented by tance from the face increases. On the other hand, in very
Button et al. [2004]. In these cases, it would be necessary large open-pit mine slopes in which blasts can involve
to develop an orientation-dependent GSI. This is a recent many tons of explosives, blast damage has been observed
idea to try to simplify the treatment of anisotropic up to 100 m or more behind the excavated slope face. This
problems. However, in view of the potential for would be a case for D=1 and there is a very large reduction
complicating the understanding of GSI, an alternative in shear strength associated with damage. Hoek and
approach may be to use an orientation-dependent UCS. Karzulovic [2000] have given some guidance on the extent
This is more logical from a physical point of view and, of this damage and its impact on rock mass properties. For
being almost completely interchangeable with GSI from a civil engineering slopes or foundation excavation, the blast
mathematical point of view, should work just as well. The damage is much more limited in both severity and extent,
GSI value in this case would be high, and the rock mass and the value of D is generally low.
strength would be determined by the orientation-dependent This problem becomes less significant in weak and
σci value. tectonically disturbed rock masses, as excavation is gener-
With the capacity of present-day microcomputers, it is ally carried out by “gentle” mechanical means and the
also possible to model anisotropy by superimposing a large amount of surface damage is negligible compared to that
number of discontinuities on an isotropic rock mass which which already exists in the rock mass.
is assigned a higher GSI value. These discontinuities can
be assigned shear strength and stiffness characteristics that Geological Strength Index at Great Depth
simulate the properties of the schistosity, bedding planes,
and joints in the rock mass. Such models have been found In hard rock at great depth (e.g., 1,000 m or more) the
to work well and give results that compare well with more rock mass structure is so tight that the mass behavior
traditional anisotropic solutions. approaches that of the intact rock. In this case, the GSI
value approaches 100 and the application of the GSI sys-
Aperture of Discontinuities tem is no longer meaningful.
The failure process that controls the stability of under-
The strength and deformation characteristics of a rock ground excavations under these conditions is dominated by
mass are dependent on the interlocking of the individual brittle fracture initiation and propagation, which leads to
pieces of intact rock that make up the mass. Obviously, the spalling, slabbing, and, in extreme cases, rock bursts.
aperture of the discontinuities that separate these indi- Considerable research effort has been devoted to the study
vidual pieces has an important influence on the rock mass of these brittle fracture processes, and Diederichs et al.
properties. [2004] provide a useful summary of this work.
There is no specific reference to the aperture of the When tectonic disturbance is important and persists
discontinuities in the GSI chart, but a “disturbance factor” with depth, these comments do not apply and the GSI
D has been provided in the most recent version of the charts may be applicable, but should be used with
Hoek-Brown failure criterion [Hoek et al. 2002] and is also caution.
used in the Hoek and Diederichs [2006] approach for
estimating deformation modulus. This factor ranges from Discontinuities With Filling Materials
D=0 for undisturbed rock masses, such as those excavated
by a tunnel boring machine, to D=1 for extremely dis- The GSI charts can be used to estimate the charac-
turbed rock masses, such as open-pit mine slopes that have teristics of rock masses with discontinuities with filling

90
materials using the descriptions in the columns for “poor” Weathered Rock Masses
or “very poor” condition of discontinuities. If the filling
material is systematic and thick (e.g., more than a few The GSI values for weathered rock masses are shifted
centimeters) or shear zones are present with clayey mate- to the right of those of the same rock masses when these
rial, then the use of the GSI chart for heterogeneous rock are unweathered. If the weathering has penetrated into the
masses (discussed below) is recommended. intact rock pieces that make up the mass (e.g., in
weathered granites), then the constant mi and the
Influence of Water unconfined strength of the σci of the Hoek-Brown criterion
must also be reduced. If the weathering has penetrated the
The shear strength of the rock mass is reduced by the rock to the extent that the discontinuities and the structure
presence of water in the discontinuities or filling materials have been lost, then the rock mass must be assessed as a
when these are prone to deterioration as a result of changes soil and the GSI system no longer applies.
in moisture content. This is particularly valid in the “fair”
to “very poor” categories of discontinuities, where a shift Heterogeneous and Lithologically Varied or
to the right may be made for wet conditions. The shift to Complex Rock Masses
the right is more substantial in the low-quality range of
rock mass (last rows and columns of the chart). GSI has been extended to accommodate the most
Water pressure is dealt with by effective stress analysis variable of rock masses, including extremely poor quality
in design, and it is independent of the GSI characterization sheared rock masses of weak schistose materials (such as
of the rock mass. siltstones, clay shales, or phyllites) often interbedded with
strong rock (such as sandstones, limestones, or quartzites).
A GSI chart for flysch, a typical heterogeneous lithological

Figure 2.—Geological Strength Index for heterogeneous rocks such as flysch.

91
formation with tectonic disturbance, was published by In design, the intact rock properties σci and the mi must
Marinos and Hoek [2001]. This chart has recently been also be considered. A “weighted average” of the properties
revised and is reproduced in Figure 2. This revision is of the strong and weak layers should be used.
based on recent experience from a number of tunnels con- Marinos et al. [2006] recently presented a quantitative
structed in Greece. It includes cases of siltstones with little description, using GSI, for rock masses within an ophi-
disturbance and a variety of cases of siltstones alternating olitic complex. Included are types with large variability
with good rock (e.g., sandstone). due to their range of petrographic types, their tectonic
For lithologically varied but tectonically undisturbed deformation, and their alternation (Figure 4). The structure
rock masses, such as the molasses, a new GSI chart was of the various masses include types from massive strong to
presented by Hoek et al. [2005] (Figure 3). For example, sheared weak, while the conditions of discontinuities are,
molasse consists of a series of tectonically undisturbed in most cases, fair to very poor due to the fact that they are
sediments of sandstones, conglomerates, siltstones, and affected by serpentinization and shearing. This description
marls produced by the erosion of mountain ranges after the allows the estimation of the range of properties and the
final phase of an orogeny. The molasses behave quite understanding of the dramatic change in tunneling, from
differently from flysch, which has the same composition stable conditions to severe squeezing within the same for-
mation at the same depth.
but was tectonically disturbed during the orogeny. They
behave as continuous rock masses when they are confined
at depth, and the bedding planes do not appear as clearly
defined discontinuity surfaces. Close to the surface the
layering of the formations is discernible, and only then
similarities may exist with the structure of some types of
flysch.

Figure 3.—Chart for confined molasse (mainly appli-

cable for tunnels).
Figure 4.—Ranges of GSI for various qualities of
peridotite-serpentinite rock masses in ophiolites.

92
 Rocks of Low Strength of Recent Age (RMR, Q, RMi [Palmström 1996], and GSI) were inves-
tigated, and all systems ratings are grouped in a common
When rocks such as marls, claystones, siltstones, and fabric index chart. The reader is reminded not to lose sight
weak sandstones are developed in geologically stable con- of the real geological world in considering such
ditions in a posttectonic environment, they usually present correlations.
a simple structure with no or few discontinuities. When
these rocks form continuous masses with no discontinu- GSI AND CONTRACT DOCUMENTS
ities, the rock mass can be treated as intact with engineer-
ing parameters given directly by laboratory testing. In such One of the most important contractual problems in
cases, the GSI classification is not applicable. rock construction and particularly in tunneling is the issue
In cases where discontinuities are present, the use of of “changed ground conditions.” There are invariably
the GSI chart for “blocky” or “massive” rock masses arguments between the owner and the contractor on the
(Figure 1) may be applicable. The discontinuities in such nature of the ground specified in the contract and that
weak rocks, although they are limited in number, cannot be actually encountered during construction. In order to over-
better than “fair” (usually “fair” or “poor”); thus, the GSI come this problem, there has been a tendency to specify
values tend to be in the range of 45–65. In these cases, the the anticipated conditions in terms of tunneling classi-
low strength of the rock mass results from low intact fications. More recently, some contracts have used the GSI
strength σci. classification for this purpose, and the authors are strongly
opposed to this trend.
PRECISION OF THE As discussed earlier in this paper, GSI was developed
GSI CLASSIFICATION SYSTEM solely for the purpose of estimating rock mass properties.
Therefore, GSI is only one element in a tunnel design
The “qualitative” GSI system works well for engineer- process and cannot be used, on its own, to specify
ing geologists since it is consistent with their experience in tunneling conditions. It must be associated with the intact
describing rocks and rock masses during logging and rock strength, the petrographic constant mi, and all of the
mapping. In some cases, engineers tend to be uncomfort- characteristics (such as anisotropy) of the rock mass that
able with the system because it does not contain param- may impose a different mode of failure than that of a
eters that can be measured in order to improve the stressed homogeneous isotropic rock mass.
precision of the estimated GSI value. The use of any classification system to specify
The authors do not share this concern, as they believe anticipated tunneling conditions is always a problem as
that it is not meaningful to attempt to assign a precise these systems are open to a variety of interpretations,
number to the GSI value for a typical rock mass. In all but depending on the experience and level of conservatism of
the very simplest of cases, GSI is best described by the observer. This can result in significant “changes” in
assigning it a range of values. For analytical purposes, this excavation or support type and can have important finan-
range may be defined by a normal distribution with mean cial consequences.
and standard deviation values assigned on the basis of The geotechnical baseline report [Essex 1997] was
common sense. GSI, with its qualitative principles of introduced in an attempt to overcome some of the diffi-
geological descriptions, is not restrained by the absence of culties and has attracted an increasing amount of inter-
good exposures or the limitations of quantitative core national attention in tunneling.
descriptions.
Although GSI is a totally independent system, in the CONCLUSIONS
earlier period of its application it was proposed that
correlation of “adjusted” RMR and Q values with GSI be Rock mass characterization has an important role, not
used for providing the necessary input for the Hoek- only to define a conceptual model of the site geology, but
Brown criterion. Although this procedure may work also for the quantification needed for analyses “to ensure
with the better-quality rock masses, it is unreliable in the that the idealization (for modeling) does not misinterpret
range of weak (e.g., GSI<35), very weak, and hetero- actuality” [Knill 2003]. If it is carried out in conjunction
geneous rock masses, where these correlations are not with numerical modeling, rock mass characterization pre-
recommended. sents the prospect of a far better understanding of the
Whenever GSI is used, a direct assessment, based on mechanics of rock mass behavior [Chandler et al. 2004].
the principles and charts presented above, is recom- The GSI system has considerable potential for use in rock
mended. Fortunately, most GSI users have no difficulty in engineering because it permits many characteristics of a
thinking of it as a totally independent system. However, in rock mass to be quantified, thereby enhancing geological
cases of comparisons or back analysis where other classi- logic and reducing engineering uncertainty. Its use allows
fication systems have been used, some kind of correlation the influence of variables, which make up a rock mass, to
with these other systems is needed. In such cases, it may be assessed and thus the behavior of rock masses to be
be useful to consult the paper by Tzamos and Sofianos explained more clearly. One of the advantages of the GSI
[in press]. The four classification-characterization systems is that the geological reasoning it embodies allows adjust-

93
 ments of its ratings to cover a wide range of rock masses Hoek E, Caranza-Torres CT, Corkum B [2002]. Hoek-
and conditions, but it also allows us to understand the Brown failure criterion. In: Bawden HRW, Curran J,
limits of its application. Telsenicki M, eds. Proceedings of the North American
Rock Mechanics Society (NARMS–TAC), Mining Innova-
ACKNOWLEDGMENTS tion and Technology (Toronto, Ontario, Canada), pp. 267–
273.
The paper is the result of a project cofunded by the Hoek E, Kaiser PK, Bawden WF [1995]. Support of
European Social Fund (75%) and Greek National underground excavations in hard rock. Rotterdam, Nether-
Resources (25%) – Operational Program for Educational lands: Balkema.
and Vocational Training II (EPEAEK II) and particularly Hoek E, Marinos P, Benissi M [1998]. Applicability of
the Program PYTHAGORAS. the geological strength index (GSI) classification for weak
and sheared rock masses: the case of the Athens schist for-
REFERENCES mation. Bull Eng Geol Env 57(2):151–160.
Hoek E, Marinos P, Marinos V [2005]. Characteriza-
Barton N, Lien R, Lunde J [1974]. Engineering classi- tion and engineering properties of tectonically undisturbed
fication of rock masses for the design of tunnel support. but lithologically varied sedimentary rock masses under
Rock Mech 6(4):189–236. publication. Int J Rock Mech Min Sci 42:277–285.
Bieniawski ZT [1973]. Engineering classification of Hoek E, Wood D, Shah S [1992]. A modified Hoek-
jointed rock masses. Trans S Afr Inst Civ Eng 15:335–344. Brown criterion for jointed rock masses. In: Hudson JA,
Button E, Riedmueller G, Schubert W, Klima K, ed. Proceedings of the Rock Mechanics Symposium
Medley E [2004]. Tunnelling in tectonic melanges: accom- (Eurock ’92). London: British Geotechnical Society, pp.
modating the impacts of geomechanical complexities and 209–214.
anisotropic rock mass fabrics. Bull Eng Geol Env 63:109– Knill J [2003]. Core values (first Hans-Closs lecture).
117. Bull Eng Geol Env 62:1–34.
Cai M, Kaiser PK, Uno H, Tasaka Y, Minami M Marinos P, Hoek E [2000]. GSI: a geologically
[2004]. Estimation of rock mass strength and deformation friendly tool for rock mass strength estimation. In: Pro-
modulus of jointed hard rock masses using the GSI system. ceedings of GeoEng 2000 at the International Conference
Int J Rock Mech Min Sci 41(1):3–19. on Geotechnical and Geological Engineering (Melbourne,
Chandler RJ, de Freitas MH, Marinos P [2004]. Geo- Victoria, Australia). Lancaster, PA: Technomic Publishers,
technical characterization of soils and rocks: a geological pp. 1422–1446.
perspective. In: Proceedings of the Advances in Geo- Marinos P, Hoek E [2001]. Estimating the geo-
technical Engineering: The Skempton Conference. Vol 1. technical properties of heterogeneous rock masses such as
London: Thomas Telford, pp. 67–102. flysch. Bull Eng Geol Env 60:82–92.
Diederichs MS, Kaiser PK, Eberhardt E [2004]. Dam- Marinos P, Hoek E, Marinos V [2006]. Variability of
age initiation and propagation in hard rock during tunnel- the engineering properties of rock masses quantified by the
ing and the influence of near-face stress rotation. Int J geological strength index: the case of ophiolites with
Rock Mech Min Sci 41(5):785–812. special emphasis on tunnelling. Bull Eng Geol Env 65:
Essex RJ. [1997]. Geotechnical baseline reports for 129–142.
underground construction. Reston, VA: American Society Marinos V, Marinos P, Hoek E [2005]. The geological
of Civil Engineers. strength index: applications and limitations. Bull Eng Geol
Hoek E [1994]. Strength of rock and rock masses. Environ 64:55–65.
News J ISRM 2(2):4–16. Palmström A [1996]. Characterizing rock masses by
Hoek E, Brown ET [1980]. Underground excavations the RMi for use in practical rock engineering. Part 1: the
in rock. London: Institution of Mining and Metallurgy. development of the rock mass index (RMi). Tunnelling
Hoek E, Brown ET [1997]. Practical estimates of rock Undergr Space Technol 11(2):175–88.
mass strength. Int J Rock Mech Min Sci Geomech Abstr Rocscience, Inc. [2007]. RocLab v1.0: rock mass
34:1165–1186. strength analysis using the generalized Hoek-Brown
Hoek E, Diederichs MS [2006]. Empirical estimation failure criterion. [http://www.rocscience.com/products/
of rock mass modulus. Int J Rock Mech Min Sci 43:203– RocLab.asp]. Date accessed: April 2007.
215. Sonmez H, Ulusay R [1999]. Modifications to the
Hoek E, Karzulovic A [2000]. Rock mass properties geological strength index (GSI) and their applicability to
for surface mines. In: Hustrulid WA, McCarter MK, van the stability of slopes. Int J Rock Mech Min Sci 36:743–
Zyl DJA, eds. Slope stability in surface mining. Littleton, 760.
CO: Society for Mining, Metallurgy, and Exploration, Inc., Tzamos S, Sofianos AI [in press]. A correlation of four
pp. 59–70. rock mass classification systems through their fabric
Hoek E, Marinos P [in press]. A brief history of the indices. Int J Rock Mech Min Sci.
development of the Hoek-Brown failure criterion. Int J
Soils Rocks.

94

View publication stats