i An update to this article is included at the end

Engineering Geology 239 (2018) 282–297

Contents lists available at ScienceDirect

Engineering Geology
journal homepage: [Link]/locate/enggeo

Maintaining geological reality in application of GSI for design of engineering T

structures in rock
⁎
Marinos Vassilis.a, , Carter Trevor.G.b
a
Aristotle University of Thessaloniki, Thessaloniki 541 24, Greece
b
TGC Geosolutions & Golder Associates, Toronto, Ontario, Canada

A R T I C LE I N FO A B S T R A C T

Keywords: In rock engineering design, significant advances have occurred in recent years in numerical modelling capability
GSI characterization with increasing trends to ever greater use of synthetic rock mass models and discrete fracture network (DFN)
Rock mass strength modelling, but all too often with little comparative improvement in geologic base data. As a consequence, there
Hoek-Brown criterion is even more need that reliable estimates be available of strength and deformation characteristics of the rock
Weathering
masses on which or within which engineering structures are to be created, be it a tunnel, a foundation or a slope.
Alteration
Tectonism
Geological Strength Index (GSI) characterization, linked with Hoek-Brown strength determination as a basis for
modelling has been widely adopted by engineers and geologists involved in design and construction of en-
gineering structures. The need for geological definition of rock mass properties required as inputs into numerical
analysis, constitutes one of the greatest reasons for application of the GSI chart, allowing characterization of
even difficult-to-describe rock masses, including tackling even the most problematic of weak and complex rock
masses. Back-analyses of tunnels, slopes and foundation behaviour using GSI and its reliable application in rock
engineering designs attest to its reliability. With continuing use worldwide, the GSI system has continued to
evolve, but greater understanding is needed in the definition of input constants, for establishing both GSI and
intact rock properties. This need for improved evaluation, particularly from a geological perspective, is ad-
dressed in this paper. Geological processes of tectonism, weathering and alteration all affect GSI. Evaluation of
these factors, which are each critical to proper GSI definition, are analyzed based on real rock mass cases.
Suggested ranges in variability of intact rock parameters uniaxial compressive strength σci and material constant
mi for common rock masses are presented in the context of a composite new GSI chart. This chart allows se-
lection of appropriate GSI ranges for any specific rock suite. Specific key engineering geological characteristics
that differentiate igneous, metamorphic and sedimentary one from each other are highlighted through discus-
sion of various example rock units (including gneisses, granites, ophiolites, limestones, schists, siltstones/
mudstones/shales, and molassic and flysch formations). Illustrations are given of how geological differentiation
dictates variability in geotechnical properties of most common rock masses.

1. Introduction foundation or a slope cut.

Mostly geological data acquisition for design purposes involves at a
Despite the fact that significant advances have occurred within al- minimum implementing one or other of the following approaches: a)
most every area of geotechnical design, with arguably the greatest laboratory testing; b) in situ testing, c) use of rock mass classifications
developments in rock engineering being in numerical modelling cap- and/or d) back analysis. However, individually, data gathered from
ability, to date similar levels of advance have not been achieved in these different approaches do not yield identical results, primarily due
improving characterization of the geological variability that exists in to the significant differences in scale that exist between each of these
natural rock masses. approaches, and also due to complications of representativeness.
Geological representativeness is key to achieving effective rock Laboratory and insitu tests because they are very location specific, may
engineering design. This requires that reliable estimates be available of not be truly representative of the overall rock mass. Better reflection is
strength and deformation characteristics of the rock masses on which or often needed of natural heterogeneity and degree of fracturing. Cost
within which engineering structures are to be created, be it a tunnel, a and time issues additionally can complicate data acquisition,

⁎
Corresponding author at: Aristotle University of Thessaloniki, School of Geology, GR-541 24 Thessaloniki, Greece.
E-mail addresses: marinosv@[Link] (V. Marinos), tcarter@[Link] (T.G. Carter).

[Link]
Received 7 July 2017; Received in revised form 11 March 2018; Accepted 21 March 2018
Available online 30 March 2018
0013-7952/ © 2018 Elsevier B.V. All rights reserved.
V. Marinos, T.G. Carter Engineering Geology 239 (2018) 282–297

Fig. 1. Basic GSI chart for visual geologic characterization of rock masses (Marinos and Hoek, 2000).

particularly when it comes to carrying out in situ tests. Back-analysis, published by Hoek et al. (1998) and Marinos and Hoek (2000), re-
provides one of the best ways to determine appropriate geotechnical sulting in the now familiar basic chart, as per Fig. 1. As initially for-
parameters, particularly when construction has started, provided the mulated, the GSI chart was structured as a step forward to better meet
analysis approach being used for the parameter definition is valid for the need for delivering more reliable, more geologically based, re-
the observed field behaviour. presentative input data.
Reasonable geotechnical parameters are needed for the design of The not infrequent lack of adequate geological definition of rock
most engineering projects, and in particular for design of tunnel support mass properties that should realistically be required as inputs into nu-
for long deep tunnels beneath high mountain ranges where drilling is merical analysis or into closed form solutions for designing tunnels,
difficult before construction starts. In such situations, where direct slopes or foundations in rock, constitutes one of prime reasons practi-
drilling and testing or insitu performance back-analysis approaches tioners, are turning more and more to use of the GSI chart rather than
cannot be executed, there is no option but to rely upon prescriptive applying other classification approaches, Arguably, use of the GSI chart
geological mapping and interpretation and then make use of some form is of particular advantage when data is limited or characterization is
of rock mass classification scheme that is correlated with the basic needed of difficult-to-describe rock masses. In fact, one of the most
parameters needed for design. significant problems of numerically based classification systems, such
The “Geological Strength Index”, GSI, was developed, initially by as the Q, RMR and RMi systems (Barton et al., 1974; Bieniawski, 1976;
Hoek (1994), but then amplified and improved in subsequent papers Palmström, 2005), as pointed out by Palmström and Broch (2006), is

283
V. Marinos, T.G. Carter Engineering Geology 239 (2018) 282–297

that they suffer markedly outside the range for which they were ori-
ginally developed. This is particularly the case for describing geological
conditions met in weak complex rock mass conditions. By contrast, the
flexibility of the GSI chart for use with many different rock types and
difficult-to-describe rock masses has been well demonstrated on nu-
merous projects, most notably by Marinos et al., 2012, describing uti-
lization of the GSI system during the construction of 62 tunnels along
the Egnatia Highway in Northern Greece. Requirements for applying
the Hoek-Brown failure criterion as basis for tunnel support in the
difficult and weak geological conditions encountered along the Egnatia
alignment led to one of the first extensions of the standard GSI geo-
technical classification chart specifically for application for complex
flysch conditions.
With the rapid growth of improved numerical design tools, which
now allow synthetic rock masses to be built, progressive failure pro-
cesses to be modelled and sequentially installed support to be analyzed,
the need for acquisition of more reliable, geologically-based rock mass
Fig. 2. Difficult Ground Conditions for undertaking rock mass classification
parameters has also grown. While introduction of classifications has (Photo from Egnatia Highway): tectonically deformed intensively folded/
greatly aided design of structures in a wide variety of rock masses, their faulted siltstone with broken and deformed sandstone layers forming an almost
use has in some cases compromised acquisition of holistic geological chaotic structure.
understanding. In difficult rock masses, changes in conditions brought
about through complex geological processes of tectonism, weathering
of the competence scale, for weak inhomogeneous rock masses, many of
and/or alteration that significantly affect GSI can be analyzed and key
the most notable improvements come from work that Hoek and Marinos
parameter characteristics modified accordingly.
had undertaken during tunnelling in difficult ground in Northern
Greece (Hoek et al., 1998, Marinos and Hoek, 2000, Marinos, 2007,
2. The Geological Strength Index Marinos 2017), (Fig. 2) and from work on saprolitic slopes (Castro
et al., 2013). At the other end of the competence scale, where brittle
The basic GSI classification chart (Fig. 1) was set up to allow any spalling dominates behaviour, key advances in understanding have
competent practitioner to address the two principal factors considered come from work by Martin, Diederichs, Cai, Kaiser and others (Cai
important influences on the mechanical properties of a rock mass – the et al., 2004; Diederichs, 2007; Hoek and Martin, 2014). Following in-
structure (or blockiness), and the condition of the joints. The chart troduction of the observational GSI chart, several attempts were made
definition approach was then recommended to be used as basis for by various authors to quantify the chart axes. Various quantitative
defining GSI for input into the now standard transfer algebra moving scales for the original GSI chart axes were suggested by, for example,
from rock mass description through to m, s and a definitions for ap- Sonmez and Ulusay (1999) and Cai et al. (2004), while a recent study
plication in the Hoek-Brown failure criterion for strength definition for about quantification of GSI was also suggest by Hoek et al. (2013), all
design, (as developed by Hoek et al., 2002), per the following equations advocating different parameter choices.
(Table 1):
For defining the rockmass deformation modulus, the following re-
lationship, proposed by Hoek and Diederichs (2006) and again depen- 3. Definition of GSI for real rock masses
dent on GSI, provides reasonable estimation:
3.1. Defining geological reality & representative GSI's
1 − D /2
Erm = Ei ⎛0.02 + ⎞
⎝ 1 + e ((60 + 20D − GSI )/12) ⎠ Defining a representative GSI value range requires the observer to
make a careful mental engineering geological “description” of the rock
With more extensive use worldwide, the GSI chart established in
mass as basis for reaching an estimate of the most applicable classifi-
1994 and its applicability in design continued to evolve, with several
cation value range appropriate for the rock mass under consideration.
new charts published for a range of different rock masses, e.g. Marinos
In dealing with specific rock masses, it is suggested that the selection of
and Hoek, 2000. The need to address the validity and applicability of
an appropriate zone location on the chart not be limited merely to
the 2002 transfer equations was also recognized. In consequence, sev-
checking the visual similarity with these sketches of the structure of the
eral additional publications, from 1998 onwards were directed towards
rock mass as they appear in the left-hand side of the charts, but rather
providing clarification of application methodology, primarily aimed at
should be based on gaining real understanding by examining actual face
addressing difficulties for practitioners, particularly with usage for both
exposures or outcrops. The most appropriate case may well lie at some
ends of the rock competence scale (Carter et al., 2008). At the low end
intermediate point between the limited number of rock mass sketches
or descriptions included around the charts. Outcrops, excavated slopes,
Table 1
tunnel faces and drill cores are the most common sources of information
Standard Hoek – Brown Transfer Equations relating intact rock and rock mass
for the estimation of a GSI value for a rock mass, but these are not of
properties with respect to GSI (Hoek et al., 2002).
equal calibre.
σ1 = σ3 + σci(mb ∗ σ3/σci + s)a Outcrops are an extremely valuable source of good observational
where
data, particularly in the initial stages of a project. However, if the
mb = mi exp ( GSI − 100
28 − 14D ) project is ultimately underground such outcrops, suffer from the dis-
and advantage that relaxation and weathering may have significantly in-
s = exp ( GSI − 100
9 − 3D ) fluenced their condition as compared with the appearance of the same
where D = Disturbance Factor due to the excavation method rock mass when exposed deep underground.
Excavated slopes and exposed tunnel faces are probably the most
1 1 −GSI /15
with a = + (e − e−20/3) reliable source of information for estimating GSI for project design
2 6
provided that these faces are representative of likely project conditions.

284
V. Marinos, T.G. Carter Engineering Geology 239 (2018) 282–297

For design of tunnels however, particularly in hard strong rock masses mines are quite often heavily altered as they tend to be mineralised, so
with limited structural dislocation, it is important that some appro- oftentimes the county rock forming the hanging wall or footwall of the
priate allowance be made for damage due to mechanical excavation or stopes for which rock engineering design is needed are also significantly
blasting that may have degraded the appearance of any mapped rock altered.
face. As the purpose of estimating GSI values for many such situations is Rock mass weathering, by contrast, particularly in glaciated regions,
aimed at assigning properties to the insitu undisturbed rock mass, typically extends only to shallow depths, with decomposition brought
failure to allow for effects of blast damage may result in assigning va- about by two different, but linked processes – mechanical and chemical
lues that may be too conservative. Attempts should therefore be made degradation. Taken in the context of the Hoek-Brown failure criteria
in such circumstances to properly estimate the degree of disturbance, and GSI, the influence that either process exerts on GSI can be sig-
and thereby define D, according with the descriptive tabulations pre- nificant, as both processes degrade not just the parent intact rock ma-
sented by Hoek et al., 2002. terial but also, they change the character and competence of the rock-
Drilled cores of reasonable diameter and core quality (e.g. by use of mass fabric.
triple tube equipment) provide one of the best sources for acquiring According to the weathering degree the discontinuity surface con-
data from depth into the rock, but one has to be cognizant that it is dition becomes poorer and the interlocking of rock blocks becomes
necessary to extrapolate the almost two-dimensional information pro- loosened. The structure on the other hand may not in principle be af-
vided by core drilling to the three-dimensional insitu rock mass. Almost fected, at least if weathering is not very advanced. The typical char-
all experienced engineering geologists are comfortable with this ex- acteristics of rock masses that have been subjected to these different
trapolation process in order to formulate an engineering geological degrees of weathering have been described by ISRM (1981) and by the
conceptual model of the investigated area. Multiple boreholes and also Engineering Group of the Geological Society of London (Anon, 1995),
inclined boreholes are of particular help in this regard as a means for with respect to certain grades (from fresh rock, W-I to clayey-sandy soil,
acquiring data necessary for achieving the most accurate 3D inter- W-VI), as listed on the left of Fig. 4.
pretation of rock mass characteristics with depth. As can be appreciated from the descriptions in this figure, both
mechanical and chemical weathering of the intact rock results not only
3.2. Interaction between GSI and tectonism in changes to the rock mass fabric (by breaking down insitu blocks into
smaller pieces), but also these processes alter properties like σci and mi,
Rock masses disturbed and broken by structural dislocation, as a consequence of their weakening the interlocking and bonding of
shearing, folding or compression can be termed “tectonized”. mineral grains within the rock skeleton. The central column in Fig. 4
When tectonism is low, jointing intensities typically are also low (after Stacey and Page, 1986), provides some guidance for defining
and GSI values are high to very high (Intact to slightly Blocky struc- estimates of the degree of intact rock strength reduction per weathering
ture). Low GSI values can however occur when even quite competent grade. As will be clear from this strength reduction column, the onset of
rock masses are crossed by multiple discontinuities, such as frequent fully pervasive weathering, such that any original intact fragments have
bedding or schistosity planes or by fault zones, often with these dis- become fully friable due to the influence of chemical/mineralogical
continuities creating Fair or Poor rock mass conditions. change, occurs only for rocks weathered to grade W-IV. At a W-III state
In tectonic areas, particularly if compressional, GSI values may be the intact rock material within the fabric still retains much of its parent
considerably reduced, since structure in such areas tends to be quite characteristics, while the condition of discontinuities may have dete-
intense with significant numbers of fractures (typically joints) present. riorated appreciably. This change from W-III to W-IV can thus con-
In some cases, the rock mass may even be sheared. These types of rock stitute a critical GSI definition boundary since there is generally con-
mass frequently behave in a more ductile manner, particularly if the siderable reduction in intact properties at this change, in addition to
parent material is also relatively weak (e.g. for mudstones, shales or block size changes resulting from reduction of incipient fragment size
siltstones). For these rock masses, joint condition on the GSI chart associated with weathering.
moves further to the right on the x-axis scale to Poor, or even to Very An indicative example of how increased degrees of weathering af-
Poor (e.g., when sheared, when slickensided or soft clay coatings are fects GSI is illustrated in Fig. 5. For rocks exhibiting weathering grades
commonly observed). For weak parent rock material, intact rock W-II and W-III, discontinuity condition is shifted to the middle or right
strength σci and mi values may also be reduced. This influence of tec- columns of the GSI chart; whereas for grades W-IV to W-V, joint con-
tonism on GSI values for two different rock types is illustrated by the ditions are likely Poor to Very Poor due to the presence of weathering
different track paths of the various zones identified within Fig. 3. products along the joints, shifting the location even further right and
downwards on the chart.
3.3. Interaction between GSI and weathering - alteration A good example of application of a modified GSI chart developed
specifically for examining weathering in gneissic rocks is shown in
GSI can be severely affected by weathering or alteration. Fig. 6. Here the horizontal joint condition scale of the standard chart
Infrequently will a project be executed entirely in fresh rock. Only in has been modified to reflect specific weathering grades, W-I through W-
deep tunnels and deep mining situations is it likely that one would be V, allowing direct definition of decreased GSI values to appropriately
entirely in essentially dry, fresh (unweathered) rock. On the other hand, reflect the effects of weathering. This chart was developed and cali-
alteration can range from none to severe, even to significant depths, as brated on a site-specific basis during the construction of six tunnels in
often observed in close proximity to ore zones in deep mining situa- Northern Greece, by comparison of GSI classifications and their re-
tions. In most construction projects, serious consideration thus needs to spective temporary support categories as designed and as implemented,
be given to the impact that these natural processes of weathering and the latter clearly showing a decrease of around 10 units from the initial
alteration can have in degrading intact rock material quality, strength “design” values. The chart in Fig. 6 was developed for GSI definition for
and deformability from the initial intact state. gneisses as well as for other rock masses with similarities in weathering
Considerable confusion exists in the engineering rock mechanics behaviour, such as granite.
(non-geological) literature regarding the two processes of weathering Actual degradation changes however, tend to be very much rock-
and alteration. This is because often the two terms are so frequently type dependent. Some rocks are resistant to intact material fabric
used interchangeably that to non-geologists there is no clarity as to the change. These typically are the rock types that the aggregate industry
difference. chooses because of their high durability, examples include diabase,
Alteration is typically brought about by deep geological processes – diorite and most quartzites. Other rocks, such as mudstones, tend to
hydrothermal alteration, metamorphic alteration etc. Ore zone rocks in decompose completely with penetrative weathering or alteration. Even

285
V. Marinos, T.G. Carter Engineering Geology 239 (2018) 282–297

Fig. 3. Indicative example of how tectonism (from low to severe) affects GSI. Shifts from Low to Severe on the left size of the chart corresponds to a more brittle
parent material and on the right size of the chart to a more ductile original parent rock material (e.g. mudstone, shale, or weak siltstone).

quite competent rocks, can degrade and decompose. Granites, for in- Degradation and disintegration can also occur as a result of laterization
stance often weather appreciably, with the feldspars turning to clay - which is typically observed in deep, tropically weathered areas. Rock
minerals. Some particular rock materials absorb water and because of material strength, particularly at the saprolite boundary can be sig-
this volume change, as a consequence of major interstitial forces within nificantly reduced. In some rocks, the intact structure may become so
the rock matrix, completely disintegrate the parent rock mass. Some pervasively degraded within the rock material itself that it is no longer
minerals disintegrate rock masses by swell processes, others by drying recognizable in terms of initial fabric.
processes. All of these weathering and alteration processes affect also parent
All of the common weathering processes cause rock mass changes material properties (notably mi, σci and Ei). It is therefore recommended
that can affect GSI estimates. Decomposition can however differ be- to always use caution if attempting to take published mi and strength
tween even the same types of rock, dependent on cementation. values for dry unweathered intact material based on tables, such as
Commonly quartz cements are stronger and more durable than calcite/ Table 2 or from Rocdata (Rocscience Inc.) listings as these could be
carbonate cements and even better than iron oxide/haematitic cements. seriously in error for actual rock mass conditions.

286
V. Marinos, T.G. Carter Engineering Geology 239 (2018) 282–297

Fig. 4. GSI and intact strength change for weathering grade W-I to grade W-VI.
Grades according to ISRM (1981) with additions after Stacey and Page (1986).

Similar to weathering, the processes resulting in rock material and from Blocky to Very Blocky or perhaps even to Disintegrated) according
rock mass alteration also affects both the intact rock properties of the to the alteration degree, as illustrated in Fig. 7. Where tectonized, this
material and joint surface conditions, both of which, again influence degradation process can also result in the formation of schistose fabrics
GSI (Fig. 7). In a fresh state, unaltered rock masses can be quite mas- within the altered rock mass. Joint conditions in such cases are almost
sive, irrespective of strength, generally with sparsely spaced dis- always Poor to Very Poor towards the lower end of the GSI scale, while
continuities. By contrast, even slightly to moderately altered rock intact properties σci and mi may also be considerably reduced.
masses can exhibit significantly degraded discontinuities, often with
recrystallized or slickensided joint surfaces evident (e.g. through ser- 4. GSI for specific rock suites
pentinization, provided that serpentinization has not fully affected the
intact part of the rock material). From the preceding discussion, it will have been realised that while
For rock masses subjected to weakening styles of alteration, GSI fabric appearance (from intact to disintegrated) can be similar for a
values can be reduced considerably with severity of alteration. In such wide range of rock masses (i.e. GSI values can remain remarkably
situations, the structure becomes progressively disturbed (changing consistent per the chart descriptions), major changes in strength and

287
V. Marinos, T.G. Carter Engineering Geology 239 (2018) 282–297

Fig. 5. Indicative example of how weathering degree (W-I to W-V) affects GSI.

deformability will occur between similarly broken rock masses because differentiation affects the geotechnical properties of the various for-
of differences in parent rock material. A wide range of variability in mations.
terms of σci and mi can thus be expected, as is evident by inspection of While Fig. 8 was basically built from data acquired during the
each of the range of small GSI charts included within the overall chart construction of numerous tunnels along the Egnatia Highway across
comprising Fig. 8. Northern Greece, it is not restricted to application solely within such
This overall chart, when read in combination with Table 2, provides locations. Although conceptually other rock type divisions (i.e., addi-
a compendium for describing most common rocks and rock masses. tional small charts) can be readily added anywhere within the overall
Most of the common GSI ranges for typical gneisses, granites, ophio- chart diagram, it is considered that most of the most common ranges of
lites, limestones, schists, siltstones/mudstones/shales and molassic and rock mass quality and rock type variability are already included within
flysch formations have been illustrated, highlighting how geological Fig. 8. Accordingly, at least at a preliminary level it can be used directly

288
V. Marinos, T.G. Carter Engineering Geology 239 (2018) 282–297

Fig. 6. Modified GSI classification chart for description of variously weathered gneisses or petrographically similar rock masses, e.g. Granite, (after Marinos, 2007).

for characterizing any particular rock masses that might be encountered additional individual rock-type specific charts can easily be developed
worldwide. For more details on each of the specific rock types char- to extend the overall conceptual GSI chart not just diagonally across the
acterized within the overall diagram of Fig. 8, the interested reader is chart but also downwards and upwards on the rock mass competence
referred to the original publications presenting the individual charts scale.
(i.e. Marinos and Hoek, 2000; Hoek et al., 2005; Marinos et al., 2005a; However, for any rock engineering design, the first issue is always
Marinos, 2017). geological characterization. This basically defines where you are within
It should also be appreciated that within this overall diagram dif- Fig. 8 and this then allows one to establish the potential range of
ferences in assigned GSI commonly occur due to different tectonism, characteristics for the rock mass of concern, based on the competence of
weathering and alteration effects. As previously discussed, these effects the parent rock material. For example, if one were working, within a
on rock material integrity and rock mass fabric may differ significantly hard-rock mine in Canada or South Africa at significant depth, one
dependent on (brittle or plastic) parent rock behaviour, which in turn would probably be looking towards the top left corner of Fig. 8, while
also affects GSI values. Care must thus be taken in picking positions for rock with the characteristics of the flysch conditions of central and
within the overall chart to make sure due consideration is given to these southern Europe then one might be looking at a point plotting in the
overarching process changes. Fortunately, as almost all such changes lower right third of the diagram. Himalayan, Andean or Alpine moun-
are visually observable, use of the standard observational GSI chart still tain belt conditions would cover the complete suite, top left to bottom
readily allows these distinguishing differences to be identified and right of Fig. 8.
quantified (specifically by reference to the small charts incorporated The flexibility that the concept sketched within Fig. 8 affords, is that
within Fig. 8). it allows new material-specific or new site-specific individual GSI charts
Together this suite of small charts within the overall diagram en- to be developed for any particularly difficult rock masses or for those
compass a very wide range of rock mass competence, covering most not already within the scope of published charts. The size and spread of
material and fabric character changes observable in typical rock masses any such individual new chart in the suite of charts can thus be tailored
commonly found around the globe. In the top left corner of the overall on a site-specific basis to capture the variation expected in local rock
chart extremely competent hard rock masses with non-degradable mass characteristics, depending on parent rock type and mineralogy,
fabrics are found, typical of the deep mines in high-strength/high stress macrofabric and overall rock mass competence.
conditions where spalling and bursts characterize behaviour. The other In the following paragraphs, each of the different groups listed in
extreme occurs in the lower right corner of the matrix diagram in Fig. 8, Table 2, and within Fig. 8 are explored. GSI characterization issues for
where rock mass competence is low and where squeezing and often each are then discussed in order to illustrate how key differences in
significant closure problems are the characteristic hallmarks (e.g. origin exert changes on the engineering geological characteristics of
flysch, shales, weak faults etc.). Within this framework, specific each geomaterial.

289
V. Marinos, T.G. Carter Engineering Geology 239 (2018) 282–297

4.1. GSI and igneous rocks

31–33
28–30

25–27

17–20

13–16
10–12

7–9
4–6
mi

4.1.1. Intrusive rocks

Two basic column divisions have been drawn on Table 2 between

Fine, (clastics) (Siltstone/Siltstone members of flysch or

Ultrafine (Claystone, Mudstone/sheared Siltstone, Shale

intrusive and extrusive igneous rocks, with the intrusive group further

Medium (Quartz cemented Sandstone) Sandstone

subdivided between felsic and mafic composition as mineralogy di-
rectly affects behaviour and engineering characteristics.

Medium carbonates (Limestone) Sandstone

members of flysch or molasse/greywacke

Fine, Calc-rock (Chalk/marl, & siltstone)

Coarse (Conglomerate - not clayey) [Link]. Felsic intrusives. Granites and diorites are typical light-coloured
igneous intrusives. When they are fresh, granitic rock masses can be
projected in the three first rows and two columns of the standard GSI
chart. According to their degree of weathering, they may shift to the
right and down to the lower rows. If weathering is slight to moderate
(weathering grades II, and to some extent III), GSI values just decrease
Typical values for σci and mi for range of igneous, metamorphic and sedimentary rocks (to be read in conjunction with Fig. 8 regarding parent rock type characteristics).

to the right, into the poor to very poor columns due to the increasing
within flysch)
molasse/tuff)
Sedimentary

presence and width of completely degraded granite (CDG in Hong-Kong

nomenclature) within the joints. If, however, weathering is more
intense (weathering grades III, IV, and V), the rock mass has started
to degrade en masse and the index also shifts downward to a poorer
overall structure. Intact rock properties also change in relation to the
weathering degree. If they are completely weathered, granitic rock
Mafic (Basalt) Intermediate (Andesite)

masses should not really be described in terms of GSI, rather they

should be treated and described as soils.
Given similarity in mineralogy, certain facets of the GSI classifica-
tion chart (Fig. 6) for gneiss presented in Section 3.3., gives parallels for
description and categorization of granites and similar intrusive rocks.
Extrusive (Volcanic)

Felsic (Rhyolite)

[Link]. Mafic intrusives. The darker, more basic suite of plutonic

igneous rocks, such as peridotites and gabbros, contain significantly
less siliceous content than the felsic suite, and as a consequence, once
exposed, are somewhat more susceptible to deterioration by weathering
than their similar grain sized siliceous (felsic) counterparts. However,
Fine (Serpentinite) (in Ophiolites)

again, like the granites, where the coarser grained and feldspathic types
Coarse (Gabbro, Peridotite) (in

Medium (Dolerite/Diabase) (in

degrade most, the coarser grained, more olivine rich, basic intrusive
rocks tend also to degrade the most.
Typically, the intermediate and the basic (mafic suite) igneous
rocks, when they are fresh, mostly are classified in the three first rows
and three first columns on a standard GSI chart. However, when they
Ophiolites)

Ophiolites)

are heavily fractured and weathered, they tend to disintegrate on

structure first and thus move to the right section of the chart, and then
Mafic

downwards to the bottom right corner when severely degraded.

[Link]. Ophiolitic complexes. Many of the suite of basic plutonic rocks

Medium (Granodiorite,

and swathes of altered and metamorphosed seafloor originated

sediments are commonly observed in ophiolite complexes. In such
Coarse (Granite)

situations, peridotites and other basic plutonic rocks are often found
interlayered with each other due to overthrusting, resulting in a wide
Intrusive

variability of characteristics. In some cases, rocks closely associated

Diorite)
Igneous

Felsic

with overthrusts can vary remarkably in geotechnical qualities, right

from excellent, through to very poor when serpentinization is extensive
Strongly Schistose (Schist, Phyllite)
Medium, amorphous (Amphibolite,

Banded, Gneissose (Biotitic Gneiss)

and/or shearing present. Mostly peridotites become more completely

serpentinized with weathering and alteration. As such, examples of
Granular Texture (Granulites,

serpentinites, schisto-serpentinites, and sheared serpentinites can be

Fine, amorphous (Homfels,

Foliated (Schists, Phyllite)

seen amid pillow lavas and chaotic masses in ophiolitic melanges. When
very high degrees of serpentinization occur, together with intense
shearing, the resulting rock mass may be difficult to identify for any
Quartz Gneiss)
Metamorphic

initial texture or fabric.

Quartzite)

Mylonites

In ophiolitic melanges, particularly within tectonic thrusts, specific

Gneiss)

individual rock units are often mixed in complete disorder with other
rocks of various origins (flysch, chert, shales etc). As a consequence
such rock masses can be so disturbed and disintegrated or sheared-la-
Typical σci (MPa)

minated that only low to very low GSI values can be assigned. The
interested reader is referred to the work by Marinos et al. (2005a) for
125–250
100–300

85–350

75–350

50–200
30–100

examples of representative GSI charts, such as that on the lower right

Table 2

20–60
10–50

hand side of Fig. 8, which were specifically created for describing and
characterizing these difficult rocks.

290
V. Marinos, T.G. Carter Engineering Geology 239 (2018) 282–297

Fig. 7. Indicative example of how typical “weakening” alteration (e.g. sericitization processes) (none to severe) affects GSI.

4.1.2. Extrusive and pyroclastic rocks thus GSI values and intact strengths can drop significantly from the
Like the intrusive rocks, the extrusive and pyroclastic volcanic as- intact undisturbed state in hours to days, to almost soil-like composi-
semblages (i.e. lavas and ejectamenta) can be of widely ranging mi- tion.
neralogy and character. Competence of original deposition can also
vary remarkably from extremely competent, hard, brittle rock types
4.2. GSI and metamorphic rocks
(dolerites, rhyolites etc.) of aggregate quality, through to weak, friable
and often swell-sensitive materials (tuffs, pumices etc.). Composition
Metamorphic rocks grade from lightly altered (meta-sandstones,
ranges from siliceous through to basic with cementation also varying
meta-conglomerates etc), which exhibit characteristics similar to the
dramatically. In consequence, some of these rocks can be the most
original rock type through to completely re-melted and re-crystallized
challenging engineering materials anywhere. Characterization for GSI,
rocks and/or sheared and dislocated foliates. Metamorphic grade and
and hence for Hoek-Brown mi requires careful calibration, as intact-
mineralogical character changes influence mi, while foliation and tex-
looking volcaniclastic rocks can have quite significant adverse de-
ture, if intrinsically weak, (such that the rock breaks and deteriorates
gradation characteristics. Many of such rocks are slake-sensitive and
along such structure, e.g. phyllites and schists) significantly influences

291
V. Marinos, T.G. Carter Engineering Geology 239 (2018) 282–297

Fig. 8. Most common GSI ranges for typical gneisses, granites, ophiolites, limestones, schists, siltstones/mudstones/shales, molassic and flysch formations in con-
junction with a range of mi and σci. (Refer to text and reference list for original papers for more details on charts).

292
V. Marinos, T.G. Carter Engineering Geology 239 (2018) 282–297

GSI values. [Link]. Schists. Schistose rock masses, including the more foliated
Two basic metamorphic rock units can thus be recognized from the gneisses, rarely plot in the top left corner of the GSI chart. They
view-point of differences in character from a GSI perspective: (i) com- usually project in the Very Blocky-Blocky/Disturbed rows. Strong
petent crystalline rocks (including most gneisses, quartzites and horn- schists (e.g. quartz-mica schists) would be expected in the left
fels etc.) and (ii) schistose, anisotropically textured rocks (including columns (Very rough to fair), while weak schists would report
schists, slates and phyllites). typically in the right columns. Extreme foliation weakness occurs in
phyllites. By contrast, slates can be remarkably competent, even though
4.2.1. Competent crystalline rock masses possessing a characteristic slaty cleavage. When weak schists are
[Link]. Gneisses. Most gneissic rock masses are generally competent sheared (also sheared phyllites) GSI's can shift to the last row and last
with, high strength. Fresh rock masses are typically sound, often two columns of the GSI chart. Anisotropy however has to be seriously
massive with minor to moderate fracture fabrics. Intact rock strength taken into account in any analysis where these foliates are involved, as
is generally very high, and structure is typically tight. Intrinsic foliation their properties can be significantly different parallel and perpendicular
can often be seen, with considerable persistence. Sometimes gneissic to foliation.
banding is only mineralogic and does not constitute a specific fabric
weakness. Banding is most often intrinsic within the rock mass and 4.3. GSI and sedimentary rocks
tightly “stitched” together, so that no clear detachable blocks are
recognizable. The decreasing quality of a slightly weathered gneissic Two basic divisions of sedimentary rock can be recognized: clastic
rock mass is however discernible from its degree of incipient fracturing, and non-clastic with the latter divided into those directly or indirectly
as weathering is typically initially confined to the discontinuities. In chemically precipitated in or associated with water (limestones, dolo-
many gneisses, as weathering progresses, foliation bands, although mites etc.) and those created by evaporation processes – gypsiferous
initially tight, can become more evident, eventually constituting basic mudstones being one example. Volcanic deposits laid down in water are
discontinuity surfaces. When more pervasively weathered such banding also sedimentary, but because of their origin such volcaniclastic rocks
sometimes opens up to create a flaky fabric, often exploiting foliation can have quite specific deleterious characteristics that markedly dif-
planes, which then separate, ultimately creating distinctly slabby rock ferentiate them from their equivalent grain size, non-volcanic coun-
blocks. terparts. Accordingly, GSI variability can be significant.
GSI values for gneissic rock masses can thus vary dramatically, with
very large ranges in quality evident from fair to very poor, controlled
4.3.1. Clastic rocks of Arenaceous and argillaceous composition
largely by weathering degree. It should be appreciated that as gneissic
As the origin of all water-borne sedimentary rocks (conglomerates,
rocks are commonly the product of intense tectonism and disturbance,
sandstones, siltstones, claystones and mudstones) varies due to the
weathering and alteration may vary markedly with depth. In con-
water velocity controlling their original deposition, different original
sequence, complications in assessing GSI values for design of under-
fabrics characterize each of these rock types over and above their basic
ground works can be expected to range from simple to extremely pro-
differences in grain size. Similarly, air-borne derived rock units - æolian
blematic, as an expected increase in quality with depth below surface
sandstones, siltstones and loess deposits, which are widespread across
may not always be apparent.
the globe also vary appreciably dependent on their parent composition
and original depositional environment. GSI's characteristically will be
[Link]. Quartzites. Quartzite and meta-sandstone rock masses mainly similar for similar rock types, irrespective of origin, but because mi-
project in the first three rows and the left two columns of a GSI chart, neralogical differences and distinctions in depositional fabrics will
due to their strong and brittle nature and their mineralogical govern engineering behaviour with weathering and alteration, extreme
composition. Typically, these highly siliceous rocks are quite resistant care needs to be taken in characterizing GSI to reflect the different
to weathering and frequently exhibit only minor degradation on the origin of the parent materials composing the rock mass.
typically smooth uncoated discontinuities that characteristically dissect
them. The exception occurs however where, frequently, quartzites exist
[Link]. Conglomerates. Conglomerates present variable quality when
alternating with phyllites, reflecting a metamorphosed original
plotted on a GSI chart depending on the cementation material and the
siltstone-sandstone sequence. In such conditions, the major inter-bed
fracturing degree. Clear distinction must be made between
bedding planes can be quite degraded, such that they are often coated
conglomerates that are of water-laid origin, and breccias - which may
with films of phyllitic material. In such situations, care must be taken in
be basically consolidated debris derived from many varied origins
characterizing such a rock mass with a ubiquitous GSI, as depending on
ranging from collapse deposits, tectonic processes to volcanic vent
the scale of the engineering problem such discontinuities may need
infills. The rock mass quality of any conglomerate depends not just on
specific definition.
the matrix material and the fracture degree after any tectonic
disturbance of the rock mass, but also on the angularity and interlock
[Link]. Marbles and skarns. Metamorphosed limestone rock masses are of the original debris. When the matrix is composed of calcite or silicate
typically more competent than, but similar in composition and the rock mass may exhibit high intact rock strength and dependent on
behaviour to strong limestones. Typically, they are often however structural dislocation would be expected to plot within the upper rows
more brittle, typically with much higher moduli values, but some of the GSI chart (from compact to blocky-very blocky), depending on
may have lower mi than expected, due to their coarser calcite crystals, the number of discontinuities present throughout the mass. At surface,
when compared with their unmetamorphosed parent carbonate rocks, many conglomeratic rock masses with competent matrix cementation
(which are discussed subsquently). tend to be relatively resistant to weathering, although some may show
loosening around original clasts within the rock mass due to differential
4.2.2. Highly anisotropic metamorphic rocks weathering and distress, while others some show dissolution
Basically, four types of schistose metamorphic rocks are of en- phenomena along joints if calcite is present. On the other hand, when
gineering significance: schistose gneisses, schists, phyllites and slates. the matrix is of low competence (e.g., silty, clayey or marly), often
Gneisses can vary from competent, almost granite-like composition disaggregation around original clasts dominates disintegration.
with minimal mineralogical foliation weakness as described in Section Commonly these types of conglomerates show little in the way of
4.2.1, right through to almost schistose. Here only the more anisotropic continuous joints with the result that the rock mass often does not
gneisses are considered along with the other foliates. appear broken into systematic blocks.

293
V. Marinos, T.G. Carter Engineering Geology 239 (2018) 282–297

Fig. 9. Revised GSI classification chart for heterogeneous rock masses such as flysch (Marinos, 2017).

[Link]. Sandstones. Sandstones are amongst the commonest rocks often becoming quite flaky due to fissility opening up parallel to bed-
worldwide, and in consequence they may have widely different ding when these rocks have been heavily weathered close to surface.
characteristics depending on parent origin. Sandstones, in general, GSI values for these types of sandstones are therefore quite low, plotting
are strong and usually exhibit quite brittle behaviour. The finer grained usually along the bottom two rows of the chart.
sandstones with calcitic or silicate cementation material are
characteristically the strongest of the sandstone suite. Depending on [Link]. Marls. Marls are typically akin to fine grained siltstone, almost
tectonic disturbance sandstones can range from compact to very blocky mudstone, but often with a percentage of calcite. Calcitic marls tend to
(with high GSI values) but many be disintegrated (with low GSI values) be the most competent and demonstrate relatively good characteristics
along, for instance, fault zones. since their intact rock strength is moderate to high and GSI values
Thickly bedded, unweathered sandstones can be massive in ap- remain high with blocky to very blocky rock mass structure, depending
pearance with brittle spalling behaviour evident in underground works on the degree of fracturing, including bedding. A characteristic example
and with such major block sizes that they plot in the top left corner of is the behaviour of blocky calcitic marls along the Isthmus of Corinth in
any GSI chart. By contrast, thinly bedded sandstones, when folded, central Greece, where 6 km long and steep (70o) cuts of 80 m height
characteristically show ductile deformation structures, and thus would have remained stable for well over than 100 years. Clayey marls, on the
project into the “seamy-disturbed” row in the GSI chart. Surface con- other hand, have low intact rock properties and show a more ductile
ditions for discontinuities present in these rock masses are generally behaviour with no persistent blocky structure. Clayey marls are also
rough to smooth, but can also be poor particularly where severe dis- generally dissected by even weaker discontinuities and thus exhibit low
location has occurred of the sandstone matrix when such rocks have GSI values. Low values are particularly apparent when such rock masses
been tectonically disturbed by thrusting or shearing mechanisms. are encountered close to surface, due to weathering and distress on
Sandstones can also be silty or marly and these varieties commonly fissility or foliation or due to shearing. When such rock masses are not
exhibit low strengths, typically of the order of about 10 MPa compared highly tectonized they may be projected into the upper few rows of the
to > 50 MPa in their typical granular form. Silty sandstones are also GSI chart but always in the right columns, due to their soft nature.
more susceptible to weathering, with the result that the rock mass tends
to loosen more quickly than more sandy varieties, with disintegration
[Link]. Shales. Shales show a fissility not present in typical mudstones.

294
V. Marinos, T.G. Carter Engineering Geology 239 (2018) 282–297

As this fissility (which most often derives from their original 40's, with cohesion still generally absent, unless inter-block cementing
consolidation and induration) controls the behaviour of these types of has developed. In such cases, GSI values range from 30 to 45. In the
rocks it is thus of critical engineering significance that it be correctly case of heavily broken limestone, with significant pelitic presence along
assessed. Typically GSI values for shales fall in the last few rows of the the joints, where the pieces are not in contact and have very poor in-
chart due to their thin stratification (akin to fine schistosity). When the terlocking, frictional properties of the rock mass are significantly re-
rock mass is sheared the structure can be described as laminated and duced. These rock mass types are characterized as Disintegrated on the
thus presents very low GSI values when plotted in the standard chart. In vertical axis of the GSI chart with Very Poor surface condition on the
tectonically undisturbed areas, at depth, their structure rating can be horizontal axis of the chart with GSI values between 20 and 25.
better than when distorted, and thus these rock masses would project
into the Blocky-Disturbed-Seamy row of the GSI chart. Typically, 4.3.3. GSI and heterogeneous rock masses
bedding planes cannot be separately distinguished within the mass. In Establishing a representative range of GSI values for complex, het-
consequence, shales, close to the surface, often present a completely erogeneous rock units comprising a strong member alternating with a
disintegrated-laminated structure, and thus would plot in the lowest weaker one, such as flysch (tectonically disturbed series) or molassic
row of the standard observational chart. Even where unstructured, formations (tectonically undisturbed sedimentary series), can be chal-
undisturbed and non-folded, anisotropy has to be considered for such lenging.
bedded rock masses.
Where claystone and/or siltstone intercalations occur amid lime- [Link]. Tectonically disturbed rock masses with alternating lithology
stones or sandstones, these more plastic members may become sheared (flysch formations). Flysch formations generally exhibit diverse
when folded, resulting in a highly disturbed rock mass where the more heterogeneity, tectonically disturbed structures and presence of
competent limestone or sandstone beds become broken, whereas the significant percentages of low strength geomaterial. In consequence,
shales become “stretched”, due to differences in their deformation be- they are often quite difficult to accurately characterize. The recently
haviour. Often parallelism between bedding planes in the more rigid updated GSI chart prepared by Marinos (2017), as presented in Fig. 9,
sandstone or limestone units, limits deformation banding within the which classifies flysch formations into 11 distinct rock mass types (I to
more shaly sequences (with boudinage structures sometimes being XI), according to its siltstone-sandstone contribution and the tectonic
apparent). In such cases, the competent members should not contribute disturbance, however greatly helps characterization.
to overall, “weighted” intact rock properties and the overall fabric GSI This new flysch chart updates the 2001 chart, proposed by Marinos
should not be rated low. and Hoek (2001) that was prepared early on during the design phase
Siliceous inclusions (cherts) are common in carbonate deformation and before the construction of more than a dozen tunnels in flysch
zones. Most often they are found intercalated with limestones; such that formation rocks along the Egnatia Highway. This new chart not only
rock mass quality can typically be projected onto the basic GSI chart updates the earlier chart, but improves its applicability on the basis of
along the Very Blocky-Blocky/Disturbed structure row and between the calibrations of behaviour with the monitoring experience gained from
second and third column. If cherts are found alternating with clayey the construction of these tunnels. Its applicability also extends to non-
shales, such rocks can be classified with the GSI chart for heterogeneous disturbed flysch rocks, provided bedding anisotropy is carefully and
rock masses, such as flysch (ref. Fig. 9, which presents the chart in- specifically addressed (Marinos, 2017). In this regard, it should be
cluded in the lower right corner of Fig. 8, but at a full-size, readable noted from this chart that high percentages of siltstone beds (when
scale). undisturbed) do not necessarily decrease the computed GSI value, in
contrast to cases where significantly, tectonically disturbed. In con-
4.3.2. Non-clastic rocks – typically carbonates sequence, in this updated, flysch-specific chart, GSI values for un-
Limestones, in general, are neither weak, nor complex formations in disturbed flysch rock mases, particularly those dominated by siltstones,
the context discussed above with respect to foliates. Limestone terranes are respectively increased from 10 to 35 units, consistent with moving
can however be very complex as a result of solution damage effects and from a Blocky to Undisturbed rating on the y-axis scale of the standard
karst. Accordingly, GSI values can be very variable, dependent on GSI chart.
proximity to solution features. Typically, though, when limestones are
massive or thickly bedded, and are distant from areas of tectonic dis- [Link]. Tectonically undisturbed heterogeneous sedimentary rock masses
turbance, or significant karst, the GSI brokenness index would be ex- (Molassic formations). When categorizing tectonically undisturbed
pected to range from Intact to Blocky-Very Blocky. Thinly to medium heterogeneous sedimentary rock masses (which in European
bedded limestones, when undisturbed (i.e., unfolded, and only perhaps terminology are defined as molassic), it is of particular concern to
slightly fractured), by contrast, can be characterized as “Blocky” with ensure that appropriate consideration is given to assessing variable
“Good” to “Fair” surface conditions and thus would be rated GSI's in the heterogeneity, the presence of low strength geo-materials within the
55 to 70 range. Calibration checks, as confirmed from tunnel excavation units and the improvement of structure that typically occurs with
behaviour, where very light support measures were applied confirm the depth. Typical tectonically undisturbed sedimentary sequences,
validity of this typical range (with GSI design values for such rock commonly comprise sandstone and siltstone or marly strata, with
masses generally > 55). alternations of different quota and layer thicknesses. Major variations
By contrast, even competent limestones, when severely distorted by in bed thicknesses can frequently be observed ranging from millimetre
thrusting and tectonic displacement, may be so broken up as to com- thick alternating laminae through to units of many tens of centimetres
prise solely a mixture of angular and rounded pieces of limestone, to even several metres; but rarely appearing sufficiently thick as to form
which would register on a GSI chart as only of the poorest rock quality. physically massive strata. Cross-bedded sedimentary units are also
In worst case conditions, such rock masses can be found to have dis- common.
integrated to such an extent due to brittle fracturing and break-up of the These types of molassic rocks differ significantly between surface
original limestone blocks that no remnants of the initial, original and depth. Bedding is the essential joint set in a molassic rock mass, but
structure remain. Rock mass quality for a heavily broken to brecciated may only be expressed as a significant weakness plane near the surface,
rock mass created by this type of major deformation may end up in the due to fissility. At depth, bedding is mainly concealed and may be
low 30's for GSI values. Characteristically, though, even though RQD's virtually healed. Weathering though, alters rock mass strength sig-
for such rock masses would be approaching zero, they still retain some nificantly, starting by degrading bedding plane strength, then the
frictional strength. parent block matrix. Siltstone (or marly) units are particularly vulner-
For less broken rock masses, GSI values may increase to the mid able to weathering, and fissility may thus develop parallel to bedding in

295
V. Marinos, T.G. Carter Engineering Geology 239 (2018) 282–297

these types of rocks, when exposed at or near ground surface. an example of a highly anisotropic rock mass, where applying a GSI
Joints in undeformed molassic rocks are generally free from all ef- value should only be done with great care if the mode of potential
fects of shear movement (slickensides) due to limited deformation of failure won't be governed by the specific shear strength of the incipient
such sedimentary sequences after deposition. Similarly, siltstone and discontinuities within the slate.
mudstone beds in undeformed sedimentary sequences, can frequently For essentially blocky rock masses, defining the relative scale of an
be found to form an almost continuous medium of mostly consistent engineering problem appropriate to the scale of the rock mass fabric is
composition. In consequence, high GSI values would be applicable for key to correct application of GSI into the Hoek-Brown criterion. While
describing and characterizing such rock masses. A value of 50–60 or there has been some criticism that GSI should only be reliably applic-
more would typically be ascribed, consistent with values validated able to heavily jointed rock masses, which at the scale of most en-
during tunnel construction. If no discontinuities at all are evident, GSI is gineering problems can be treated as homogeneous and isotropic, this
even higher (potentially > 75); allowing the rock mass in such situa- limitation can be readily overcome with a little judgment, as the issue is
tions to be treated as more or less intact. really one of relative block size. Basically, as long as incipient block size
Even though GSI's for such rock masses would be expected to be within the rock mass fabric is less than about 1/10th of the controlling
high, the fact that many of these types of sedimentary rocks are of low dimension of the engineering construction of concern, GSI and the
UCS reduces their overall rock mass strength, resulting in them being Hoek-Brown equations can be applied. Specific weakness zones (such as
only a fair to weak rock mass from a tunnelling performance perspec- faults) can and should then be considered with separately described
tive. parameters. Such an approach to consider specific anisotropy in de-
At the surface, the heterogeneity of such rock masses is much more formation behaviour using GSI is described in Fortsakis et al. (2012),
discernible. As definition of GSI for such near-surface conditions can be wherein a single well-defined shear zone constituting a major geolo-
quite confusing, especially for weathered conditions where description gical structure, has been modelled as an equivalent Hoek–Brown ma-
may be difficult when the rock appear quite heterogeneous. Rather, terial within the layered rock mass, thereby superimposing the specific
such rocks should be described using the specifically prepared version discontinuity as a significantly weaker element. In such an analysis
of the observational chart for fissile molassic rocks presented by Hoek framework, a different GSI value needs to be assigned to the rock mass,
et al. (2005). ignoring the unique major discontinuity. While this type of approach
For construction at depth in fresh, essentially unweathered, mo- can be effective, in other situations it may be more appropriate to
lassic rocks, where, typically evidence of sedimentary sequence varia- characterize the specific properties of the particular discontinuity uti-
bility is only apparent as colouration layering, and the rocks generally lizing laboratory shear testing data or using the Barton and Bandis
appear massive, it is recommended that the basic observation GSI chart approach (1990).
be used. As discussed in Hoek et al. (2005) and Marinos et al. (2013), On a global scale, GSI description and application of the Hoek-
characterization of these types of undeformed rock are better handled Brown equivalent continuity approach can however be further ex-
using the standard chart, resulting in much more appropriate (and tended to suggest that, when dealing at very large scale, rock mass
lighter) support measures than would be envisaged, based on applica- strength will reach a constant value when the sizes of individual rock
tion of the flysch-like GSI classification approach. blocks are sufficiently small in relation to the overall size of the en-
gineering structure being considered. By contrast, where block sizes of a
rock-mass are close to the same order of size of the engineering struc-
5. GSI limitations ture being analyzed, GSI and the Hoek-Brown criterion should not be
used. Rather, the overall stability of the engineering structure itself
Care should always be taken in appropriately applying the defined should be analyzed by considering the behaviour of the individual
GSI value as basis for ascribing Hoek-Brown material parameters, as the blocks and/or wedges defined by the intersection of the geological
basic GSI correlation equations may not be directly appropriate for all structural features considered controlling stability.
rock mass situations (Marinos et al., 2005b). The standard expressions An example of this transition is frequently seen in large open pit
relating GSI and the Hoek-Brown criteria are based upon the assump- mines (such as shown in Fig. 10), in which individual structures or
tion that the rock mass contains a sufficiently large number of inter- intersecting structural features control bench scale stability, while the
secting discontinuities that it can be considered to behave as an iso- overall slope can oftentimes be considered as an equivalent homo-
tropic mass. As a consequence, the basic GSI equations should therefore geneous rock mass. The Hoek-Brown criterion can thus be applied at pit
not be applied for design of engineering projects in rock masses with scale, while at bench scale the blocks must be considered kinematically
either (i) a dominant (i.e., strongly anisotropic) structural fabric or (ii) as discrete units. These basic differences in scale, with respect to the
with a sparsely jointed fabric, where large scale kinematics or spalling engineering structures of importance at each scale are readily apparent
behaviour may respectively dominate conditions. Undisturbed slate is

Fig. 10. Examples of different scales of fabric - sta-

bility at bench scale is clearly controlled by inter-
secting structural features, while at overall slope
scale the rock mass can be characterized as a Hoek-
Brown material within which major faults and shear
zones must be superimposed.
Photo from 1000 m deep Chuquicamata open pit
mine in Chile, Hoek, pers comm

296
V. Marinos, T.G. Carter Engineering Geology 239 (2018) 282–297

from the photograph and sketch diagrams in Fig. 10. In this case many References
scales of geological structure are involved, and in consequence, for
much of the mechanics controlling slope stability of the overall slopes ANON, 1995. The description and classification of weathered rocks for engineering
of the > 1000 m deep Chuquicamata open pit mine in Chile, the rock purposes. Geological Society Engineering. Group Working Party Report. Quart. J.
Eng. Geol. 28, 207–242.
mass can be considered as a Hoek-Brown material. Barton, N.R., Bandis, S., 1990. Review of predictive capabilities of JRC-JCS model in
engineering practice. In: Barton, N., Stephansson, O. (Eds.), Rock Joints, Proc. Int.
6. Conclusions Symp. On Rock Joints, Loen, Norway. Balkema, Rotterdam, pp. 603–610.
Barton, N.R., Lien, R., Lunde, J., 1974. Engineering classification of rock masses for the
design of tunnel support. Rock Mech. 6 (4), 189–239.
Use of the GSI rock mass classification system (and extension Bieniawski, Z.T., 1976. Rock mass classification in rock engineering. In: Bieniawski, Z.T.
through the associated m, s and a parameter relationships linking GSI (Ed.), Exploration for Rock Engineering. A.A. Balkema, Johannesburg, pp. 97–106.
Cai, M., Kaiser, P.K., Uno, H., Tasaka, Y., Minami, M., 2004. Estimation of rock mass
with the Hoek-Brown failure criterion) provides a proven, effective and strength and deformation modulus of jointed hard rock masses using the GSI system.
reliable approach for defining rock mass characteristics. It has been Int. J. Rock Mech. Min. Sci. 41 (1), 3–19.
widely adopted by engineers and geologists for strength prediction for Carter, T.G., Diederichs, M.S., Carvalho, J.L., 2008. Application of modified Hoek-Brown
transition relationships for assessing strength and post yield behaviour at both ends of
surface and underground excavation design for a wide range of most
the rock competence scale. In: In Proc. The 6th International Symposium on Ground
“normal” blocky rock masses for which the Hoek-Brown criteria re- Support in Mining and Civil Engineering Construction, 30 March – 3 April 2008. Cape
mains fully valid. Analytic or practical back-analyses of rock masses Town, South Africa. 108. Journal of the Southern African Institute of Mining and
where the GSI system has been applied, in conjunction with the basic H- Metallurgy, pp. 37–59 pp 325–338.
Castro, L.A.M., Carvalho, J., Sá, G., 2013. Discussion on how to classify and estimate
B equations, continues to certify its appropriateness when applied strength of weak rock masses. In: Dight, P.M. (Ed.), Slope Stability 2013. Perth,
within the correct range of blockiness. The GSI system itself however Australian Centre for Geomechanics.
remains effective for describing rock masses right out to either end of Diederichs, M.S., 2007. Mechanistic validation and practical application of damage and
spalling prediction criteria for deep Tunnelling. The 2003 Canadian geotechnical
the rock competence scale, allowing characterization also of rock colloquium. Can. Geotech. J. 44 (9), 1082–1116.
masses requiring spalling or squeezing analysis utilizing modified Hoek- Fortsakis, P., Nikas, K., Marinos, V., Marinos, P., 2012. Anisotropic behaviour of stratified
Brown or other failure criteria relationships. rock masses in tunnelling. Eng. Geol. 141–142 (19), 74–83.
Hoek, E., 1994. Strength of rock and rock masses. News J. Int. Soc. Rock Mech. 2 (2),
GSI permits a wide variety of rock mass fabrics to be easily and 4–16.
reliably quantified, significantly enhancing geological logic and redu- Hoek, E., Diederichs, M.S., 2006. Empirical estimation of rock mass modulus. Int. J. Rock
cing geo-engineering uncertainty. The influence of various engineering Mech. Min. Sci. 43, 203–215.
Hoek, E., Martin, C.D., 2014. Fracture initiation and propagation in intact rock. J. Rock
geological characteristics can be better defined than perhaps can be Mech. Geotech. Eng. 6 (4), 287–300.
achieved through generically coded parameter summations or multi- Hoek, E., Marinos, P., Benissi, M., 1998. Applicability of the geological strength index
plications as the sole means for definitively ranking and characterizing (GSI) classification for weak and sheared rock-masses – the case of the Athens schist
formation. Bull. Eng. Geol. Environ. 57 (2), 151–160.
a rock mass. The process of plotting an envelope on a GSI chart not only
Hoek, E., Caranza-Torres, C.T., Corcum, B., 2002. Hoek-Brown failure criterion. In:
allows better understanding to be gained of rock mass variability, but Bawden, H.R.W., Curran, J., Telesnicki, M. (Eds.), Proc. North American Rock
also allows better appreciation of rock mass character and behaviour, Mechanics Society (NARMS-TAC 2002), 2002 edition. Mining Innovation and
necessary for consideration in defining parameters for numerical Technology, Toronto, Canada, pp. 267–273.
Hoek, E., Marinos, P., Marinos, V., 2005. Characterisation and engineering properties of
modelling. tectonically undisturbed but lithologically varied sedimentary rock masses. Int. J.
One of the key advantages of the Geological Strength Index is that Rock Mech. Min. Sci. 42 (2), 277–285.
the geological reasoning it embodies allows characterization to be made Hoek, E., Carter, T.G., Diederichs, M.S., 2013. Quantification of the Geological Strength
Index chart. In: 47th US Rock Me-chanics/Geomechanics Symposium. AR-MA, San
of a very wide range of rock masses and conditions, including both Francisco, pp. 13–672.
weak and complex situations, but always maintaining care to keep ISRM, 1981. Rock characterization, testing and monitoring – ISRM suggested methods. In:
within valid applicability limits. Brown, E.T. (Ed.), International Society of Rock Mechanics. Pergamon, Oxford.
Marinos, V., 2017. A revised, geotechnical classification GSI system for tectonically dis-
While “structure” appearance (from intact to disintegrated) can be turbed heterogeneous rock masses, such as flysch. Bull. Eng. Geol. Environ. https://
similar for a wide range of rock masses, major changes in strength and [Link]/10.1007/s10064-017-1151-z. (online first).
deformability can occur between similarly fractured rock masses, solely Marinos, V., 2007. Geotechnical Classification and Engineering Geological Behaviour of
Weak and Complex Rock Masses in Tunneling. School of Civil Engineering,
because of differences in parent rock material competence. This varia-
Geotechnical Engineering Department, National Technical University of Athens
bility in terms of σci and mi, as schematically delineated in the overall (NTUA), Athens (Doctoral thesis, In greek).
chart diagram included in Fig. 8, is of critical importance for correct Marinos, P., Hoek, E., 2000. GSI: A geologically friendly tool for rock-mass strength es-
timation. In: Proc. Geo. Eng. 2000 at the Int. Conf. on Geotechnical and Geological
definition of rock mass strength and deformability behaviour in an
Engineering, Melbourne. Technomic Publishers, Lancaster, Pennsylvania, pp.
overall, comparative context. As can be observed from the global matrix 1422–1446.
chart of Fig. 8, in relation to each of the already published GSI charts Marinos, P., Hoek, E., 2001. Estimating the geotechnical properties of heterogeneous rock
encompassed within the overall matrix chart, complete coverage of any masses such as flysch. Bull. Eng. Geol. Environ. 60, 82–92.
Marinos, P., Hoek, E., Marinos, V., 2005a. Variability of the engineering properties of
range of GSI can be developed on a site-specific basis for most common rock masses quantified by the geological strength index: the case of ophiolites with
rock types and conditions. Gneisses, granites, ophiolites, limestones, special emphasis on tunnelling. Bull. Eng. Geol. Environ. 65 (2), 129–142.
schists, siltstones/mudstones/shales, as well as molassic and flysch Marinos, V., Marinos, P., Hoek, E., 2005b. The geological strength index – applications
and limitations. Bull. Eng. Geol. Environ. 64 (1), 55–65.
formations are each represented, highlighting how geological differ- Marinos, V., Prountzopoulos, G., Fortsakis, P., Koumoutsakos, D., Papouli, D., 2012.
entiation amongst different rock types affects overall geotechnical Tunnel information and analysis system: A geotechnical database for tunnels.
properties on a comparative basis. However, as it is emphasized in the Geotech. Geol. Eng. [Link]
Marinos, V., Fortsakis, P., Prountzopoulos, G., 2013. Tunnel behaviour and support in
paper, every rock mass has its own features and the presented specific molassic rocks. The experiences from 12 tunnels in Greece. In: Proceedings of the
GSI ranges should be used with caution. ISRM International Symposium EUROCK2013.
Palmström, A., 2005. Measurements of and correlations between block size and rock
quality designation. In: Tunnelling & Underground Space Tech. 20(4). pp. 362–377.
Acknowledgements
Palmström, A., Broch, E., 2006. Use and misuse of rock mass classification systems with
particular reference to the Q-system. Tunnel. Underground Space Tech. 21, 575–593.
Thanks are due to Evert Hoek and Paul Marinos for their insight in Sonmez, H., Ulusay, R., 1999. Modifications to the geo-logical strength index (GSI) and
their applicability to sta-bility of slopes. Int. J. Rock Mech. Min. Sci. 36, 743–760.
initially developing the GSI approach and to Mark Diederichs and
Stacey, T.R., Page, C.H., 1986. Practical Handbook for Underground Rock Mechanics.
others for their critiques over recent years of its applicability over the Trans Tech. Publications, Clausthal-Zellerfeld publ.
full range of the rock competence scale.

297
 Update 1 of 2
Engineering Geology
Volume 242, Issue , 14 August 2018, Page 222

DOI: [Link]
 Engineering Geology 242 (2018) 222

Contents lists available at ScienceDirect

Engineering Geology
journal homepage: [Link]/locate/enggeo

Corrigendum

Corrigendum to “Maintaining geological reality in application of GSI for T

design of engineering structures in Rock” [Engineering Geology 239 (2018)
282–297]
⁎
Marinos V.a, , Carter T.G.b
a
Aristotle University of Thessaloniki, 541 24, Thessaloniki, Greece
b
TGC Geosolutions & Golder Associates, Toronto, Ontario, Canada

The authors regret for the errors regarding some figure captions (for granite, (after Marinos, 2007).
Figs. 4,5,6) that have been scrambled in the published manuscript. Fig. 6. GSI and intact strength change for weathering grade W-I to
Though the figure captions of Figs. 4, 5 and 6 were submitted in their grade W-VI. Grades according to ISRM (1981) with additions after
accurate order, there was some problem in the published version. The Stacey and Page (1986).
suitable captions for every figure are: Fig. 7. Indicative example of how alteration (none to severe) affects
Fig. 1. Basic GSI chart for visual geological characterization of rock GSI.
masses (Marinos and Hoek, 2000). Fig. 8. Most common GSI ranges for typical gneisses, granites,
Fig. 2. Difficult Ground Conditions for undertaking rock mass clas- ophiolites, limestones, schists, siltstones/mudstones/shales, molassic
sification (Photo from Egnatia Highway): Here, tectonically deformed and flysch formations in conjunction with a range of mi and σci. (Refer
intensively folded/faulted siltstone with broken and deformed sand- to text and reference list for original papers for more details on charts).
stone layers forming an almost chaotic structure. Fig. 9. Updated GSI classification chart for heterogeneous rock
Fig. 3. Indicative example of how tectonism (from low to severe) masses such as flysch (Marinos 2017).
affects GSI. Shifts from Low to Severe on the left size of the chart cor- Fig. 10. Examples of different scales of fabric - stability of individual
responds to a more brittle parent material and on the right size of the benches clearly controlled by intersecting structural features while the
chart to a more ductile rock (e.g. mudstones, shales, siltstones). overall slope can be characterized as a Hoek-Brown material on which
Fig. 4. Indicative example of how weathering degree (W-I to W-V) major faults and shear zones must be superimposed. Photo from 1000 m
affects GSI. deep Chuquicamata open pit mine in Chile, Hoek, pers comm.
Fig. 5. Modified GSI classification chart for description of variously The authors would like to apologise for any inconvenience caused.
weathered gneisses or petrographically similar rock masses, e.g.

DOI of original article: [Link]

⁎
Corresponding author.
E-mail address: marinosv@[Link] (V. Marinos).

[Link]

Available online 09 July 2018

0013-7952/ © 2018 Elsevier B.V. All rights reserved.
 Update 2 of 2
Engineering Geology
Volume 248, Issue , 8 January 2019, Page 357–360

DOI: [Link]
 Engineering Geology 248 (2019) 357–360

Contents lists available at ScienceDirect

Engineering Geology
journal homepage: [Link]/locate/enggeo

Corrigendum

Corrigendum to “Maintaining Geological Reality in Application of GSI for T

Design of Engineering Structures in Rock” [Engineering Geology 239 (2018)
282–297]
Marinos V.a, , Carter T.G.b
⁎

a
Aristotle University of Thessaloniki, 541 24 Thessaloniki, Greece
b
TGC Geosolutions & Golder Associates, Toronto, Ontario, Canada

The Authors regret that certain figure diagrams were incorrectly captions in discussions throughout the originally published paper were
placed with respect to their captions during type-setting the previously correct. The revised sequence of diagrams for Figs. 4, 5 and 6 as now
published manuscript and this was not completely corrected in the amended in this version, now properly also agree with all original cross-
subsequently issued Corrigendum. This Amendment corrects these er- referencing.
rors and properly matches the diagrams for Figs. 4, 5 and 6 with their The correct sequence of diagrams along with their associated cap-
respective captions. NOTE: Only the diagrams were incorrectly posi- tions, as should have appeared in the original publication for Figs. 4, 5
tioned. The captions in the original publication were correctly located and 6, follow:
with respect to the original text. Also, all cross-referencing of the Figure The Authors apologise for any inconvenience caused.

DOI of original article: [Link]

⁎
Corresponding author.
E-mail address: marinosv@[Link] (V. Marinos).

[Link]

Available online 20 December 2018

0013-7952/ © 2018 Elsevier B.V. All rights reserved.
V. Marinos, T.G. Carter Engineering Geology 248 (2019) 357–360

Fig. 4. GSI and intact strength change for weathering grade W-I to grade W-VI. Grades according to ISRM (1981) with additions after Stacey and Page (1986).

358
V. Marinos, T.G. Carter Engineering Geology 248 (2019) 357–360

Fig. 5. Indicative example of how weathering degree (W-I to W-V) affects GSI.

359
V. Marinos, T.G. Carter Engineering Geology 248 (2019) 357–360

Fig. 6. Modified GSI classification chart for description of variously weathered gneisses or petrographically similar rock masses, e.g. granite, (after Marinos, 2007).

360