Engineering Geology 202 (2016) 24–35

Contents lists available at ScienceDirect

Engineering Geology

journal homepage: [Link]/locate/enggeo

Comparison of quantified and chart GSI for four rock masses

Robert Bertuzzi a,⁎, Kurt Douglas b, Garry Mostyn a
a
Pells Sullivan Meynink, UNSW, Australia
b
School of Civil & Environmental Engineering, UNSW, Australia

a r t i c l e i n f o a b s t r a c t

Article history: This paper discusses the Geological Strength Index (GSI). It provides data from four different rock masses to ex-
Received 1 June 2015 tend the case history offered by Hoek et al. (2013) in their paper that proposes to quantify GSI with RQD and the
Received in revised form 4 January 2016 joint condition rating of RMR.
Accepted 4 January 2016
The correlation between the GSI assessed from its chart and the quantified GSI was found to be fair for the
Available online 7 January 2016
datasets from the four rock masses; the Hawkesbury Sandstone and Ashfield Shale of Sydney; the Greenland
Keywords:
Group greywacke and argillite of South Island, New Zealand; and the Otago Schist of South Island, New
GSI Zealand. The two methods produce data points that are generally within ± 10 of each other. The exceptions
Case studies are rock masses whose blockiness may not be captured well by RQD.
Quantification It is recommended that the quantified GSI approach be used to supplement and check the visually
Engineering geology assessed chart GSI.
The authors also suggest different GSI values that may be needed to cater for different numerical methods.
Crown Copyright © 2016 Published by Elsevier B.V. All rights reserved.

1. Introduction masses. This inherent difficulty with all classification systems, not
only GSI, means that correlations and ensuing results must be used
A recent paper by Hoek et al. (2013) proposed quantifying the Geo- cautiously. Care therefore is needed in the attempts that make the
logical Strength Index (GSI) with the Rock Quality Designation (RQD) systems more robust and accessible to inexperienced professionals
(Deere, 1968) and with the joint condition rating (JCond89) of the so that the natural variability and the complexity of geology is not
Rock Mass Rating system (RMR) (Bieniawski, 1989). The reason given masked.
by Hoek et al. (2013) for quantifying GSI is their observation that the This paper is divided into two parts. The first presents a discus-
GSI chart is being used by inexperienced professionals ‘who are less com- sion of the GSI system. The second provides data from four different
fortable with these qualitative descriptions’. The issue of inexperience is rock masses to supplement the case history offered by Hoek et al.
perennial; one shared by many leading rock engineering practitioners, (2013).
for example Terzaghi (1946), Bieniawski (1976), Hoek (1994), Fookes
(1997), Stacey (2000), Edelbro et al. (2007) and Pells (2008). The best 2. Discussion OF GSI
outcomes are derived from experienced engineering geologists and
geotechnical engineers working together. However, it is recognised 2.1. Background
that industry relies on young geologists and engineers for routine field
work. Quantifying the GSI may provide a means of reducing inadvertent Hoek (1994) introduced GSI as an alternative to RMR as a means to
errors and inconsistencies by these inexperienced professionals in determine rock mass strengths primarily:
classifying a rock mass.
The GSI uses geological and geometric parameters to arrive at a • to overcome the limitation that RMR does not work for very poor rock
quantitative value to represent the quality of the rock mass. Putting masses
aside the subjectivity of the person using the method, large scatter • to avoid double counting the influence of the groundwater and joint
exists in the data simply due to the natural variability of rock orientations in RMR or the joint water reduction factor (Jw) and the
stress reduction factor (SRF) in Q.

⁎ Corresponding author.
In its original formulation GSI ranged from “about 10, for ex-
E-mail addresses: [Link]@[Link] (R. Bertuzzi), [Link]@[Link] tremely poor rock masses, to 100 for intact rock” (Hoek, 1994). The
(K. Douglas), [Link]@[Link] (G. Mostyn). 1976 formulation of RMR 76 (Bieniawski, 1976) was the basic

[Link]
0013-7952/Crown Copyright © 2016 Published by Elsevier B.V. All rights reserved.
 R. Bertuzzi et al. / Engineering Geology 202 (2016) 24–35 25

reference for GSI though the index could be calculated using the Table 1
1989 version of RMR89 (Bieniawski, 1989) and part of the Q system Parameters used in RMR from Bieniawski (1976) and Bieniawski (1989).

(Barton et al., 1974) as: Parameter Rating

• GSI = RMR89 − 5 for RMR89 N 23 and RMR76 RMR89

• GSI = 9LogeQ′ + 44 for RMR89 b 23. 1 Strength of the intact rock material 0–15 0–15
2 Drill core quality (RQD) 3–20 3–20
3 Spacing of discontinuities 5–30 5–20
The parameters that make up RMR are listed in Table 1. The param- 4 Condition of discontinuities (JCond) 0–35 0–30
eter JCond89 is calculated according to its components as given in 5 Groundwater 0–10 0–15
6 Adjustment for discontinuity −60 – 0 −60–0
Table 2.
orientation
GSI was therefore a trimmed version of RMR in that it used only
the first four parameters from Table 1, assigning the dry score of
10 or 15 to ‘Groundwater’ for RMR 76 and RMR 89 , respectively, and
the ‘very favourable’ score of 0 for ‘Orientation’, to avoid ‘double
counting’ (Hoek, 1994). So the original GSI was a number based on • Updated flysch, molasse, limestone, ophiolite and disturbed/weath-
four parameters; intact rock strength, RQD, spacing of discontinu- ered gneiss Marinos (2010)
ities and JCond89. • Jointed, blocky rock mass for tunnels of about 10 m span and slopes
As with all classification systems, there are limitations with GSI, b20 m high (Hoek et al., 2013).
including the subjectivity of the person using the system. Others
also noted that the original GSI may not adequately consider the
scale of the rock mass problem as its parameters, particularly RQD Each of the individual GSI charts describes the particular rock mass
and JCond, are scale-dependent, e.g. Priest and Hudson (1976), type in greater detail than the generic description originally provided.
Hudson and Harrison (1997), Mostyn and Douglas (2000). Carter and Marinos (2014) prepared a matrix showing the relationships
Following on from the seminal work of Hoek and Brown between the GSI charts and the Hoek–Brown strength criterion
(1980), in which a table was published that linked a geological parameters of σ c and m i for different lithologies. This highlights
description to an RMR value and to the Hoek-Brown strength the importance of developing the geological model to estimate GSI.
criterion parameters, Hoek (1994) published a chart showing The authors suggest different sets of material properties and by
that linkage with GSI (Fig. 1). This paved the way for assessing extension GSI values may be needed to cater for different numerical
GSI directly from a chart rather than quantifying through RMR. methods as well as for different problem scales. This complicates
An early chart was published in Hoek and Brown (1997) and is the system but it is an important consideration when deciding ap-
reproduced here as Fig. 2. propriate GSI values and properties. For example, for the same
Significantly, Marinos and Hoek (2000) largely through work done rock mass:
in Greece on tectonically disturbed and/or weathered rock masses,
extended the range covered by the chart and suggested that GSI should • It may only be possible to include a few specific geological structures
be estimated directly from it. Estimating GSI directly from the chart was in a large numerical model of an open pit. Hence, properties to be used
also the recommendation independently reached by Mostyn and in continuum analyses e.g. FLAC, Phase2, PLAXIS, Abaqus, are likely to be
Douglas (2000). appropriate.
A number of implications arise from estimating the GSI value directly • For a smaller problem, such as tunnel scale or multiple bench scale in
from the chart instead of calculating it through RMR. The first is that an open pit, it may be practicable to include several geological struc-
though intact rock strength is broadly implied by selecting a particular tures in the continuum analysis. Hence, different material properties
chart e.g. sandstone or granite (Marinos and Hoek, 2000), strength is to those used in the first problem are likely.
not explicitly included. In contrast, intact rock strength is the first • If that problem was analysed using discontinuum techniques,
parameter required when estimating RMR. The second is that prob- e.g. UDEC, or ‘jointed’ continuum analyses with numerous geolog-
lem scale needs to be considered by the experienced practitioner ical structures explicitly modelled e.g. jointed Phase 2 , it may be
when selecting the vertical axis of the chart. By so doing, one of appropriate that material properties reflecting the behaviour of
the RMR/GSI limitations is addressed. This follows on from the a smaller representative rock mass are used.
notion expounded in the original work of Hoek and Brown (1980)
as reproduced here in Fig. 3. The third is that the interlocking of
rock blocks was introduced. The authors also note that the GSI chart's axes are not necessarily in-
It is worth noting that as the problem scale increases further from dependent. Often the blockiness of the rock mass (vertical axis) is linked
those shown in Fig. 3, the rock mass may again be dominated by one to the surface quality of a rock mass (horizontal axis). This is commonly
or multiple large geological structures such as regional faults. That is
the ‘intact rock’ is replaced by ‘rock mass’ separated by regional faults
or persistent defect sets. Table 2
There is actually a third issue with estimating GSI directly from the Components of JCond89 after Bieniawski (1989).
chart, and that is selecting which chart to use. Length b1 m 1–3 m 3–10 m 10–20 m N20 m
It is important to note that there are currently nine GSI charts known Rating 6 4 2 1 0
to the authors. Separation None b0.1 mm 0.1–1 mm 1–5 mm N5 mm
Rating 6 5 4 1 0
• General rock mass as extended to cover schist (Hoek et al., 1998), in- Roughness Very rough Rough Slightly Smooth Slickensided
tact rock (Hoek, 1999) and flysch (Marinos and Hoek, 2001). This is Rating 6 6 rough 1 0
6
covered in two charts which are included in the RocLab programme1
Infilling None Hard b5 Hard N5 Soft b5 Soft N5 mm
Rating 6 mm mm mm 0
4 2 2
Weathering Unweathered Slightly Moderately Highly Decomposed
1 Rating 6 5 3 1 0
Version 1.033 build date 8 July 2013 RocScience Inc.
26 R. Bertuzzi et al. / Engineering Geology 202 (2016) 24–35

Fig. 1. Chart linking geological description, GSI and the Hoek–Brown criterion Hoek (1994).

seen as a diagonal trend from top left to bottom right when different 2.2. Quantification
classes of the same rock mass type are overlain on one chart. This
trend is observed in the charts published by Marinos and Hoek (2000) Several authors have suggested methods to quantify the GSI chart ‘to
and Marinos et al. (2005), who also noted that not all combinations facilitate use of the system especially by inexperienced engineers’ (Cai et al.,
are possible and ‘blocked off’ the top right and bottom left cells in the 2004).
GSI chart (Fig. 4). The same trend is observed in Figs. 5 to 8 which Sonmez and Ulusay (1999) suggested linking the vertical axis of the
show the visually assessed chart GSI values for the four rock masses GSI chart to ‘volumetric joint count’ (Jv) and the horizontal axis to three
presented in the second part of this paper. of the five components of the RMR's JCond parameter. Cai et al. (2004)
 R. Bertuzzi et al. / Engineering Geology 202 (2016) 24–35 27

and the horizontal axis ‘joint condition’. A similar approach

was taken by Russo (2009) who linked GSI with Vb and the ‘joint
condition factor’ contained within the RMi classification system
(Palmström, 1995).
Hoek et al. (2013) quantify GSI as equal to 1.5 × JCond89 + RQD/2 and
as shown in Fig. 9, set the chart's vertical axis to RQD/2 and its horizontal
axis to 1.5 × JCond89. It can be seen that the top and bottom rows are
removed compared with the GSI chart of Figs. 4 to 8. Fig. 9 reverts
to the original rock mass structure range in Hoek (1994) and Hoek
and Brown (1997) but with slightly lower GSI values (compare Fig.
9 with Figs. 1 and 2). It is important to note that the applicable
range for a reliable quantified GSI is limited to RQD ≤ 80.
Defining the vertical axis as equivalent to Jv, Vb or RQD/2 implicitly
re-introduces problem scale though. The same rock mass generates
the same GSI irrespective of scale. Hoek et al. (2013) caution that
using RQD, Jv or Vb ‘limits the definition of rock structure to the dimension
of the blocks’. Cundall et al. (2008) suggested addressing this shortcom-
ing by expressing the vertical axis “as the number of blocks across the
scale of interest.”
Also RQD, Jv or Vb essentially make the assumption that all defects
contribute equally to the rock mass properties. However, as the scale
of the problem increases, the rock mass behaviour will be increasingly
controlled by the large, persistent defects. Further, neither of these
‘blockiness’ measures capture the degree to which rock blocks are
interlocked, which is an important part of the descriptions in the GSI
chart. For example, the photograph of the rock mass in Fig. 10 shows a
very well interlocked blocky rock mass, which plots between GSI = 50
to 70. However, the rock mass has essentially RQD = 0.
It is important therefore to re-iterate the recommendation of Cai
et al. (2004) and Russo (2009) and implied in Hoek et al. (2013) that
the quantified GSI approach is supplementary to and not necessarily a
replacement of the visually assessed GSI.
The next section presents data from the following four different rock
masses, which were specifically chosen as they are within the groups
Fig. 2. Original GSI chart from Hoek and Brown (1997).
assessed by Marinos and Hoek (2000), to add to the case history offered
in Hoek et al. (2013).

noted that by adding measurable data, the system becomes less

• The Hawkesbury Sandstone and Ashfield Shale of Sydney, Australia
dependent on the practitioner's experience. They proposed quanti-
region
fying the GSI chart by making the vertical axis ‘block volume’ (Vb)
• The Greenland Group greywacke and argillite of Reefton, New Zealand
• The Otago Schist of Macraes, New Zealand.

3. Case studies

3.1. Hawkesbury Sandstone and Ashfield Shale

The Hawkesbury Sandstone is usually described as a medium to

coarse grained, quartzose sandstone which had been deposited in
sub-horizontal, 0.5 to 5 m, but typically 1 to 2 m, thick beds that
exhibit either massive or cross-bedded facies. The Ashfield Shale
grades from a claystone, siltstone to a siltstone/fine grained sand-
stone laminite. The Ashfield Shale comprises sub-horizontal, well-
developed, thinly laminated rock with wider spaced bedding at
about 0.1 to 0.5 m spacing.
Borehole data collected for several tunnelling projects in Sydney has
been used to characterise Sydney's Hawkesbury Sandstone and Ashfield
Shale based on the five class (I to V) Sydney classification system (Pells
et al., 1998). The database comprises 6 km of borehole logs containing
over 7200 records of defect characteristics as well as field estimated
strengths, point load index tests and laboratory uniaxial compressive
strength (UCS) tests. The data has been collected by several different
geotechnical organisations and various geotechnical engineers and
engineering geologists. Hence, any bias of particular geotechnical
Fig. 3. Various rock mass problem scales from Hoek and Brown (1980). organisations or individuals should be balanced out in the collated data.
28 R. Bertuzzi et al. / Engineering Geology 202 (2016) 24–35

Fig. 4. Ranges of GSI for various qualities of peridotite–serpentinite rock masses in ophiolitic complexes from Marinos et al. (2005). The chart shows the typical diagonal trend from top-left
to bottom-right for decreasing rock mass quality.

Details of the classification and ensuing rock mass properties are core logging. The resulting GSI is listed in Table 7. Fig. 9 shows Class III
given Bertuzzi (2014a, 2014b). The typically observed defect character- Ashfield Shale, for example, which shows GSI range of 36 to 65, and
istics and those characteristics which can occur but represent adverse mean of 55.
conditions, as contained within the borehole database, are summarised Table 7 also lists the chart GSI which is visually assessed from tunnel
in Table 3 for Hawkesbury Sandstone and Table 4 for Ashfield Shale. The and cutting exposures using the sketches on the left-hand side of the GSI
data was used to calculate the JCond89 rating following Table 2 for each charts shown in Figs. 5 and 6.
class. Not surprisingly, it was found that JCond89 is not one number but
varies in each class. Fig. 11 shows that variation in Class II Hawkesbury 3.2. Greenland Group greywacke and argillite
Sandstone, for example.
The JCond89 rating and borehole RQD were plotted onto the quanti- The greywacke and argillite rock mass data of the Greenland Group
fied GSI chart as recommended in the text accompanying the chart of the South Island of New Zealand is based on mapping of rock expo-
(see Fig. 9). An ellipse is drawn to cover the range of GSI represented sures and borehole core logging carried out by the lead author and
by the various combinations of JCond89 and RQD as obtained from the staff from OceanaGold's Reefton open pit gold mine.
 R. Bertuzzi et al. / Engineering Geology 202 (2016) 24–35 29

Fig. 5. GSI for Hawkesbury Sandstone; the arrow shows the progression from Class I to V.

Fig. 6. GSI for Ashfield Shale; the arrow shows the progression from Class I to V.
30 R. Bertuzzi et al. / Engineering Geology 202 (2016) 24–35

Fig. 7. GSI for Reefton greywacke and argillite; the arrow shows the progression from Class A to E.

Fig. 8. GSI values for Otago Schist; the arrow shows the progression from Class A to D.
 R. Bertuzzi et al. / Engineering Geology 202 (2016) 24–35 31

Fig. 9. Revised GSI chart from Hoek et al. (2013) with the range shown for Ashfield Shale Class III.

The mine is located in the northwest of the South Island approxi- UCS of greywacke is 80 MPa (varying between 20 and 318 MPa) and
mately 7 km southeast of the township of Reefton (Fig. 12). The area is for argillite, the average UCS is 50 MPa measured perpendicular to bed-
a highly active seismic zone, being approximately 35 km from New ding and 15 MPa measured parallel to bedding.
Zealand's Alpine Fault. The area comprises Ordovician-aged interbedded At the mine, the sequence of greywacke and argillite is folded into a
high strength greywacke and argillite of the Greenland Group. Average series of tight, north–south trending, gently, plunging folds. The shape

Table 3
Typical and adverse conditions of defects in Hawkesbury Sandstone from the borehole
database.

Parameter Typical Adverse

Class I to V Bedding Persistence [m] N20 N20

Roughness Rough Slightly rough
Shape Undulating Planar
Aperture [mm] b1 N10
Infill None/limonite Sandy clay
Cross-bedding Persistence [m] N10 N10
Roughness Rough Slightly rough
Shape Undulating Planar
Aperture [mm] b1 b1
Infill None/limonite None
Jointing Persistence [m] 5 N10
Roughness Rough Slightly rough
Shape Planar Planar
Aperture [mm] b1 N10
Infill None/limonite Sandy clay
Fig. 10. Quartzite rock mass downstream of the Ord Diversion Dam, Western Australia.
32 R. Bertuzzi et al. / Engineering Geology 202 (2016) 24–35

Table 4 north of Dunedin in the Otago region of the South Island of New
Typical and adverse conditions of defects in Ashfield Shale from the borehole database. Zealand (Fig. 13).
Parameter Typical Adverse The schist comprises a sequence of interlayered and gradational
Class I, II & III Bedding Persistence [m] N20 N20
psammitic and pelitic lithologies that have been derived by metamor-
Roughness Slightly rough Smooth phism of Mesozoic sandstone and mudstone respectively. The rocks
Shape Undulating Planar are strongly foliated and depending on origin are either light grey,
Aperture [mm] b1 N10 quartz rich and laminated (psammite) or dark grey to green, micaceous
Infill None/limonite Silty clay
and finely laminated (pelite).
Jointing Persistence [m] 3 N5
Roughness Slightly rough Smooth The intact rock strength does vary across the rock types but it is not
Shape Planar Planar distinctive and is very dependent upon the degree of weathering and
Aperture [mm] b1 N10 whether it is tested parallel or perpendicular to foliation. Typically, the
Infill None/limonite Silty clay schist is of high strength (UCS = 25 to 50 MPa) perpendicular to folia-
Class IV & V Bedding Persistence [m] N20 N20
Roughness Slightly rough Smooth
tion but is of low to medium strength parallel to foliation (UCS = 10
Shape Undulating Planar to 25 MPa).
Aperture [mm] b1 50 The same process as that for Reefton was followed to define four
Infill Clay coating Silty clay classes of schist rock mass (A to D) with each class showing similar
Jointing Persistence [m] 3 N5
geotechnical characteristics (Table 6). Similarly, the same process as
Roughness Slightly rough Smooth
Shape Planar Planar that for the Reefton was used to estimate the quantified and chart GSI
Aperture [mm] b1 50 as listed in Table 7.
Infill Clay coating Silty clay

3.4. Comparison

of the folds results in the overall bedding generally dipping between 20 The GSI values based on:
and 70° towards the west, although the local orientation is dependent
• the chart assessed by visual interpretation of the rock mass at a scale
upon the folding. The folds are offset by several faults. Bedding in the
suitable for a tunnel or multiple benches in an open pit mine using
greywacke is typically 0.5 to 3 m thick; and in the argillite is less than
the sketches on the left-hand side of the GSI charts shown in Figs. 4
2 m.
to 8, and
Thirty-one boreholes were geotechnically logged. Borehole intervals
• the quantification approach of Hoek et al. (2013).
assessed to have similar geotechnical characteristics were then
interrogated to provide histograms of lithology, RQD, weathering,
alteration, field estimated strength and defect characteristics. are listed in Table 7 and graphed in Fig. 14. The overall correla-
Following this process, five classes of rock mass were defined (A tion between the two sets of values is fair as measured by r 2 =
to E) with each class showing similar geotechnical characteristics 0.68.
(Table 5). JCond 89 was calculated using the defect characteristics Most of the data from the two methods plot within a band defined
from the borehole logs. around GSI ± 10. This is in keeping with the data plotted by Russo
The same process as described for Hawkesbury Sandstone and (2009) as shown in Fig. 15.
Ashfield Shale was followed. That is, the quantified GSI was found by The noticeable exception is the poorer correlation for the Ashfield
plotting an ellipse to cover the JCond89 and RQD ranges contained in Shale and for the top class of both the Greenland Group greywacke
the core logging database and the chart GSI was visually assessed from and Otago Schist. The Ashfield Shale is a laminated to thinly bedded
exposures in the open pit using the sketches on the left-hand side of siltstone but its RQD from borehole core of Class I and II is excellent.
the GSI chart shown in Fig. 7. Both estimates of GSI are listed in Table 7. Both the Greenland Group greywacke and Otago Schist are foliated
metamorphic rocks with the better quality rocks having high RQD.
High RQD tends to give the poorer correlation as seen in Fig. 16
3.3. Otago Schist which presents the same data as in Fig. 14 but grouped according
to RQD and JCond89. The figure suggests that the correlation between
The data of the Otago Schist is sourced from OceanaGold's the GSI assessed from its chart and the quantified GSI becomes poorer
Macraes Flat gold mine which is located approximately 60 km as RQD increases. There does not appear to be a clear trend with
JCond89.
Priest and Hudson (1976) showed that RQD is insensitive when
the mean discontinuity spacing is above 0.3 m. The RQD is there-
fore insensitive to the range of lamination/foliation and bedding
spacing of the Ashfield Shale, Greenland Group greywacke and
Otago Schist. An experienced practitioner using the GSI chart
would be able to identify this issue and choose an appropriate
value.

4. Conclusions

The correlation between the GSI assessed from its chart and the
quantified GSI following the methodology of Hoek et al. (2013) was
found to be fair for the combined datasets from four rock masses pre-
sented herein; the Hawkesbury Sandstone and Ashfield Shale of Syd-
ney; the Greenland Group greywacke and argillite of South Island,
New Zealand; and Otago Schist of South Island, New Zealand. Most of
Fig. 11. Variation in JCond89 as calculated for Hawkesbury Sandstone Class II. the data plot within ±10 from the two methods.
 R. Bertuzzi et al. / Engineering Geology 202 (2016) 24–35 33

Fig. 12. Map of Reefton Goldfield showing location of open pit and other historically important mining centres (modified from OceanaGold, 2008).

The rock mass that has poorer correlation between the quantified It is therefore recommended that the quantified GSI approach of
and chart GSI, also has a mismatch between RQD and the problem Hoek et al. (2013) be used to supplement and check the visually
scale. The ability of the quantified approach to estimate GSI reduces as assessed chart GSI. A difference between the two approaches of GSI ±
RQD, or for that matter Jv or Vb as noted by Hoek et al. (2013), becomes 10 is probably likely.
less able to measure the blockiness of a rock mass. The two approaches are needed because of the complexity of geolo-
It might be possible to address this shortcoming of the quantified GSI gy and the natural variation of rock masses. The authors caution against
approach by adopting the suggestion of Cundall et al. (2008) and ex-
press the vertical axis “as the number of blocks across the scale of interest”
rather than RQD/2. This means that the problem scale becomes an ex-
plicit input parameter to GSI. However, this still would not address the
issue that block size alone does not capture the degree to which the
rock blocks are interlocked, which is an important part of the descrip-
tions in the GSI chart.

Table 5
Greenland Group greywacke and argillite rock mass classes.

Class General description Intact rock Typical RQD

strength

A Lithified rock with widely Medium to high 60–100%

spaced defects and rare
shearing
B Fractured rock with Low to medium 30–70%
frequent defects and rare
shearing
C Highly fractured rock Low to medium 20–60%
with frequent defects and with zones of 0–10%
some shearing
D Fractured to fragmented Extremely low to 10–40%
rock with frequent medium with frequent zones of low
shearing quality 0–10%
E Fragmented and sheared Extremely low to b15%
low grade rock–soil low
Fig. 13. Map of Macraes showing location of Macraes.
34 R. Bertuzzi et al. / Engineering Geology 202 (2016) 24–35

Table 6
Macraes rock mass classes.

Class General description Intact rock Typical RQD

strength

A Lithified rock with frequent defects High 60–100%

and rare shearing
B Fractured rock with frequent defects Low to high 30–70% with zones
and some shearing of 25–50%
C Fractured to fragmented rock with Low to high 30–40% with zones
frequent shearing of 0–10%
D Fragmented/sheared rock Extremely low b20%
to very low

Table 7
Comparison of quantified GSI and that obtained from its chart.

Class Typical JCond89 Chart GSI Quantified

RQD Mean Mean GSI Mean
(range) (range) (Range)

Hawkesbury I 90–100 20 (11–33) 75 (65–85) 70 (57–75)

Sandstone II 90–100 15 (9–22) 65 (55–75) 63 (55–73)
III 75–100 12 (8–19) 55 (45–65) 58 (55–68)
IV 50–75 7 (4–16) 45 (35–55) 42 (34–56)
V b50 6 (3–12) 35 (25–45) 20 (10–38)
Fig. 15. Comparison of a quantified and chart GSI from Russo (2009). According to Russo
Ashfield Shale I 90–100 22 (12–26) 55 (45–65) 73 (60–80)
(2009), the red triangles are data taken by him from papers by Hoek; and the black
II 80–100 22 (7–26) 50 (40–60) 73 (55–78)
circles are data from his mapping. Most points fall within a band defined by GSI ± 10
III 40–70 18 (7–23) 40 (30–50) 55 (36–65)
which is drawn on by the authors. (For interpretation of the references to color in this
IV b40 10 (2–21) 30 (25–40) 25 (12–45)
figure legend, the reader is referred to the web version of this article.)
V b40 6 (1–19) 25 (15–30) 16 (8–35)
Greenland Group A 60–100 17 (13–20) 80 (75–85) 57 (47–67)
Greywacke/argillite B 30–70 17 (9–20) 55 (45–65) 43 (34–55)
C 20–60 15 (9–19) 40 (35–45) 42 (30–55) using the quantified GSI approach directly from borehole data without
D 10–40 11 (7–19) 30 (25–35) 23 (16–35) the necessary development of a geotechnical model. Statistical analysis
E b15 13 (1–21) 20 (15–25) 16 (5–24)
could be performed on the quantified GSI values which will give an un-
Otago Schist A 60–100 18 (7–25) 75 (70–80) 66 (50–80)
B 30–70 18 (6–22) 60 (45–70) 53 (32–62) founded impression of accuracy. Yet the purely visual GSI chart ap-
C b40 11 (2–22) 30 (25–35) 27 (11–46) proach is subjective and dependent on the practitioner's experience. It
D b20 12 (4–19) 15 (12–20) 23 (10–35) is very difficult to visualise a rock mass solely from borehole logs.
As Hoek in the Second Glossop Lecture elegantly wrote: “A good
engineering geologist and a good geotechnical engineer, working as a

Fig. 14. Comparison of quantified GSI and the GSI assessed from the chart. The highlighted region shows a band Quantified GSI = Chart GSI ± 10.
 R. Bertuzzi et al. / Engineering Geology 202 (2016) 24–35 35

Fig. 16. Comparison of quantified and chart GSI. This is the same data as Fig. 14 except grouped according to RQD (left) and JCond89 (right).

team, can usually make realistic educated guesses for each of the parame- Hoek, E., Brown, E.T., 1997. Practical estimates of rock mass strength. Int. J. Rock Mech.
Min. Sci. 34, 1165–1186.
ters required for a particular engineering analysis” (Hoek, 1999). Hoek, E., Marinos, P., Benissi, M., 1998. Applicability of the geological strength index (GSI)
The authors also suggest that different GSI values may be needed to classification for very weak and sheared rock masses. The case of the Athens Schist
cater for different numerical methods. Formation. Bull. Eng. Geol. Environ. 57, 151–160.
Hoek, E., Carter, T.G., Diederichs, M.S., 2013. Quantification of the Geological Strength
Index chart. 47th US Rock Mechanics/Geomechanics Symposium (ARMA 13-672,
Acknowledgment San Francisco, CA, USA).
Hudson, J.A., Harrison, J.P., 1997. An Introduction to the Principals. Engineering Rock Me-
chanics, Pergamon, Oxford, England.
The authors thank the anonymous reviewer for their time and in-
Marinos, P.V., 2010. New proposed GSI classification charts for weak or for complex rock
valuable input which has greatly contributed to this paper. masses. 12th International Congress. Bulletin of the Geological Society of Greece, Pa-
tras, pp. 1248–1258.
Marinos, P., Hoek, E., 2000. GSI: A Geologically Friendly Tool for Rock Mass Strength
References
Estimation.
Marinos, P., Hoek, E., 2001. Estimating the geotechnical properties of heterogeneous rock
Barton, N., Lien, R., Lunde, J., 1974. Engineering classification of rock masses for the design
masses such as flysch. Bull. Eng. Geol. Environ. 60, 85–92.
of tunnel support. Rock Mech. 6, 189–236.
Marinos, P., Hoek, E., Marinos, V., 2005. Variability of the engineering properties of rock
Bertuzzi, R., 2014a. Sydney sandstone and shale parameters for tunnel design. Aust.
masses quantified by the geological strength index: the case of ophiolites with special
Geomech. J. 49, 95–104.
emphasis on tunnelling. Bull. Eng. Geol. Environ. 65, 129–142.
Bertuzzi, R., 2014b. Sydney sandstone and shale parameters for tunnel design — corrigen-
Mostyn, G.R., Douglas, K.J., 2000. Strength of Intact Rock and Rock Masses, GeoEng2000.
dum 1. Aust. Geomech. 49, 4–39.
AGS, Melbourne, Australia, pp. 1389–1421.
Bieniawski, Z.T., 1976. Rock mass classifications in rock engineering. Proceedings of the
OceanaGold Ltd, 2008. Life of Mine review, internal report.
Symposium on Exploration for Rock Engineering, Johannesburg.
Palmström, A., 1995. RMi — A Rock Mass Characterisation System for Rock Engineering
Bieniawski, Z.T., 1989. Engineering Rock Mass Classifications. John Wiley & Sons.
Purposes. University of Oslo, Norway.
Cai, M., Kaiser, P.K., Uno, H., Tasaka, Y., Minami, M., 2004. Estimation of rock mass defor-
Pells, P.J.N., 2008. What happened to the mechanics in rock mechanics and the geology in
mation modulus and strength of jointed hard rock masses using the GSI system. Int.
engineering geology? 6th International Symposium on Ground Support in Mining
J. Rock Mech. Min. Sci. 41, 3–19.
and Civil Engineering Construction
Carter, T.G., Marinos, V., 2014. Use of GSI for Rock Engineering Design. Proceedings 1st In-
Pells, P.J.N., Mostyn, G.R., Walker, B.F., 1998. Foundations on sandstone and shale in the
ternational Conference on Applied Emprical Design Methods in Mining, Lima, Peru,
Sydney region. Aust. Geomech. 33, 17–29.
pp. 1–19.
Priest, S.D., Hudson, J.A., 1976. Discontinuity spacings in rock. Int. J. Rock Mech. Min. Sci.
Cundall, P.A., Pierce, M.E., Mas Ivars, D., 2008. Quantifying the size effect of rock mass
Geomech. Abstr. 13, 135–148.
strength. In: Potvin, Y., Carter, J., Dyskin, A., Jeffrey, R. (Eds.), 1st Southern Hemi-
Russo, G., 2009. A new rational method for calculating the GSI. Tunn. Undergr. Space
sphere International Rock Mechanics Symposium. Australian Centre for
Technol. 24, 103–111.
Geomechanics, Perth, pp. 3–15.
Sonmez, H., Ulusay, R., 1999. Modifications to the geological strength index (GSI) and
Deere, D.U., 1968. Geological considerations. In: Stagg, K.C., Zienkiewicz, O.C. (Eds.), Rock
their applicability to stability of slopes. Int. J. Rock Mech. Min. Sci. 36, 743–760.
Mechanics in Engineering Practice. John wiley & Sons, pp. 1–20.
Stacey, T.R., 2000. Reservations regarding the use of rock mass classifications in rock en-
Edelbro, C., Sjöberg, J., Nordlund, E., 2007. A quantitative comparison of strength criteria
gineering. Proceedings Bergmekanikdag 2000, SveBeFo. Swedish Rock Engineering
for hard rock masses. Tunn. Undergr. Space Technol. 22, 57–68.
Research and National Group ISRM, pp. 1–18.
Fookes, P.G., 1997. Geology for engineers: the geological model, prediction and perfor-
Terzaghi, K., 1946. Rock defects and loads in tunnel supports. In: Proctor, R.V., White, T.L.
mance. Q. J. Eng. Geol. 30, 293–424.
(Eds.), Rock Tunneling with Steel Supports. The Commercial Shearing and Stamping
Hoek, E., 1994. Strength of rock and rock masses. ISRM News J. 4–16.
Co., Youngstown, Ohio, pp. 47–83.
Hoek, E., 1999. Putting numbers to geology — an engineer's viewpoint. Q. J. Eng. Geol. 32,
1–19.
Hoek, E., Brown, E.T., 1980. Empirical strength and criterion for rock masses. J. Geotech.
Eng. 106, 1013–1035.