Вы находитесь на странице: 1из 15

See

discussions, stats, and author profiles for this publication at: https://www.researchgate.net/publication/276207717

INTERPRETATION OF UCS TEST RESULTS FOR


ENGINEERING DESIGN

Conference Paper · May 2015

CITATIONS READS

4 1,275

4 authors:

Robert Bewick Florian Amann


Golder Associates RWTH Aachen University
24 PUBLICATIONS 124 CITATIONS 114 PUBLICATIONS 594 CITATIONS

SEE PROFILE SEE PROFILE

P. K. Kaiser Derek Martin


Laurentian University University of Alberta
187 PUBLICATIONS 4,338 CITATIONS 176 PUBLICATIONS 3,140 CITATIONS

SEE PROFILE SEE PROFILE

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

Structural design of shaft for hydropower construction View project

Paraglacial, periglacial rock slope stability View project

All content following this page was uploaded by Robert Bewick on 13 May 2015.

The user has requested enhancement of the downloaded file.


ISRM Congress 2015 Proceedings - Int’l Symposium on Rock Mechanics - ISBN: 978-1-926872-25-4

INTERPRETATION OF UCS TEST RESULTS FOR ENGINEERING DESIGN

*R. P. Bewick
Golder Associates Ltd.
1010 Lorne Street
Sudbury, Canada P3C 4R9
(*Corresponding author: rbewick@golder.com)

F. Amann
ETH Zurich, Department of Earth Science
Sonneggstrasse 5
8092 Zürich

P. K. Kaiser
Laurentian University
935 Ramsey Lake Rd
Sudbury, Canada P3E 2C6

C. D. Martin
University of Alberta
Department of Civil and Environmental Engineering
Edmonton, Alberta T6G 2W2

1
ISRM Congress 2015 Proceedings - Int’l Symposium on Rock Mechanics - ISBN: 978-1-926872-25-4

INTERPRETATION OF THE UCS TEST FOR ENGINEERING DESIGN

ABSTRACT

This article focuses on the uniaxial compressive strength (UCS) of homogeneous and
heterogeneous rocks. Critical factors impacting the UCS of rocks are reviewed and then separate
assessments of UCS data from homogeneous and heterogeneous rocks are presented. The variability
(coefficient of variation) of UCS test results for homogeneous rocks is found to be generally <25% while
for heterogeneous rocks the variability is >35%. Failure mode variation is found to be the cause of the
variability of UCS test results in heterogeneous rocks. Empirical engineering design methods that use UCS
as an input parameter are discussed and suggestions are provided for selecting appropriate UCS values.

KEYWORDS

UCS, compressive strength, laboratory testing, interpretation, design

INTRODUCTION

The UCS is one of the most commonly used rock engineering parameters whether for rock mass
classification or for strength determination. The mean UCS and its variability are often assumed to
represent a reliable rock material property. For the reasons discussed in this article, UCS is rarely
representative of the intact rock UCS. The UCS test simply records the collapse load during uniaxial
loading of a cylindrical specimen. As suggested by many, the UCS is an index rather than a unique
engineering parameter. The UCS is thus a proxy for rock strength which depends on the loading rate (e.g.,
Bieniawski 1967), specimen geometry (e.g., Hudson et al., 1971), specimen size (e.g., Bieniawski 1968),
and many other factors. Furthermore, the UCS is not the same as the Hoek-Brown strength criterion
parameter σci.

There are differences in opinion on what is commonly considered to be ‘intact rock’. When
sampling ‘intact’ pieces of core from drillholes crossing blocks of rock bound by block forming
discontinuities, samples will be ‘intact’ but some samples will be veined, others will be damaged, and
others will be neither. According to Hoek (1983), intact rock pieces are “the unfractured blocks occurring
between structural discontinuities in a typical rock mass” and the σci parameter for intact rock is determined
using the fitting procedure proposed by Hoek and Brown (1997) based on the data requirements outlined
by Hoek (1983). Studies (e.g., Bewick and Kaiser 2014; Kaiser and Kim 2014) have shown that
independently determined UCS and σci values for the same rock type are not equivalent. Thus, UCS data
cannot be used as a replacement for σci.

In this article, we focus on the UCS from unconfined tests and differentiate between UCS with
failure through intact rock of predominately homogeneous specimens (called UCSho=homogeneous) and failure
through combined failure modes such as shear on defects (such as veins) along with through intact rock
(called UCShe=heterogeneous). Both UCSho and UCShe are representative of the strength distribution in intact
rock blocks forming a rock mass. The UCS data from homogenous rock specimens generally provide a
good estimate of the strength of the least defected part of the rock because the failure mode is minimally
impacted by visually obvious heterogeneity and typically occurs through mostly intact rock material. For
this reason, UCSho is often called the ‘intact strength’ and labelled UCSi . However, as stated above, UCSi
is not to be confused with σci.

2
ISRM Congress 2015 Proceedings - Int’l Symposium on Rock Mechanics - ISBN: 978-1-926872-25-4

The UCS data from heterogeneous parts of the rock blocks, due to their inherent variability at
both the field and laboratory scales, often are obtained from specimens that do not predominately fail
through ‘intact’ rock. Heterogeneous rocks exhibit a variety of failure modes such as axial splitting,
combined shear and splitting rupture, shear along a single pre-existing discontinuity, and combinations of
the previously listed failure modes as shown in Figure 1. As a consequence, the UCS data from
heterogeneous rocks almost always includes a mixture of failure modes resulting in more variable UCS
values (typically with Coefficient of Variation, CoV >35%). Thus, the UCS test results from
heterogeneous rocks are representative of the UCS distribution in a heterogeneous rock block if a
sufficiently large number of specimens are tested. Primarily for the variability in failure mode, leading to
highly variable UCS data, the reported UCShe for heterogeneous rocks is often confusing, misleading and
challenging to apply to many empirical design methods.

One of the sources for this confusion originates in testing standards and suggested methods such
as ASTM D 2938-86 and ISRM (1979; 1999) which are the industry standards for conducting and
reporting laboratory uniaxial tests on rock cylinders for the purpose of estimating a rock’s intact
compressive strength. While these standards and suggested methods provide excellent guidance on how to
prepare rock specimens for testing and how to monitor and conduct the tests so that the testing procedures
are reliable, they provide insufficient guidance on how to: (1) report the results in a meaningful manner
such that they can be properly processed outside the laboratory; and (2) interpret the results for the
determination of reliable engineering parameters. Since the focus of the standards was the testing of
homogeneous rocks that predominantly fail through intact rock material, modifications to the standards and
suggested methods are required for heterogeneous rocks.

In this article, focus is placed on Uniaxial Compressive Strength (UCS) testing procedures, test
results and their interpretation, and guidance on the use of UCS data in some empirical design approaches
for strong (UCS >25MPa) hard brittle rocks. The terminology adopted for this article is outlined in
Appendix A. The reasons for variability in UCS and the number of specimens needed to obtain
representative, statistically valid UCS values are investigated. It is shown that the UCS is not necessarily a
unique engineering parameter for a given rock type and, depending on a rock’s heterogeneity, different
UCS values may need to be specified for various design aspects.

Figure 1. Examples of common failure modes observed in hard brittle rocks loaded in compression
showing before and after testing photographs (core specimens are ~63 mm in diameter); (a-b) intact – axial
splitting; (c-d) intact – shear rupture; (e-f) combination of intact rock and pre-existing discontinuities; (g-h)
failure along a pre-existing discontinuity; (i-j) failure around clasts or other specimen scale matrix
elements; and (k-l) non-valid test result where failure occurred due to end sloughing.

3
ISRM Congress 2015 Proceedings - Int’l Symposium on Rock Mechanics - ISBN: 978-1-926872-25-4

UCS TESTING REQUIREMENTS

ISRM (1979; 1999) provide suggestions for standardized testing of cylindrical rock specimens
under laterally unconfined conditions with respect to specimen end preparation, slenderness, and diameter
to grain size ratio. The significance of specimen end-face preparation and frictional constraints is often
underestimated in routinely performed tests. Imperfections such as irregularities on both the platen and
specimen face lead inevitably to local stress concentrations (e.g., point loads or line loads) that
substantially influence the failure behaviour and decrease the peak strength of a rock specimen. This is
illustrated by Figure 2 showing a fused quartz specimen (i.e., UCS=1000 MPa, E=72 GPa). The specimen
was loaded under unconfined conditions. At a load of 50 MPa (i.e., 1/20th of the UCS) tensile cracks
emanated from an imperfection (i.e., a piece of particulate on loading platen) at the specimen end-face.
The greater the stiffness and strength of the rock the less forgiving a UCS test is to imperfections in
specimen preparation showing the need for carful adherence to testing procedures.

Figure 2. Fused quartz specimen fractured at 1/20th of its UCS due to a small particulate on the loading
platen/specimen interface. (a) Isometric view and (b) Top view. (Photograph curtesy of F. Amann).

Peng (1970) and Peng and Johnson (1972) analyzed experimentally and numerically the stress
state within a rock specimen exposed to uniaxial loading considering various end-boundary conditions.
They found that for increased friction between the platen and the rock specimen the non-uniformity of the
stress state in the specimen increased, even though a uniform stress was applied. In compression, a rock
specimen tends to expand laterally as it shortens axially. Frictional constraints at the specimen´s end-faces
tend to prevent lateral expansion causing the specimen’s ends to be subjected to shear stress. As a
consequence, the stress state in the specimen is neither uniform nor uniaxial (i.e., it is actually a triaxial
stress state). In practice it is almost impossible to fully eliminate frictional constraints at the platen.
However, the effect of frictional constraints on the non-uniformity of the stress state in the center of the
specimen can be reduced by adjusting the slenderness ratio (height/diameter ratio). The effect of
slenderness ratio on Young´s Modulus, UCS, and strain at peak stress was analyzed in several studies on
concrete and rock (e.g., Mogi 1966; Dhir and Sangha 1973). Figure 3a shows the relative strength in
percent (i.e., UCS normalized by UCS with a slenderness ratio of 2.5) for different rock types and
slenderness ratios (Mogi 1966; Dhir and Sangha 1973). For a slenderness ratio > 2.5, UCS changes
slightly (i.e., by less than 2%). For smaller ratios the relative strength increases considerably depending on
the tested material. The ISRM suggested methods account for this important effect by suggesting a height-
to-diameter ratio between 2.5-3.0 and the ASTM standard by suggesting a height-to-diameter ratio >2.0.

4
ISRM Congress 2015 Proceedings - Int’l Symposium on Rock Mechanics - ISBN: 978-1-926872-25-4

Figure 3. (a) Impact of slenderness ratio (height/diameter) on UCS relative to the UCS at a slenderness
ratio of 2.5. The data shows that UCS is minimally influenced by slenderness at ratios >2.5. (b) Impact of
the largest grain to specimen diameter ratio on UCS for concrete (modified from Blanks and McNamara
1935). The plot shows that at largest grain to specimen diameter ratios >10, UCS is minimally impacted.

ISRM (1979) suggests that the specimen diameter should be 10 times larger than the size of the
largest grain (i.e., 10:1). For determining the full stress strain curve ISRM (1999) suggests 20:1. Blanks
and McNamara (1935) performed tests on cylindrical mass concrete specimens with a height to diameter
ratio of 2:1, diameters ranging from 50 mm to 900 mm and ratios between the largest grain to specimen
diameter ranging from 2:1 to 48:1 (Figure 3b). This unique series of tests suggests that for a ratio larger
than 10:1 the UCS is only slightly affected by the grain size while for smaller ratios a large variability in
UCS is to be anticipated. The reason for increasing the suggested ratio from 10:1 to 20:1 is not explored in
ISRM (1999) but is possibly related to the fact that for determining the full stress-strain response, strain
gauges mounted on the specimen surface are required.

Even if the above listed effects on UCS and its variability have been minimized, most rocks show
a natural variability that is relevant for engineering design purposes. The ISRM (1979) states that the
number of tests should be determined from practical considerations but at least five are preferred;
irrespectively if the rock is homogeneous or heterogeneous. To investigate the impact of specimen
population effects on the mean and distribution of UCS values, a dataset of damaged Lac du Bonnet granite
(Martin 1993) containing 69 UCS tests (mean 139 MPa; Standard Deviation 10 MPa; CoV 7%) was
randomly sampled to obtain populations of 5, 10, 15, 20, and 30 UCS test results. The results of the
random sampling are shown in Figure 4 as histograms of each of the data populations and a fit normal
distribution to the data. While the mean and standard deviation values are generally comparable for the
presented sample populations, visually a distribution of test results becomes apparent when the sample
population is between 10 and 20. Based on this preliminary assessment, greater than five UCS tests are
likely required to provide a meaningful representation of the UCS value and distribution for a given rock
type. Ten tests appear to be a reasonable minimum for this relatively homogeneous rock (i.e., CoV 7%).
While not presented for space considerations, more tests would be needed to define a reliable mean,
standard deviation, and distribution of more heterogeneous rocks with CoV’s greater than 25%. Initial
testing of greater than 10 specimens should be considered for heterogeneous rocks.

5
ISRM Congress 2015 Proceedings - Int’l Symposium on Rock Mechanics - ISBN: 978-1-926872-25-4

Figure 4. Impact of specimen population on Lac du Bonnet Granite mean, standard deviation, and
distribution of UCS test results with low CoV. (Stdev – standard deviation; PDF – Probability density
function).

UCS TESTING RESULTS FROM HOMOGENEOUS AND HETEROGENEOUS ROCKS

Homogeneous

Homogeneous rocks typically fail predominately through ‘intact’ rock material and are well suited
for the current standard and suggested methods as outlined above and ISRM (1979; 1999). Thus, the
homogeneous test results (UCSho) in this section are representative of UCSi. To emphasize this, UCS test
results for four homogeneous rock types are shown in Figure 5 with fitted normal distributions. The four
homogeneous rock types are also summarized in Table 1. While these UCS test results were completed
following standard and careful testing procedures, some but relatively constrained variability is evident in
the results. This is highlighted by the CoV in Table 1 which ranges from 2% for Carrera marble, the most
homogenous rock type presented in this article, to 15% for the Sandstone. For these four homogeneous
rock types, based on specimen populations of 10 to 13, the CoV is ≤ 15%.

The variability observed in homogeneous rocks is primarily due to two sources: (1) the quality of
the standard and suggested testing procedures; and (2) variability in the homogeneous rock specimens.
Based on the Carrera marble dataset containing 10 specimens, limited variability in the testing results is
evident. This suggests that if the testing standards and suggested methods are adhered to, there should be
minimal effects of the testing procedure on the UCS value and the resulting UCSho = UCSi is a reliable
engineering strength index. The variability in homogeneous rocks is thus predominately due to inherent
specimen variability that depends on rock type.

6
ISRM Congress 2015 Proceedings - Int’l Symposium on Rock Mechanics - ISBN: 978-1-926872-25-4

8 Carrera Marble 0.16


Granite (Franklin and Hoek 1970)
7 Quartz Dolerite (Franklin and Hoek 1970) 0.14
Sandstone (Franklin and Hoek 1970)
6 0.12

5 0.1
Frequency

PDF
4 0.08

3 0.06

2 0.04

1 0.02

0 0
0
15
30
45
60
75
90
105
120
135
150
165
180
195
210
225
240
255
270
285
300
315
330
345
UCS (MPa)
Figure 5. Histograms of UCS test results completed on some homogeneous rock datasets. Showing fit
normal distributions to each dataset. Statistical summary provided in Table 1. PDF – Probability Density
Function.

Table 1. Statistical summary of UCS results for compiled homogenous rocks in Figure 5.
Lithology Average Standard CoV Minimum Maximum Population
UCS (MPa) Deviation (%) UCS (MPa) UCS (MPa)
(MPa)
Carrera Marble 96 2 2 91 98 10
Sandstonea 53 8 15 37 68 13
Quartz Doleritea 313 9 3 211 332 12
Granitea 194 25 13 111 214 10
a
Franklin and Hoek (1970)

Heterogeneous

Unlike homogeneous rocks, heterogeneous rocks when tested exhibit a number of different failure
modes as previously discussed and shown in Figure 1. Consequently, UCS test results for heterogeneous
rocks are highly variable and are representative of UCShe (see Appendix A) and not of the homogeneous
part UCSho. As shown in the previous section for homogeneous rocks, if the standard and suggested testing
methods are conformed to, variability in UCS test results appear directly related to variability in the
material tested. In the case of heterogeneous rocks, the potential for various failure modes leads to
increased variability. UCS test results for four heterogeneous rock types are shown in Figure 6 with fitted
normal distributions and summarized in Table 2. While these UCS test results were completed following
standard and suggested testing procedures, high variability is evident in the results (CoV in Table 2 ranging
from 33% for Quintner Limestone to 84% for Sulfate-rich clay rock). For the three rock types with
reasonable CoV values (i.e., <50%), the CoV is on average 38% (ranging from 33 to 42%). This is typical
for heterogeneous rock types. The excessive CoV of 84% for the sulfate rich clay rock flags that the
sample population is made of at least two unique groups and the reported mean and CoV are of little
practical value.

One of the main limitations of the standard and suggested methods is that they were
predominately written for homogeneous rocks where failure is generally through homogeneous rock.
When these standards and suggested methods are applied to heterogeneous rock test result summaries
where only the mean, standard deviation, and CoV are to be reported for the core specimens tested (as per
recommendations in ISRM 1979; ASTM D 2938-86), high variability is obtained with a relatively low
average UCS value. If the heterogeneous rock test results are divided by failure types such as breaks along
discrete planes of weakness, combinations of intact rock and defects, and through intact rock, the mean

7
ISRM Congress 2015 Proceedings - Int’l Symposium on Rock Mechanics - ISBN: 978-1-926872-25-4

intact compressive strength increases and the CoV for the intact failure mode tends to those found for the
homogeneous part of the rocks. An example of this is shown in Table 3 where the UCS test results for a
heterogeneous limestone have been summarized by the available UCS tests regardless of failure mode,
UCS test results with breaks along discrete discontinuity planes removed, and finally failure involving
predominately intact rock alone. As is evident in Table 3, the mean UCS increases and the CoV decreases
to a value similar to the range of the presented homogeneous test results in Table 1.

Figure 6. Histograms of UCS test results completed on some heterogeneous rock datasets. Showing fit
normal distributions. Statistical summary provided in Table 1. Note that negative UCS values are
suggested by the fit normal distributions. In reality, any fit distribution to UCS data needs to be constrained
to values greater than 0 MPa.

Table 2. Statistical summary of UCS results for compiled heterogeneous rocks in Figure 6.
Lithology Average Standard CoV Minimum Maximum Population
UCShe (MPa) Deviation (%) UCS (MPa) UCS (MPa)
(MPa)
Limestone 91 38 42 33 152 12
Quintner 99 33 33 55 149 15
Limestoneb
Sulfate-rich clay 43 37 84 6 151 21
rockc
Quartz Monzonite 142 53 38 35 222 12
b c
Perras et al. (2012); Amann et al. (2013)

Table 3. Heterogeneous rock – Impact of failure mode grouping on UCS and its variability.
Parameter Complete Discrete discontinuity Intact breaks
dataset breaks removed only (UCSho)

Average UCS (MPa) 91 107 120


Standard Deviation (MPa) 38 28 25
CoV (%) 42 26 20
Population 12 9 6

The following example illustrates the need for careful data filtering. For the Quartzite described
by Bewick et al. (2011), the UCS data is shown in Figure 7. The mean UCS for the available data is 118
MPa with a CoV of 59% leading to the normal distribution shown with black dashed line. This distribution,
unless replaced by a skewed fit, suggests a high likelihood for extremely low (near zero) strengths or weak
rock and a maximum strength of about 300 MPa. Upon examination and filtering by failure mode, it was

8
ISRM Congress 2015 Proceedings - Int’l Symposium on Rock Mechanics - ISBN: 978-1-926872-25-4

found that the dataset can be represented by two failure mode groupings with respective mean UCS and
CoV. Respective normal distributions are shown by dotted and dashed green lines and the combined bi-
modal distribution is shown by the full green line. This distribution conforms to the binned data well and
further illustrates the need for data filtering by failure mode. This example illustrates that the mean value
and CoV for the available data is highly misleading as the rock is made up of a rock with high
homogeneous strength (UCSho = 215 MPa/37%) and a matrix of lower strength (UCShe = 95 MPa/42%).
The practical consequences of this misrepresentation are significant as discussed in the section on UCS in
rock engineering.

Figure 7. Heterogeneous quartzite showing bi-modal distribution due to different failure modes in the
dataset.

Summary and recommendations

The standard and suggested testing methods for the determination of the UCS provide excellent
guidance on specimen preparation and testing procedure such that the UCS testing procedures minimally
impact the variability of the test results. The methods also provide reliable results for homogeneous rocks
where failure is through predominately intact rock thus producing UCSho, which in this case is synonymous
with the intact strength UCSi. For heterogeneous rocks, the UCS tests results when available failure modes
are considered should be reported as a UCS for heterogeneous rock, UCShe. When the CoV exceeds 30 to
35%, further investigations into the cause for the high variability are suggested. It is likely that the
underlying data is bi- or multi-modal and the source of the variability needs to be identified and reported.
Filtering of the UCS test data is required to obtain estimates for each group of data making up the entire
dataset family. If data is affected by defects causing various failure modes, each mode should be analyzed
separately and reported separately. If other factors are identified, the data should be reported in a similar
manner for each identified group.

UCS IN ROCK ENGINEERING

The basis for rock engineering designs are rooted in empirical approaches. One of the common
input parameters used in these empirical approaches is the UCS and the UCS value is either used directly
or indirectly. In this section, the role of the UCS in 3 commonly used empirical approaches in rock
engineering is discussed and suggestions are provided for selecting appropriate UCS values.

Empirical spalling assessment

When the stresses on the boundary of an underground excavation reach the rock mass strength,
failure occurs. In good quality hard rock, such as that found in many underground environments, the
failure process is described as ‘spalling’ and the associated rock mass strength as the ‘spalling strength’.

9
ISRM Congress 2015 Proceedings - Int’l Symposium on Rock Mechanics - ISBN: 978-1-926872-25-4

The term spalling is purposely used to indicate that the failure process involves extensional
splitting/cracking first analyzed by Fairhurst and Cook (1966). This process is fundamentally different
from that of shearing which is commonly observed in weak rocks. Consequently, the stress state that leads
to spalling is in the region of low confinement where the laboratory rock strength is typically defined by
the tensile strength and the uniaxial compressive strength.

From a tunnel design perspective two issues must be addressed: (1) what procedure should be
used during the site characterization phase to assess if spalling should be anticipated during construction;
and (2) if spalling is expected, what design procedure should be used to assess the lateral and radial extent
of the spalled zone. Martin (2014) reviewed the various approaches and suggested that spalling strength
can be estimated by using the crack initiation (CI) determined using UCS tests. Nicksiar and Martin
(2013) compiled the crack initiation stress for 336 specimens of low porosity Igneous, Metamorphic and
Sedimentary rocks. They found the CI stress to UCS strength ratio (CIR) ranged from 0.42 to 0.47
regardless of the rock type in uniaxial compression and when confined CIR ranged from 0.50 to 0.54. If
the maximum tangential stress on the boundary of a tunnel exceeds CI, spalling should be expected. To
predict the extent of spalling (‫ܦ‬௙ ) the empirical equation is often used:

஽೑

= 0.49 + 1.25(ߪ௠௔௫ /ܷ‫( )ܵܥ‬1)

where ߪ௠௔௫ is the maximum tangential stress on the boundary of the tunnel and a is the tunnel radius.

When establishing the CIR, Nicksiar and Martin (2013) used rocks that ranged in UCS values
from 15 MPa to 375 MPa, yet the CIR only ranged from 0.42 to 0.47. Hence it would appear that CIR is
insensitive to the rock homogeneity. Spalling initiates as a strain localization phenomenon, which means
that the spalling volume on the boundary of the tunnel is relatively small. If the rock contains defects the
depth of spalling should be assessed using the UCS for the defected rock (UCShe) and filtered for UCSho to
provide a bound to the range of potential strengths and depth of spalling. If homogeneous, the UCS of the
homogeneous, intact rock, UCSho, should be used. Experience selecting the UCS value for application in
Equation 1, suggests it should be the mean value (Rojat et al., 2009).

Empirical pillar design

The extensive research of Salamon and Munro (1967), using 125 case histories involving coal
pillar collapses, led to the well-known pillar strength formula:

ௐഀ
ߪ௣ = ‫ ܭ‬ு ഁ (2)

where σp (MPa) is the pillar strength, K (MPa) is the strength of a unit volume of coal, and W and H are the
pillar width and height in metres, respectively, and α and β are empirical constants.

The formula in equation 2 expresses the pillar strength as a collapse load that is a function of
geometry of the specimen and the strength of a unit volume of the material. The similarities between the
loading path in a UCS test and the loading path experienced by a pillar in a mine, as extraction proceeds,
are obvious. The major differences being, the rate of loading and the scale of the specimens with the UCS
volume being only a small fraction of the pillar volume.

While equation 2 was originally designed for coal, similar forms of the empirical equation have
been proposed for various rock masses (Martin and Maybee, 2000). Figure 8a shows the various pillar
formulas compiled by Martin and Maybee (2000) and possible pillar stability curves for pillars in stressed
hard brittle rock (Figure 8b) with the pillar strength normalized to the uniaxial compressive strength. It is
evident from Figure 8a that despite the uncertainty in empirical formulas derived from back analyses,
scaling the pillar strength to the UCS shows considerable consistency between the pillar design curves for

10
ISRM Congress 2015 Proceedings - Int’l Symposium on Rock Mechanics - ISBN: 978-1-926872-25-4

narrow pillars (W/H<2). Hence there is an expectation that the pillar strength may be linked to the UCS
strength.

a) b)

Figure 8. (a) Comparison of various empirical pillar strength formulas. (b) Empirical pillar database
showing Hoek-Brown brittle parameter derived pillar stability curves (both [a] and [b] modified from
Martin and Maybee 2000).

The notion that the UCS could be linked to the pillar strength was explored by Košták (1971).
Košták proposed that pillar strength should be based on a cumulative distribution of UCS values for
heterogeneous rock. Košták recommended that a minimum of 40 specimens of rock that preserve defects
in the correct proportion to the intact (homogeneous) rock was needed. By combining both the UCS
results for the defected and intact specimens, Košták suggested that the strength magnitude corresponding
to p = 11.5% for the cumulative distribution, could be taken as representative of the pillar strength for any
particular rock. Using the data from Košták, the pillar strength at a cumulative probability of 11.5% is 116
MPa, which is approximately 57% of the mean UCSho, 94% of the mean UCShe and 63% of mean,
combined UCSho and UCShe. While none of these values are in agreement with the empirical strength
formulas provided in Figure 8a, using the UCShe gives the closest agreement. Hence in order to use the
existing empirical pillar formulas it appears that UCShe provides the closest agreement with the empirical
experience.

Tunnel boring machine penetration

Empirical penetration prediction models (PPM) have been derived through correlating rock
mechanical data (i.e., UCS, degree of jointing, orientation of joints) and TBM specific parameters (i.e.,
cutter diameter, cutter spacing) with achieved penetration rates (Gehring 1995; 1997; Bruland 2000). In
modern tunneling, where TBMs are frequently used, penetration prediction models are of key relevance for
estimating the construction time and costs. In general, empirical PPMs are based upon a prediction of the
base penetration rate (mm/rev) which is subsequently reduced to account for rock mass and TBM
characteristics (Lislerud 1988; Gehring 1995; Bruland 2000). Many PPM´s consider the UCS of the rock
as one key parameter for predicting the base penetration rate. Since the cutting process itself occurs on the
cm-scale, Gehring (1995) discusses the relevance of an accurately determined UCS where failure of the
rock specimens is unaffected by any macroscopic flaws or discontinuities. Test results where flaws lead to
an obvious strength reduction shall not be considered for penetration prediction. Gehring (1995) therefore
used specimens with a slenderness ratio of 1.0 to minimize strength reduction effects associated with flaws.
However, as discussed previously, for specimens with a slenderness ratio <2.5 the strength tends to
increase substantially and thus the increase in UCS found by Gehring (1995) is a combined influence of
eliminated flawed specimen results and an unknown stress state.

TBM penetration is dependent on the cutting process which is not only affected by the rock
strength on the cm-scale, but also on the spatial distribution of rock strength exposed on the tunnel face.
The strongest rock at the tunnel face often determines the penetration rate. Hence, the relevant strength for
TBM performance assessment should be chosen as the upper limit of strength data. The average strength
would lead to a too optimistic prediction of the base penetration rate. Only UCSho values or UCShe filtered

11
ISRM Congress 2015 Proceedings - Int’l Symposium on Rock Mechanics - ISBN: 978-1-926872-25-4

for the homogeneous parts with failure through homogeneous intact rock should be utilized. For defining a
representative UCS for TBM penetration prediction in heterogeneous rocks a sufficient number (i.e.,
typically >10) of representative specimens per rock unit is required.

CONCLUSIONS

While the UCS test is one of the most common tests in site characterisation programs, producing
meaningful results remains a challenge. It is evident from the previous sections that the indiscriminate use
of UCS data reported without careful screening may be misleading. Based on the findings from the work
used in this paper, the following recommendations should be considered when undertaking a testing
program to determine the UCS:
• All testing should be carried out by a qualified laboratory using appropriate QA/QC
procedures using the ISRM or ASTM methods and procedures.
• UCS from homogeneous rock:
o Use a minimum of 5 to10 specimens to define UCSho=i mean and CoV.
• UCS from heterogeneous rock:
o A minimum of 10 to 40 specimens may be required to define UCShe mean and CoV.
o If CoV exceeds 30 to 35%, the data is likely multi-modal and one needs to identify
sources contributing to the multi-modal behavior and report respective mean
strengths and CoVs for each unique dataset.
o Filtering of testing data is required to get the intact strength (UCSho) of the
homogeneous component of a rock block while the tested dataset is representative of
the strength (UCShe) and variability of a heterogeneous rock block.
o Depending on the empirical design approach, either (UCSho or UCShe) or both may
be relevant.

REFERENCES

Amann, F., Undul, O., Kaiser, P.K. (2013) Crack initiation and crack propagation in heterogeneous sulfate-
rich clay rocks. Rock Mech Rock Eng. doi:10.1007/s00603-013-0495-3.
ASTM D 2938-86 (1991) Standard test method for unconfined compressive strenght of intact rock core
specimens.
Bewick, R.P., Kaiser, P.K., Valley, B. (2011) Interpretation of triaxial testing data for estimation of the
Hoek-Brown strength parameter mi. Paper 11-347. 45th US Rock Mechanics / Geomechanics
Symposium held in San Francisco, CA, June 26-29.
Bewick, R.P., Kaiser, P.K. (2013) Discussion on “An Empirical Failure Criterion for Intact Rocks” by
Peng et al. (2013). Rock Mechanics and Rock Engineering. DOI 10.1007/s00603-013-0514-4.
Bieniawski, Z.T. (1967) Mechanism of brittle fracture of rock, parts I, II, III. Int. J. Rock Mech. Min. Sci.
4(4), 395-430.
Bieniawski, Z.T. (1968) The effect of specimen size on compressive strength of coal. Int. J. Rock Mech.
Min. Sci. 5, 325-335.
Blanks, R., McNamara, C. (1935) Mass concrete tests in large cylinders. ACI Journal Proceedings, 31(1),
280–303.
Bruland, A. (2000) Hard Rock Tunnel Boring. Vol. 1. Trondheim: NTNU Trondheim.
Dhir, R.K., Sangha, C.M. (1973) Relationships between size, deformation and strength for cylindrical
specimens loaded in uniaxial compression. Int. J. Rock Mech. Min. Sci. 10, 699-712.
Franklin, J.A., Hoek, E. (1970) Developments in triaxial testing equipment. Rock Mech. 2, 223-228.
Fairhurst, C., N. G. W. Cook (1966). The phenomenon of rock splitting parallel to the direction of
maximum compression in the neighbourhood of a surface. In Proc. of the 1st Congress of the
International Society of Rock Mechanics, Lisbon, pp. 687–692.
Gehring, K. (1995) Leistungs- Und Verschleißprognosen Im Maschinellen Tunnelbau. Felsbau 3.

12
ISRM Congress 2015 Proceedings - Int’l Symposium on Rock Mechanics - ISBN: 978-1-926872-25-4

Gehring, K. (1997) Classification of Drillability, Cuttability, Boreability and Abrasiveness in Tunnelling.


Felsbau 3.
Hoek, E., Brown, E.T. (1980) Underground excavations in rock. The Institution of Mining and Metallurgy,
London.
Hoek, E. (1983) Strength of jointed rock masses. Geotechnique, 23(3). 187-223.
Hoek, E., Brown. E.T. (1997) Practical estimates of rock mass strength. Int. J. Rock Mech. Min. Sci. 34(8),
1165-1186.
Hudson, J.A., Brown, E.T., Fairhurst, C. (1971) Shape of the complete stress strain curve for rock. Proc.
Symp. Rock Mech. 13th, Univ. Illinois, Urbana, IL.
International Society for Rock Mechanics (1979) Suggested methods for determining the uniaxial
compressive stregnth and deformability of rock materials.
International Society for Rock Mechanics (1999) ISRM suggested method for the complete stress-strain
curve for intact rock in uniaxial compression. Int. J. Rock Mech. Min. Sci. 36, 279-289.
Kaiser, P.K., Kim, B. (2014) Characterization of strength of intact brittle rock considering confinement
dependent failure processes. Rock Mech Rock Eng. DOI 10.1007/s00603-014-0545-5.
Kaiser, P.K., Amann, F., Bewick, R.P. (2015) Overcoming challenges of rock mass characterization for
underground construction in deep mines. 13th International ISRM Congress 2015. Montreal, QC.
Košták, B. (1971). Pillar strength prediction from representative sample of hard rock. Int. J Rock Mech,
Min Sci. 8(5), 523–526. doi:10.1016/1365-1609(71)90016-5.
Lislerud, A. (1988) Hard Rock Tunnel Boring, Prognosis and Costs. Tunneling and Underground Space
Technology 3 (1): 9–17.
Martin, C.D. (1993) Strength of massive Lac du Bonnet Granite around underground openings. Ph.D.
Thesis. University of Manitoba, MB.
Martin, C. D. (2014) The impact of brittle behaviour of rocks on tunnel excavation design. In L. R.
Alejano, Á. Perucho, C. Olalla, & R. Jiménez, (pp. 51–62). In Proceedings EUROCK 2014, Rock
Engineering and Rock Mechanics: Structures in and on Rock Masses, Vigo. Taylor and Francis
Group, London, UK.
Martin, C.D., Maybee, W. (2000) The strength of hard-rock pillars. Int J. Rock Mech. Min. Sci., 37(8),
1239–1246.
Mogi, K. (1966) Some precise measurements of fracture strength of rocks under uniform compressive
stress. Rock Mech. Eng. Geol., 4: 175-186.
Nicksiar, M. & C. D. Martin (2013) Crack initiation stress in low porosity crystalline and sedimentary
rocks. Engineering Geology 154(64–76).
Peng, S.D. (1970) Failure and Fracture of Chelmsford Granite, PhD Thesis, Stanford University.
Peng, S., Johnson, A.M. (1972) Crack growth and faulting in cylindrical specimens of Chelmsford
Granite.Int. J. Rock Mech. Min. Sci. 9, 37-86.
Perras, M.A., Diederichs, M.S., Amann, F. (2012) Fracture initiation and propagation in the Quintner
Limestone. ARMA 12-401. 46th US Rock Mechanics Symp. Chicago, IL. June 24-27, 2012.
Rojat, F., V. Labiouse, P. K. Kaiser, & F. Descoeudres (2009) Brittle rock failure in Steg Lateral Adit of
the Lotschberg Base Tunnel. Rock Mech Rock Eng. 42, 341–359.
Salamon MDG, Munro AH. (1967) A study of the strength of coal pillars. J S Afr Inst Min Metall.
68:55267.

APPENDIX A - TERMINOLOGY

For clarity the following terminology is used in this article starting from the definition of a rock
mass through to UCS and specimen types.

Rock mass: made up of rock blocks defined by block forming open discontinuities.
Rock block: may be homogeneous (i.e., relatively similar properties throughout) or heterogeneous (i.e.,
with properties changing spatially within the rock block).
Intact test specimen: typically obtained by coring, is a coherent piece of rock that could be sampled
without breakage during drilling; it may be discontinuous and damaged at the micro-scale but it is intact.

13
ISRM Congress 2015 Proceedings - Int’l Symposium on Rock Mechanics - ISBN: 978-1-926872-25-4

Intact homogeneous rock (Figure A1a): Cylindrical rock specimens which visually appear to be
homogeneous with limited variability in grain size distribution, mineral grain composition, or extent of
micro-fracturing, such as Carrera Marble or Lac du Bonnet Granite (Martin 1993).
Intact heterogeneous rock (Figure A1b): This satisfies the intact homogeneous rock definition except
homogeneity. Cylindrical rock specimens which visually appear heterogeneous with evident veinlet
stockworks or non-uniform grain size distribution at the specimen scale, variable mineral grain
composition, or extent of micro-fracturing, such as quartz monzonite or sulfate rich clay rocks (Amann et
al., 2013).
UCS test: A test on a cylindrical rock core specimen that records the collapse load for a given loading
condition that meets ASTM standard or ISRM suggested method requirements of: height to diameter ratio
of 2.5-3.0; diameter preferably of not less than NX core size (~54 mm); Specimens shall be tested at
its natural water content; Loading rate 0.5-1.0 MPa/s (total testing/loading time 5-10min).
UCS value: The collapse stress regardless of failure mode calculated from the collapse load and cylindrical
specimen loading area.
Homogeneous rock UCS (UCSho): The UCS value obtained from failure through predominately
homogeneous, intact parts of the rock. A small scatter (CoV <25%) in UCS values should be anticipated.
UCSho is often called UCSi, the strength of the intact part of the defected rock.
Heterogeneous rock UCS (UCShe): The strength based on UCS test results where available valid failure
modes are considered regardless of failure type. A large scatter (CoV >25%) in UCS values should be
anticipated.
Coefficient of Variation (CoV): a measure of relative variation of a dataset given as the standard deviation
as a percentage of the mean.

Figure A1 Examples of homogeneous (a) and heterogeneous (63mm in diameter) (b) rock core specimens.

14

View publication stats

Вам также может понравиться