Вы находитесь на странице: 1из 10

Continuum and Distinct Element Numerical Modeling in Geomechanics – 2011 – Sainsbury, Hart, Detournay & Nelson (eds.

) Paper: 08-01
©2011 Itasca International Inc., Minneapolis, ISBN 978-0-9767577-2-6

Physical and numerical investigation of conglomeratic rocks

M.S. Akram
The University of New South Wales, Australia
Institute of Geology, University of the Punjab, Lahore, Pakistan
G. Sharrock & R. Mitra
The University of New South Wales, Australia

ABSTRACT: Conglomeratic rocks are natural materials comprised of macro particles with size
varying from gravels to cobbles and boulders (diameter >2 mm) bonded by relatively fine
grained cement particles with diameter <2 mm. Sampling and laboratory strength testing of such
rocks remains a challenge owing to the variation in strength and elastic properties of particles
and the cement matrix. The mechanical behavior of such materials has been studied in physical
tests on synthetic conglomerates composed of steel spheres and Portland cement. Accordingly,
equivalent numerical simulations were conducted with PFC3D to study its mechanical response
(Akram & Sharrock 2009, 2010). In this paper, the mechanical response of a range of
conglomeratic rocks is examined. Macro-particles are considered to be from sedimentary (i.e.
sandstone) or igneous (i.e. granite) origins and have Hertzian contacts. The cementing materials
considered are Portland, argillaceous and arrenaceous (quartzitic) cements with UCS strengths
in the range of 12.7 to 70 MPa. The necessary micromechanical parameters for numerical
simulation were derived from published literature. The mechanical responses of the numerical
assemblies were studied in uniaxial, triaxial and Brazilian tensile tests. The macroscopic
behavior of each conglomerate was then analyzed in terms of Hoek-Brown and Mohr-Coulomb
peak strength criteria.

1 INTRODUCTION

Conglomerates are sedimentary rocks made up of coarse grained clasts cemented with fine
grained cement matrix. The clasts are generally the detritus of pre-existing rocks and vary in
mineral composition, sizes and shapes. These may belong to sedimentary, metamorphic or
igneous origins. Clast sizes range from gravels to boulders with angular to rounded shapes,
depending on the degree of transportation prior to deposition. Likewise, the cement matrix has a
variable mineral composition and depositional history. The cement may be siliceous (quartzitic),
calcitic or gipsiferous depending on the original mineral composition, or argillaceous or
arrenaceous subject to grain sizes. Hence, given the wide variation in the mechanical properties
of clasts and cements, conglomerates exhibit highly variable and heterogeneous mechanical
behavior.
The determination of intact rock strength parameters of conglomerates is constrained by the
difficulties of extracting undisturbed samples (when cementing material is very weak) and the
validity of the test results as per the requirement of the standard laboratory testing procedures
(e.g. Hoek 1977, Brown 1981, ISRM 1983). A valid test must have a minimum specimen to
clast size ratio of 10 (ISRM 1983). This ratio can be achieved with gravel sized clasts, however,
it is nearly impossible to obtain a ratio of 10 for cobble or boulder sized clasts with conventional
coring methods.
Even in valid laboratory testing, the mechanical response of conglomerates has been observed
to vary widely subject to the properties of the cement and clasts, and the packing or arrangement
of clasts. For example, the uniaxial compressive strength (UCS) of a conglomerate with a weak
clayey cement matrix, (Boyum 1961) or a high strength quartzitic cement may range from 1.2 to
239 MPa (Shafiei & Dusseault 2008). The variation in the mechanical response of
conglomeratic rocks is considered to be a complex combination of composition and mechanical
properties of the clasts and matrix, the size distribution and arrangement of clasts, and the
mechanisms occurring at the micro scale. In natural conditions, the dependence of a specific
parameter on the macroscopic behavior of a conglomerate is difficult to examine due to many
contributing factors and parameters. Consequently, the measurement of strength and
deformation parameters for conglomerates remains a challenge at laboratory scale due to their
heterogeneous and composite nature, and sampling and testing constraints and requirements.
Alternatively, Akram (2010) conducted a study on idealized conglomerates by physical and
numerical modeling to investigate the mechanics of conglomeratic rocks. In this study, physical
conglomerate specimens were prepared from steel spheres and Portland cement paste. The tests
included uniaxial, triaxial, Brazilian tensile and shear box tests. Equivalent numerical
conglomerate specimens were prepared in PFC3D (Itasca 2008) with measured and known micro
parameters. The responses of the physical and numerical assemblies were compared in an effort
to extend numerical simulations to examine the sensitivity of various factors affecting the
response of a conglomerate. The comparison (Akram & Sharrock 2009, 2010) showed both
similarities and contrasts in similar testing methods. For example, in uniaxial compression, the
numerical tests had good agreement with physical tests for peak strengths, damage thresholds
and failure mechanisms. Similarly, good agreement was obtained between numerical and
physical triaxial tests. The numerical tensile strength was also found to lie within the variation
range of physical test results, showing reasonable agreement. Similarly, in the shear box tests, a
good agreement of physical and numerical conglomerates was observed in terms of cohesion
and angle of friction. However, the numerical conglomerates did not reproduce the elastic
response and post peak behavior of physical models in uniaxial and triaxial testing. Similarly, a
well developed distinct crack through the specimen was not obtained in the numerical Brazilian
tensile test.
Despite the contrasts observed in the elastic response, numerical simulation was used to
investigate the sensitivity of the various parameters of an idealized conglomerate that were kept
constant in the comparison study. These included the effect of specimen size, particle size
distribution, sensitivity of the particle and cementing materials (Akram 2010). However, this
paper present the results of the sensitivity of the particle and cementing materials in an idealized
conglomerate consisting of uniformly sized spheres and a homogeneous cement matrix.

2 METHODOLGY
2.1 Composition of conglomerates
In the present study, three particle and three cementing materials were considered to fabricate
the conglomerates. The particle materials were steel, granite and sandstone (Table 1) and
cements were Portland, argillaceous and arrenaceous (Table 2). The properties of steel and
Portland cement are documented in previous studies by Akram & Sharrock (2009, 2010). The
properties of granite, sandstone, and argillaceous and arrenaceous cements were sourced from
the literature (see Tables 1 & 2 for references). The numerical conglomerates were prepared
using all particle and cementing materials. The particle material properties from Table 1 were
used for the generated clast particles. The cement properties from Table 2 were used to derive
micro parameters for parallel bonds representing interparticle cement in conglomerates (Table 4).
In total, nine (9) conglomerates were prepared for uniaxial, triaxial and Brazilian tensile
testing. The prepared numerical models represent idealized natural conglomerates comprised of
spherical, uniformly sized clasts and a homogeneous cement matrix of uniform strength. The
strength of the cement and clast-cement interface is constant, as the parallel bond represents a
uniform strength across the bonded particles. In this study, rigid body particles we used, and
hence failure occurred in the cement matrix through breakage of parallel bonds. Hence, the
results of this investigation are applicable to the clast supported conglomerates with high
strength and stiffness particles and low strength and stiffness cements, in which failure always
occurs through the cement. The details of the conglomerates and the nomenclature used to
denote them are summarized in Table 3.

2
It should be noted that in this study, the strength and elastic parameters of the real (physical)
materials were used (Tables 1 & 2). Any particle material or interparticle cement inducing a
variation in the mechanical response of a conglomerate will be the function of the stiffness (E)
and strength (UCS) of the cement, and the stiffness of the particles (with the assumption of rigid
particles). The UCS is equal to E times the modulus ratio (MR) which is 300-400 for all particle
and cement materials used in this study. Therefore, the results of all conglomerates can be
analyzed and interpreted in terms of UCS and E contrasts of particle materials and cements (i.e.
UCScem / UCSpart and Ecem / Epart). However, in this paper, the mechanical behavior of the
conglomerates is analyzed in terms of the uniaxial compressive strength of cement to the
stiffness of the particle material (UCScem / Epart) and the stiffness ratio of interparticle cement to
particle material (Ecem / Epart).

Table 1. Summary of particle materials and their properties used to create numerical conglomerates.
________________________________________________________________________________________________________
Density (ρ) Young’s modulus (E) Shear modulus (G) Poisson’s
Particle materials Kg/m 3
GPa GPa ratio (ν )
________________________________________________________________________________________________________
Steel STL 7800 200 74.5 0.31
Granite1 GR 2600 73 30 0.21
2
Sandstone SST 2300 33 13 0.25
________________________________________________________________________________________________________
1
Properties outsourced from Lac du Bonnet granite (after Duevel & Haimson 1997).
2
Properties outsourced from Fell sandstone, Northamberland, UK (after Bell 1978, 2007).

Table 2. Summary of interparticle cements and their properties used for parallel bonds to create numerical
conglomerates.
________________________________________________________________________________________________________
UCS (σc) Shear strength (τ) Tensile strength (σt) Young’s modulus (E)
Cementing materials MPa MPa MPa GPa
________________________________________________________________________________________________________
Portland cement PC 12.7 4.0 1.35 3.18
Argillaceous cement1 ARG 27.4 4.9 1.80 4.20
Arrenaceous cement2 ARN 70.0 12.5 5.40 30.0
________________________________________________________________________________________________________
1
Properties outsourced from Hawkesbury sandstone, Sydney, Australia (after Sharrock et al. 2009).
2
Properties outsourced from Fell sandstone, Northamberland, UK (after Bell 1978, 2007).

Table 3. Nomenclature and details of the conglomerates prepared using different particle and cementing
materials.
Cementing materials
Portland cement Argillaceous cement Arrenaceous cement
(PC) (ARG) (ARN)
Steel (STL) [STL+PC] [STL+ARG] [STL+ARN]
Materials
Particle

Granite (GR) [GR+PC] [GR+ARG] [GR+ARN]


Sandstone (SST) [SST+PC] [SST+ARG] [SST+ARN]

2.2 Specimen diagenesis and testing


A uniformly sized (diameter = 4.75 mm) array of elastic particles were generated inside
cylindrical containers bounded by frictionless walls. The length and diameter of the uniaxial
specimens were set to 188 mm and 94 mm respectively, while, 94 mm and 47 mm were adopted
as diameter and thickness of the Brazilian discs. Since, the present study is the part of the
physical and numerical studies of the conglomerates presented in Akram & Sharrock (2009,
2010), and Akram (2010), the adopted dimensions of the numerical specimens are the same as
that of the prepared physical models. The porosity of the specimens was also kept the same as
that of physical models (i.e. 39.7%). The particles were provided with the shear modulus and
Poisson’s ratio to implement a Hertz-Mindlin contact model representing the physical

3
laboratory. The gravity induced packing was adopted in all conglomeratic specimens similar to
physical models (Akram & Sharrock 2009, 2010, and Akram 2010). Parallel bonds were
installed at the particle contacts to mimic interparticle cementation (Table 4). In all
conglomerates, an interparticle friction of 78° was adopted, being the outcome of micro to
assembly friction sensitivity study (after Akram & Sharrock 2009, 2010). The testing was
conducted by controlling the wall velocities with servo-mechanism. All conglomerates were
subjected to uniaxial, triaxial (at 5.0 & 10.0 MPa confining pressures) and Brazilian tensile tests.
A detailed description of the specimen diagenesis and testing techniques used in the present
study can be found elsewhere, for example, in Akram & Sharrock (2009, 2010) and in Akram
(2010).
A summary of the test results of all the numerical conglomerates is shown in Table 5. A
discussion of the test results is given in the following sections.

Table 4. Summary of parallel bond parameters derived from Table 2.


________________________________________________________________________________________________________
Cementing Normal Strength Shear Strength Normal Stiffness Shear Stiffness
Materials ( σ c ) MPa ( τ c ) MPa ( k n ) Pa/m ( k s ) Pa/m
________________________________________________________________________________________________________
PC 1.5 4.0 6.69e11 6.69e11
ARG 2.0 4.9 8.84e11 8.84e11
ARN 6.0 12.5 6.32e12 6.32e12
________________________________________________________________________________________________________

Table 5. Summary of test results on numerical conglomerates.


________________________________________________________________________________________________________
Peak Strength (MPa) Young’s Tensile
Uniaxial Triaxial Triaxial modulus Poisson’s Strength
Composition of conglomerates (UCS) (5 MPa) (10 MPa) (GPa) ratio (MPa)
________________________________________________________________________________________________________
[STL+PC] 3.12 20.8 44.5 2.98 0.074 0.102
[STL+ARG] 3.87 20.9 44.7 3.55 0.067 0.176
[STL+ARN] 7.37 20.2 44.6 14.9 0.043 0.431
[GR+PC] 2.87 23.3 50.6 2.30 0.054 0.134
[GR+ARG] 3.54 23.3 50.4 2.84 0.052 0.184
[GR+ARN] 6.77 22.5 50.4 12.2 0.035 0.485
[SST+PC] 2.64 25.2 55.2 1.98 0.047 0.131
[SST+ARG] 3.26 25.2 55.7 2.49 0.046 0.199
[SST+ARN] 6.45 27.4 55.8 9.58 0.027 0.744
________________________________________________________________________________________________________

3 DISCUSSION
3.1 Peak strengths
The peak strength results of the uniaxial and triaxial tests are shown in
Figure 1. These results were analyzed to explore the peak strength dependence on particle
material and interparticle cement. In uniaxial testing, the interparticle cement was observed to
influence the peak strength significantly, irrespective of the particle material. The peak strengths
of the numerical conglomerates were observed to increase as cement strength is increased, as
shown in Figure 1a for steel particles, Figure 1b for granitic particles, and Figure 1c for
sandstone particles. However, interestingly, this influence gradually decreased with the increase
of confining pressure (Figs. 1a-c).
On the other hand, particle material had negligible effect on the uniaxial strengths for all
cementing materials. However, the particle material did the influence peak strength at higher
confining pressures; that is, peak strength increased with a decrease in the stiffness (and
strength) of the particles. This effect is shown for Portland cement in Figure 1d, for argillaceous
cement in Figure 1e and for arrenaceous cement in Figure 1f. These observations suggest that
the properties of interparticle cement mainly influence peak strength in uniaxial, or close to
uniaxial conditions. By contrast, the properties of the particle material have a more significant
effect in confined conditions.

4
The peak strength of all numerical conglomerates was also observed to change with the
variation of cement strength (unconfined compressive strength- UCS) and the stiffness (E) of the
particle and the cementing materials. These changes were studied in terms of the ratio of UCS of
the cementing material to the E of the particle material (i.e., UCScem/ Epart). The peak strengths of
the conglomerates were found to generally increase with the increase of UCScem/ Epart ratio
(Figure 2a)
The tensile strength of conglomerates was determined in Brazilian tensile testing. These
results are plotted against the UCScem/ Epart ratio in Figure 2b. Generally, an increase in tensile
strengths was noted with the increase of UCScem/ Epart ratio similar to that of peak strengths in
uniaxial and triaxial tests (Fig. 2a).
60

a d
50
Peak Strength [MPa]

40

σ1 30

20

[STL+ARN] [SST+PC]
10 [STL+ARG] [GR+PC]
[STL+PC] [STL+PC]

60

50 b e
Peak Strength [MPa]

40

σ1
30

20

[ GR+ARN] [SST+ARG]
10 [ GR+ARG]
[GR+ARG]
[ GR+PC] [STL+ARG]
0

60

50
c f
Peak Strength [MPa]

40

σ1
30

20
[SST+ARN]
[SST+ARG] [SST+ARN]
10
[SST+PC] [GR+ARN]
[STL+ARN]
0

σ σ3
0 2 4 6 8 10 12 0 2 4 6 8 10 12

3
Confinem ent [MPa]
Figure 1. Sensitivity of the peak strengths of various conglomerates with particle and cementing materials
in σ1 − σ3 space. Plots a, b and c show a relative variation of peak strengths with the variation of
cementing material for: a) steel particles, b) granitic particles and c) sandstone particles. Plots d, e and f
show a relative variation of peak strengths with the variation of particle materials with: d) Portland
cement, e) argillaceous cement and f) arrenaceous cement.

The compressive to tensile strength ratio is an important mechanical characteristic of a


particular conglomerate. Based on uniaxial and Brazilian tensile test results, the uniaxial
compressive to tensile strength ratios (UCS/T) of all conglomerates were determined and plotted
against the UCScem/ Epart ratio and the stiffness ratio of the cement to particle material (i.e. Ecem/
Epart ). Generally, a decreasing trend of UCS/T was observed as the UCScem/ Epart ratio increased
(Fig. 3a). A similar decrease in UCS/T was noted with the increase of the Ecem/ Epart ratio (Fig.
3b). These observations indicate that the compressive to tensile strength ratio of a conglomerate
decreases with the decrease of the cement strength to particle stiffness ratio and the contrast in
the stiffness.

5
70.0 0.8
UCS
TRX (5M P a) 0.7
60.0
TRX (10M P a)

Tensile Strength [MPa]


0.6

Peak Strength [MPa]


50.0 R2 = 0.5418
0.5 R2 = 0.5951
40.0 0.4
R2 = 0.5228
30.0 0.3

0.2
20.0
R = 0.6210
2 0.1
10.0
0.0

0.0
a -0.1 b
0.00001 0.0001 0.001 0.01 0.00001 0.0001 0.001 0.01
UCScem/Epart UCScem/Epart

Figure 2. Variation of; a) peak strengths, and b) tensile strength of various conglomerates against the ratio
of the uniaxial strength of the cement to particle material.
35
35
30
30 UCS/T

R2 = 0.9194 25
25

20 20
UCS/T

15 15 R2 = 0.8739
10 10

5 a 5 b
0 0
0.00001 0.0001 0.001 0.01
0.01 0.10 1.00 10.0
UCScem/Epart
Ecem/Epart
Figure 3. Variation of uniaxial compressive strength to tensile strength ratio: a) with uniaxial strength of
cement to stiffness of particle materials (UCScem/ Epart) and, b) with Young’s modulii ratio of cement to
particle materials (Ecem/ Epart).

The variation of UCS/T with cement type is shown in Figure 4a, and with the particle material
in Figure 4b. It is observed that the UCS/T ratio is highest for Portland cement and lowest for
arrenaceous cement (Fig. 4a). Similarly, UCS/T is highest for steel particles and lowest for
sandstone particles (Fig. 4b). In combining the effect of particle and cementing materials, it is
clear that the UCS/T ratio is highest for conglomerate [STL+PC] consisting of steel particles
glued together with Portland cement (corresponding to UCScem/ Epart of about 0.00006) and is
lowest for conglomerate [SST+ARN] having sandstone particles and arrenaceous interparticle
cement (corresponding to UCScem/ Epart of about 0.002).
35

30 a R2 = 0.8809
30
b
R2 = 0.9542

25 25
R2 = 0.9645
20
UCS/T

20 R2 = 0.9936
UCS/T

15 15 R2 = 0.9802

10 10
A rrenaceo us Cement R2 = 0.9562 Sandsto ne P articles
5 A rgillaceo us Cement 5 Grainite P articles
P o rtland Cement Steel particles
0 0
0.00001 0.0001 0.001 0.01 0.00001 0.0001 0.001 0.01
UCScem/Epart UCScem/Epart
Figure 4. Variation of uniaxial compressive strength to tensile strength ratio (UCS/T) with uniaxial
strength of cement to stiffness of particles (UCScem/ Epart); a) for interparticle cements and, b) with particle
materials.

Another factor that is believed to control the UCS/T ratio is the ratio of the shear to normal
(tensile) strength (i.e., τ / σ n ) of the interparticle cement (parallel bond). The shear and tensile
strengths of the cementing materials used for numerical simulations are summarized in Table 2.
The UCS/T ratio was plotted against the shear to tensile strength ratio of the interparticle cement
in Figure 5. It was observed that the UCS/T ratio increases with the increase of the shear to
normal strength ratio ( τ / σ n ) of the interparticle cement, irrespective of the particle material.

6
35

Sandstone Part icles


30
Grainit e Particles
Steel particles
25

20
UCS/T

15 Arrenaceous Cement

10 Argillaceous Cement

5
Port land Cement

0
2.00 2.10 2.20 2.30 2.40 2.50 2.60 2.70 2.80

τ /σ n
Figure 5. Variation of uniaxial compressive strength to tensile strength ratio (UCS/T) of conlgomerates
with shear to normal strength ratio ( τ / σ n ) of the interparticle cement (parallel bond) for steel, granitic
and sandstone particles.

3.2 Young’s modulus and Poisson’s ratio


The sensitivity of elastic response to particle and cementing materials was investigated across
the nine conglomerates. Although the response of the numerical conglomerates was stiffer in
comparison to the physical conglomerates, as discussed in Akram & Sharrock (2009, 2010) and
Akram (2010), here, the relative variation of the elastic response of the numerical assemblies
with particle and cementing material is presented. Young’s modulus and Poisson’s ratio were
determined at 50% of the peak strength in uniaxial testing. The recorded values were plotted
against the modulus ratio of cementing material to particle material (Ecem/ Epart). The plot of the
Young’s modulii versus Ecem/ Epart ratio is shown in Figure 6.
Young’s modulus was found to increase as the stiffness of particles or interparticle cement
increases. For a fixed particle material, the modulus increases by Ecem/ Epart ratio. That is, by
increasing the stiffness of the cementing material, the modulus of the conglomeratic rock
increases. For a fixed cementing material, the increasing trend is steep with steel particles
(E=200 GPa) and relatively shallow for sandstone particles (E=33 GPa). The granitic particles
(E=73 GPa) show an intermediate increasing trend. Similarly, with a particular cementing
material, modulii decrease with an increase of the Ecem/ Epart ratio; that is, by decreasing the
stiffness of the particles (Fig. 6). This decreasing trend is sharper for arrenaceous cement
(E=30 GPa) than for argillaceous (E=4.8 GPa) or Portland cement (E=3.18 GPa).

16.0

12.0 R2 = 0.9961
R2 = 0.9949
E [GPa]

R2 = 0.9958
8.0

R2 = 0.9971

Granite Particles
4.0 Sandstone Particles
R2 = 0.9845 Steel Particles
Arrenaceous Cement
R2 = 0.9798 Argillaceous Cement
Portland Cement
0.0
0.01 0.10 1.00 10.00
E cem /E part
Figure 6. Variation of Young’s modulii (E) of conglomerates with particle and cementing materials
corresponding to the modulus ratio of cementing material to particle material (Ecem/ Epart).

7
The Poisson’s ratios of all the conglomerate assemblies were determined in uniaxial testing at
50% of the peak strengths. Although the Poisson’s ratios in the numerical simulations are lower
compared to the synthetic conglomerates, as discussed in Akram & Sharrock (2009, 2010) and
Akram (2010), here a relative variation was found to exist with respect to particle and cementing
materials. The variation of the Poisson’s ratio was assessed with the stiffness ratio of cement to
particle material (Ecem/ Epart) and is shown in Figure 7. Poisson’s ratio was found to decrease
with the increase of (Ecem/ Epart) ratio, irrespective of the particle or cementing material.

0.08

0.07

0.06
Poisson's Ratio

0.05

0.04

0.03 R2 = 0.96

0.02

0.01

0.00
0.01 0.10 1.00
E cem/E part
Figure 7. Variation of Young’s modulii (E) of conglomerates with particle and cementing materials
corresponding to the modulus ratio of cementing material to particle material (Ecem/ Epart).

4 STRENGTH CRITERIA OF NUMERICAL CONGLOMERATES

The test results of all numerical conglomerates were subjected to strength criteria to observe the
effect of particle and cementing materials. Mohr-Coulomb and Hoek-Brown criteria were
applied on the test results using Rocdata 4.0 (Rocscience 2010).

4.1 Mohr-Coulomb criterion


In previous studies (e.g. Akram 2010), it was observed that numerical conglomerates exhibit
Mohr-Coulomb behavior in contrast to that of physical conglomerates. Hence, Mohr-Coulomb
criterion was applied to the uniaxial and triaxial test results on all numerical conglomerate
assemblies (Table 5). The Mohr-Coulomb fitting curves for all conglomerates are shown in
Figure 8 in major-minor principal stress space along with curve fitting parameters. Generally,
the sum square of errors (residuals) (i.e. SSE[R]) is less than 10 for most of the conglomerates,
demonstrating a reasonable fit. It was noted that conglomerates with the same particle material
show similar trends, irrespective of cementing material. The slope of the regression lines is steep
for low strength particles (i.e. sandstone particles) and gentle for high strength particles (i.e.
steel particles), yielding high and low internal friction angles respectively.

4.2 Hoek-Brown criterion


Hoek-Brown criterion was applied to the results of uniaxial, triaxial and Brazilian tensile tests of
the nine conglomerates (Table 5) to observe the relative variation of Hoek-Brown parameters
with respect to particle and cementing material. The Hoek-Brown curves for all conglomerates
are shown in Figure 9 in major-minor principal stress space along with curve fitting parameters,
that is, the sum square of errors (residuals) (SSE[R]) and material constant (mi). The variation of
mi induced by the properties of the particle and cementing materials can be clearly seen in
Figure 9. In the literature (e.g. Rocscience 2008, 2010), the typical value of mi for natural
conglomerates is 21±3 with lower and upper values of 18 and 24 respectively. However, it is
noteworthy that these values are based on very limited and often invalid tests on specific
conglomerates that may not represent the real strength at a valid specimen to clast size ratio. In
numerical conglomerates, the value of mi was observed to increase with the decrease in the
strength and stiffness of the particle materials and cement matrices as well (Fig. 9).

8
120
110 SSE[R]
[STL+PC] 5.82

Major Principal Stress (MPa)


100
[GR+PC] 7.82
90 [SST+PC] 9.41
[STL+ARG] 7.62
80 [GR+ARG] 8.93
[SST+ARG] 12.23
70 [STL+ARN] 22.31
[GR+ARN] 24.52
60
[SST+ARN] 9.43
50
40
30
20
10
0
0 5 10 15 20 25

Minor Principal Stress (MPa)


Figure 8. Mohr-Coulomb criterion curves for various conglomerates in major-minor principal stress
space.

70
SSE [R] mi
[STL+PC] 49.80 29.92
60
Major Principal Stress (MPa)

[GR+PC] 76.04 30.46


[SST+PC] 104.13 45.25
50 [STL+ARG] 51.78 22.76
[GR+ARG] 74.79 25.83
[SST+ARG] 116.23 36.66
40 [STL+ARN] 64.20 14.18
[GR+ARN] 90.58 15.40
30 [SST+ARN] 98.65 13.98

20

10

0
-2 0 2 4 6 8 10 12

Minor Principal Stress (MPa)


Figure 9. Hoek-Brown peak strength criteria for the nine numerical conglomerates in major-minor
principal stress space.

5 CONCLUSIONS

The mechanical behavior of an idealized clast supported conglomerate (comprised of spherical


and uniformly sized clasts) was examined with reference to the clast (or particle) material and
interparticle cement in ISRM tests. The study indicates that the mechanical parameters of
conglomerates are significantly influenced by the strength and stiffness of the particles and
cementing materials.
In uniaxial testing, the interparticle cement was observed to influence peak strength of the
conglomerate significantly, irrespective of particle material. The peak strength of the numerical
conglomerate was observed to increase as the cement strength (UCS) was increased. However,
this influence gradually decreased with the increase of confining pressure. On the other hand,
the particle material was observed to have a significant influence in triaxial conditions and
negligible effect in uniaxial conditions for all cementing materials. The peak strength of
conglomerates increased as the stiffness of the particles decreased. Hence, it is concluded that
the strength of a conglomeratic rock is a function of the strength and elastic properties of the
cement matrix predominantly in uniaxial conditions with no less effect from characteristics of
clasts. However, properties of the clasts become more important at higher confining pressures,
and here the contribution is less significant from the interparticle cement.
It was also observed that the peak strengths of the conglomerates generally increased with the
increase of UCScem/Epart ratio. Similarly, the tensile strengths of the conglomerates determined in
Brazilian tensile testing were also found to increase with the increase of the UCScem/ Epart ratio.

9
The effect of the particle and interparticle cementing materials was also analyzed for the non-
dimensional ratio of uniaxial compressive strength to tensile strength (UCS/T). The UCS/T ratio
was found to decrease with the increase of UCScem/Epart. However, the decreasing trend depends
on the particular particle and cementing material.
Young’s modulii were found to increase with the increase of the stiffness of both the particles
and the interparticle cement. The increasing trend depends on the stiffness of the particles and
cement as well.
The Mohr-Coulomb criterion was found to be a better model for numerical conglomerates
than the Hoek-Brown criterion. However, both Mohr-Coulomb and Hoek-Brown criteria
showed similar trends of the fitting curves for the same particle materials. Steep slopes of the
fitting curves were obtained for low strength particles (i.e. sandstone particles) and gentle slopes
were obtained for high strength particles (i.e. steel particles), which yielded high and low
internal friction angles (Mohr-Coulomb criterion) and, low and high values of the material
constant (Hoek-Brown criterion) respectively.
The present investigation and previous studies (e.g. Akram & Sharrock 2009, 2010, Akram
2010) involving DEM modeling for conglomerates highlight DEM’s potential to model and
investigate the response of coarse grained rocks such as conglomerates, breccias and
agglomerates that are difficult to study in physical laboratory because of the sampling and
testing constraints. However, more work is required in this area, such as, sensitivity of the clast
shape, incorporating post failure interlocking of cementation and variation of the clast-cement
interface strength, to precisely model the behavior of such rocks.

REFERENCES

Akram, M.S. 2010. Physical and numerical investigation of conglomeratic rocks. PhD Thesis. School of
Mining Engineering, The University of New South Wales, Sydney.
Akram, M.S. & Sharrock, G. 2009. Physical and numerical investigation of a cemented granular assembly
under uniaxial and triaxial compression. The 43rd US Rock Mechanics Symposium and 4th U.S.-
Canada Rock Mechanics Symposium June 28 – July 1, 2009, Asheville, American Rock Mechanics
Association (ARMA), (CD-ROM) 09-024.
Akram, M. S. & Sharrock, G. 2010. Physical and numerical investigation of a cemented granular
assembly of steel spheres. Int. J. Numer. Anal. Meth. Geomech. Published online in Wiley InterScience
(www.interscience.wiley.com). DOI: 10.1002/nag.885.
Bell, F.G. 1978. Physical and mechanical properties of the Fell sandstones, Northumberland, England.
Engineering Geology. 12 (1): 1-29.
Bell, F.G. 2007. Engineering Geology. Butterworth-Heinemann, ELSEVIER, Oxford.
Boyum, B.H. 1961. Subsidence case studies in Michigan mines. Symposium on Rock Mechanics,
University Park, Penn., USA. 19-57.
Brown, E.T., Ed. 1981. Rock Characterization, Testing and Monitoring - ISRM Suggested Methods.
Pergamon Press, London, London.
Duevel, B. & Haimson, B. 1997. Mechanical characterization of pink lac du Bonnet granite: Evidence of
nonlinearity and anisotropy. Int. J. Rock Mech. Min. Sci. 34 (3-4): 117.e1-117.e18.
Hoek, E. 1977. Rock mechanics laboratory testing in the context of a consulting engineering
organization. Int. J. Rock Mech. Min. Sci. Geomech. Abstr. 14: 93-101.
ISRM 1983. Suggested methods for determining the strength of rock materials in triaxial compression:
Revised version. Int. J. Rock Mech. Min. Sci. Geomech. Abstr. 20 (6): 285-290.
Itasca Consulting Group, Inc. 2008. PFC3D - Particle Flow Code in 3 Dimensions, Ver. 4.0 User's
Manual. Minneapolis: Itasca.
Rocscience 2008. RocLab1.0. Rocscience (www.rocscience.com), Toronto.
Rocscience 2010. RocData 4.0. Rocscience (www.rocscience.com), Toronto.
Shafiei, A. & Dusseault, M.B. 2008. Geomechanical properties of a conglomerate from Iran. 42nd US
Rock Mechs Symp. & 2nd U.S.-Canada Rock Mechs Symp. June 29- July 2, San Francisco, ARMA 08-
053.
Sharrock, G., Akram, M.S. & Mitra, R. 2009. Application of synthetic rock mass modeling to estimate
the strength of jointed sandstone. The 43rd US Rock Mechs Symp & 4th U.S.-Canada Rock Mechs
Symp. June 28 – July 1, 2009, Asheville, American Rock Mechanics Assoc. (ARMA), (CD-ROM) 09-
59.

10

Вам также может понравиться