Numerical Simulations of Compressive

Tests on Bimrock

Monica Barbero
Assistant Professor, Department of Structural, Geotechnical and Building
Engineering, Politecnico di Torino, Italy
e-mail: monica.barbero@polito.it

Mariacristina Bonini
Research Assistant, Department of Structural, Geotechnical and Building
Engineering, Politecnico di Torino, Italy
e-mail: mariacristina.bonini@polito.it

Mauro Borri-Brunetto
Assistant Professor, Department of Structural, Geotechnical and Building
Engineering, Politecnico di Torino, Italy
e-mail: mauro.borri@polito.it

ABSTRACT
The paper deals with the mechanical behaviour of “bimrock” (block in matrix
rock), characterized by a heterogeneous structure, constituted by rock blocks
included in a small-grained, well-cemented material (matrix) with lower
mechanical characteristics. The bimrock is a very peculiar material that may give
rise to serious problems in engineering applications. To properly model this kind
of problems it is necessary to suitably characterize the bimrock. In order to
contribute to the understanding of the bimrock mechanical behaviour and to
individuate strength and deformability laws suitable for the bimrock as an
equivalent continuum, a number of unconfined or laterally confined tests on
bimrock specimens have been numerically simulated. The finite difference and the
finite element methods, implemented in the FLAC 2D and the ADINA 2D and 3D
codes, respectively, have been used to investigate possible model-dependent
responses. Novel approaches have been used, in which the size, orientation and
position of rock blocks contained in a two and three-dimensional specimen are
generated according to a random process that produces representative samples of
block populations with assigned statistical properties. Different volumetric
proportions and aspect ratios of the blocks have been considered and the effects on
the strength and deformability of bimrock are presented. The numerical results
from both 2D and 3D analyses show the influence of the volume proportion of the
blocks on the strength and deformability of bimrock. Furthermore, the results

- 3397 -
 Vol. 17 [2012], Bund. X 3398

suggest that the interfaces between the blocks and the matrix play an important
role, and may have an influence on the mechanical behaviour of bimrock.

KEYWORDS: bimrock, melange, heterogeneous materials, numerical modelling

INTRODUCTION
Bimrock, block in matrix rock, is a particular type of rock that is characterized by a
heterogeneous structure. Bimrock has been well defined by several Authors (Medley 1994,
Lindquist 1994, Lindquist and Goodman 1994) as a material that is constituted by rock blocks,
contained in a small-grained, well cemented material (matrix) with lower mechanical
characteristics.
Bimrock is a very peculiar material, which can give rise to serious engineering application
problems. In particular, it is difficult to estimate the strength and deformability parameters to use
in numerical analyses that are needed to assess, for instance, the safety of tunnel excavations
(Barpi et al. 2011, Bonini and Barla 2012, Bonini et al. 2012) or the stability of slopes (Barbero
and Barla 2010, Barbero and Barpi, 2012).
The studies reported in literature have pointed out some of the main characteristics of a rock
mass that can be classified as a typical bimrock (Medley 2002):
- The block size distribution is scale independent, that is, the lab and site size
scale behaviour is the same.
- The ratio between the block and matrix strength is greater than 2.
- The maximum and minimum sizes of the blocks are conventionally
assumed to be 75% and 5% of the characteristic engineering length (tunnel
diameter, thickness of a landslide, laboratory rock specimen diameter, etc.).
In this range, the block size distribution has a power law with an exponent
of –2.3.
- The strength and deformability of a bimrock depend on the volumetric
proportion and orientation of the blocks (Sönmez).
Different methods are available to evaluate the volumetric proportion of the blocks (BVP in
the following): one-dimensional (drilling and outcrop mapping), two-dimensional (photographs
and maps analysis) and three-dimensional (laboratory analysis). It has been observed that the first
two methods underestimate the BVP, and lead to significant mistakes.
A numerical approach has been used in the present study to analyze the influence of the BVP
on the strength and deformability characteristics of bimrock, as an equivalent continuum. First, a
number of two-dimensional bimrock specimens, with or without lateral confinement pressure,
have been numerically tested using finite difference and finite elements methods, implemented in
the FLAC and the ADINA codes (Barbero et al. 2007). A novel approach has been used to
introduce the rock blocks in the specimens, according to given statistical parameters. After, three-
dimensional bimrock specimens, with or without lateral confinement pressure, have been
numerically tested using ADINA code (Barbero et al. 2008). By using a procedure implemented
in Matlab, spherical rock blocks are introduced in the volume of a cylindrical specimen, with a
random process in which size and position are generated according to the statistical properties of
their distributions, providing a domain constituted of matrix and rock blocks. This geometry is the
base for generating the mesh to be analysed, after defining the material model and the boundary
conditions, with Adina.
 Vol. 17 [2012], Bund. X 3399

2D ANALYSES
Generation of the specimens
The generation of a two-dimensional bimrock sample starts with the assigning of the
statistical distribution parameters of the elliptical inclusion properties, namely: minimum and
maximum size, exponent of the power law, ratio of the maximum axes, range of orientation.
Each inclusion is added to a sample by firstly extracting its major axis, by means of a power-
law random numbers generator, then, obtaining the minor axis and the axis rotation with respect
to a fixed reference frame, using uniform random number generators. The insertion of inclusions
ends when the chosen BVP is attained.
A second phase of the algorithm defines the position of the ellipses via a uniform random
generation of the centre coordinates. In order to retain only feasible inclusion configurations, a
test on the interpenetration of the ellipses was included in the algorithm (Donev et al. 2005 and
references therein). The inclusions are added to the specimen in reverse-size order, the position of
each new block is checked against the previous ones, and the extraction is repeated until a
suitable arrangement is obtained.
The algorithm that generates the ellipses is the basis for the discretisation that is needed for
the subsequent numerical analyses. A rectangular grid, with a sufficiently small cell size is
generated to build a FLAC model. The points defining a polygonal approximation of the ellipse
boundary are then superimposed onto the grid, so that an adjustment of the relevant grid-points is
introduced, and the mechanical properties of the inclusions are given to the enclosed cells.
A different approach was followed to build the ADINA models. In this case, the points
defining the boundaries of the ellipses, together with the boundary points of the sample, are used
to generate a Delaunay triangulation of the domain (Schewchuck 1996). Triangular three-noded
finite element mesh is obtained, and, as in the previous case, the mechanical properties of the
matrix and inclusions are then assigned.

Numerical model
Five cm wide and 10 cm high rectangular specimens were modelled. The characteristic
engineering length is 5 cm (the width of the specimen), therefore the maximum and minimum
sizes of the blocks are 0.25 cm and 3.75 cm, respectively. The following assumptions were made:
- The blocks have circular or elliptical shapes.
- The matrix and blocks are described by elastic-perfectly-plastic behaviour,
with a Mohr-Coulomb yield criterion (the same values of the mechanical
parameters used by Lindquist and Goodman (1994) were used; they are
collected in Table 1).
As far as the boundary conditions are concerned, the base of the specimen is fixed and the top
can only move in the vertical direction. In unconfined tests, the axial displacement (or the vertical
load) of the upper edge of the specimen is imposed, while the side edges are free to move. In
confined tests, a constant uniform pressure is applied to the side edges.
 Vol. 17 [2012], Bund. X 3400

Table 1: Mechanical properties of the materials

E = Young’s modulus, ν = Poisson’s ratio, σc = uniaxial compressive strength, σt = tensile
strength, ϕ′= friction angle, c′ = cohesion.

E ν σc σt φ′ c′
[GPa] [−] [MPa] [MPa] [deg] [MPa]
Matrix 2.5 0.25 7.12 0.712 24.7 2.28
Inclusions 5.0 0.25 12.68 1.268 38.3 3.07

FLAC analyses
Samples with circular and elliptical blocks with axis ratio of 2 and random orientation were
examined. The specimens were discretised with quadrilateral, possibly distorted cells (Figure 1)
and analysed in plane strain conditions. Four different BVP were considered for specimens with
circular blocks: 12%, 24%, 38%, and 54%. Three BVP were considered for specimens with
elliptical blocks: 12%, 24%, and 37%. The following tests were simulated for each BVP:
- Three unconfined tests, on specimens with circular and elliptical blocks
- Two confined tests for each lateral pressure (0.5, 1.5, 3 MPa) on specimens with
elliptical blocks
Homogeneous specimens, constituted by only matrix or block material, were also analyzed
for reference and model validation.
The tests were simulated as a loading-controlled process until the axial stress was close to the
peak value, then displacement-controlled conditions were adopted.

Figure 1: Examples of the specimens with circular (top) and elliptical (bottom)
blocks in the FLAC analyses (BVP 12%).
 Vol. 17 [2012], Bund. X 3401

ADINA analyses
FEM analyses were only carried out on specimens with circular blocks. The specimens were
discretised with triangular three-noded elements (Figure 2) in plane strain conditions. Seven
different BVP were chosen: 14%, 22%, 34%, 48%, 61%, 68%, and 72%. The following tests
were simulated for each BVP:
- One unconfined test
- One confined test for each lateral pressure (0.5, 1, and 2 MPa)
Only specimens constituted by matrix or block material were also analyzed in this case. The
tests were simulated in displacement-controlled conditions.

Figure 2. Examples of the specimens with circular blocks in the ADINA analyses
(BVP 22%).

Results
The load-displacement curves were plotted for each numerical test, both in the FLAC and
ADINA analyses (Figure 3). In all cases the maximum load attained during the test was used to
calculate the strength of the specimen. In what follows, the term unconfined strength is adopted to
indicate the value determined in the absence of lateral pressure.
 Vol. 17 [2012], Bund. X 3402

14
12
Deviatoric stress

10
(MPa)

8
6 0.5 MPa
4 1.5 MPa
2 3 MPa
0
0.000 0.002 0.004 0.006 0.008 0.010
Axial strain (-)
Figure 3: Typical stress-strain curves for specimens in confined conditions (FLAC
results).

The analyses carried out using the two numerical methods produced quite comparable results
so that, in most cases, it was possible to interpret all the data together, even though some
differences emerged, as discussed below. As a general remark, it is important to state that the
geometry of the blocks was more accurate in the ADINA model, thanks to the triangular element
mesh. Instead, due to grid limitations, it was only possible to simulate low values of BVP with
the FLAC code.
In unconfined loading conditions, plasticization starts and spreads in the matrix, and tensile
strength is reached in some points along the block-matrix interface. When the applied load is
close to the peak value, yielded points also appear within the blocks (Figure 4).

Figure 4: Example of the evolution of yielding (lighter-grey zones) at 60% (left)

and 95% (right) of the maximum load (ADINA results).
 Vol. 17 [2012], Bund. X 3403

The influence of the BVP on the unconfined compressive strength of bimrock is shown in
Figure 5. It should be noted that different responses are obtained from the two numerical models.
FLAC results are limited to a maximum BVP equal to 37%, due to limitations in the adaptation of
the quadrangular grid to the underlying block population.
Beyond this value, only ADINA results are available, and these show that the compressive
strength increases more rapidly with higher values of the BVP.
The FLAC specimens generally offer an underestimation of the strength, possibly due to the
fact that the mesh is not sufficient refined to take into account the smallest inclusions. The shape
of the blocks does not seem to influence the compressive strength trend.
The tangent deformation modulus, evaluated at 50% of the compressive shear strength,
increases with the BVP, in a linear trend. The trend is always the same irrespective of the block
shape (Figure 6).

15
blocks
Unconfined compressive
strength (MPa)

10
matrix

FLAC - circular blocks

5
FLAC - elliptical blocks
ADINA - circular blocks

0
0 20 40 60 80 100
BVP (%)

Figure 5: Influence of BVP on the unconfined compressive strength of bimrock

specimens.
 Vol. 17 [2012], Bund. X 3404

6
blocks
Deformation modulus 5
(GPa) 4

3
matrix
2 FLAC - circular blocks
1 FLAC - elliptical blocks
ADINA - circular blocks
0
0 20 40 60 80 100
BVP (%)

Figure 6: Influence of BVP on the deformation modulus of bimrock specimens.

The Hoek-Brown strength criterion (Hoek & Brown, 1997) was used to interpret the tests
results obtained for bimrock with circular and elliptic blocks. The results from all the available
tests were considered, and the specimens were grouped into BVP classes in the 0-20%, 20-40%,
40-60%, and 60-80% ranges. The parameters m and σci were calculated for each class by
regression of all the relevant data.
The points in Figure 7 show the stress at failure, while the lines represent the corresponding
Hoek-Brown failure envelope. The results obtained for homogeneous specimens are also
reported. The positive influence of the BVP on the strength of bimrock is made more evident by
plotting the Hoek-Brown parameters m and σci for the four BVP classes (Figure 8). It should be
noted that the values of the same parameters are σci,m = 7.0 MPa, mm = 3.8 and σci,b = 12.0 MPa,
mb = 9.9 for the matrix and block, respectively.
 Vol. 17 [2012], Bund. X 3405

30
matrix
25
<20%
20
σ 1 (MPa)

20-40%
15
40-60%
10
60-80%
5
block

0
0 1 2 3
σ 3 (MPa)

Figure 7: Failure data for all the tests. The lines represent the Hoek-Brown
envelope for the indicated BVP ranges.

12
m Unconfined compressive strength
10
m (-), σ ci (MPa)

0
1 <20% 2 20-40% 3 40-60% 4 60-80%
BVP

Figure 8: Values of the m parameter and of the unconfined compressive strength σci of
the Hoek-Brown criterion.
 Vol. 17 [2012], Bund. X 3406

3D ANALYSES
Generation of the specimens
The set of spherical inclusions incorporated in a cylindrical specimen has been generated by
means of a computer code written in Matlab (The Mathworks, Inc.), which produces a sample
that can be considered as an extraction from a population of spherical inclusions, with given
statistical parameters. The probability density function for the diameter d of the inclusions,
adopted in the paper, is a truncated power law, given by

1− q
f (d ) = − 1+ q 1+ q
dq (1)
a −b

where q = –2.3, and a, b, are the minimum and maximum diameter, respectively. They have been
chosen equal to 5% and 75% of the specimen diameter, according to the description given in the
introduction. According to this assumption, the probability distribution reads as:

a 1+ q − d 1+ q
F (d ) = − (2)
a 1+ q − b 1+ q

Once the input data are assigned, i.e., diameter and height of the specimen (0.10 and 0.05 m
respectively), BVP and BVP tolerance, the inclusions are created inside the cylindrical volume,
enforcing the requirements of non-interpenetration and the attainment of the prescribed BVP.
The algorithm starts with the generation of a sample of spheres, according to the distribution
defined by Eq. (2), up to the given BVP. Then, the position of each sphere is chosen at random
inside the volume of the specimen, repeating the extraction until no inter-penetration with other
spheres, or with the sample boundaries, is detected. More in details, to minimize the number of
rejected trials, the addition of the spheres is made according to their size, from the largest to the
smallest one, and, in order to ensure a convenient separation of the inclusions, which have to be
completely embedded in the matrix material, a slightly enlarged radius is considered when
verifying the possibility of insertion of a new sphere.
The output of the generation procedure, for a given specimen, consists in a list that gives, for
each sphere, the coordinates of the centre and the radius. The format of the file containing these
data is compatible with the requirements of the pre-processor of the finite element code Adina
(ADINA R&D, Inc.), which has been used for the numerical simulations.
The construction of the finite element mesh follows two steps, using the three-dimensional
solid modeller Adina-M:
- Geometry definition: at first, the outer cylinder is generated, according to the
specimen dimensions, and then a set of spherical bodies representing the
inclusions is introduced, subtracting their interior volume from the cylinder, so
that different material properties can be assigned to the domains occupied by
the inclusions and by the matrix, respectively.
- Meshing: the interior of all the spheres is discretised using four-noded
tetrahedral elements, with a size depending on the diameter, to ensure that a
sufficiently fine subdivision is obtained, also for the smallest inclusions.
Afterwards, a certain mesh density is chosen for the outer surface of the
 Vol. 17 [2012], Bund. X 3407

cylinder, and then the matrix volume is subdivided into tetrahedral elements
(Figure 9).

(a) (b)
Figure 9: Specimen generation sequence, involving discretization and meshing of
surfaces and solids (a); final mesh with refining around smaller blocks (b).

The numerical study has been conducted with the objective of reproducing an experimental
campaign aimed at the determination of strength and deformability parameters of bimrock
samples subjected to uniaxial and triaxial compression tests. A set of three specimens for any of
the values BVP = 10%, 20%, 30%, and 40% has been generated. In the following, reference will
be made to a given specimen with a code consisting of a two-digit number, indicating the BVP, a
letter (a, b, c) referring to a particular inclusion placing, followed by other two digits, which refer
to the lateral applied pressure (00, 05, 10, 20, corresponding to 0.0, 0.5, 1.0, and 2.0 MPa,
respectively). For instance, specimen 40b05 is the second of the samples with BVP = 40%, tested
at the confining pressure of 0.5 MPa.
In addition to the specimens made of two materials, in order to assess the validity of the
meshing procedure, also a homogeneous specimen consisting of inclusions made of the same
matrix material has been tested.
It has to be noted that, differently from real experimentation, a numerical test on a given
specimen can be repeated at different confining pressures. For example, the only difference
between tests 30a10 and 30a20 is the lateral pressure, being the set of inclusions in the samples
exactly the same.

Numerical model
In all the numerical tests, the bottom face of the specimen has been considered as fixed, while
the nodes on the upper face are constrained to move in the vertical direction, according to a linear
time history. The final displacement is attained through equal steps, each one corresponding to
the axial strain increment Δε = 0.001.
On the lateral surface, the normal pressure p has been applied, to simulate the confinement in
a triaxial cell. The cases p = 0.0 (uniaxial compression), 0.5, 1.0, and 2.0 MPa have been
considered.
Both materials are described according to the Drucker-Prager elastic perfectly-plastic model,
with associated flow rule. The yield function f is defined as follows:
 Vol. 17 [2012], Bund. X 3408

f = α J1 + J 2D − k (3)

where J1 and J2 are the first stress invariant and the second deviatoric stress invariant,
respectively, α and k are the Drucker-Prager parameters, which can be related to the Mohr-
Coulomb parameters, c′ and ϕ′ by the expressions, valid in case of compression failure (Bathe,
1996, p. 604):

2 sin φ ′ 6 c ′ cos φ ′
α= , k= (4)
3 (3 − sin φ ′) 3 (3 − sin φ ′)

The same values of the mechanical parameters used in 2D analyses were assumed (Table 1).

Results
The numerical testing campaign has been performed on 12 specimens, each one subjected to
4 levels of lateral pressure, obtaining the stress-strain curves up to failure. In all cases, at a certain
value of the imposed axial displacement, the numerical procedure did not converge anymore.
Nevertheless, the state of the specimen at the last converged step has been considered as
representative of the ultimate conditions, and the corresponding deviatoric stress has been
considered in the subsequent analyses as the failure stress. As an example, Figure 10 shows, with
a darker shade of grey, the yielded zones at failure for 4 different BVPs.
By summing the vertical nodal reactions calculated at the base of the specimen, for each
loading step, the load-displacement curves can be plotted. For example, Figure 11 shows the
deviatoric stress σa –σr vs. the axial strain εa, for the cases of specimens 10c, 20b, 30b and 40a.
It can be observed that, for low BVPs (e.g., 10%, Figure 11 (a)), the strength of the samples is
similar to the matrix one (see σc value in Table 1) and yielding is rather uniform (Figures 10 (a)
and 10 (b)). In general it has been noted that, given a certain BVP, the stress distribution in the
specimen with the presence of a few large blocks may change the localisation of the yielded
zones, as they may alter significantly the stress distribution in the specimens.
Furthermore, greater BVP induces higher strength (Figures 11 (b)-(d)). From the results
obtained, one can infer that between BVP = 10% and BVP = 20% there is the threshold above
which the influence of blocks becomes significant. Above this value, yielding starts around the
surface of blocks, which initially carry a significant amount of the applied load, then it propagates
to the whole specimen. At the end of the test, the blocks are almost completely plastic and only
part of the matrix shows an elastic behaviour (Figure 12).
It is interesting to note the presence of yielding bands on the lateral surface of the specimen
(Figure 13).
The influence of the blocks may be summarized as follows:
- BVP being equal, a uniform arrangement of the blocks induces a higher strength
- The onset of yielding zones is more likely in thin sections of matrix isolated by
blocks located near to the boundary of the specimen
- Large blocks alter significantly the stress distribution.
- The influence of the BVP on the unconfined compressive strength σc of the
specimens is shown in Figure 14 (a). It should be noted that the results of different
 Vol. 17 [2012], Bund. X 3409

specimens characterized by the same BVP show a good reproducibility. The

unconfined compressive strength of the matrix, σc, is obtained also for the
specimens with BVP = 10%. For increasing BVP, σc increases according to a
linear trend up to he maximum BVP.
A similar behaviour has been pointed out for deformability. The tangent deformation
modulus, evaluated at 50% of the compressive strength of the specimen, increases with the BVP,
with a linear trend (Figure 14 (b)).

(a) (b)

(c) (d)
Figure 10: Yielded zones at failure (dark grey) for specimens 10a20 (a), 20b20 (b), 30c20
(c) and 40a20 (d).
 Vol. 17 [2012], Bund. X 3410

14
14
12
12
10
σa-σr (MPa)

10

σa-σr (MPa)
8
8
6 10c00
6 20b00
10c05
4 20b05
10c10 4
20b10
2 10c20 2 20b20
0 0
0.000 0.002 0.004 0.006 0.008 0.010 0.000 0.002 0.004 0.006 0.008 0.010
εa (-) εa (-)

(a) (b)

14 14
12 12
10 10
σa-σr (MPa)

σa-σr (MPa)
8 8
30b00
6 6 40a00
30b05
4 40a05
4
30b10
40a10
2 30b20 2
40a20
0 0
0.000 0.002 0.004 0.006 0.008 0.010 0.000 0.002 0.004 0.006 0.008 0.010
εa (-) εa (-)

(c) (d)
Figure 11: Typical stress-strain curves for specimens in unconfined and
confined conditions: 10c (a), 20b (b), 30b (c), and 40a (d) specimens.

Figure 12: Onset and propagation of yielding up to failure for the specimen 40c20.
 Vol. 17 [2012], Bund. X 3411

Figure 13: Yielding zones on the lateral surface of the specimen 40c.

14 blocks
12

10
σc (MPa)

8 matrix

4 Specimen a
Specimen b
2 Specimen c

0
0 10 20 30 40 50 60
BVP (%)

blocks
5000

4000
E (MPa)

3000
matrix

2000

1000

0
0 10 20 30 40 50 60
BVP (%)

(a) (b)
Figure 14: Influence of the BVP on the unconfined compressive strength (a),
and on the average deformation modulus (b) of bimrock specimens.

The Hoek-Brown strength criterion (Hoek and Brown, 1997) has been used to interpret the
results of the numerical analyses. The data from all the available tests have been considered, and
the parameters m and σci have been calculated for each BVP by regression. Figure 15 (a) shows
the clear influence of the BVP; the dashed and dash-dot curves represent the envelopes of matrix
and block, respectively, while the continuous curves represent the limit curves for the numerical
specimens. The Hoek-Brown failure envelopes are in the range characterised by the limit curves
 Vol. 17 [2012], Bund. X 3412

and a higher BVP corresponds to a higher strength. The specimens with BVP = 10% show a
strength envelope practically coincident with the matrix one.
The same data have been interpreted according to the Mohr-Colulomb failure criterion
(Figure 15 (b)) by linearization of the Hoek & Brown curves in the range 0.0-2.0 MPa. In the
same figure, the dashed and dash-dot curves refer to the input values c′, ϕ′ given in Table 1.

25
matrix
block
20 40%
30%
σ 1 (MPa)

15

10 20% 10%

0
0.0 0.5 1.0 1.5 2.0 2.5 3.0
σ 3 (M Pa)

6
matrix
5 block
40%
4 30%
τ (MPa)

3
20% 10%
2

0
0.0 0.5 1.0 1.5 2.0 2.5 3.0
σ (MPa)

Figure 15: Influence of the BVP on strength properties: Hoek-Brown (a) and
Mohr-Coulomb (b) failure criteria.

CONCLUSIONS
The results of the numerical modelling of compressive tests on bimrock specimens
highlighted some interesting aspects concerning the behaviour of the material:
- The compressive strength increases with the BVP, with a more marked effect for higher BVP
values.
- Yielding starts and spreads in the matrix, and tensile strength is reached in some points along
the block-matrix interface. When the applied load is close to the peak value, yielded points
also appear within the blocks.
- In confined conditions, the blocks are less affected by yielding, and the matrix also shows
tensile failure along the block-matrix interface.
- The bimrock deformation modulus is proportional to the BVP, according to a simple rule of
mixtures.
- The analyses on bimrock with elliptical blocks provided results that are very similar to those
 Vol. 17 [2012], Bund. X 3413

obtained for circular blocks. It should be noticed that, due to limitations of the generation
process, only low values of BVP were considered in the case of elliptical blocks. Therefore,
the interpretation of the behaviour of simulated bimrock with a high volumetric proportion of
elliptical blocks is not reliable.
- Comparing the results of the analyses performed with FLAC and ADINA codes, it is possible
to observe that:
- FLAC allows both the pre and post peak behaviour of bimrock specimens to be easily
modelled, but the quadrangular mesh, with almost equal cells, precludes a good
representation of the block geometry.
- ADINA can use a triangular mesh, therefore the shape of the blocks, even the smallest ones,
is well reproduced. However, some difficulties arise when the post peak behaviour, with large
plastic strains, has to be modelled.
- Furthermore, the results obtained from the 3D analyses allow to assert that:
- For low BVP (e.g., 10%), the mechanical response of the samples is similar to the behaviour
of the matrix alone and yielding is rather uniform, even if the particular arrangement of the
blocks may change the localisation of the yielded zones.
- It seems that BVP = 20% is a threshold value over which the influence of the blocks becomes
significant.
- For increasing BVP, the compressive strength of the bimrock increases in a linear trend. This
observation is valid for BVP up to 40% because of the lack of data referring to higher BVPs.
- A similar behaviour has been pointed out for deformability. The tangent deformation
modulus, evaluated at 50% of the compressive shear strength, increases with the BVP, in a
linear trend after the threshold.
The bimrock strength characteristics have been interpreted according to the Hoek-Brown and
Mohr-Coulomb criteria. In both cases the envelopes lie between those of the matrix and the
blocks and a higher strength corresponds to a higher BVP.
Both 2D and 3D results show that the response of bimrock to compression seems to be deeply
influenced by the behaviour of the material at the interfaces between the matrix and the blocks.
For this reason appropriate contact elements should be introduced into the model to simulate the
interaction, which could allow for discontinuities of the displacement field around the blocks.
Furthermore, as the bimrock response to compressive load history is very complex, a better
understanding of the problem could be possible by comparing numerical results with
experimental ones. On the basis of this remark an experimental campaign of laboratory tests on
artificial bimrock samples has been carried out and the comparison with numerical results is
under study.

REFERENCES
1. Barbero M. and Barla G. (2010) “Stability Analysis of a Rock Column in Seismic
Conditions” Rock Mechanics and Rock Engineering (43), 845-855. ISSN: 0723-2632, DOI:
10.1007/s00603-010-0097-2
2. Barbero M. and Barpi F. (2012) “Geomechanical modeling to study the effects of slope
instability on buildings: a case study in northern Italy” GEAM Geoingegneria Ambientale e
Mineraria (2), 5-18. Pàtron, ISSN: 1121-9041
 Vol. 17 [2012], Bund. X 3414

3. Barbero M., Bonini M. and Borri-Brunetto M. (2007) “Numerical modelling of the

mechanical behaviour of bimrock.” Proc. 11th ISRM Int. Congress, 377-380. Taylor &
Francis Ltd.
4. Barbero M., Borri-Brunetto M. and Bonini M. (2008) “Three-dimensional finite element
simulations of compression tests on bimrock” In: Geomechanics in the Emerging Social and
Technological Age, 12th Int. Conf. of the Int. Ass. for Computer Methods and Advances in
Geomechanics (IACMAG 12). Goa, India 1-6 October 2008
5. Barpi F., Barbero M. and Peila D. (2011) “Numerical modelling of ground-tunnel support
interaction using bedded-beam-spring model with fuzzy parameters.” Gospodarka Surowcami
Mineralnymi, Polish Academy of Sciences. Committee of Mineral Economy Sustainable
Development Istytut Gospodarki surowcami mineralnymi i energia, (27), ISSN: 0860-0953
6. Bathe K.J. (1996) “Finite Element Procedures.” Prentice-Hall, Englewood Cliffs, NJ, USA.
7. Bonini M. and Barla G. (2012) “The Saint-Martin-La-Porte Access Adit (Lyon - Turin Base
Tunnel) Revisited” Tunnelling and Underground Space Technology (30), 38-54.
8. Bonini M., Lancellotta G. and Barla G. (2012) “State of stress in tunnel lining in squeezing
rock conditions” Rock Mechanics and Rock Engineering, OnLine First, DOI
10.1007/s00603-012-0326-y.
9. Donev A., Torquato S. and Stillinger F. H. (2005) “Neighbor list collision-driven molecular
dynamics simulation for nonspherical hard particles: II. applications to ellipses and
ellipsoids.” Journal of Computational Physics 202 (2), 765-793.
10. Hoek E. and Brown E.T. (1997) “Practical estimates of rock mass strength.” International
Journal of Rock Mechanics and Mining Sciences 34 (8), 1165-1186.
11. Lindquist E.S. and Goodman R.E. (1994) “The strength and deformation properties of a
physical model melange.” In P.P Nelson, P.P. & S.E. Laubach (eds), Proc. 1st North
American Rock Mechanics Conference (NARMS). Rotterdam, Balkema.
12. Lindquist E.S. (1994) “The mechanical properties of a physical model melange.” Proceedings
of the 7th Congress of the International Association of Engineering Geologists, Lisbon.
13. Medley E.W. (1994) “Engineering characterization of melanges and similar block-in-matrix
rocks (bimrocks).” PhD Dissertation, Dept. Civil Engineering, Univ. California at Berkeley.
UMI Inc., Ann Arbor (USA).
14. Medley E.W. (2002) “Estimating block size distribution of melanges and similar block-in-
matrix rocks (bimrocks).” In R. Hammah, W. Bawden, J. Curran & M. Telesnicki (eds), Proc.
5th North American Rock Mechanics Symposium (NARMS), 599-606. Toronto, University
of Toronto Press.
15. Shewchuk J.R. (1996) “Triangle: Engineering a 2D quality mesh generator and Delaunay
triangulator.” In M.C. Lin & D. Manocha (eds), Applied Computational Geometry: Towards
Geometric Engineering, Lecture Notes in Computer Sciences 1148, 203-222. Berlin,
Springer-Verlag.
16. Sönmez H., Gokceoglu C., Tuncay E., Medley E.W., Nefeslioglu H. A. (2004) “Relationships
between Volumetric Block Proportions and Overall UCS of a Volcanic Bimrock.” Felsbau
22(5), 27-34.

© 2012 ejge