Professional Documents
Culture Documents
Abstract
The intersection of discontinuities in a jointed rockmass creates in situ blocks of variable three-dimensional (3D) geometry. The size
and shape of rock blocks in a rockmass assembly have a dominant influence on the engineering properties of a rockmass, and control, for
example the failure geometry of a rock face and the optimum associated support and surface restraint strategy. The distribution of block
volumes within a rockmass can be effectively analyzed and communicated using a cumulative volume distribution curve. Integrated with
this analysis, a methodology has been developed to characterize the shape and volumetric shape distribution of rock blocks in 3D models
of jointed rockmasses. The proposed Block Shape Characterization Method takes into account two factors: the first, a, describes the
shortening of the minor principal axis of the block while b describes the elongation of the major axis. When merged with the block
volume distribution, the Block Shape Characterization Method can effectively describe and classify both the size and shape distributions
of any jointed rockmass and is not limited to orthogonal blocks or hexahedra. This methodology can classify any polyhedra and has been
developed and calibrated based on synthetic joint data and simulated block assemblies. A demonstration is given of its effectiveness in
characterizing block geometries in the field using mapped fracture data.
r 2006 Elsevier Ltd. All rights reserved.
1365-1609/$ - see front matter r 2006 Elsevier Ltd. All rights reserved.
doi:10.1016/j.ijrmms.2006.04.004
ARTICLE IN PRESS
K.S. Kalenchuk et al. / International Journal of Rock Mechanics & Mining Sciences 43 (2006) 1212–1225 1213
ore in a cave mining operation. The geometry of the block tool that characterizes block sizes and shapes within
controls the mechanisms of instability leading to dilution. rockmasses. The method is developed and calibrated using
The size and shape of a waste block may significantly three-dimensional (3D) simulations and tested on field
influence the migration of this dilution to a drawpoint. outcrops.
Fig. 1 shows an example of dilution mode controlled by
block shape. The blocks below point A are elongated 2. Model development
vertically while the blocks below point B are platy. The
difference is due to a difference, from one side of a The joint models used in this study have been generated
geological contact to the other, in joint spacing on one of using the discrete element modeling software 3 Dimen-
several joint sets. Work is ongoing to understand the role sional Distinct Element Code (3DEC) developed by Itasca
of block size and shape in controlling the wall dilution [6]. A jointed rockmass is simulated by generating a joint
mode and migration behavior of waste blocks within the pattern that is statistically based on joint spacing,
broken ore volume during muck draw in this setting. To orientation and persistence. It should be noted that the
this end, this research paper describes the development of a techniques used and methodology introduced in this study
Fig. 1. View of an open pit above a caving ore zone with a confluence of faults and a geological contact intersecting in the lower center of the image. Bench
height is 15 m. Lower image shows details of wall rock highlighting block shape differences between zone A (elongated) to the left of the intersection region
and zone B (platy) to the right.
ARTICLE IN PRESS
1214 K.S. Kalenchuk et al. / International Journal of Rock Mechanics & Mining Sciences 43 (2006) 1212–1225
of rock blocks for practical engineering purposes. Of these, schemes. Zingg [22] introduced a Cartesian coordinate
Franklin and Dusseault [15] describe quantitative aspect system using S=I an I=L ratios. Sneed and Folk [23] use a
ratio information using terminology such as cubic, slabby, triangular diagram to plot a disk:rod index (DRI ¼ (L1)/
prismatic or columnar. For instance, a rock block with (LS)) and S=L. Palmstrom [18] also proposed a graphical
aspect 1:5:7 would be slabby. Sen and Eissa [16] relate Jv, division of block shape domains. Fig. 3 presents a
RQD and block volumes for idealize block shapes such as summary of several previously proposed shape diagrams.
cubes, plates or bars. Smith [17] discusses the representa- The Zingg diagram has four shape fields designated as
tion of block size and shape on a stereograph highlighting spherical, flat, elongated-flat and elongated (corresponding
the dihedral angle versus spacing for sequential joint pairs. to sphere, disc, blade and rod, respectively). Data displayed
Palmström [18] suggested a division of block shapes based on the Zingg diagram is generally not effectively distrib-
on the ratio of short, medium and long edge lengths for an uted throughout the plottable space and is problematic for
orthogonal hexahedral block. many classification applications [21,23,24]. This ineffective
Rock block shape can also be described in terms of the use of the entire grid and the use of a fourth and arbitrary
order of block shape, defined by the number of surfaces on shape class (blade is not a unique end-member shape)
a block. For example, shapes such as tetrahedral, reflect the need for an alternative to the Cartesian
pentahedral, hexahedral or heptahedral. This block order coordinate system as an effective means to represent shape
can be used to asses the stability of an exposed rock face; data. Triangular diagrams, as generated through different
higher order blocks are generally less removable because parametric means by [23–25], appear to be a more
they are less likely to fully form and more likely to be stable appropriate method for presentation of particle shape.
since non-parallel translation on shear surfaces and minor Particles plot on a continuum between equidimensional,
non-convexities aid stability [3]. planar and elongated shapes that mark the corners of a
Wang [19] developed a 3D block shape index, l, to triangular diagram.
describe the shape of ore fragments. This block shape index Despite the numerous methods proposed for shape
is the ratio of block volume to the volume of a sphere classification, no single method has been widely accepted.
whose diameter equals the maximum block size: All of these methods to classify rock block shape, with the
6V exception of Smith [17], rely on the assumption that a block
l¼ , (1) is approximately orthogonal, an assumption that is
pðl max Þ3 acceptable in only a few rockmasses. Typical block shapes
where V is the block volume, and lmax is the greatest are tetrahedral [26] or polyhedral [4]. The method proposed
distance between two vertex points on the block. Wang
then classifies the shape of blocks into five categories
according to 3D shape index value: bar (o0.001), plate
(0.001–0.077), block (0.007–0.22), cube (0.22–0.37) and
sphere (0.37–1.0).
Palmström [20] introduced a shape factor b used to
determine the equivalent block volume in rockmasses with
few joint sets that do not form discrete blocks. There are
various equations that can be used to calculate this block
shape factor, such as
b ¼ 20 þ 7ðSmax =S min Þ, (2)
by Smith [17] provides a good indication of rock block The factor in Eq. (4) is related to a power function of the
shape with a description of size based on joint spacing; average inter-vertex direction cosine (dot product divided
however, this method relies heavily on manual stereonet by product of vector lengths). The squaring of terms within
techniques and is difficult to automate. The proposed the summations results in a weighting in favor of longer
Block Shape Characterization Method is appropriate for chord inter-relationships while the external squaring is a
determining in situ block shapes in 3D modeling applica- scaling operation for graphical representation. The b value
tions, and can be applied to blocks of any shape, from for the example block in Fig. 4 is 7.3.
perfectly orthogonal to highly irregular. A triangular For elongated shapes, the longest vertex-to-vertex
diagram will be used as it is more appropriate to display coordinates are nearly parallel returning a maximum b
shape data with three end-member shapes. value of 10. In a blocky equidimensional object, the longest
inter-vertex contacts are distributed in three dimensions;
5. The Block Shape Characterization Method for a perfect cube, b is equal to 0.82. In a flat planar object
(i.e., a slab), the longest vertex-to-vertex contacts are co-
The Block Shape Characterization Method takes into planar, returning a range of b values and therefore a
account the co-linearity of the longest vertex-to-vertex second function is needed to differentiate between ‘‘cubes’’
distances, as well as a unit-less relationship between surface and ‘‘slabs’’.
area, average chord length (inter-vertex distance) and
volume. The application of this methodology to 3D 5.2. Relating surface area and volume
stochastic modeling is easy to automate.
Slabs have a higher surface area to volume ratio than
cubes. A relationship between the surface area and volume
5.1. Inter-vertex co-linearity
of an arbitrary object has been defined as
The first stage of the Block Shape Characterization As l avg
a¼ , (5)
Method is to quantify the average angle between vectors 7:7V
defined by the longest inter-vertex dimensions (chord which is a dimensionless parameter that relates block
lengths) for a rock block (including consideration of all volume, V, surface area, As, and the average chord length,
edges, face diagonals and internal diagonals). Angular lavg. A numerical factor of 7.7 is used to normalize a to a
relationships between the shortest of chords within a block value of 1 for a cube.
are not indicative of the block shape while long chords and The a parameter reflects the flatness of an object. A
the angular relationships between them can be used to perfect cube has a ¼ 1, as the aspect ratio of an object
determine whether the block is rod like (elongated). Once increases the a value can approach infinite bounds. It is
all chord lengths are calculated, the median chord length is possible to achieve ao1;a values for equidimensional
determined and all orientation vectors with chord lengths polyhedra with more than 6 sides fall between 0.52 and
less than the median are disregarded. For the example 1. A perfect sphere yields a ¼ 0:52, as the average chord
block in Fig. 4, the median unit chord length is 17.8 units. length of a sphere is 4r/3 [27]. If the a value was normalized
The remaining chords are used to generate the factor for a sphere, the denominator constant would have to be 4.
P 2 For comparison, the normalizing denominator for a
ða bÞ2
b ¼ 10 P . (4) tetrahedron would be 12O(3/2). However, in terms of rock
jjajj2 jjbjj2 block shape, a sphere would be an extreme case, and for
engineering simplicity, the cube was selected as the limiting
shape for practical application. For the simple block
example in Fig. 4, a is equal to 5.
The triplets of values for spacing refer to the examples in Figs. 9, 10 and 11
respectively.
sets of the same orientations, varying only the spacing tion of that rockmass. This is to ensure that distribution of
parameters in order to simulate jointed rockmasses block geometry for blocks defined by discrete joints is not
dominated by elongated, slabby and equidimensional block skewed by the outer blocks truncated by the model bounds.
shapes. The orientation and spacing parameters are Fig. 9 illustrates a rockmass model dominated by
summarized in Table 2. Once again it should be noted elongated block shapes. The histogram indicates that as
that no rock blocks touching the outer boundaries of any volume increases, there is less variance in block shape. If
model are incorporated in the size and shape characteriza- the spacing of joint set 3 was increased relative to the
Fig. 9. Illustration of a simulated rockmass dominated by elongated bar- Fig. 10. Illustration of a simulated platy rockmass. While most blocks
shaped rock blocks. The upper portion of the Block Shape Diagram is cluster in the lower left-hand portion of the Block Shape Diagram, there is
densely clustered with data points. The histogram of block shape distribution a slight trend towards elongated shapes, these blocks are generated due to
within each block volume bin indicates that smaller blocks occur in all shape variances in dip and dip direction of individual joints, these blocks fall
types, while larger blocks are completely dominated by elongated shapes. primarily in the smallest block volume bin.
ARTICLE IN PRESS
1220 K.S. Kalenchuk et al. / International Journal of Rock Mechanics & Mining Sciences 43 (2006) 1212–1225
spacing of joint sets 1 and 2, the aspect ratio of the rock plane of the joint orientation, and the larger blocks would
blocks would increase forcing the larger blocks even closer cluster closer to the lower right-hand portion of the Block
to the upper corner of the Block Shape Diagram. Shape Diagram.
The rockmass illustrated in Fig. 10 is dominated by platy Fig. 11 shows a rockmass with three equally spaced joint
shapes with some blocks trending towards more elongated sets. The smallest blocks range between all shapes and the
shapes. The elongated blocks are generally smaller than the larger blocks are primarily equidimensional-blocky. This
platy blocks due to the intersection of non-parallel joints. variation in shape between different block volumes once
Increasing the spacing of joint sets 2 and 3 relative to joint again illustrates the influence that variability in joint
set 1, the aspect ratio of the blocks would increase in the orientation has on rock block geometry.
6. Influence of persistence on block shape and volume on its accessibility and the quality of joint exposure. A total
of 114 joints were mapped at Site 1, then statistically
The persistence of joints strongly influences the size and analyzed using DIPS [29] and a symmetrically normal
shape of in situ rock blocks, as well as the likelihood of distribution about the mean pole orientation [10]. Fig. 13
discrete blocks even forming [28]. To illustrate the shows a stereographic projection of the sampled data, the
application of the Block Shape Characterization Method, mean joint set planes and the distribution envelopes used
a brief case study compares the influence of joint for stochastic modeling. Table 3 summarizes the joint
persistence on block volume and shape distributions. Using orientation and spacing data for Site 1.
the same joint orientation and spacing data as the example For the 3DEC simulation of the rockmass joint
model illustrated in Fig. 2, Fig. 12 shows the size and shape orientation, data were assigned a normal distribution with
distributions for persistence values that are not equivalent maximum deviation truncated at a variability limit of
for all joint sets. Two of the joint sets have persistence 68.28%. Unlike the artificial rockmass described earlier,
values of 1.0 while the third has a persistence equal to 0.5. the joint spacing data for Site 1 was not negative
For this model, the D40 and D60 volumes increase slightly exponentially distributed. In this case, the joint spacing
and the block shapes become more elongated (compare data are better fit to a normal distribution. This distribu-
Fig. 12 to Fig. 8). This change in block geometry reflects tion for joint spacing was also used by Sen and Eissa [16].
the impact that joint persistence has on the aspect ratio of The best-fit normal distribution was truncated at the
the resulting rock blocks. As a joint set becomes less recorded minimum and maximum values as in Fig. 14. All
persistent, the aspect ratio of rock blocks increases in the joints have been assigned a persistence value of 1.0 for this
direction normal to that joint set orientation, the block study as most joint traces mapped were continuous over
volumes also increase accordingly. the mapping surface.
Fig. 15. Block Shape Diagram for Site 1. The size and shape distributions
Fig. 14. Truncated normal distributions of joint set spacing for Site 1. of Site 1 are confirmed by visual observation, Fig. 16.
ARTICLE IN PRESS
K.S. Kalenchuk et al. / International Journal of Rock Mechanics & Mining Sciences 43 (2006) 1212–1225 1223
Fig. 16. (a) Upper image pair shows site 1 of test case. (b) Lower image shows rock face detail in Site 2.
Acknowledgments
References
[14] Peaker SM. Development of a simple Block Size Distribution Model [22] Zingg T. Beitrag zur Schotteranalyse. Schweiz Mineral Petrol Mitt
for the classification of rock masses. MSc thesis, Department of Civil 1935;15:39–140.
Engineering, University of Toronto, Ontario, Canada, 1990. [23] Sneed ED, Folk RL. Pebbles in the lower Colorado River, Texas, A
[15] Franklin JA, Dusseault MB. Rock engineering. New York: McGraw- study of particle morphogenesis. J Geol 1958;66:114–50.
Hill; 1989. [24] Hofmann HJ. Grain-shape indices and isometric graphs. J Sediment
[16] Sen Z, Eissa E. Rock quality charts for log-normally distributed Res 1994;A64:916–20.
block sizes. Int J Rock Mech Min Sci 1992;29:1–12. [25] Graham DJ, Midgley NG. Graphical representation of particle shape
[17] Smith JV. Determining the size and shape of blocks from linear using triangular diagrams: an Excel sreadsheet method. Earth Surf
sampling for geotechnical rock mass classification and assessment. Proc Landforms 2000;25:1473–7.
J Struct Geol 2004;26:1317–39. [26] Grenon M, Hadjigeorgiou J. Drift reinforcement design based on
[18] Palmström AA. Measurement and characterization of rock mass discontinuity network modeling. Int J Rock Mech Min Sci
jointing. In: Sharma VM, Saxena KR, editors. In-situ characteriza- 2003;40:833–45.
tion of rocks. Rotterdam: Balkema; 2001. p. 49–97. [27] Jokisch DW, Patton OW, Pajon DA, Inglis BA, Bolch WE. Chord
[19] Wang L, Yamashita F, Sugimoto F, Tan G. A methodology for distributions across 3D digital images of a human thoracic vertebra.
predicting the in situ size and shape distribution for rock blocks. Med Phys 2001;28:1493–504.
Rock Mech Rock Eng 2003;36:121–41. [28] Grenon M, Hadjigeorgiou J. Open stope stability using 3D joint
[20] Palmström A. Rmi—a rock mass characterization system for rock networks. Rock Mech Rock Eng 2003;36:183–208.
engineering purposes. PhD thesis, Oslo University, Oslo, 1995. [29] Rocscience. DIPS: geographical & statistical analysis of orientation
[21] Oakey RJ, Green M, Carling PA, Lee MWE, Sear DA, Warburton J. data. Geotechnical analysis software, version 5.1. Rocscience; 2004.
Grain shape analysis—a new method for determining representative [30] Zimmerman RW, Bodvarsson GS. Effective block size for imbibition
particle shapes for populations of natural grains. J Sediment Res or absorption in dual-porosity media. Geophys Res Lett 1995;22:
2005;75:1065–73. 1461–4.