Sei sulla pagina 1di 16

A.H. WILSON, C.A. JERMY, M. RIDGEWAY AND G.

CHUNNETT 525

Rock-strength and physical properties of


Norites of the Merensky and Bastard Units,
western Bushveld Complex
A.H. Wilson, C.A. Jermy and M. Ridgeway
School of Geological Sciences, University of KwaZulu-Natal, Durban, South Africa, 4041
e-mail: wilsona@ukzn.ac.za; jermy@ukzn.ac.za; ridgeway@ukzn.ac.za;

G. Chunnett
School of Geological Sciences, University of KwaZulu-Natal, Durban, South Africa, 4041
Present address: Anglo American Platinum Corporation P.O. Box 62179,
Johannesburg, South Africa
e-mail: gchunnett@angloplat.com
© 2005 December Geological Society of South Africa

ABSTRACT
Platinum mining in the Bushveld Complex has expanded greatly over the last decade with many operations planned for, or already
at depths greater than 1000 m. Mine infrastructures, rock fragmentation properties, rock failures and safety issues are important
aspects of deeper mining operations and require an understanding of controls and properties which influence rock strength. This
study, which includes rock classification of both fresh and altered material, chemical compositions, fabrics and structure, was carried
out to investigate mainly the noritic rocks in the vicinity of the Merensky reef in the western Bushveld Complex.
Uniaxial compressive strength (UCS) and tensile strength (TS) are shown to be broadly dependent on rock type, and strongly
dependent on the degree of structural alteration as reflected by the Rock Quality Designation (RQD). In addition, there is an
unexpectedly strong correlation between UCS values and both normative quartz and Zr contents of the whole rock. Both of these
compositional parameters relate to the entrapment of a late-stage liquid which when solidified acted as cement and influenced the
propagation of fractures under conditions of failure. Micrographic observations indicate that interstitial quartz, even in relatively
small amounts, is effective in arresting fracture propagation. Relatively small variations in the quartz content could therefore
significantly affect the rock strength of the norites. Other possible controls on rock strength include the relative development of
foliation fabric, the distribution of crystal sizes, and the way that the rocks have annealed at high temperature.

Introduction measurements with quantification of textures and


The Bushveld Complex is the world’s largest layered chemical compositions of the same samples.
intrusion and has been mined for nearly 100 years for a
wide variety of metal commodities. It is the single largest The Bushveld Complex
resource and producer of platinum and chromium. The Bushveld Complex is the largest mafic-ultramafic
While most of the mineable resources for the past layered intrusion on Earth covering an area of
50 years have been close to the surface, the scale of approximately 67 000 km2 and has a broadly oval shape
mining operations in the early part of the 21st Century extending some 375 km in an east-west direction and
has necessitated consideration of developing mines to 300 km in a north-south direction. The intrusion
depths of greater than 1000 m below surface. comprises four lobes which dip inward towards each
The development of mine infrastructures, efficiency of other forming an elongated deep basin structure
rock breaking, rock failures and safety aspects are reflecting its emplacement into the Transvaal
important considerations in these deeper mines. The full Supergroup sedimentary basin at 2058.9 ± 0.8 Ma (Buick
assessment of these parameters requires an under- et al, 2001). The mafic and ultramafic rocks of the
standing of rock-strength and rock integrity of the Bushveld Complex (Rustenburg Layered Suite) form a
stratigraphic sequences which enclose the ore zones. succession more than 10 000 m thick (Eales and
This study relates to that part of the stratigraphy Cawthorn, 1996) which is made up of four major
overlying the Merensky reef, and in particular to the stratigraphic components, these being from the base
norites which comprise the hanging-wall to the reef. upwards: the Lower Zone, Critical Zone, Main Zone and
While these rock types generally fall into the category of Upper Zone. The Critical Zone is economically the most
‘strong rock’, strength ranges and physical attributes important and also represents the transition from
which impact on geotechnical properties have not ultramafic rocks (harzburgites and pyroxenites) to mafic
previously been documented. rocks (norites and gabbronorites).
The aim of the study is to investigate the systematic The Merensky reef, one of several major economic
variation of rock strengths of fresh norites immediately horizons in the Bushveld Complex mined for the
above the Merensky reef and to integrate these platinum group elements (PGE) and base metals, is part

SOUTH AFRICAN JOURNAL OF GEOLOGY, 2005, VOLUME 108 PAGE 525-540


526 ROCK-STRENGTH AND PHYSICAL PROPERTIES OF NORITES OF THE MERENSKY
AND BASTARD UNITS

Figure 1. Geological map of the western Bushveld Complex indicating the area in which this study was carried out and the major
stratigraphic zones in relation to the Merensky and Bastard Units. The heavy dashed line is the outcrop of the Merensky reef. Detailed
stratigraphy for core GC2 is shown together with the legend. Abbreviation: MP - Merensky pyroxenite; BP - Bastard pyroxenite. Other drill
holes used in this study are GC7 and GC8.

of a series of cyclic units located close to the Critical Bushveld norites, can be defined as planes of weakness
Zone – Main Zone boundary. The geographic location of within confined regions of the rock. These are normally
the study area and stratigraphy are shown in Figure 1. of a shear type and can be either continuous or
Mining operations on the Merensky reef have expanded discontinuous under differing stress conditions. Crack
greatly since 1994 with seven new mines and projects propagation by failure caused by stress concentrations at
being brought into, or approaching, production. At the the ends of minute cracks (Griffith, 1921) has long been
same time, existing mines have developed operations to recognized as important in rock failure and in fluid
greater depths. Extensive mining of the Merensky reef, induced stress. In most igneous rocks, stresses are
as well as development in the hanging-wall, requires induced primarily as a result of internal strain during
understanding of the controls of rock strength of the cooling. Failure in rocks is considered to take place
norites which overlie the reef. Many parameters initially along crystal boundaries (Brace, 1961; Hoek,
influence rock strength and the manner in which rocks 1965) but crack propogation will penetrate crystals at the
fail. This paper reports on an investigation into the point of failure. In the context of rock strength, most
influences of rock composition, textures (including igneous rocks are regarded as being relatively isotropic
mineral orientation and crystal size), on the geotechnical (in contrast to sedimentary and metamorphic rocks
properties of rocks in the vicinity of the Merensky reef which are usually anisotropic) and any deviation from
and in particular, the hanging-wall norites where that ideal situation will have an important influence on
extensive mining development takes place. rock strength. In the simplest case of planar anisotropy,
the rock will have a set of parallel planes of weakness,
Mechanisms of Failure and Fracture in Brittle Rocks the strengths of which are related to the coefficient of
Cracks in strong rocks, which as a category include the internal friction, which can be expressed quantitatively

SOUTH AFRICAN JOURNAL OF GEOLOGY


A.H. WILSON, C.A. JERMY, M. RIDGEWAY AND G. CHUNNETT 527

Figure 2. 60 mm diameter core which has undergone UCS testing. Core samples ideally fail along a plane of less than 45o to the vertical
axis of compression. Failure types observed in the tests are: (a) Failure along a single plane; (b) failure along a single plane together with
high angle longitudinal fractures; (c) failure along two fractures symmetrically inclined to the longitudinal axis.

(Jaeger and Cooke, 1977). Planes of failure may also Young’s modulus and the average crystal surface area
intersect the principal planes of weakness, a situation per unit volume (Onodera and Kumara, 1980). However,
which can readily be expressed from the Mohr circle as pointed out by Bell (2000), defects superimposed on
diagram. Crack propagation during failure of anisotropic the coherent internal structure of a rock also have a
rocks may take place in both the principal plane of strong effect on its ultimate strength, and the strength of
weakness and also in other sets which may be randomly some crystalline rocks is inversely proportional to the
orientated. frequency of defects (Willard and McWilliams, 1969).
The basic assumption of the initial stages of rock Furthermore, many rocks have a primary fabric (as
failure is that the condition is attained in which the opposed to defect planes), which in layered mafic
maximum tensile strength around the most critical crack igneous rocks can be of several types. Materials such as
reaches some critical value. It is likely that this situation sandstones, slates and marbles have compressive
arises by the interaction of several small cracks in the strengths which are highly directional (Donath, 1961;
form of an array. The crack propagates further with an Griggs, 1951). The strength of rock is also dependent on
increase in the applied stress while at the same time temperature and pressure which influences the
failures start from other micro-cracks which were not transition from brittle to ductile deformation.
initially in critical orientations. In general, major fractures Pore fluids have a major influence on compressive
will be parallel to the axis of the intermediate principal strength of both weathered rock (Turk and Dearman,
stress but this need not be the case for other minor 1986) and fresh rock (Broch, 1979) as a result of
cracks. Therefore an array of micro-cracks can develop lowering of the molecular cohesive strength, as
in a number of orientations which collectively lead to expressed by the general equation (Turk and Dearman,
rock weakness in several planes and directions. There is 1986):
currently little information as to how the microfracture
set is controlled by mineralogy and texture and it is the  =  - 
aim of this study to investigate controls on rock failure Where  = effective rock strength
related to texture and composition.  = total rock strength
An important controlling factor on the physical and  = pore water pressure
mechanical behaviour of igneous rocks is the manner in
which crystals interlock. Fracture is considered more Wetting may also cause replacement of strong
likely to take place along crystal boundaries resulting in oxygen-silicon bonds by weaker hydrogen bonds and if
intergranular fracture rather than within crystals resulting this takes place at the apex of a microfracture it will
in transgranular fracture (Bell, 2000). Therefore irregular lower the strength required for failure at that point
and strongly interlocking or sutured crystal boundaries resulting in rapid crack propagation (Atkinson, 1984).
make fracture more difficult. It has been shown for Most of the Bushveld Complex rocks contain small but
granite that there is a linear dependence between significant amounts of hydrous minerals which may

SOUTH AFRICAN JOURNAL OF GEOLOGY


528 ROCK-STRENGTH AND PHYSICAL PROPERTIES OF NORITES OF THE MERENSKY
AND BASTARD UNITS

release water under certain conditions of local heating, development of fractures is likely to be dependent
stress or during unloading. on many inter-related factors such as grain size,
grain shape, crystal packing, mutual degree of grain
Rock Strength of Crystalline Igneous Rocks interlockment, degree of annealing as a result of cooling
This study is mainly concerned with norites in the from high temperature, and the nature and type of the
western Bushveld Complex, specifically in the sequence cementing medium. These factors apply to igneous
overlying the Merensky reef, a situation in common with rocks as well as sedimentary rocks. The complete
many other layered intrusions which host major quantification of the influence of all such parameters is
platinum group element (PGE) deposits (e.g. MSZ of the currently not possible and empirical formulation allows
Great Dyke and the J-M Reef of the Stillwater Complex). preliminary evaluation of rock strengths in design (Hoek
In general, rocks of this type can be described as and Brown, 1997). Sample size is also a consideration
granular, of relatively homogeneous composition and during testing of rock strengths as Hoek and Brown
massive, of low deformability and of generally high (1997) noted that testing under laboratory conditions
strength (Anon, 1977; Bell, 2000). The rock strength tended to overestimate strength values. This phenomena
properties are also strongly dependent on weathering was explained as a result of damage due to micro
(usually relating to near surface exposure or proximity cracking developing at critical intensities at lower stress
to faults and shear zones), as a result of the increased levels in the field compared to laboratory conditions
clay content and porosity. Furthermore, joints and where smaller samples were tested at high loading rates
fractures (often in multiple sets) relating to regional (Martin and Chandler, 1994).
stresses and unloading of the igneous body due to The Uniaxial Compressive Strength (UCS) test carried
exhumation, also create planes of weakness and can out in this study used equipment and procedures as
facilitate passage of fluids with concomitant alteration. recommended by the Commission on Testing Methods,
In contrast to igneous rocks, laminated sedimentary International Society for Rock Mechanics (ISRM)
rocks, usually have a preferred mineral orientation and (Brown, 1981). The diameter of the core used is 60 mm
are markedly anisotropic regarding unconfined (TNW core) and conformed with standard procedure
compressive strength in relation to the orientation of the using length:diameter ratio of 2:1. The UCS of a rock is
sample ( Donath, 1961; Attewell and Sandford, 1974). calculated by dividing the load at which failure occurs
This weakness is also apparent when loading is at high by the cross sectional area of the sample. The rate of
angles to the foliation plane (Brown et al., 1977). For loading of the sample has to be consistent and to
brittle foliated rocks the maximum failure strengths comply with the standard rate of loading as set by the
occur where the foliation is perpendicular to the axis of ISRM (Bieniawski and Hawkes, 1978). The samples used
loading and is at a minimum where the angle is 30o in the geotechnical tests are representative of the intact
(Ramamurthy et al., 1993). The maximum uniaxial strength of the rock (Hoek and Brown, 1997), as this
compressive strength (UCS) and tensile strengths (TS) form of testing does not take the presence of
values for mica phyllites are in the order of 80 and 20 discontinuities into account. However, the geotechnical
MPa respectively, and that for quartz schists and test results should reflect the inherent mineralogy and
phyllites 190 and 100 MPa, respectively. Massive igneous any physical or chemical alteration of those minerals
rocks (granites and basalts) exhibit UCS strengths within the samples.
of 150 to 300 MPa and are described in the category of UCS testing, in which cylinders of rock are
being ‘very strong to extremely strong’ (Bell, 2000). compressed parallel to the long axis (Figure 2), is one of
Many medium to coarse-grained igneous rocks, such as the most useful methods of determining rock-strength
the norites and pyroxenites of the Bushveld Complex, properties. The response to the test is strongly
possess variable degrees of mineral alignments, both dependent on the contact of the platens between the
linear and foliated, but the influence of mineral fabrics rock surfaces which are subject to the compression.
has not previously been investigated in these rock types. The surfaces need to be flat and at right angles to the
long axis of the specimen. The strength of the sample
Testing of Strength and Failure in Brittle Rocks decreases with increasing length in relation to diameter.
Brittle failure, as defined by Bell (2000), occurs as a Failure starts where the rock is in contact with the
result of a sudden loss of cohesion in a plane that is not platens at a strength level somewhat lower than that of
preceded by appreciable permanent deformation, and the maximum uniaxial strength of the sample (Jaeger
can take place on a variety of scales. While theory and Cooke, 1977). Failure under uniaxial compression is
predicts that the relationship between principal stresses usually violent as a result of the rapid release of energy
at rupture is linear, in actual practice these are non-linear which may also induce secondary fractures at the point
due to the area of grains in frictional contact as the of breaking. Errors in the test can arise from stresses
pressure increases. This concept was taken into account set up by the frictional effects of the platen contact
in the Grifftith’s theory of crack propagation (Griffith’s which may allow lateral expansion. In the ideal
Cracks), and further expanded by Brace (1964) who case a frictionless surface should exist between the
showed that fracture in hard rock is initiated at grain rock and the platen. In this study, samples which
boundaries. However, rock strength and the underwent solely longitudinal splitting as a result of end-

SOUTH AFRICAN JOURNAL OF GEOLOGY


A.H. WILSON, C.A. JERMY, M. RIDGEWAY AND G. CHUNNETT 529

effects (and therefore not adequately representing the mineral proportions and applying standard criteria for
inherent rock properties) were excluded from the rock naming as recommended by the International
overall assessment. In very strong materials (as in the Union of Geological Sciences Subcommission (IUGS) on
present case) both longitudinal splitting and shear the Systematics of Igneous Rocks (Le Maitre, 1989).
fractures are common (Fairhurst and Cooke, 1966) The IUGS system defines anorthosites as mafic rocks
(Figure 2). with greater than or equal to 90% plagioclase, and
pyroxenites as having less than 10% plagioclase. In the
UCS values are calculated from the equation: lower part of the Main Zone in question there are few,
if any, true anorthosites which are compositionally
UCS (MPa) = load at failure (kN)/1000 homogeneous, although small-scale components of
cross-sectional area of sample (m2 ) these rocks may be true anorthosites. The only persistent
layer of anorthosite (between a few mm and 60 cm,
The direct testing of the tensile strength (TS) of a depending in reef width) is at the base of the Merensky
rock is complex and time-consuming and therefore it is reef (and even then, the equivalent layer is not always
common practice to employ the indirect Brazilian Disc present in all areas of the Bushveld Complex). In most
test. For this test, a constant rate of loading is required other cases, plagioclase is concentrated in irregular
(as recommended by the ISRM) (Bieniawski and zones on a scale of 3 to 10 cm but these zones are not
Hawkes, 1978). The diameter to length ratio of the persistent layers and therefore only give an impression
samples is 0.5:1. The sample is loaded diametrically of being anorthosites, particularly when viewed in drill
until failure occurs under tension. The tensile strength core.
is calculated by means of the following equation, Therefore, the rigid application of the IUGS system
where P is the compression load at failure, d is the based on modal mineralogy would classify most rocks of
disc diameter and t is the length or thickness of the disc the Merensky Unit (including much of the Merensky
sample. pyroxenite and Bastard pyroxenite) as norites or
gabbronorites (to include mela- or leuco- varieties as
TS (MPa) = 0.636 P (kN) / 1000 (as applied to 60 mm diameter core) indicating more mafic and felsic varieties respectively).
2 A further difficulty in the practical implementation of
d t (m )
estimating accurate modal compositions is the generally
These indirect measurements of tensile strength heterogeneous nature of these rocks where the scale of
suffer from the limitation that the tests involve variation is commonly in the order of 1 to 5 cm, or
inhomogeneous stresses and are therefore not a true greater, particularly for the pegmatoidal rocks in the
measure of the tensile strength. The tensile strength of Merensky reef. Compositions of these rocks may,
rock is more variable than UCS and is strongly however, be reliably determined by converting whole
influenced by specimen size (Jaeger and Cooke, 1977). rock major element compositions to volume percentage
It is therefore necessary to make a number of repeated normative minerals (here using the CIPW norm) because
measurements of the same material. of the relatively simple mineralogical assemblage.
The affect of fluid pressure in the pores of the In applying such a normative classification, a small
specimen have a great influence on rock strength. Small correction needs to be applied for the most mafic rocks
amounts of fluid on grain boundaries can result in so that about 1.5% Al2O3 contained in orthopyroxene
viscous flow of material and therefore significantly affect is not converted into plagioclase. Approximately
the measured rock strength. While the rocks in the study 2.5% Al2O3 is contained in clinopyroxene but for the
may be characterized as massive with very low porosity western Bushveld Complex this mineral constitutes less
levels (Brace, 1965), early entrapment of fluids giving than 7% (by volume) and the correction is minor.
rise to small amounts of hydrous minerals, as well as The correction is dependent on the amount of
secondary alteration resulting in the formation of normative pyroxene and for 95% normative pyroxene
hydrous minerals can influence the measurements. the correction is a subtraction of 6% plagioclase from the
Unlike soils, in which pore spaces may be norm and an addition of 4% pyroxene followed by
interconnected, it may be assumed in massive unaltered recalculation of the total norm back to 100%.
igneous rocks that there is minimal interconnectivity and The correction is applied on a sliding scale and becomes
therefore fluids in non-porous igneous rocks will only minimal for normative pyroxene contents less than 25%.
affect short range propagation of fractures. For rocks in the Merensky reef environment, sulphide
and chromite should be included in the norm
Classification of rock-types in the Merensky Unit calculation.
Rock names and normative compositions There is currently little agreement on the
Visual inspection allows a qualitative assessment of rock formalization of rock names relating to the Merensky
types, cumulus status of minerals, and texture (including Unit. The commonly identified ‘pyroxenites’ or
crystal size), essential for rock recognition and ‘feldspathic pyroxenites’ are widely recognized mainly
description in the field and in mining applications. on the basis of their being more mafic than the
Identification of rock types is usually on the basis of underlying and overlying norites, and on textural form,

SOUTH AFRICAN JOURNAL OF GEOLOGY


530 ROCK-STRENGTH AND PHYSICAL PROPERTIES OF NORITES OF THE MERENSKY
AND BASTARD UNITS

may not exceed 10%, and is therefore inappropriate for


some of these rocks in this sequence. While the term
‘pyroxenite’ is retained for rocks with greater than
90% pyroxene, a further category of ‘feldspathic
pyroxenite’, or more correctly ‘plagioclase pyroxenite’
(Gillespie and Styles, 1999; Brown, 2004), is applied
here where the pyroxene content is less than 90%, but
where plagioclase is interstitial. Further distinctions are
required in the plagioclase-rich cumulates. The term
‘anorthosite’ is retained for all rocks with plagioclase
greater than or equal to 90% by volume, and ‘pyroxene
anorthosite’ for rocks with plagioclase less than 90% and
greater than 80% by volume, with the remainder being
pyroxenes. The suggested scheme for the Merensky Unit
based on normative mineral constituents (illustrated in
Figure 3 and in Table 1), formalizes fields of pyroxenite,
plagioclase pyroxenite, norite/gabbronorite, pyroxene
anorthosite and anorthosite on a quantitative basis.
A comparison is made with the IUGS fields (Figure 3a)
with those proposed here (Figure 3b), on which are
superimposed compositions of the rocks analysed in this
study. The data set is derived from 167 analyses from
four different borehole sections in the same area of the
western Bushveld Complex. This classification also
identifies clearly defined compositional groupings
pertaining to the nomenclature.
Minor mineral components which are not
represented but are important in the genesis of these
rocks include sulphide, chromite, amphibole,
phlogopite, quartz, olivine, apatite and zircon.

Textural types
Textures also play an important role in the classification
and recognition of distinct rock-types in the Merensky
and Bastard Units. The variable textural forms of both
pyroxene and plagioclase are well known and described
from many zones of the Bushveld Complex. The so-
called ‘poikilitic anorthosites’ are characterized by large
pyroxene crystals enclosing smaller crystals of
plagioclase. A further term in common usage is
‘oikocryst’ which is defined as ‘matrix or host crystals
through which smaller crystals of other minerals are
scattered as poikilitic inclusions’ (Rice, 1963). More
recently this term has been used synonymously with the
Figure 3. Normative mineral compositions for 167 samples from term ‘heteradcumulate’ (Wager and Brown, 1968) which
the Merensky and Bastard Units in relation to fields of classification has mechanistic implications for both the enclosed and
and density. (a) Compositional limits on the basis of normative the enclosing crystals (e.g. Wilson, 1992). This term
pyroxene content (volume. %) with limits conforming to the modal implies nucleation and growth of the various phases at
fields as defined by Le Maitre (1989). (b) Proposed compositional an early stage of the crystallization history and the
fields as being more appropriate for the Merensky and Bastard difference in textures probably arises from physical and
Units. (c) Relation of density and normative plagioclase content kinetic controls rather than by the phase controlled
(volume %) together with the compositional limits shown in (b). sequence of crystallization. In many cases the term
‘oikocryst’ cannot be applied to the pyroxenes in this
even though, as pointed out above some of these rocks region of the Bushveld Complex because they are
classify as varieties of melanorite on modal mineralogy commonly an agglomeration of smaller discrete crystals
alone. Therefore, it is important that such a distinction is and the term refers strictly to single crystallographically
retained in any revised classification scheme. The IUGS continuous mineral crystals which enclose crystals
system has the term ‘plagioclase-bearing pyroxenite’, but of a different mineral. Other descriptive textural terms
this is not clearly defined other than that plagioclase in common use include ‘mottled anorthosite’, ‘spotted

SOUTH AFRICAN JOURNAL OF GEOLOGY


A.H. WILSON, C.A. JERMY, M. RIDGEWAY AND G. CHUNNETT 531

Table 1. Limits of normative mineral compositions in defining rock types of the Merensky and Bastard Units in the western Bushveld
Complex.

IUGS classification Normative-based classification used here


Vol. % Vol. % Vol. % Vol. % Vol. % Vol. %
Plagioclase Orthopyroxene Clinopyroxene Plagioclase Total Pyroxene Clinopyroxene
Anorthosite >90 <10 >90 <10
Pyroxene anorthosite Term not used 80-90 10-20
Norite <90 >10 <5 40-80 20-60 <5
Gabbronorite <90 >10 >5 40-80 20-60 >5
Clinopyroxene Norite <90 >10 >5 40-80 20-60 >5
Opx > Cpx Opx > Cpx
Plagioclase pyroxenite <10 >90 <5 10-40 60-90 <10
For plagioclase
orthopyroxenite
Pyroxenite <10 >90 <5 <10 >90 <10
For plagioclase
orthopyroxenite
Prefixes mela- and leuco- can be applied respectively for more mafic and felsic varieties of norite and gabbronorite, the limits of which are
defined in Figure 3b. For the suite of rocks used in this study normative clinopyroxene is less than 10% and therefore the term gabbronorite
has been included in the table.

Figure 4. Distribution of mafic minerals and plagioclase for four typical sections of drill core GC2 determined by high contrast image
analysis of polished core. (a) Norite at 61.66 m; (b) Leuconorite at 14.16 m; (c) Anorthosite at 48.62 m; (d) Anorthosite at 51.49 m.
The textural types do not reflect the mineralogical composition but there is close agreement between the normative plagioclase content
(volume %) and the plagioclase content determined from image analysis. The narrow dashed lines are the average foliation direction as
determined from best-fit ellipses of plagioclase crystals and the heavy dashed line represents high angle dislocations considered to have
originated as a result of loading and compaction shearing (also see Figure 5).

anorthosite’, ‘granular norite’ etc. none of which and while this is broadly correct, there exist zones in the
have been rigorously defined or characterized and for Merensky reef where the plagioclase occurs as discrete
which there exist no firm criteria for their use (Brown, unzoned single oikocryts becoming highly zoned at the
2004). crystal margins. In the felsic rocks the orthopyroxene
It is commonly accepted that in the ‘pyroxenite’ ranges from individual granular crystals to large poikilitic
components of the Merensky and Bastard Units, the crystals, or agglomerations of smaller crystals which are
orthopyroxene is cumulus and plagioclase interstitial, interstitial to the plagioclase.

SOUTH AFRICAN JOURNAL OF GEOLOGY


532 ROCK-STRENGTH AND PHYSICAL PROPERTIES OF NORITES OF THE MERENSKY
AND BASTARD UNITS

Image Analysis Textural forms studied by the image analysis


Image analysis (Figure 4) provides a rapid and effective technique indicate that widely differing textures have
method for classifying both the textures and very similar mineralogical compositions. This is
compositions of rocks of this type (dominated by particularly evident in the granular textured norite
plagioclase and orthopyroxene) using high contrast (GC2 263) and the poikilitic norite (GC2 56). The degree
grey-scale scans of polished core followed by image to which pyroxene is dispersed between cumulus
processing to create a black-white pixel map. plagioclase is also highly variable even for samples of
The method requires that only dark and light minerals similar composition, as indicated for samples GC2 226
(in this case mainly pyroxene and plagioclase) in the and GC2 214. Samples were orientated such that the dip
plane of the polished surface be counted and that muted direction of the layering was at a maximum and were cut
tones from angular intersections at edge contacts of perpendicular to the layering plane. Many samples
pyroxene and plagioclase be eliminated by correct analyzed indicate a major and minor structural grain at
selection of grey-scale levels. As many of these rocks are higher angles to the vertical than the plane of layering,
texturally inhomogeneous on scales of 1 to 10 cm, and which may have developed as a result of compaction
even on larger scales in some cases, this method has shearing. High angle dislocations (Figure 5) developed
proved to be superior and more consistent compared during late-stages of crystallization in a semi-rigid mass
with the conventional method of grid point counting, of crystals and liquid by loading and shearing.
which is also very time consuming. Major textural types One of the objectives of the current study is to
can therefore be readily identified as well as obtaining ascertain the extent that rock types, based on mineral
an estimate of the mineral proportions. There is good proportions, and textures influence the geotechnical
correspondence between normative plagioclase content properties.
and the volume of plagioclase estimated from the image
analysis, and the observed deviation of 1 to 3% is well Rock Types and Density Relations
within the accuracy required for general classification The densities of unaltered mafic and ultramafic rocks in
purposes. The poikilitic nature of the pyroxene is also this part of the Bushveld Complex are largely dependent
clearly evident, except for GC2 263 in which the on the proportions of the dominant minerals present
pyroxene is largely cumulate with filamental rather than on variances of the mineral compositions,
overgrowths formed by late-stage growth providing a which have relatively narrow ranges. For rocks in the
cement to the plagioclase crystals in close proximity. compositional range from pyroxenite to anorthosite
A limitation of this method is the inability to distinguish (and excluding chromitite and those highly enriched in
between ortho- and clinopyroxene and, while less of a sulphide), the whole-rock density is strongly dependent
problem for the norites in the western Bushveld on proportions of plagioclase and pyroxene. For the
Complex, this becomes more important on the east side section studied, a high degree of correlation between
where clinopyroxene is in some rocks a major mineral normative plagioclase with density (measured by the
constituent. water immersion method using core samples weighing

Figure 5. Thin section of leuconorite showing high angle dislocations in relation to the average foliation in the plane of the layering.
The dislocations probably originated as a result of loading and compaction shearing in the semi-rigid mass of crystals at high temperature.
The shearing caused fracturing of relatively few crystals with only a few showing healed fractures but also caused reduction in crystal size
close to the shear plane. This indicates that crystals continued to grow in the shear regime. Continued stress in the completely solid regime
is reflected by bent and deformed crystals. These planes of weakness can influence the manner in which the rocks break.

SOUTH AFRICAN JOURNAL OF GEOLOGY


A.H. WILSON, C.A. JERMY, M. RIDGEWAY AND G. CHUNNETT 533

Figure 6. Frequency distribution diagrams for the results of the UCS (uniaxial compressive strength), TS (tensile strength) and RQD (rock
quality designation) tests for three drill cores (GC2, GC7 and GC8). GC2 shows the most consistent distribution for UCS but also has the
lowest median value for TS. GC2 has the most consistently high value for RQD. Cores GC7 and GC8 have a wide range of values including
very low values.

200 to 400 gm) is observed (correlation coefficient The RQD value (Rock Quality Designation) is a
r2=0.995). Therefore density determined rapidly and semi-quantitative assessment of the incidences of
easily by the immersion method provides a potentially discontinuities within given intervals in the core and
accurate method of determining rock compositions. provides a measure of rock mass. Core GC2 was visually
At this stage it is not known to what extent this noted as being more coherent and less jointed than GC7
relationship varies in different areas of the Merensky and GC8. Core GC7 had some degree of fracturing (with
Unit. one zone of extensive fracturing) while GC8 was more
extensively fractured, and in some zones very heavily
Comparison of UCS and TS Test Results fractured. The frequency of RQD values is recorded for
Samples from three borehole cores were used in the specific intervals in each core judged to be of a
study which showed measurable variation in a number consistent nature as reflecting the measured intensity of
of primary and secondary physical properties such as, discontinuities and fractures. On this basis core GC2 is
rock-types based on mineral proportions, textural types, coherent and competent with an RQD of about 90,
mineral fabrics, fracture density and degree of whereas for the other two cores which have similar
weathering. A total of 182 samples were used for UCS ranges, the distribution of data is generally to much
testing and 174 for TS tests (Figure 6). While the three lower values. The high frequency peak for all three
cores were located in the same general area of the cores is 80-100% (Figure 6), indicating that the cores
western Bushveld Complex, separated by less than have both fractured and unfractured zones.
1 km, and represent the same stratigraphic interval UCS measurements for the three cores are consistent
(Figure 1), some consistent variations in primary textures with the differences in RQD. GC2 has a symmetrical
were noted. distribution of rock strengths ranging from 80 to

SOUTH AFRICAN JOURNAL OF GEOLOGY


534 ROCK-STRENGTH AND PHYSICAL PROPERTIES OF NORITES OF THE MERENSKY
AND BASTARD UNITS

Figure 7. Variation of UCS and TS values through the zone of Merensky and Bastard Units in relation to normative plagioclase contents
for drill core GC2. (a) Variation of normative plagioclase composition highlighting the cyclic characteristics of the section. There is little
variation in plagioclase content above the 40 m mark. (b) Values of UCS vary in a systematic manner indicating primary controls relating
to rock properties. (c) Variation of TS which has some correlation to peak positions for UCS. The TS value immediately above the Merensky
pyroxenite is the lowest measured value for the entire sequence. (d) Correlation of UCS and TS measurements as relating to identified cyclic
variation observed in (b). The units A, B and C from the top of the section lie on a different trend to the rest of the data. This is likely to
represent an inherent property of the rock sequence, but may also be the result of near-surface alteration.

220 MPa with a relatively narrow standard deviation. pyroxenite. This pattern is repeated above the Bastard
In contrast GC7 and GC8 have lower values and a wider pyroxenite but the composition remains remarkably
distribution, with GC8 being significantly degraded. constant higher in the norites. Consistent with the
This is in contrast with the TS properties which show compositional changes, however, distinct textural
the inverse trend with GC2 having overall lower values. changes are observed giving a range of rock types which
Core GC7, and even more so for GC8, have a include granular norite and poikilitic leuconorite.
distribution of lower values which is consistent with the Upward in the Merensky Unit there is an overall increase
lower RQD rating of the cores. However, the lower in UCS values (Figure 7b) indicating that rock strength is
standard deviation for GC2 is indicative of other controls dependent on the gradual change of rock-type from
on tensile strength. plagioclase pyroxenite to leuconorite. However, on
approaching the base of the Bastard pyroxenite the TS
Stratigraphic Controls on UCS and TS (Figure 7c) and UCS decrease dramatically indicating
The high RQD ratings for core GC2 indicate that the rock controls other than rock type. Above the Bastard
is fresh and unaltered and suitable for investigations on pyroxenite there is a marked and consistent increase in
stratigraphic controls on rock strength properties. plagioclase content which corresponds to a slight
Variations in bulk compositions (Figure 7a) are reflected increase in both TS and UCS. Above the 40 m mark the
by the normative plagioclase content with a gradual amount of plagioclase is relatively constant, slightly
increase in the plagioclase content from the top of the decreasing upwards in the sequence, while at the same
Merensky pyroxenite to immediately below the Bastard time there are major changes of both TS and UCS over

SOUTH AFRICAN JOURNAL OF GEOLOGY


A.H. WILSON, C.A. JERMY, M. RIDGEWAY AND G. CHUNNETT 535

Figure 8. Variation of: (a) UCS values and (b) Zr content, with stratigraphic height in core GC2; (c) UCS with Zr content; (d) UCS with
percentage of normative quartz. A strong dependence for peak values is observed between UCS and Zr content and with normative quartz
through the section and on the basis of the variation plots. Different parts of the section suggest slightly different controls as indicated by
the symbols.

narrow vertical intervals indicating controls other than Trace Element and Minor Mineral Controls on Rock
bulk rock composition. This change indicates that the Strength Properties
poikilitic leuconorites are considerably weaker than An unexpected result is the correlation of rock strength
the granular leuconorites. A broad linear dependence is (and in particular UCS) on minor mineral component
observed between TS and UCS for the lower part of the and trace element composition of the whole rocks as
section (rock packages E to J in Figure 7d). Above 20 m expressed by the Zr content in the whole rock and
(rock packages A to C in Figure 7d) there is a shift to normative quartz (volume %). Normative quartz in these
lower TS values giving a slightly different array on this rocks indicates an excess of silica which cannot be
diagram which is not reflected in the bulk composition accommodated by the major silicate mineral assemblage.
or rock type. While precise correlation does not occur for the entire
Relative rock strengths vary from 10 to 60% for TS section investigated, over certain intervals there is a
and between 40 to 100% for UCS over intervals of about high degree of correlation, and the overall sense of
5 m. The smooth variation of values within these variation for Zr content and UCS is remarkably similar
intervals is also indicative of primary controls on (Figures 8a; b). The Zr content is not a function of the
rock strength properties which systematically vary through general mineralogical composition because completely
the sequence. Therefore, although controls relating to contrasting patterns are observed for the variation of
rock type and plagioclase content are observed, the normative plagioclase and Zr content (compare
variation is complex and cannot simply be explained by Figures 7a; 8b). Both Zr and normative quartz content
variations in bulk rock composition or textural form. reflect a particular stage in the early crystallization of

SOUTH AFRICAN JOURNAL OF GEOLOGY


536 ROCK-STRENGTH AND PHYSICAL PROPERTIES OF NORITES OF THE MERENSKY
AND BASTARD UNITS

cumulus minerals (Campbell, 1987) and are commonly


reflected by interstitial or poikilitic textures. These late-
forming minerals may be more evolved feldspars,
quartz, amphiboles and quartz-feldspar micrographic
intergrowths. In many cases these cementing phases are
in small amounts occurring as thin films rimming the
cumulus minerals. The distribution of these minerals is
also strongly influenced by surface energy and ‘wetting’
effects (Hunter, 1996). Normative quartz reflects excess
silica in the whole rock and arises as a result of
fractionation taking place on a small-scale with local
migration of evolved interstitial liquid. Larger amounts of
trapped liquid, which may also have migrated over
limited distances, gave rise to more abundant quartz
forming in the interstices between the cumulus crystals
eventually acting as cement to those crystals. Zr is an
element which is largely incompatible with cumulus
minerals (plagioclase and pyroxene) and tends to
remain in the late-stage liquid as solidification proceeds,
eventually crystallizing as small crystals of zircon
together with other trace minerals such as apatite, rutile
and monazite. This element is also commonly used as a
measure of the amount of trapped liquid initially
contained in a cumulate rock (Cawthorn, 1996; Wilson,
2001). Therefore, both Zr and normative quartz reflect
the development of small amounts of late-stage minerals
(or the cementing phases) in cumulates.
A cementing medium, or even small amounts of
minerals which are concentrated at the margins
of cumulate crystals are likely to affect the way the rock
fractures under stress because of its influence on grain-
boundary fracture propagation. Fracture is thought to
develop initially along grain boundaries before the stress
migrates into the interior of the crystals and therefore
even small amounts of competent minerals located at
crystal boundaries could influence rock strength. While
these relationships are a strong indication of the controls
on the strength properties of these rocks, the fact that
Figure 9. Photomicrographs of core samples that have undergone the patterns for UCS and Zr (and normative quartz) are
UCS testing. Fractures resulting from the tests are highlighted by different, while the sense of variation is similar, again
the white lines. Solid white areas are interstitial quartz. (a) indicates that the cementing medium is only one
Fractures penetrate several feldspar crystals and follow crystal contributing factor.
boundaries but terminate along annealed zones of finer-grained To investigate likely controls in the fracture
crystals which also contain small amounts of cementing quartz. propagation, samples which had previously undergone
Note the deformed twin lamellae in the centre lower edge of the UCS testing were studied texturally in thin section to
image. (b) Fracture penetration of crystals which terminate against monitor fracture patterns. Fractures were observed to be
an interstitial quartz crystal. (c) High fracture penetration of controlled by a number of mineralogical and textural
plagioclase crystal terminating against a crystal of well-developed forms summarized as follows: 1) Cementing quartz
interstitial quartz. crystals clearly had a tendency to act as non-penetrative
barriers to fractures by which several fractures would be
seen to terminate at quartz-plagioclase boundaries.
these rocks in which trapped silicate liquid was Some of these fractures had penetrated several feldspar
solidifying to form a matrix to the cumulate phases crystals (Figures 9a to c). 2) Fracture transmission is
(Wilson, 1992). preferentially developed along cleavages in feldspar,
The explanation of the relationship between rock rather than along crystal boundaries as proposed by Bell
strength and composition is postulated to lie with the (2000). 3) Agglomerations of finer grained feldspar
role of the trapped liquid in the initial formation of the crystals are associated with interstitial quartz and did not
rock. Late-stage liquid crystallizes small amounts of readily transmit fractures through those zones (Figure
new mineral phases which act as a cement to the 9a). 4). Fractures tended to be deflected around quartz

SOUTH AFRICAN JOURNAL OF GEOLOGY


A.H. WILSON, C.A. JERMY, M. RIDGEWAY AND G. CHUNNETT 537

Figure 10. Crystal size and crystal orientation analysis for samples 15 (at 15 m) and 52.73 (52.73 m) from GC2 core. The left series of panes
shows the results of image analysis (outlines are plagioclase and black areas are pyroxene) and crystal size distributions on the same scale.
Sections shown are perpendicular to the layering plane. Both samples have a normal distribution of crystal sizes but sample GC2-15 is
overall coarser grained. Orientation ellipses show the distribution of alignments of long crystal axes for the entire population and for
selected coarser-grained distributions to represent the larger crystal population. In the finer-grained sample GC-2 52.73 the larger crystals
have a better foliation because of the higher shape anisotropy (a/b=1.52) in contrast with the coarser sample GC-2 15 in which the larger
crystals are essentially equidimensional (a/b=1.17). The right series of panes are histograms of long axis alignments for each of the size

SOUTH AFRICAN JOURNAL OF GEOLOGY


538 ROCK-STRENGTH AND PHYSICAL PROPERTIES OF NORITES OF THE MERENSKY
AND BASTARD UNITS

lithologic layering but crystal imbrication can result from


banking-up of crystals during flow.
Fabric styles and degree of development of crystal
orientations can therefore potentially exert an influence
on rock strength. A strong lamination fabric, particularly
where this is related to small-scale lithologic layers,
could provide planes of weakness leading to reduction
of rock strength, particularly TS. This would be further
accentuated if the mineral type had a cleavage or
parting parallel to the direction of strain propagation.
A combination of poorly developed fabric interlaced
with a matrix of fine interlocking crystals could result in
a very strong structure. The strongest rocks would be
expected to be those that have a completely interlocking
network of randomly orientated prismatic crystals.
The mafic rocks of the Merensky Unit have a wide
variety of fabric types. Sections were cut at right angles
and parallel to the core axis and orientated in relation to
Figure 11. Variation of UCS values for core GC2 in relation to
the layering plane which is observable by fine-scale
volume. % normative plagioclase contents for samples with higher
modal layering in the core. Norites which have granular
degrees of alignments (called GC2 52.73 type) and for samples
textured pyroxene crystals were analyzed using polished
with lower degrees of crystal alignments (called GC2 15 type).
sections of core by image analysis techniques. Outlines
The more feldspathic samples have overall greater degrees of
of cumulus pyroxene and plagioclase were converted
crystal alignments and in both populations there is a general
into best fit ellipses. Strong overall alignments lie
tendency for UCS values to increase with the amount of
between 70 to 90o corresponding to the dip of the
plagioclase. The outlined fields emphasize the general trends.
layering, and a further weaker fabric developed at a
higher angle over a spread of about 30o. Two samples,
crystals bounded by feldspar rather than along the which on a visual basis represent contrasting fabric
boundaries of the crystals (Figure 9c). types, were studied in detail using image analysis
techniques (Figure 10). Grain-boundary recognition
Fabric and Crystal Size Analysis software was utilized in orientated thin sections and
Crystal orientation and mineral fabrics could potentially analyzed by fitting ellipses to the defined crystal forms.
also influence the physical properties of these rocks and This enabled a size-analysis of the crystals to be carried
the way they resist stress. Alignment of tabular crystals out as well as an analysis of the orientation of the
in a horizontal or layering plane would have the ellipses.
same affect as platy minerals in metamorphic rocks Sample GC2 52.73 is finer grained than sample
(as previously described). It is common for cumulus GC2 15 and possesses a well-developed lamination
minerals in mafic rocks to adopt a preferred orientation fabric (Figure 10). Fabric analysis of the two samples
or even several orientations for the same mineral type showed that GC2 52.73 has a well-developed lamination
but such orientations are seldom perfect. Preferred fabric restricted to the larger crystals and therefore by
orientation of crystals can arise from early-stage settling removing the smaller crystals from the analysis this
and flowage in the magma regime, by compaction and fabric was even more strongly developed. Sample GC2
rotation during the mid-stages of solidification and from 15 possesses a weaker fabric which is even less
stress in the semi-brittle stage (Benn and Allard, 1989; developed when the finer-grained crystals were
Nicolas, 1992). The greater the shape anisotropy of the removed.The crystal size distribution shows that the
minerals, the more likely it is that a primary orientation coarser grained rock (GC2 15) has a fine-grained
fabric will be established (Wilson, 1982; Wilson, 1992). component and a wider relative crystal size distribution
A tabular shaped crystal type may give rise to a which is asymmetric. In spite of these significant
lamination fabric whereas an elongate crystal type could differences in distribution, the average crystal size of
result in a lineation-type fabric. The type of fabric also both samples is very similar. Sample GC2 52.73, has a
depends on the degree of packing and interaction of the significantly higher UCS value relative to GC2 15 (186
crystals and mass flowage of a mush, the latter and 144 MPa respectively) and a lower TS (8.8
commonly resulting in soft-sediment type deformation, compared with 9.2 MPa). As a general rule, UCS values
equivalent to slumping. It is also likely that magma flow are approximately eight times the TS values indicating
patterns will change during the solidification of the that tensile strength has been significantly reduced by an
chamber in response to convection systems, influx of inherent physical property of the rocks. This is likely to
new magma and the breakdown of double diffusive be caused by the relatively well-developed lamination
convection cells (Campbell, 1996). Usually, but not fabric in both these samples. The lower UCS value of
always, the preferred orientation is in the plane of the sample GC2 15 is attributed to the greater propagation

SOUTH AFRICAN JOURNAL OF GEOLOGY


A.H. WILSON, C.A. JERMY, M. RIDGEWAY AND G. CHUNNETT 539

of critical fractures along cleavages in the larger crystals, some broad correspondence in trends for UCS and TS
and the lamination fabric may have compromised the TS over specific vertical intervals, in detail each of these
in GC2 52,73. tests has different variations through the sequence
The effects of normative plagioclase content, crystal indicating different controls.
size and fabric type are examined in relation to rock UCS values are lower in poikilitic leuconorites
strength. In the plot of normative plagioclase versus UCS compared with the more granular varieties. There is also
two groups are identified relating to textural type. strong correspondence between UCS values in relation
The group containing plagioclase crystals which are to both normative quartz contents of the norites and the
finer-grained and show a normal size distribution Zr contents. Both of these compositional parameters
(similar to that of GC2 52.73) are shown to have overall relate to entrapment of evolved liquid in the later stages
higher UCS values (Figure 11), whereas types similar to of crystallization giving rise to small amounts of
GC2 15 generally have lower UCS values and show less important late-stage minerals. Such late-stage crystallized
dependence on the plagioclase content. material acted as cement and therefore influences the
rock strength. Detailed evaluation of core which failed
Discussion under laboratory tests showed that fractures which arose
The determination of rock strength is important in as a result of the imposed strength tests were
mining operations because it influences mine planning propagated through crystals, as well as along crystal
and development, methods used in rock and ore boundaries, but commonly terminated against interstitial
extraction, and safety issues. Information on rock quartz crystals. These observations therefore explain the
strength is also important in the design of processing control of the cementing medium on rock strength.
and crushing plants. The matter of rock strength has, up Within the controls of the cementing medium there are
to this stage, not been a major limiting factor in the also other influences which give rise to displaced trends
mining of Merensky reef, but as mining operations in the geochemical relationships which also cannot be
extend to considerably greater depths, more detailed explained by bulk rock compositions. Within these
knowledge of rock properties and their controlling variations it is shown that mineral fabric analyses have
factors is becoming increasingly important. In this study, an influence on rock strength. Rocks with well
three borehole cores of variable degrees of preservation developed foliation fabrics (typically the poikilitic
were investigated from the aspects of rock-strength leuconorites) tend to be weaker than those with a
properties by measuring UCS and TS under laboratory random fabric. However, crystal size also affects rock
conditions. Rock- strength properties of unaltered core strength and may counteract the influence of fabric.
were then related to fundamental petrological properties High-angle shear zones which formed initially at high
such as textures (including crystal size and fabric), temperature and then annealed during the cooling
mineralogical and chemical compositions. process may also affect rock strength although the direct
UCS values for the norites in the stratigraphic influence of this structure on rock strength has yet to be
environment of the Merensky reef range from <100 to investigated.
>200 MPa with freshest rock having a mean value about In conclusion, rock strengths of the norites in the
150 MPa. The range for TS is from <5 to 15 MPa with a environment of the Merensky reef exhibit considerable
mean of 11 MPa. These values confirm that most of the variation which is dependent on a number of controls.
norites rocks fall into the category of ‘strong’ to ‘very These include bulk rock composition, textural relations
strong’. However, the distribution extends to relatively and mineral associations, the nature and amount of
low values for some samples of fresh rock (80 MPa for cementing medium, fabric and the stress history of the
UCS and 5 MPa for TS) which if related to specific rock rock from early stage compaction to late-stage cooling.
types or rock properties in particular stratigraphic The relative roles of these controls are a fertile area for
horizons, may need to be considered in mining applied research and should be investigated further.
operations. Where the RQD (Rock Quality Designation)
is low, reflecting observable alteration and shearing, Acknowledgements
both the UCS and TS values are skewed to significantly Anglo Platinum is thanked for providing the drill core
lower values. Bimodal distributions in rock strengths are used for this study and for permission to publish this
predictably observed where cores have zones of both paper. The National Research Foundation (NRF) of
fresh and altered rock. In some such cases the rock South Africa and THRIP (Technology and Human
would be categorized as of ‘moderate’ or even ‘weak’ Resources Industry Programme) are acknowledged for
strength. financial support. Dr. Chris Ware is acknowledged
Even in unaltered core there are zones where UCS for providing helpful comments. Helpful comments
and TS are relatively low and systematic variations are were received from reviewers Dr. A. Bye, Dr. J.L. van
observed for both test results over specific vertical Rooy and Ray Brown.
intervals indicating changes in controls relating to
fundamental rock properties. The norites of the References
Merensky and Bastard Units are complexly layered both Anon, (1977). The description of rock masses for engineering purposes.
Working Party Report. Quarterly Journal of Engineering Geology,
compositionally and mineralogically. While there is

SOUTH AFRICAN JOURNAL OF GEOLOGY


540 ROCK-STRENGTH AND PHYSICAL PROPERTIES OF NORITES OF THE MERENSKY
AND BASTARD UNITS
10, 355-388. Volume 1: Classification of Igneous Rocks, British Geological Survey
Atkinson, B.K. (1984). Subcritical crack growth in geological materials. Research Report, RR99-06, 52pp.
Journal of Geophysical Research, 89, 4077-4114. Griffith, E.W. (1921). The phenomena of rupture and flow in solids.
Attewell, P.B. and Sandford, M.R. (1974). Intrinsic shear strength of a brittlle Philosophical Transactions of the Royal Society of London, A221, 163-198.
anisotropic rock 1. Experimental and mechanical interpretation. Griggs, D.T. (1951). Deformation of Yule Marble. Bulletin of the Geological
International Journal of Rock Mechanics and Mining Science, 11, 423-430. Society of America, 62, 853-862.
Bell, F.G. (2000). Engineering Properties of Soils and Rocks. Blackwell Hoek, E. (1965). Rock fracture under static stress conditions, National
Science, Oxford, United Kingdom. Mechanical Engineering Research Institute C.S.I.R., Pretoria, South Africa,
Benn, A. and Allard, B. (1989). Preferred mineral orientations related to 383pp.
magmatic flow in ophiolite layered gabbros. Journal of Petrology, Hoek, E. and Brown, E.T. (1997). Empirical strength criterion for rock mass.
30, 925-946. Proceedings of the American Society for Civil Engineers, Journal of
Bieniawski, Z. T. and Hawkes, I. (1978). Suggested methods for determining Geotechnical Engineering Division, 106, 1013-1035.
tensile strength of rock materials, for ISRM Commision of Laboratory and Hunter, R.H. (1996). Texture development in cumulate rocks. In: R.G.
Field Tests. International Journal of Rock Mechanics and Minerals Science, Cawthorn (Editor), Layered Intrusions, Elsevier, The Netherlands, 77-102.
15, 99-103. Jaeger, J.C. and Cooke, N.G.W. (1977). Fundamentals of Rock Mechanics.
Brace, W.F. (1961). Dependence of fracture strength of rocks on grain size. Chapman and Hall, New York, United States of America, 585pp.
Bulletin of Industrial Minerals, 76, 99-103. Le Maitre, R.W. (1989). A Classification of Igneous Rocks and Glossary of
Brace, W.F. (1965). Some new measurements on linear compressibility of Terms, Blackwell Scientific, Oxford, United Kingdom, 193pp.
rocks. Journal of Geophysical Research, 70, 391-398. Martin, C.D. and Chandler, N.A. (1994). The progressive failure of Lac duc
Broch, E. (1979). Changes in rock strength caused by water. In: Proceedings Bonnet granite. The International Journal of Rock Mechanics, Mining
of the 4th International Conference of Rock Mechanics, Montreaux, Science and Geomechanical Abstracts, 30, 643-659.
France, 71-76. Nicolas, A. (1992). Kinematics in magmatic rocks with special reference to
Brown, E.T. (1981). Rock Characterization and Testing and Monitoring, gabbros. Journal of Petrology, 33, 891-916.
Pergamon Press, Oxford, United Kingdom, 211pp. Onodera, T.F. and Kumara, A.H.M. (1980). Relationship between texture and
Brown, E.T. Richards, L.R. and Barr, M.V., (1977). Shear strength mechanical properties of crystalline rocks. Bulletin International
characteristics of the Delabole Slates. In: Proceedings of the Conference of Association of Engineering Geology, 22, 173-177.
Rock Engineering, Newcastle upon Tyne, United Kingdom, 33-51. Ramamurthy, T., Venkatappa, R.G. and Singh, J. (1993). Engineering
Brown, R.T. (2004). Mafic and ultramafic rocks of the Bushveld Complex: a behaviour of phyllites. Engineering Geology, 33, 209-225.
note on their classification and nomenclature, 77pp., Anglo Platinum Rice, C. M. (1963). Dictionary of Geological Terms. Pinceton University Press,
Internal Report. Princeton, New Jersey. 465pp.
Buick, I.S., Maas, R. and Gibson, R. (2001). Precise U-Pb titanite age Turk, N. and Dearman, W. R. (1986). Influence of water on engineering
constraints on the emplacement age of the Bushveld Complex, South properties of weathered rock. In: J.C. Cripps, F.G. Bell and M.G. Culshaw
Africa. Journal of the Geological Society of London, 158, 3-6. (Editors), Groundwater in Engineering Geology, Geological Society,
Campbell, I.H. (1987). Distribution of orthocumulate textures in the London, United Kingdom, 131-139.
Jimberlana intrusion. Journal of Geology, 95, 35-54. Wager, L.R. and Brown, G.M. (1968). Layered igneous rocks. Oliver and
Campbell, I.H. (1996). Fluid dynamic processes in magma chambers. Boyd, Edinburgh. 588pp.
In: R. G. Cawthorn (Editor), Layered Intrusions, Elsevier, The Netherlands, Willard, R.J. and McWilliams, J.R. (1969). Microstructural techniques in the
45-76. study of physical properties of rocks. International Journal of Rock
Cawthorn, R.G. (1996). Models for incompatible trace element abundances Mechanics and Mining Sciences, 6, 1-12.
in cumulus minerals and their application to plagioclase and pyroxenes in Wilson, A.H. (1982). The geology of the Great Dyke, Zimbabwe: the
the Bushveld Complex. Contributions to Mineralogy and Petrology, ultramafic rocks. Journal of Petrology, 23, 240-292.
123, 109-115. Wilson, A.H. (1992). The geology of the Great Dyke, Zimbabwe:
Donath, F.A. (1961). Experimental study of shear failure in anisotropic rocks. Crystallization, layering, and cumulate formation in the P1 Pyroxenite of
Bulletin of the Geological Society of America, 72, 985-991. cyclic unit 1 of the Darwendale Subchamber. Journal of Petrology,
Eales, H.V. and Cawthorn, R. G. (1996). The Bushveld complex. In: 33, 611-663.
R.G. Cawthorn (Editor), Layered Intrusions, Elsevier, The Netherlands, Wilson, A.H. (2001). Compositional and lithological controls on the PGE -
181-230. bearing sulphide zones in the Selukwe Subchamber, Great Dyke: a
Fairhurst, C. and Cooke, N.G.W. (1966). The phenomenon of rock splitting combined equilibrium - Rayleigh fractionation model. Journal of Petrology,
parallel to a free surface under compressive stress. Proceedings of the First 42, 1845-1867.
Congress of the International Society of Rock Mechanics, Lisbon, Portugal,
1, 687-692.
Gillespie, M.R. and Styles, M.T. (1999). BGS Rock Classifiaction Scheme Editorial handling: J. M. Barton

SOUTH AFRICAN JOURNAL OF GEOLOGY

Potrebbero piacerti anche