You are on page 1of 7

ARMA 11- 282

The implication of high block heights on cave flow and recoveries


van As, A.
Rio Tinto, Brisbane, QLD, Australia
Power, G.P
Power Geotechnical, Diamond Valley, QLD, Australia
Van Hout, G.J.
Rio Tinto, Antwerp, , Belgium

Copyright 2011 ARMA, American Rock Mechanics Association


th
This paper was prepared for presentation at the 45 US Rock Mechanics / Geomechanics Symposium held in San Francisco, CA, June 26–29,
2011.
This paper was selected for presentation at the symposium by an ARMA Technical Program Committee based on a technical and critical review of
the paper by a minimum of two technical reviewers. The material, as presented, does not necessarily reflect any position of ARMA, its officers, or
members. Electronic reproduction, distribution, or storage of any part of this paper for commercial purposes without the written consent of ARMA
is prohibited. Permission to reproduce in print is restricted to an abstract of not more than 300 words; illustrations may not be copied. The
abstract must contain conspicuous acknowledgement of where and by whom the paper was presented.

ABSTRACT: Over the past decade mining companies have adopted aggressive mining strategies when designing their new block
cave mines, ultimately driven by NPV. Block heights have typically doubled or even quadrupled compared to those of the past
and similarly drawpoint spacings have increased by up to 30%. The poor track record experienced from several of these new
mines brings into question whether the ‘state of the art’ in cave flow and recovery prediction tools can be confidently applied, both
empirical and numerical methods. Empirical models are by definition only applicable when applied within the constraints of the
data that supports them. Thus there is an urgent need to both re-examine and expand the empirical models to incorporate the
experiences from these ‘outlier’ mines or to develop new models. Numerical models, on the other hand, face different challenges,
the greatest being the inability to model the required level of detail on a mine-wide scale, i.e. computational limitations.

1. INTRODUCTION
Unlike most mining methods, block cave mines demand
a large capital investment for the development and
construction of a significant portion (if not all) of the
mine prior to the commencement of any production.
Thus mining engineers are tasked to design and optimize
the resource recovery to ensure the maximum return on
investment whilst still tempered by practical mining and
geotechnical constraints. In the past, one of these
mining constraints has been the height of the block to be
caved. The main reasoning behind the imposed block
height constraint is to ensure that the orebody is able to
cave easily, that dilution entry is minimized and that the
drawpoint brow integrity is maintained until all of the Figure 1. Evolution through time of the trend for the block
economic ore above is extracted. Benchmarking studies height in block/panel caving mines (after ISC-II 2004).
conducted by Florez and Karzulavic for the International
Caving Study II [1], clearly show the increasing trend to derived from observations of laboratory scale physical
model experiments (using sand or rock aggregates),
develop caves with higher block heights. The latest
limited mine scale field trials (using marker recovery
proposed block heights being up to twice the maximum
experiments), mine observations and numerical methods
presented on the chart in Figure 1.
(numerical and stochastic models).
The main mechanisms of cave flow and mixing within
The main reason for relying on empirical rather than
the caved ore column are arguably either poorly
mine-calibrated numerical methods has simply been due
understood or over simplified in the predictive
modelling tools used in estimating the resource recovery. to the vast scale of the problem, as mine scale modelling
To date the basis for most cave flow and resource of interparticular flow behaviour demands extensive
recovery prediction tools are based on empirical rules computation power.
However recent experiences from modern, higher block experiments [6] have demonstrated that draw zones may
height caves as well as recent advances in cave flow also eventually form as cylinders with a constant
research in conjunction with the development of diameter dependant on the drawpoint width and the
improved modelling tools and sophisticated flow fragmentation distribution.
monitoring systems have challenged the conventional
Pierce [7] has commented that in coarse material the
wisdom regarding cave mass flow. Consequently the
occurrence of stable arching plays and important role in
suitability of using existing empirical rules, for recovery
narrowing the IMZs. Pierce postulates that under such
predictions where the block heights are ‘poles apart’
conditions it is possible for the stagnant material outside
from those from which they have been derived, is
the IMZ perimeter to form a stable hoop and restrict
justifiably questionable.
lateral growth of the IMZ. This phenomenon could well
explain why in operational caves, which typically have a
bi-modal fragmentation distribution at the drawpoint,
2. CURRENT THEORY AND DEBATE
that the observed IMZs are narrower than that predicted
Empirical theory for cave flow is founded on three main by flow codes based on ellipsoidal IMZs.
flow mechanisms, namely granular flow, mass flow and
Pierce [7] has also suggested that this phenomenon of
void diffusion.
cylindrical IMZ shapes should be expected to develop in
2.1. Dynamic Ellipsoids or Static Cylinders of Draw? caves where the column heights exceed several hundred
Granular flow is described by Kvapil [2] as the particle diameters, a real scenario for most modern block
movement of material within an ellipsoidal volume caves.
beneath a flowing drawpoint. These ellipsoids of
2.2. Does Draw Zone Interaction and Mass Flow
motion are characterised by higher velocities in the
Occur at the Cave Scale?
centre, with lower velocities closer to the boundary at
Laubscher [8] describes mass flow as the uniform
which these become zero. The ellipsoid of material
downward movement of material without any form of
movement is termed either the ‘limit ellipsoid’, draw
mixing. According to Laubscher, mass flow occurs
zone or ‘isolated movement zone’ (IMZ) whilst the
above the height at which the draw zones above all
ellipsoid of material extraction is termed either the
operating drawpoints coalesce and interact. Below this
‘ellipsoid of draw’ or ‘isolated extraction zone (IEZ),
zone of draw zone interaction, termed the height of
refer to Figure 2. Kvapil [2] also established a direct
interaction zone, (HIZ), is the region characterised by
relationship between the diameter to height ratio of the
granular flow, as illustrated in Figure 3.
ellipsoids and the average fragmentation size.
Surface Rill

Dilution
Ore

Limit Ellipsoid

Mass Flow
Even draw down
Rapid Draw in Centre Height of Interaction Zone

Interactive zone Low pressure


Elipsoid of draw

High pressure

Figure 3. Illustration of Mass Flow and Interactive Gravity


Flow (after Laubscher 1992).

Thus in order to maximise recovery and minimise


dilution it is imperative that drawpoints are spaced
sufficiently close to ensure that draw zones interact and
mass flow prevails. However if draw zone interaction
fails to occur or if poor drawcontrol results in highly
differential rates of draw between adjacent drawpoints,
Figure 2. Terminology used to describe the zones of material then mass flow will not eventuate and any overlying
movement and material extraction waste will rapidly flow down into the drawpoints, in
particular if it is finer than the underlying ore. This early
It is important to note that numerous other physical dilution entry has the obvious unwanted effect of
modelling studies [3,4,5] and mine-scale marker
reducing the value of the ore, or in some cases even Isolated & interactive draw
plan view at 1600 mm L
prematurely sterilising the orebody.
800
Thus the concept of the HIZ is fundamental to the Interactive draw
Isolated drawbell
600
successful prediction of ore recovery. As stated by
Laubscher [8], “If interactive theory is accepted then the 400

higher the draw column the lower the dilution. This is 200

on the basis that the ore/dilution interface will be

S-N axis (mm)


0

maintained as a distinct zone and that dilution will only -600 -400 -200 0 200 400 600 800 1000 1200 1400 1600

-200
enter the ore column when the ore/waste contact reaches
the height of the interaction zone”. For the successful -400

prediction of ore recovery the design engineer must -600

answer the non-trivial question of what height above the -800

extraction level the HIZ occurs (or whether it occurs at -1000

all). The higher the interaction zone, the earlier granular E-W axis (mm)

flow, material mixing and consequently dilution occurs.


Figure 5. Plan view above the physical model showing the
The question being debated though is whether drawpoint negligible difference between measured isolated widths and
interaction implies that the IMZs must overlap for mass draw zone widths when concurrently drawing drawpoints
flow to occur? The early work by Kvapil [2] suggested spaced at a distance equal to 1.5 times the IMZ
IMZs must overlap yet later studies by Laubscher [5, 8]
et al argued that IMZs need only be spaced within 1.5 2.3. Is Void Diffusion a More Appropriate Mechanism
times the IMZ width. Furthermore Laubscher postulated at the Cave Scale?
that interacting draw zones within the same drawbell Laubscher also described a third flow mechanism that
increase to up to 1.5 times their combined diameter, can take place in non-uniform material where a high
similar interaction across the minor apex, between porosity, highly mobile zone above a drawpoint is
drawbells also caused an 1.5 times increase in combined bounded by compacted, stagnant material. Essentially
diameter, thereby arguing that interaction across the this is a form of isolated draw, conceptualised as the
major apex was indeed possible if drawpoints were upward movement of voids and thus described using
drawn in unison. ‘void diffusion’ theory. The hypothesis is that this type
of mechanism dominates flow during the early stages of
cave development and early draw, until granular flow
develops. “It is postulated that chimneying, piping or
funnelling may occur when a succession of voids work
their way through the caved material to surface. Once
the path is established and filled with mobile fine
material, it becomes a preferred flow channel providing
a path for the migration of dilution” [8].

3. LEARNINGS FROM OPERATING CAVES


Experience from modern operations with relatively high
block heights (though considerably shorter than those
currently proposed) has lead to valuable insights that
Figure 4. Progressive draw zone interaction between adjacent have sparked the debate over the correct evaluation of
drawpoints, and drawbells (Laubscher 2000). the ‘super’ high column caves. Due to the sensitivity of
the subject the operations are kept anonymous and only
Contrary to Laubscher’s theory for the progressive a summary of the learnings are itemised below.
development of interacting draw zones and drawbells,
recent physical modelling experiments by Castro [9], • Where a void forms between the cave back and the
using aggregate material, demonstrated that interaction caved muck pile, free surface rilling is a
of these movement zones did not increase the size of the significant mechanism of material flow,
extraction zones, see Figure 5. particularly where differential draw prevails.
Under these conditions the shear dilation of the
material leads to a porosity increase causing IMZs
to grow quite wide and rilling occurring as the
predominant mechanism for lateral movements.
This scenario is very different to that for high rather that rapid cave propagation tends to lead to
column caves with no airgap. the formation of narrower draw zones.
• Similarly, an important observation from borehole • IMZ diameters do not appear to grow laterally
cameras and flow marker recoveries is that with height in the far field (i.e. away from the
surface/interface rilling is a dominant mechanism influence of the drawpoint). There have been
of lateral movement where the cave boundaries numerous observations in the field where draw
(cave back or cave walls) are inclined and develop zone geometries resemble narrow cylinders rather
‘overhangs’. Interestingly, it appears that than ellipsoids.
boundary drawpoints are often characterised by a
distinct large proportion of both fines and coarse • Generally speaking, resource recovery modelling
fragmentation (i.e. a bimodal fragmentation results (which are based on ellipsoidal IMZs and
distribution). A plausible explanation is that the IEZs) tend to understate initial grades and
material up against solid boundary walls tends to overstate grades in the latter mine life. The likely
dilate and shear, thereby increasing the porosity of reason being that in the models the lateral IMZ
the material and promoting fines migration (well ellipsoidal growth occurs too early in the
described by void diffusion theory). An example production cycle thereby stinting the early vertical
of the extremely mobility of this zone is illustrated IMZ growth and forcing early interactive draw.
by the recovery of a physical marker which As a result any overlying dilution is intercepted
travelled around 420m in just 6 days. and mobilised too late thereby under-predicting
the real impact of dilution.
• Cave boundaries must be incorporated as a
constraint into all cave flow / resource recovery • For short block heights (particularly in panel
models. Best results will undoubtedly be caves), rilling may well prove the dominant flow
produced from coupling cave propagation and mechanism occurring at the advancing cave front
cave flow models where the dynamic relationships and thus mask or avert the effects of narrow draw
between the two can be simulated iteratively. zones. However in caves with high block heights
and overlying waste, narrower (than modelled)
• Cave flow markers installed in areas unaffected by IMZs will inevitably accelerate dilution entry and
surface rilling tend to reveal near vertical if no draw interaction occurs before this time then
movement in the cave column. the potential for early drawpoint shut-off is a
significant risk.
• Cave markers have reaffirmed the obvious
conclusion that IMZ width is directly related to the
competency of the rock mass (i.e. average 4. MODELLING THE RECOVERIES FOR
fragmentation size).
MULITPLE CAVE BLOCK HEIGHTS
• Dilution entry, measured from unambiguous Numerical models offer significant advantages in flow
marker horizons and physical markers, has modelling, as they enable the user to apply flow rules to
revealed that draw zone diameters (IEZ) are block model data and produce recovery results for scales
significantly narrower than predicted or modelled. up to life of mine. For most modern mines, budgeted
Back-analysis of draw zone diameters (assuming a grade recovery is predicted through the use of numerical
constant bulking factor of 20%) have verified flow models. The way these models work and how
widths of as low as 6m-9m (or greater than 50% those are used therefore has a significant influence on
reduction from predicted and modelled). the assessment of the viability of a project.
This finding, although alarming, is contrary to Inexperienced model users may incorrectly believe that
findings arrived from back-analysed end-of-life these models are capable of predicting for the user what
resource recovery calculations which typically the draw zone (IEZ) diameters will be. In fact, it is the
demonstrate high metal recovers overall. It is not user of the model who inputs parameters which
uncommon for caving operations to claim over determine the flow width to be modelled. As such, these
95% recovery from around 120% of the tonnes. models are the means by which assumptions on IEZ
• Where IMZs appear narrow the rate of cave width are translated into revenue forecasts. This can
propagation is more rapid. This is not to say that lead to significant problems if insufficient scrutiny is
narrow IMZs result in rapid propagation (in fact placed on the modelling parameters being input into
there is considerable evidence to prove that these models, or on the capacity of the models
propagation is often unrelated to production), but themselves to model flow and recovery in a complex
environment.
The choice of model to be used for a specific task fine and very fine fragmentation profiles (fragmentation
depends on the requirements of the user but calibration was modelled as uniform) and for a single (600m block
for forward prediction is required for each of the height), double (2 x 300m block heights) and triple (3 x
methodologies chosen. 200m block heights) block caves (as depicted in Figures
6-8).
• If a large number of modelling simulations are
needed under non-complex cave propagation and coarse Medium fine Very fine
flow circumstances, or the generation of
schedules, empirical rules based models can be
appropriate. Where possible in Greenfields
situations, models should be calibrated against
appropriate benchmarks. Rule based models are
most effective at brown-fields operations when
site calibration data is well established from
periods of previous operation, allowing reasonable
calibration of the model.
• If it is required to determine realistic flow shapes
or which need to retain some integrity with
physical properties then hybrid empirical-physical
laws models can be used These models produce Figure 6. CA model illustrating the collective IEZs that
less rapid results, but can calibrate results to develop in a single lift cave for varying fragmentation profiles.
realistic flow profiles (dependent on the skills of
coarse Medium fine Very fine
the user).
• If more detailed understanding of the particle to
particle interaction is required, leading to an
understanding of mixing and variability, or the
impact of cave propagation variations, particle to
particle based models may be more appropriate
These programs have the capacity to model such
particle to particle interactions, and produce output
that is more in keeping with what is understood
about flow in full scale. Unfortunately these
models demand significant computer processing
power and as such the size of the model is limited
to simulate a handful of drawpoints at best.
4.1. Example Evaluation Of an Orebody Using a
Cellular Automata Flow Model Figure 7. CA model illustrating the collective IEZs that
Model Design develop in a two-lift cave for varying fragmentation profiles.
coarse Medium fine Very fine
To demonstrate how modelling is used to evaluate the
viability of a high block orebody, several simplistic,
cellular automata (CA) models of a hypothetical orebody
were constructed and run for various block heights and
various fragmentation profiles.
The hypothetical model comprised of a 200m wide by
200m long by 1200m high block, the lower 600m
orebody contained a uniform grade throughout and was
overlain by 600m of waste at zero grade.
The mining footprint contained an area of 150m x 150m
and comprised of 64 drawpoints spaced on a 14m x 14m,
offset herring bone layout. Production assumed uniform
draw down and excellent draw control (rarely achieved).
Figure 8. CA model illustrating the collective IEZs that
The objective of the modelling was to evaluate the develop in a three-lift cave for varying fragmentation profiles.
percentage recovery of the orebody for coarse, medium,
It should be noted that the CA model implicitly
Grade Profile For 1, 2 & 3 Three Lift Caves
simulates the effects of fragmentation size as a function
100
of the drawzone dimensions, i.e. coarser fragmentation is
represented by a short, wide ellipsoid whilst finer
fragmentation is represented by a taller, narrower

Grade %
ellipsoid. This is controlled through the use of a 80
probability factor setting in the CA modelling software.

60
Model Results
27 29 31 33 35
Production Draw 37 39 41
The results proved typical of all flow models based 1 Lift Very Fine 2 Lifts Medium 3 Lifts Coarse
on the concept of ellipsoidal draw zones. The final
recoveries are tabulated in Table 1 below. Figure 10. CA Modelled recoveries for more realistic
fragmentation profiles associated with comparative multi-lift
# Lifts block caves.
Fragmentation
1 2 3
coarse 99.99 99.57 98.79
medium 99.82 98.50 96.76 5. THE EFFECTS OF CAVE COLUMN HEIGHT
fine 99.45 96.93 94.29
Very fine 97.53 91.32 87.49 Assuming interactive theory is accepted, to determine
the height of the HIZ the engineer must calculate the
Table 1. CA modelling recovery results run for multi-lift caves average width of the draw zones to determine where and
with varying fragmentation profiles. when they will interact. In doing so the engineer must
first decide whether the draw zones develop as
expanding ellipsoidal IMZs or cylinders with a constant
In essence, the results demonstrate that the greater the diameter as under draw.
number of lifts the lower the recovery, particularly for
orebodies with fine fragmentation, the reason being that A summary of the important issues and risks to be
the waste is ‘tapped’ sooner and dilutes the orebody considered in the evaluation of high lift caving projects
earlier (refer to Figure 9). are:
However if one assumes that the fragmentation for the
single lift cave is finer due to the greater overburden (i) Interactive draw may never be fully achieved for
loads and greater comminution in the column, then high column caves, particularly across the major
recoveries may prove contrary, as indicated by the apices. And void diffusion may prove the
hypothetical line in Figure 9. Furthermore, Figure 10 dominant flow mechanism.
illustrates that although dilution entry may be delayed by (ii) Where interactive draw does occur, it could well
reducing the number of lifts, the rate of dilution disappear in time with increasing draw column
increases with lift height. heights due to extensive fining of material through
secondary fragmentation and higher confining
Modelled Recoveries For 1, 2 & 3 Lift  (tangential) stresses around the draw zone. Thus
the draw zone diameters could ‘shrink’ as a
Block Caves function of the diminishing fragmentation and any
100 uniform mass flow (i.e. the HIZ) will disappear.
Modelled Block  Recovery %

95 (iii) The possible reduction in draw zone width over


time would serve to accelerate the vertical
90 movement of material from above, if this is waste
then dilution entry may well occur earlier than
85 predicted by models that assume interactive draw
1 2 3 over the life of the cave.
# Lifts
coarse medium fine Very fine Anticipated Trend
This reduction in draw zone widths over time
could sterilise large portions of the ore column.
Figure 9. Modelled recoveries for multi-level block cave lifts
with varying fragmentation profiles. (iv) High abutment stresses are inevitable, thus an
advanced undercut methodology with minimal
extraction level development ahead of the
undercut must be adopted to minimise/avoid the amount of development, provided there are sufficient
abutment stress damage. drawpoints. This all is only permissible if an
economically viable quantity of ore is recovered at an
(v) Vertical point loading from the overlying column
acceptable dilution content” [8].
will prove a common phenomenon that can only
be controlled through strict draw control practice
The ultimate decision facing the design engineer is what
and uncompromising draw compliance.
drawpoint spacing to adopt that will ensure the greatest
(vi) The drawpoint and brow should be very well recovery of the resource over the life of the cave. It may
supported with a strong lining to minimise brow well be that other constraints (such as rock mass strength
wear. Remedial support activities often lead to and stress) ultimately dictate the final drawpoint spacing
undesirable consequences such as exacerbating so as to ensure the integrity of the production and/or
point loading and irregular draw ultimately undercut levels over the life of the cave, nevertheless the
leading to poor recoveries. models used to predict resource recovery, which
ultimately feed into the financial evaluation, must
(vii) The shape of the cave boundaries may vary
represent reality so that an informed assessment is made
significantly due to the orientation and magnitude
of the project risk. Experience from operating caves
of the induced stresses and thus a vertical
suggest that this has not been the case and as a result
projection of the cave walls from the footprint is
significant reserves have been unrecovered.
highly unlikely and unrealistic.

6. CONCLUSIONS REFERENCES
Current flow models are an integral part of the reserve 1. International Caving Study II, 2005. End of Project
estimation process, but presently do not completely ICSII Report. JKMRC and Itasca Consulting Group,
explain outcomes we see in operating block caves and Inc: Brisbane.
thus warrant further development. For example, the 2. Kvapil, R., 1965. Gravity flow of granular material in
importance of the propagating cave back and cave walls hoppers and bins-Part 1. International Journal of Rock
on cave flow is clearly evident and is instrumental in Mechanics and Mining Sciences. Vol. 2,. 35-41.
promoting the rilling mechanism often observed in 3. McCormick, R.J., 1968. How wide does a drawpoint
operating mines. In addition, the development of static, draw? Engineering and Mining Journal, June. 106-116.
cylindrical draw zones or ellipsoids that develop
adequately narrow draw columns close to cylindrical, 4. Marano, G. 1980. The interaction between adjoining
drawpoints in free flowing materials and its application
must also be incorporated into flow models so that their
to mining, Chamber of Mines Journal, Zimbabwe, May
effects on recoveries can be quantified and financial 2001. 25-32.
implications evaluated.
5. Laubscher, D.H. 1994. Cave mining – the state of the
The arguments presented for a critical review of the art, The Journal of The South African Institute of
empirical methods used for designing and evaluating the Mining and Metallurgy. October 1994.
viability of high block caves is not a criticism of the
6. Janelid, I. 1972. Study of the Gravity Flow Process in
pioneering work conducted in the past. On the contrary Sublevel Caving. International Sublevel Caving
it is a criticism of the ‘blind application’ of the empirical Symposium, Atlas Copco, Stockholm.
methods and failure to incorporate new learning’s to
improve the tools for current applications. Laubscher 7. Pierce M, E. 2009. A model for gravity flow of
fragmented rock in block caving mines. PhD thesis,
clearly understood and articulated the problems that face
University of Queensland.
design engineers and his following statements ring truer
now for modern, high lift caves than for those during the 8. Laubscher, D, H. 2000. Chapter 26, Drawpoint
time at which it was written: Spacing, Block Caving Manual. Prepared for
International Caving Study. JKMRC and Itasca
“Drawpoint spacing is one of the most important and Consulting Group, Inc: Brisbane.
controversial items in cave mining. Often mine planners 9. Castro, R. 2006. Study of the mechanisms of granular
will only consider spacings that have been or are in use flow for block caving‘, PhD thesis, University of
regardless of the mining environment or the rock mass. Queensland.
The object of this investigation is to present sound
design parameters backed by underground experience.
There are good reasons to increase drawpoint spacing so
as to improve the strength of the extraction level, larger
and longer drawpoints for larger LHD’s and to reduce

You might also like