BLOCK CAVE PRODUCTION SCHEDULING USING PCBC

Tony Diering, Gemcom Software International Inc., Vancouver, Canada

Otto Richter, Gemcom Australia, Perth, Australia
Daniel Villa, Gemcom Software International Inc., Vancouver, Canada

over the years. These companies are listed in the

Abstract acknowledgements section of this paper.

Gemcom PCBCTM is a software package which Typical project workflow

has been developed over the last 22 years for the
planning and scheduling of block cave mines. This A significant number of block cave projects
paper presents an update of the various research and have been studied using PCBC over its 20 year
development activities done to PCBC recently. It history. During that time, a well used work flow has
also provides an overview of the current capabilities evolved which is usually used as a guideline for new
of the software including tools for both feasibility projects.
type studies as well as tools for operating mines. Conceptually, the steps are as follows:
PCBC is used extensively by prospective and • Figure out what is in the ground (geological
operating block cave mines and some of the recent model)
applications are described. • Work out where you want to mine (X,Y,Z limits)
• Work out the tons and grade that you will get
Introduction from those limits
• Work out how long it will take (time)
History • Optimize and iterate to add further value to the
PCBC was first developed in 1988 for the project.
Premier Diamond Mine in South Africa(Diering, • Repeat the whole process every time a new
2000). In 1992, the first production scheduler was geological model is produced as the project
added to the system and in 1994 a significant upgrade evolves.
was done for Northparkes Mine in preparation for The overall steps in the process are described
their Lift 1. PCBC was upgraded to Microsoft® below:
Windows® operating system in 1996. In 2002, the
Cave Management System (CMS) was developed for • Footprint Finder. This is an application which
Freeport DOZ mine and this was upgraded to provide works off the geological block model and whose
SQL Server® database support in 2003 for Finsch primary objective is to help assess the best
Diamond mine(Diering, 2004). A new algorithm elevation (or elevations) for the block caving
called Template Mixing (Diering, 2007) was added to footprint (Z extent of the mine)
provide better and alternate flow modeling options to • Generate draw points. Setting up draw points
users. requires assessment of draw point spacing,
tunnel orientation etc. (X,Y extents)
More recently, various enhancements to the
• Construct slice file. This is the process of
production scheduler and other areas of the program conversion of a geological model to be aligned
have been completed some of which are described with the draw points such that each draw point
here. has an in-situ (un-mixed) resource above it.
It is worth noting that the block cave market (in • Compute best Height of Draw (HOD). Each
terms of software) is very limited. As such, it is draw column is evaluated to assess the best or
difficult to fund high quality research and program highest dollar value which can be achieved for a
development. We at Gemcom Software International given set of mining costs and product revenue
Inc. have been very fortunate to have companies and recovery factors.
sponsor custom development activities within PCBC • Production scheduling. This is the heart of the
PCBC system. It is important to distinguish

1
 between production (tons and grade produced) framework has proven invaluable over the years and
and development scheduling (tunneling and has allowed our development efforts to focus on the
development). PCBC does production block cave part of the problem minimizing the need
scheduling. This provides the tons and grade to develop and maintain the underlying graphical and
forecasts for the project which has been database subsystems.
described as “the mine planners promise to the
shareholders as to what the mine can produce”.
• Advanced schedules. No schedule is ever
complete or final. During the project evaluation
stage, new pricing or geometrical options will be
considered and new geological models generated
as the exploration drilling progresses. During
production, new schedules are generated
whenever the actual production varies from the
plan (which is always). So the need for a
scheduler which can run in typically less than 20
to 30 minutes per run is important.
• Operating mine set up. Once a mine is going
into production, then it is possible to set up a
Figure 1 Typical view of PCBC running inside the GEMS
database to store production tons and draw point
general mine planning package
assay and other observational data. The
importance of accurately recording and Components of PCBC are described below
managing the tonnages extracted from each draw including initial assessment of footprint location,
point has long been recognized. model set up and mineable reserve assessment, then
• CMS can be used to help manage the daily (or scheduling and production management.
shift based) draw order. This is the daily
tonnage target set for each draw point. This is Foundation
essential if a managed block cave is to be
maintained. This is the framework within which PCBC
operates. The various components are summarized
• Geological/geotechnical monitoring. Tools have
as follows:
been developed within PCBC to help store,
display and analyze observed data • Graphical interface
• Least Squares (LSQ) and grade reconciliation • Blocks
can be used to base schedules on observed assay • Lines
data instead of block model data for more • Points
accurate schedules • Triangulations
• During the above process, it is essential to have • SQL database / workspaces
appropriate tools to interrogate and query the • Profile editors / parameter management
results generated. Over the years, a substantial
toolbox has evolved based on project and user Footprint Finder
requirements.
Input for Footprint Finder utility comes from a
In this paper geological block model together with mining costs,
revenue factors, etc. The program will look at each
This paper describes the various components in level in the block model and then construct vertical
the PCBC product. It is not intended to provide any columns accumulating the dollar value. Vertical
explanation as to how these components work or are mixing is applied to each column using an algorithm
used. based on Laubscher’s mixing method (Laubscher,
1994).
Components of PCBC
This is very useful to obtain an initial idea of
The various components of PCBC have evolved where to locate a footprint and what the initial
to support the above project workflow. PCBC runs footprint shape might be. Figure 2 shows
inside the Gemcom GEMSTM mine planning package accumulated columns plotted according to value.
developed by Gemcom Software International Inc.
(Figure 1) The ability of PCBC to work inside of this

2
 with the process of defining a reasonable
economically and geotechnically feasible outline.

PCBC
Overall steps of a typical project (from the
program, not project perspective) are as follows:
• Set up the initial working environment inside a
GEMS project.
• Slice file construction (Figure 5). This is an
integral part of the process. Utilizing user-
defined draw cone shapes, a column of rock
above each draw point is simulated and stored in
what is termed a slice file. The term slice as the
total column is broken into slices which match
the vertical spacing of the geological block
model.
Figure 2 Footprint Finder example on one level

Repetition of this process on each level allows

the tons and value generated to be plotted as shown
in Figure 3.
Tons Dollar value

1,000 9,000
900 8,000
800 7,000
Dollar Value (M$)

700 6,000
Tonnage (Mt)

600
5,000
500
4,000
400
300 3,000
200 2,000
100 1,000
- -
Figure 5 Schematic of block model to slice file conversion
2560
2620
2680
2740
2800
2860
2920
2980
3040
3100
3160
3220
3280
3340
3400
3460
3520
3580
3640

Elevation Draw point locations are used to construct a

vertical column which is then intersected with the
Figure 3 Footprint Finder : Tons and dollar value vs block model. The various overlaps of the draw cones
footprint elevation are resolved so as to not double count material and
this is accumulated into the slice file for each draw
An example from Freeport’s DOZ mine is shown in point. This is referred to as a “NoMix” slice file,
Figure 4. since no material mixing has yet been applied.
Figure 6 shows a section of a block model and
the resulting “NoMix” slice file with one column per
draw point.

Figure 4 Footprint Finder example (DOZ mine)

The higher grade zones are shown in warmer

colors. This type of value plot assists considerably

3
 Figure 7 shows that the results from Footprint
Finder and Best HOD tools are typically quite similar
as one would hope.
Once the basic preparation work has been done,
production schedules can be generated.
A typical schedule requires input of the
following key components (Figure 8):
• Sequence to develop the draw points (and
undercut)
• Constraints on the maximum draw rate which
can be applied to draw points
• Tonnages required in each scheduling period
• Information to control the cave shape. It is usual
to look at different strategies and compare
• Numerous other inputs, constraints and reporting
control options

Figure 6 Block model to slice file conversion

• A variety of material mixing algorithms may

then be applied to the slice file to simulate the
actual material mixing which takes place as
material is extracted from the draw points.
• Best HOD. The Best HOD utility will
accumulate tons, grade and dollar value in each
draw column (after application of vertical
mixing) to provide an estimate of mineable
reserves for different footprint shapes. As this Figure 8 Production scheduling components in PCBC
process uses actual draw point locations and
assumed draw column shapes, it is generally A basic schedule will open draw points
considered more geometrically accurate than the according to the sequence and deplete tons from each
Footprint Finder. according the Production Rate Curve (PRC), apply
material mixing if required and report tons and grade
mined in a variety of formats.
An advanced schedule could look at changing
parameters for individual or groups of draw points,
adjusting the schedule to past tons mined, having the
HOD profile follow a given cave shape and adding
information to report undercut tons separately from
production tons.
Experience has shown that it is very useful to
have a clean and efficient interface between PCBC
and Microsoft® Excel®. This allows reports to be
generated in a format which can quickly be further
analyzed by engineers.
In addition, when dozens or hundreds of
schedules are being run, it is useful to have what is
called a “playback tool”. This allows various aspects
of the schedule to be studied visually to look for
Figure 7 Best HOD based footprint (bottom) vs Footprint trends (or data input errors!)
finder result (top) (Cadia East)

4
 The production scheduler can just as easily be Each draw point can be categorized in a variety
used for forward looking schedules or for analysis of of ways (Figure 10), including over-draw, under-
past performance. This is very useful for grade draw, normal, draw-bell development, wet muck
calibration and reconciliation purposes. (which is a safely concern) or as requiring special
treatment. The tonnage for each category is set
Cave Management System accordingly.
CMS was originally developed for Freeport
DOZ mine and then further refined for use at Finsch
mine. Currently there are seven mines using or
planning to use CMS. CMS aims to generate a draw
order for each draw point every day or shift. It uses
the recent historical (actual) tonnages to adjust and
manage the draw and provides the supporting
database, reporting and user interface to facilitate this
process. At De Beers Finsch mine, CMS has been
closely integrated with the Sandvik Automine® Figure 11 Excel “map” format for draw point result display
system.
Figure 11 shows an example of daily production
data displayed using Excel. It is important to have a
clean interface between the CMS database and Excel
for ease of analysis by the draw control personnel.

LSQ
The LSQ tool is intended for operating mines.
Once a mine has been in operation for a few years, it
will likely have a draw point sampling program. The
draw point assay values can be stored and sorted per
draw point and then composited into 10m or 15m
intervals to provide some averaging of the highly
variable assays.
Figure 9 How CMS fits in between historical tons and
future plans Subject to a variety of constraints, a least
squares trend line is put through the composites and
Figure 9 shows how CMS fits in between the then this can be extrapolated for a short distance up
historical tons mined and the requirement to adjust the draw column into what is essentially the un-
the plan of the next few months (using PCBC mined part of the column (Figure 12).
schedules) in a process called “Catch-up” to fit in
with the long term plan. (Diering, 2004) This becomes particularly useful when the draw
point assays suggest that the draw point should
remain open (usually after 100% draw) when the
slice file values suggest that the draw point should be
closed. For draw points where the sample trend
differs from the slice file, then the slice file values are
replaced with the sampled values for selected draw
points. This is somewhat similar to the open pit
practice of taking blast-hole samples to improve the
local grade of a bench about to be blasted and mined.
Figure 12 shows a single draw column with
sampled values at various heights (HOD) above the
draw point. The graph shows these together with the
trend line and some extrapolated points. Maximum
and minimum grade values are set so that steep up or
Figure 10 Categorization of draw points for priority downward trends do not generate unrealistic grade
assignment values.

5
Figure 12 Sample compositing and trend line analysis in
LSQ

The LSQ tool can either be run as a stand-alone

Figure 13 Example of graphical display of geological data
tool or right within the PCBC production scheduler.

Operations tools and reconciliation

Once a mine is in operation, there is a variety of
useful ways in which draw point sampling data
(grade, geotechnical and geological) can be displayed
and analyzed. A key reason for doing this type of
work is so that we can better understand if or where
there is irregular behavior with the cave itself. Some Figure 14 Example of residual slice file to block model
of the analysis types are listed below: conversion
• Use of draw point assays for grade reconciliation Figure 14 shows an example in which the
and for calibration of the model residual slice file (after simulation of mining the full
• Use of draw point assays for improved short block cave tons) is used to re-estimate a geological
term forecasting using the LSQ tool already block model. This block model can then be used for
mentioned. planning of another future mining block.
• Use of geological samples to supplement the
reconciliation process or to better understand Recent Developments
horizontal and vertical migration of material
within the cave (Figure 13)
PCBC is over 20 years old and hence should be
• Use of geotechnical (fragmentation) data to considered as a mature product. As such the basic
better understand the relationship between draw planning and scheduling work flow is well covered.
rate at draw points and rock type or mining area. On the other hand, being mature and with a good
• Use of convergence data in production tunnels to foundation, PCBC has provided a useful foundation
help prevent excessive closure (or collapse) of for a number of recent developments which are
these tunnels. Freeport has shown quite summarized in this section.
convincingly at their DOZ mine that a diligent
program of monitoring convergence in these • Upgrades to material flow tools
tunnels is beneficial. In areas where high • Playback utility
convergence rates are observed, adjacent draw • Display tools
points have an increased tonnage target which
tends to relieve the high stresses. Each of these is considered in more detail
• Seismic data and/or extensometers can be used to Material flow upgrades
help predict the location of the cave back which
can in turn be used to set up surfaces for Material flow is an integral part of the block
simulation of the “rilling” process which cave mining (and material depletion) process. The
migrates material non-vertically. entry of dilution is a significant factor in the planning
• The residual slice file model can be used to re- process and modeling of this has proven to be
estimate a block model which can be used either difficult. As such, a variety of different mixing tools
in a multi-lift mining situation or as part of a new have been developed within PCBC and users are then
block model for a “super-pit” which some mines given the choice as to which approach they would
are considering. like to adopt. This is summarized in Table 1.

6
 Method Ease Linear? Comment
of use

No mixing Easy Yes No mixing base

case (In-situ)

Pre-Vertical Easy Yes PCBC “Default”

(includes pre-
mixing
erosion)
Figure 15 Movement mechanisms in a block cave
Laubscher Easy Yes Uses Laubscher
tables It differs from other material flow algorithms
mixing such as discrete particle and cellular automaton
methods. A major advantage of Template Mixing is
Sequential Harder No Older method, its speed. Figure 16 shows a few steps in a depletion
mixing includes simulation. Blue represents dilution, yellow is ore
toppling and the intermediate colors represent progressive
mixing as the ore is extracted.
Template Harder No Most flexible
Mixing option available

REBOP™ Harder No Not yet

interface generally
available

Table 1 Material mixing options in PCBC

The linear methods can be applied with the Best

HOD utility to find mineable reserves before the
schedule is run. For the non-linear methods, the
mineable reserve will be a function of the mining
sequence and draw strategy. Therefore, mixing has Figure 16 Template Mixing 2D example
to be built right into the production scheduler. This is
one of the key differentiators between PCBC and Figure 17 shows an example for our “sand-box”
other commercial scheduling tools. project using toppling, rilling and normal mixing.
The pre-mix option in PCBC was recently
upgraded to allow for the inclusion of a draw cone
“erosion” mechanism. This is useful as there is
increasing evidence that draw cone radii may not be
as large as is often hoped and also that the draw cone
radius changes with time. A fraction of each slice is
frozen and then an erosion rate is specified which
allows this material to be mixed with material higher
up each draw column.
Template Mixing was introduced in to PCBC in
2006. (Diering, 2007). It allows a variety of mixing
mechanisms to be simulated including vertical
mixing, rilling, toppling and fines migration. (Figure
15)

7
 Figure 19 Block model used for PCBC/REBOP calibration

Results from the calibration exercise were really

encouraging, suggesting ways to improve both the
PCBC and REBOP modeling processes. A sample
Figure 17 Sandbox example with toppling, rilling and calibration curve of Cu grade is shown in Figure 20.
vertical mixing
Cu% Rebop vs PCBC (N2E5)
In 2008, a joint initiative was done with Rio
2.00
Tinto, Itasca and Gemcom to provide the potential to 1.80
combine the PCBC and REBOP programs. This was 1.60
done using a hand-shake mechanism so as to 1.40
1.20
minimize the changes required to each program and
Cu %

1.00
to keep them as independent as possible to facilitate 0.80
Cu%_M3P3
future development. 0.60 Cu% Tm4
0.40
0.20
0.00
Jan-08

Jan-09

Jan-10

Jan-11

Jan-12

Jan-13

Jan-14
Jul-08

Jul-09

Jul-10

Jul-11

Jul-12

Jul-13

Jul-14

Figure 20 Example calibration curve for PCBC (orange) vs

REBOP (green)

Figure 18 REBOP results displayed in PCBC (Markers left The original version of PCBC used what we
and cones to right) term “Laubscher mixing” (Laubscher, 1994). This
was replaced by pre-vertical and sequential mixing
Figure 18 shows two examples of REBOP
options in 1994 in PCBC. However, there are still
results plotted within the PCBC program. The results
projects (or people) who like to be able to compare
from REBOP are used directly in the production
back against the Laubscher mixing. So it was re-
scheduler and also for modification of the slice file.
introduced into PCBC in 2008. It is also useful for
As a separate, but related project, Gemcom comparison against Footprint Finder results which
worked with Rio Tinto to calibrate PCBC and use the same mixing.
REBOP against one another. Figure 19 shows the
geometry of the calibration problem. 50 fictitious
draw points were located in this block model for
testing purposes.

8
 Figure 22 Excel map transfer utility example

The Excel map format is useful for a single

attribute per draw point such as HOD. However, if
Figure 21 Dilution entry. Pre-vertical mixing (top) vs one has multiple attributes (which sum to 100%), a
Laubscher mixing (bottom)
very useful display option is the pie chart. The
Figure 21 shows an example comparing dilution program will plot a pie chart at each draw point
entry from a single draw column. PCBC pre-vertical location using data directly from the underlying
mixing has more of an S-curve dilution entry where database from what is called multi-bucket format.
Laubscher mixing has a straight line dilution entry (Figure 23)

Playback Utility
This tool is used to “playback” or study results
from a production schedule. Display options include
Triangular Irregular Networks (TINs), contours, pie
charting, and 3D columns. Playback examples are
shown in the later sections on Freeport DOZ and
Palabora.

Display tools
Over the years, a variety of different graphical
display tools have been developed. The more recent
ones are the Excel interface which allows any draw
point related data to be exported directly into Excel in
the correct cell row and column positions for direct
display in Excel as shown in Figure 22 Figure 23 Example of Pie chart display

In addition to the static displays available, a

more dynamic display of selected information for
individual draw points by right click or mouse
movement over draw points can be very useful
(Figure 24)

9
 Figure 26 Height of draw profile at Freeport DOZ mine
from Playback tool

Freeport Grasberg
The Grasberg block cave is scheduled to start
Figure 24 CMS “control panel” with right click and production as the Grasberg open pit slows down at
display information
the end of its life. (Figure 27) (Brannon, Casten, &
Other options are size based plots (Figure 13), Johnson, 2004) This will be a very large block cave
3D draw columns (Figure 14) and plotting of draw with production up to 160,000t/d. Numerous
points in appropriate shapes. scheduling options have been evaluated using PCBC
and particular emphasis has been placed on effective
modeling of large open pit failures which will
Project examples generate additional dilution material.
Freeport DOZ
PT Freeport Indonesia has been using PCBC
since around 2000. They are currently mining close
to 80,000 t/d, making it a large block cave mine (T.
Casten, 2008). PCBC and CMS are used extensively
for planning and scheduling at the DOZ mine as well
as for daily draw control. Figure 25 shows a plot of
forecast rock types at one step during a production
schedule. Figure 26 shows a plot of HOD for the
same mining step. Figure 27 Grasberg block cave in close proximity to the
large open pit

Figure 27 shows the proximity of the block cave

draw columns to the large open pit.

Figure 25 Forecast rock types at Freeport DOZ mine

generated in Playback tool

10
 Salvador
The Salvador mine in Chile has used PCBC
both for the detailed scheduling of individual mining
panels (Figure 30) as well as for combined
scheduling of multiple mining blocks (Figure 31).

Figure 28 Column values from Footprint Finder used for

footprint assessment

Figure 28 gives an idea of the variability of the Figure 30 Slice file display and layout at Salvador Mine
orebody edges and also alludes to the difficulties in
sequencing and scheduling such a large orebody (grid
size above is 200m!).

Northparkes
PCBC was first used for Northparkes E26 Lift 1
around 1994 and then for Lift 2 planning and
currently for Lift 2 North (Figure 29) (Ross, 2008)
and E 48. Each lift has provided surprises and
challenges from a modeling perspective.

Figure 31 Scheduling of multiple mining panels at

Salvador Mine

As this is an older mine, current work is looking

to re-estimate the residual grades in older mined out
areas for use with future planning.

Figure 29 Northparkes E26 mining

11
Andina

Figure 32 View of three panels (lifts) at Andina mine Figure 34 Monthly tonnage display (poor draw control
(top) and good draw control (bottom)

Palabora went through a period during which it

was difficult to achieve good draw control.
However, more recently, the draw control has been
much improved. (Pretorius & Ngidi, 2008) This is
shown clearly in Figure 34.

Figure 33 Plan of Andina third panel showing grizzly and

LHD sectors and existing development
Figure 35 Seismic data display example from Palabora
PCBC has been used extensively at Codelco’s
Andina mine for a number of years. Challenges in Figure 35, also from Palabora shows a plot of
modeling this deposit include the multi-lift aspect seismic events for one month together with draw
together with reliable estimation of residual grades of points and the cave Height of Draw profile (which is
mined out blocks (Figure 32), the effective different from cave back profile).
scheduling of grizzly and LHD sectors (Figure 33),
the sheer size of the project and caving issues related Ridgeway
to primary and secondary rock types.
The Ridgeway Deeps Mine of Newcrest used
Palabora PCBC with the Template Mixing option to study
rilling and how this impacted the mineable reserve
The Rio Tinto Palabora mine in South Africa and overall production schedule (Burgio & Diering,
started block cave production in 2000. (Moss, 2008) (Figure 36)
Russell, & Jones, 2004)
As the scheduler can work with historical
tonnages as easily as forward looking tonnages, the
playback tool can thus also be used for historical
analysis or reconciliation purposes.

12
 Figure 38 Schematic of open pit, cave zone and Block 4
draw points at De Beers Finsch mine

Figure 36 Section of Ridgeway deeps block cave model Figure 39 shows a section with some of the
showing irregular cave propagation on right side residual draw columns. These are trimmed against
the known topography and the new failure material
Different scenarios were modeled to see the
(red) then starts to mix with the existing material
effect of limited cave propagation on the East side of
(blue). The mixing zone is shown by the
the cave.
intermediate colors.
Cadia East
The Cadia East project of Newcrest provided
interesting modeling challenges as it is a large multi-
lift project. (Figure 37) Extensive use has been made
of the Footprint Finder tool to assist with
determination of elevations together with more
accurate schedules from PCBC.

Figure 39 Addition of new failure material and mixing of

this material with existing cave rock mass

Figure 40 is similar to Figure 39 except that the

sequential mixing in PCBC is turned off. This
example shows the importance of being able to
Figure 37 Multi-lift example from Cadia East, Newcrest model this process in a non-linear manner. The final
mineable reserve is required to be adjusted monthly
Finsch or every time the failure surface is modified.
Additional tools in PCBC allow for the addition of
The Finsch Block 4 block cave is an example of
anticipated material for the remainder of the life of
mining beneath an old open pit (Richter & Diering,
the Block 4 block cave as well.
2004) (Figure 38). As mining progresses, additional
pit wall material is failing into the developing cave.
The remaining ore and ore/waste combination has to
be continually updated as additional material fails
into the cave.

13
 Geological observations are made routinely at
draw points of up to 8 different rock types. These
were also fed into the geological block model so that
comparisons could be made of the observed vs model
rock types. Figure 43 shows the modeled rock types
vs time and Figure 44 the observed rock types vs
time. A detailed study of the differences between the
two can be very informative and lead to ways to
improve the model which may not be apparent from
the grade model / assays. The geological modeling
thus provides another “dimension” into the
calibration process.
Figure 40 Addition of new failure material without
additional mixing

Calibration examples
Freeport DOZ
A detailed description of the calibration curves
in figures Figure 41 and Figure 42 is beyond the
scope of this paper. (Villa, Prasetyo, & Diering,
2008) Figure 41 is for grade and it shows the extent
to which the PCBC model can be changed to improve
the fit actual against actual observations.
Figure 43 Geological composition from block model and
PCBC production schedule

Figure 41 Freeport DOZ. Calibration of grade

Figure 42 is for the Marble rock type. It shows

how the original PCBC marble curve (bottom) can be
changed to more closely approximate the geological Figure 44 Geological composition based on draw point
draw point observations (higher curves). observations

Palabora
Work has been done at Palabora to improve the
short term grade estimates using the LSQ tool
described above. Figure 45 shows the improvement
in the short term comparing the PCBC LSQ forecast
vs Samples.

Figure 42 Freeport DOZ. Calibration of Marble rock type

14
 Figure 47 Various PCBC runs vs assayed results for IW
sector, Salvador mine for 8 years

Figure 45 Measured vs PCBC standard and LSQ adjusted

grades

Figure 46 shows a similar set of graphs, but Figure 48 Various PCBC runs vs assayed results for ICE
comparing hang up frequency. In this case, there was sector, Salvador mine for 17 months
no initial model forecast for hang-ups, but based on
the LSQ approach, a reasonable forecast for short Example of block model adjustment
term hang up frequency was achieved. In this example (Figure 49Figure 35), various
attempts were made to calibrate the PCBC results
with the observed mill feed grades. However, the
PCBC grades were too high irrespective of the
mixing parameters used. This is an example in which
the underlying block model is “at fault”. Re-
estimation of the block model with different
interpolation parameters has largely resolved this
discrepancy.

Figure 46 Measured vs LSQ modeled Hang-ups

Salvador
Various calibration runs were done at Salvador
mine. Two examples are shown in Figure 47 and
Figure 48. A detailed explanation of the curves is
beyond the scope of this paper, but the graphs show
how mixing parameters were adjusted to improve
both the model results and the confidence in other
Figure 49 Calibration example in which the block model
forecast results.
required re-estimation

Example involving old mining areas

This example considered two separate runs. In
the first (Figure 50), the block model was not
adjusted correctly for historic mining. Once this had
been recognized and appropriate changes made to the
area in which mining had taken place, a much better
fit between PCBC model grades and observed mill
grades was obtained. (Figure 51)

15
 The calibration examples also clearly show the
benefits to be gained from doing a calibration
exercise using both grades and rock types. In each
case, a clearer understanding caving mechanisms is
gained from the work. This type of calibration also
strongly justifies the effort of taking draw point
samples for grade and rock types.
The development process for PCBC has been
significantly enhanced by collaborative projects with
key clients and this assistance is gratefully
acknowledged.

Acknowledgements
The authors would like to thank the following
mining companies for permissions to publish
information and figures pertaining to their projects in
Figure 50 Grade curves measured vs PCBC before
adjustment for mined out area
this paper: Freeport-McMoRan Copper & Gold Inc.,
Rio Tinto, Newcrest Mining Limited, De Beers
Consolidated Mines – Finsch mine, Codelco División
Salvador, Codelco División Andina and Palabora
Mining Company.
The authors also gratefully acknowledge
assistance with the development of the software from
Freeport-McMoRan Copper & Gold Inc., PT
Freeport Indonesia, Rio Tinto, De Beers
Consolidated Mines – Finsch mine, Codelco División
Andina and Palabora Mining Company.

References
1. Brannon, C., Casten, T., & Johnson, M. (2004).
Design of the Grasberg block cave mine.
MassMin, (pp. 623 - 628). Santiago.
Figure 51 Grade curves measured vs PCBC after 2. Burgio, N., & Diering, T. (2008). Simulating
adjustment for mined out area irregular cave propagation using PCBC.
MassMin, (pp. 1033 - 1042). Lulea.
Concluding remarks
3. Diering, T. (2004). Combining long term
scheduling and daily draw control for block cave
PCBC has been applied to a variety of different
mines. MassMin, (pp. 486 - 490). Santiago.
block cave projects and mines over the last 20 years.
Every project has its own unique challenges some of 4. Diering, T. (2000). PC-BC: A block cave design
which have been described in this paper. As the and draw control system. MassMin, (pp. 469-
program has evolved to meet these new problems, its 484). Brisbane.
capability has been enhanced.
5. Diering, T. (2007). Template Mixing: A
A key component of the modeling and program Depletion Engine for Block Cave Scheduling.
development process has been the ongoing APCOM, (pp. 313 - 320). Santiago.
calibration of PCBC against observations / sampling.
This process has clearly indicated that it is not always 6. Laubscher, D. (1994). Cave Mining: State of the
the material mixing which required the most Art. SAIMM , October, 279 - 293.
adjustment. Careful attention is also required in areas 7. Moss, A., Russell, F., & Jones, C. (2004).
of past mining, or for open pit failure material or Caving and Fragmentation at Palabora:
even to the geological block model itself.

16
 Prediction to Production. MassMin, (pp. 585 -
590). Santiago.
8. Pretorius, D., & Ngidi, S. (2008). Cave
management ensuring optimal life of mine at
Palabora. MassMin, (pp. 63 - 72). Lulea.
9. Richter, O., & Diering, T. (2004). Production
Scheduling at Finsch Diamond Mine. MassMin,
(pp. 453 - 458). Santiago.
10. Ross, I. (2008). Northparkes E26 Lift 2 block
cave – A case study. MassMin, (pp. 25 - 34).
Lulea.
11. T. Casten, L. R. (2008). P.T. Freeport Indonesia's
Deep Ore Zone mine - expanding to 80,000
tonnes per day. MassMin. Lulea.
12. Villa, D., Prasetyo, R., & Diering, T. (2008).
Calibration of mixing model to predict grade at
Freeport’s DOZ Mine. Massmin, (pp. 1053 -
1062). Lulea.

17