Optimized Open Pit Mine Planning and Scheduling

I.K.Kapageridis
Maptek/KRJA Systems Ltd., United Kingdom

ABSTRACT: A complete Mine Planning solution often requires the integration of different “specialist”
packages. It is necessary to understand what each package brings into a system and how they relate to each
other in the mine planning and scheduling process. In isolation, each package may fall short of the final solu-
tion. However, when integrated with strategy, they form a powerful Mine Planning and Scheduling system. A
step-by-step procedure is presented as a “model” for the mining engineer to follow in arriving at an optimized
mine plan. The optimization process presented in the paper is capable of achieving all possible schedule ob-
jectives given the limitations in design and deposit. Obviously, as the deposit is depleted, the possibility that
certain constraints can be met is reduced. This requires careful consideration in the scheduling optimization
cycle.

1 INTRODUCTION basic block model preparation, manipulation, as a

data transfer medium, and for schedule visualization;
1.1 The Concept
Pit Optimization and first-pass Schedule Analysis
In recent years, it has become a common approach in is performed using Whittle 4X Multi-Element (Ana-
mine planning to use multiple software packages to lyzer). Various features are showcased including the
achieve better results. The approach described in this user friendly Proteus GUI and the new Blending
paper consists of a general mining package and two Module. The Milawa Scheduling algorithm is used
mine scheduling and optimizing packages that oper- in an example case study.
ate on different stages of the optimization and Final Schedule optimization and blending is per-
scheduling process (Figure 1). formed in Vulcan’s Chronos Scheduling and Opti-
mization package.

1.3 The Model

There are three “streams” in the mine planning and
scheduling model:
Conceptual Design and Scheduling in Whittle:
This involves block model preparation, variable ex-
port/import, pit optimization, mine design, schedul-
ing and visualization;
Optimizing and Scheduling the Conceptual Whit-
tle Design in Chronos: This involves passing the pit
Figure l: Optimized mine planning and scheduling model. design back to Vulcan through push-back variables,
reserving the block model against these variables,
1.2 Software Components building a Chronos scheduling workbook in
Chronos, and scheduling using the Chronos Optimi-
Three major software packages are discussed and il- zation module. Results are transferred back to Vul-
lustrated here: can for visualization.
Vulcan’s Envisage program is used as the General Final Pit Design in Vulcan and Final Schedule
Mining Package (GMP). This software is used in optimization in Chronos: This is the most detailed
and time consuming stream in the model. It involves
the use of all previous results to:
i) create a final phased pit design in Vulcan;
ii) transfer to a Chronos scheduling workbook; iii)
optimize and schedule in Chronos; and iv) visualiza-
tion in Vulcan.
In the following paragraphs we will discuss these
streams in detail using an example case study from a
Banded Iron Formation (BIF) deposit in the US
(Slade, 2001).

2 CONCEPTUAL DESIGN AND

SCHEDULING IN WHITTLE
2.1 Preparation of the Block Model in VULCAN Figure 2: Formats tab in Whittle 4X.

Whittle requires a number of variables to exist in the

ELEMENTS
VULCAN block model. These variables necessary
for pit optimization are the following: WTRC = Product Wt. Recov.
1. A pit slope variable containing “slope zone” EXWR = Exploration Wt. Recov.
numbers can be used. Alternatively, slopes S102 = Silica in Concentrate
can be set by rock type. MGFE = Mag. Suscept Crude
2. A rock type variable is necessary for the dif- SFER = Indication of Oxidation
ferent material types, e.g., OVB, WST, ORE,
HGR, LGR, RATO = Indication of Oxidation
3. Whittle 4X Multi-Element does not require a KWHT = Grindability measure
Net $ Value. Product element grades (e.g., TACH = Ore Haul unit Cost
Au, Ag, Cu, Fe, and Recovery) are passed di- WSTH = Waste Haul unit Cost
rectly from Vulcan to Whittle and used with- ROYL = Royalty Cost
in the program to calculate “block values”.
4. Ten element variables can be set up for just In our BIF example the primary product grade is
about any function you require. The obvious WTRC and is described as “recovered product tons”
product elements might include Au, Ag, Cu, divided by “processed ore tons”. Rock types LLTC
or Fe. Less obvious elements might include: and LHTC correspond to the mineable “low grade”
PIT (used for haulage calculations), ROYL and “high grade” ores respectively. UHTC and
(used to calculate royalty costs), RECV (re- ULTC are un-leased ore grade material and therefore
covery variable used to determine recovered excluded from the ultimate pit and scheduled min-
product). The advantage of this Multi- Ele- ing. All other rock types are waste material (Figure
ment feature is that it is easy to perform 3).
“What-if’ and “sensitivity” variations with-
out having to go back to the original block
model. Figure 3: The Summary tab shows the 14 rock types exported.

2.2 Vulcan Model Export to Whittle 2.3 Pit Optimization in Whittle 4XPE
In this step we export pit optimization variables to Setting up of the relevant expressions, cost model
Whittle .MOD format. The Formats tab in 4X shows and revenue factors is required to run pit optimiza-
the 10 elements exported from Vulcan (Figure 2). tion in Whittle 4X Proteus Environment. Once the
setup is complete optimization can take place in a
step by step procedure that is displayed graphically
in a tree-like graph (Figure 4).

Figure 4: Optimization in 4X.

2.4 Whittle Pit Design Selection

This step constitutes the “conceptual” mine design
sequence.
1. Ultimate Pit: we select an Ultimate Pit from the
Pit-by-Pit Graph. There are many ideas on how to 2. Milawa NPV Maximizer Algorithm
achieve this. Some include: This scheduling method mines the benches in an op-
• Max Best Case Value; timum sequence, maximizing NPV. Note the in-
• Required Tonnage; crease over “Best Case” is +36%.
• Product Price Point;
• Max Specified Case Value. 3a. Milawa Balanced
This technique mines the benches in an optimum se-
quence, attempting to balance out mining require-
Figure 5: Pit by pit graph in Whittle 4X. ments, while trying to maximizing NPV. Note that
this has come at a cost of 21% in NPV compared
2. Push Backs (Phases): we select a sequence of with the Milawa NPV method.
Push Backs (Pit Phases) from the Ph-by-Pit graph. 3.
Minimum Mining Width: Whittle 4X allows the user 3b. Milawa NPV and Sequential Lead Schedules
to apply the concept of “Minimum Mining Width” with a Mining Limit
(MMW). This has the effect of “redistributing” ton- Apply a Total Mining Limit.
nage between the chosen Push Backs in order to ac-
commodate the MMW. Sequential Lead
Also, using MMW renames the Push Backs from Pit • Fails to meet Product tons in early years;
Number to Push Back number starting from 1. • Fails to meet ore tons in early years;
• Fails to blend ore types in early years;
The above report shows the tonnages present in • Unworkable in the critical early years.
each Push Back — originally after selecting the pits,
and after the MMW is applied. It is a “juggling” act Milawa NPV
to apply MMW and find a balance between the Push • Stripping minimized in early years;
Backs. In this case, a balance was found using Pits 9, • Fails to meet ore production in later years;
16, 25, and 37. A MMW of 300 ft was used in this • Fails to meet Product tons in later years;
instance. • Ore type blend is very poor;
Care must be taken when applying the MMW • Basically unworkable — yet makes the most mon-
function, especially when optimizing existing pits. ey?
Re-distributing tonnages can have the unexpected 3c. Milawa Balanced with Mining Limit
result of “covering” up exposed inventory present in Limit total mining to Z MT. Note that we now have
operating benches. a “balanced schedule” which meets the #1 objective
of making required product tons. Objectives #3 (ore
mining) and #4 (ore type blend) are still off target.
2.5 Scheduling in Whittle
Using the Push Backs (Pit Phases) selected from Pit
Optimization, schedule using three techniques avail-
able in Whittle 4X Analyser. In the case of the BIF 2.6 Blending in Whittle
Project scheduling objectives were: In 4X Version 2.2, Whittle have added a new blend-
1. Must achieve stated product output — X ing module. The parameters for the example BIF
MT; problem were entered and the module applied. We
2. Minimize and balance total mining (elimi- are able to make the 30:70 Low:High grade ore
nate stripping spikes); blend. However, it is at the expense of other sched-
3. Limit the ore throughput to a maximum of Y ule objectives, i.e., Product and Ore Tonnage targets.
MT;
4. Blend the High:Low grade ores to 70%:30%.
5. Maximize NPV.
Whittle allows the scheduling of the conceptual 2.7 Visualization in Whittle
mine design inherent in the Push Backs and Bench- Whittle 4X Proteus Environment includes a 3D Vis-
es. The physical problem is set up, with all the spa- ualization module. A “Mining Sequence and 3D
tial integrity of pit slopes and Bench-Push Back Visualization file” is specified on the Definition tab
Precedence, automatically in the Whittle model. for Whittle Schedule graphs. The mining sequence
1. Fixed Lead file produced (.MSQ) can be read by the 3D Visuali-
This schedule mines out the Phases sequentially, zation module. A pit shell representing the mining
with a fixed lead between benches in adjacent Push progress to the end of each scheduled period is dis-
Backs (PB). A zero lead specified in Whittle mines played and can be viewed interactively. It is also
out and completes each PB sequentially. This tech- possible to display colored block model slices along
nique is often termed “Best Case”. with the pit shell surface.
Figure 13: Visualization in Whittle 4X. Figure 14: Visualization of Whittle schedule in VULCAN.

2.8 Whittle Schedule Visualization in Vulcan 2.9 Conclusions

The Mining Sequence File (MSQ) generated in Whittle is a very powerful tool when the LG pit op-
Whittle contains all the information necessary to timization tools are combined with the outstanding
create a Vulcan Animation. However, currently the time-value-of-money (DCF) and scheduling analysis
relevant information must be extracted using C Shell routines. They are very easy to set-up and use within
and other programming utilities. Pre-processing is as the new Proteus Environment.
follows: Many different types of schedules can be run on
1. Export the MSQ file from Whittle 4XPE. the chosen conceptual pit design (Push Backs or
2. Remove all the “parcel” records using GREP Phases). These allow the user to focus on the various
with a reverse search on the PROCESSING scheduling objectives. The execution of the various
path field (last field). schedules is so fast that a user can run as many as
3. Run the “periodsplit.pl” PERL script on the necessary to draw initial conclusions on the value of
“PP37MW300.rep” file to produce a series of the prospect in hand.
mining “touch” files: “fileOl.rep”, “fileO2 The “Milawa Balanced” schedule has been shown
rep” “fileO3 .rep”, “fileO4.rep”. The number to get closest to the schedule objectives for our ex-
of files produced, which contain data, repre- ample Banded Iron Formation project. Whittle 4X
sents the maximum number of benchlblock now provides a simple, yet powerful method for vis-
“mining touches” found in the schedule ualizing the Pit design and Schedule. The scheduling
(benches mined over a series of production output from Whittle can be imported into Vulcan for
periods). This procedure only handles up to 4 animation. Whittle’s new blending routine enables
mining “touches” per bench. the user to quickly establish the effect of blending on
4. Run the “convert ijk.csh” CShell script the schedule.
which executes the “schedule.awk” AWK Whittle scheduling does not provide a final solu-
script to convert the ijk coords to xyz coords. tion to multi-variable scheduling constraints. There-
The “schedule.awk” is customized to the fore, the next Model Stream is to export the Whittle
translation implied in the Block Model head- design to Chronos where it is possible to apply con-
er used and adjusted for the ijk offsets to straints to multiple variables, divide the conceptual
block centroids. design up into multiple pits, and solve complex
5. Import the ASCII files into the GMP block blending problems.
model. You will need 2 block model varia-
bles per “touch” file. One to contain the peri-
od in which a block is first touched (eg, 3 OPTIMIZING AND SCHEDULING THE
msqOl), another to contain the % mined in CONCEPTUAL WHITTLE DESIGN IN
that period, and so on, for touch 2, 3, and 4. CHRONOS
6. Once the import is complete, the engineer
can contour (Grade Shell) the “first touch”
blocks variable. This shows the first period in
which a block is “touched” by mining and 3.1 Preparing the Vulcan Block Model for
gives the engineer some idea of the sequen- Chronos
tial order of “bench by bench” mining. This step provides Chronos more choices as it seeks
an optimum solution to the multiple constraints ap-
plied. The actions included are:
 1. Import the Whittle Pit Optimization results “hands on” procedure. It is a cyclical process in
into the Vulcan model. This constitutes the which the engineer:
Conceptual Mine design. 1. Identifies a set of schedule constraints, ac-
2. Add other “key” variables (in addition to 2. cording to the schedule objectives, and saves
Builds a list of available mining benches in a them in named Constraint Sheets.
pushback) required by Chronos for reserve 2. Builds a list of available mining benches in a
block location. These variables are pit and Chronos Task.
bench. This will allow the engineer to sepa- 3. Sets up the problem, specifying the periods
rate the logical pits and allocate blocks to the to be optimized, the constraint tables to be
benches identified in whittle. used in each period, and other conditions on
3. Divide the deposit into pits using solids flag- the optimization, eg, whether to allow partial
ging in Envisage. mining, Push Back Ratio tables to be used,
4. Add the block “value” variable (Cash Flow). etc.
Use the same formulation as that used in 4. Creates the “bci” and “bco” optimization
Whittle 4X. This must be verified. files which contain the problem (bci) and
5. Generate a Reserve Inventory for Chronos provide a file to record the result (bco).
using Block — Reserves Advanced. This 5. Solves the optimization.
will include the key breakdown fields: 6. Loads the result (bco) into Chronos Result
Pit_Pushback_Bench, the products to be Tables (CRT).
scheduled, the grades to be reported, and the 7. Schedules the result (from the CRT) and ob-
cashflow resulting from the mining of each serves the schedule in the Summary General
bench. Sheet.
Schedule Optimization in Chronos is an iterative
process. The engineer needs to be constantly review-
3.2 Set up a Chronos Scheduling Workbook ing and testing the result while optimizing. Rarely is
A brief outline of the steps required in setting up a the problem solved by pushing a button and walking
workbook follows: away.
1. Start Chronos and open a new Workbook.
2. Import the Vulcan Reserve Dump file
(.DMP) into a Reserve Sheet.
3. Insert derived columns into the Reserve
Sheet.
4. Format the inserted columns and set the type
and weighting fields.
5. Generate a Period Calendar and set the
scheduling periods.
6. Create a single Chronos Destination called
MINE. This is a key step in that all mission
critical variables must be defined here.
7. Create a single Chronos Process called
EXCAVATE.
8. Create a Process Lookup Table called
TABLE. Figure 15: Chronos schedule optimization cycle.
9. Create a Push Back Ratio table called
RATIO. This table sets, i) the minimum pro- It appeared that there was further potential to reduce
portion of any one bench to be mined in any the upper mining constraint, thereby increasing
one period, ii) the proportion of the bench NPV. The only constraints used on this schedule are
above to be mined in advance of the current the range of Gold Metal production, and Ore less
bench. than 4.75Mt. The objective is to test what the project
10. Create a General Sheet and populate the low- can theoretically create in NPV.
er half with an automatic Period Summa ry. Attempt to balance the Total Mining tons by care-
11. Set Block Precedences and check in the Re- fully reviewing the existing schedule and finding the
serve Sheet. Convert Precedences to Block “hurdle” points of the schedule. The result of some
Ordering Numbers. trial and error optimization runs, using constraint
variations, is a set of Total Mining constraints which
3.3 Run Optimization and Schedule the Results solve. The illustration below shows the Optimized
Schedule 02.
The secret of successful Optimization and Schedul-
ing in Chronos is to approach it as an iterative
3.4 Create an Animation in Vulcan for Schedule ahead so the detail design work can follow previous
Visualization work based on sound principles (Rodriguez, 2001).
The user can maximize NPV (Net Present Value)
VULCAN provides functionality for creating sched-
and at the same time it can follow blending con-
ule sequence animations. In our example we need to
straints, keeping stripping ratio at a constant rate un-
grade shell each individual mining bench per pit,
til the later years in the life of the mine where it
pushback, and bench. Because the Chronos reserve
drops off and keep positive cash flow.
was taken directly from the block model, keying off
the three variables: pit, pushback, and bench, trian-
gulations do not exist for the scheduled entities. A C
Shell script can be used to batch the Grade Shelling
routine which creates the triangulations all in one
step.
The next step is to rename the Pit-Pushback-Bench
triangulations resulting from the Grade Shelling pro-
cess to conform to the key field naming convention
in the Chronos Reserve Sheet. The naming conven-
tion used on the BIF project is:
P<pit#>_<pb#>_<bench#> e.g., POl_0 1_Ol .OOt.
Finally we run a Perl script which builds a Vulcan
animation file by matching the Chronos Reserve Figure 17: Graphical display of schedule in VULCAN.
Sheet keys with the triangulation and displaying it in
the period in which it was mined. The Display At-
tributes option in Chronos is used to color the bench REFERENCES
triangulations by Period.
Rodriguez, F., 2001. Implementing Whittle results to Pit De-
sign, Long Term Scheduling and Maximizing NPV using
3.5 Conclusions Chronos Optimization Package. VI Vulcan Users Confer-
Using the designs generated in Whittle and the ence, Maptek South America.
Slade, R., 2001. Optimized Open Pit Mine Planning and
schedules generated in 4X Analyser, schedules can Scheduling using Vulcan, Whittle, and Chronos. VI Vulcan
be optimized using any number of constraints in Users Conference, Maptek South America.
Chronos. The up-front scheduling work in Whittle is
invaluable in setting the scene for the fmal optimiza-
tion work in Chronos.
Vulcan provides powerful tools for easily subdi-
viding the Whittle Push Backs into Multiple pit
models. This provides more alternatives for optimi-
zation as illustrated in the BIF project.
The Chronos Optimization-Scheduling process is
cyclic and should be viewed as an iterative process.
Chronos Optimization is capable of achieving all of
the Schedule objectives given the limitations in de-
sign and deposit. Obviously, as the deposit is deplet-
ed, the possibility that certain constraints can be met
is reduced. This requires careful consideration in the
Scheduling Optimization cycle.
Vulcan provides the engineer many ways in which
to animate schedules from Chronos for visualization,
checking and presentation.

4 FINAL PIT DESIGN IN VULCAN AND

FINAL SCHEDULE OPTIMIZATION IN
CHRONOS

All the previous schedule optimization is done to

give the engineer an understanding of the problem,
the possibilities, and the result to expect once this
“stream” is complete. The results illuminate the path