Sei sulla pagina 1di 37

UNIVERSIDADE FEDERAL DE MINAS GERAIS

ESCOLA DE ENGENHARIA
DEPARTAMENTO DE ENGENHARIA DE MINAS

MATHEUS MASCARENHAS DE ULHÔA XAVIER

A CASE STUDY OF A GOLD DEPOSIT THROUGH


DIRECT BLOCK SCHEDULING

BELO HORIZONTE
2016
MATHEUS MASCARENHAS DE ULHÔA XAVIER

A CASE STUDY OF A GOLD DEPOSIT THROUGH


DIRECT BLOCK SCHEDULING: THE CURRENT
STATE OF DEVELOPMENT OF A PIONEER
COMMERCIAL SOFTWARE

Trabalho de Conclusão de Curso


apresentado ao Departamento de
Engenharia de Minas da Universidade
Federal de Minas Gerais como parte
integrante dos métodos de avaliação para
obtenção de título de Bacharel em
Engenharia de Minas.

Orientador: Prof. Dr. Alizeibek Saleimen


Nader.

BELO HORIZONTE
2016
RESUMO

Sequenciamento é o processo que determina uma sequência de extração de um modelo


de blocos tridimensional, bloco a bloco, em adição à definição de um destino para cada
bloco em conformidade com o objetivo econômico de maximizar o Valor Presente
Líquido de um projeto mineral. Este problema foi abordado de várias formas desde os
anos 60, comumente através de programação inteira, de um modo inteligente para lidar
com as limitações computacionais da época por meio de um procedimento fragmentado
em etapas. Durante décadas, em vista do avanço tecnológico, pesquisas foram
realizadas sobre como lidar com o problema do sequenciamento considerando uma
visão holística, usando métodos baseados em agregação, heurísticas baseadas em
otimização, relaxação Lagrangeana, entre outros, sejam em separado ou combinados.
O presente conjunto de estudos de caso objetiva entender o desenvolvimento atual do
Sequenciamento Direto de Blocos, analizando o comportamento do Depósito
McLaughlin e as implicações no Valor Presente Líquido e na Produção de Material, ao
longo da vida útil da mina, frente à adoção de restrições operacionais cumulativas
usando um método híbrido composto por MILP e heurísticas. Os cenários
parametrizados compreendem um caso base, um caso de restrição de lavra, e um caso
lavra restrita e forçada. Em complementaridade, uma análise preliminar de cenários
simplificados usando a metodologia estocástica serão aplicados considerando incerteza
econômica, como guia para futuros estudos aprofundados. O software comercial provou
suas capacidades para lidar com uma base de dados de 2.1 milhões de blocos em um
tempo razoável e respeitou as restrições impostas.

Palavras-chave: sequenciamento direto de blocos, programação inteira mista,


planejamento de lavra, planejamento estratégico, otimização global de ativos.
ABSTRACT

Scheduling is the process that determines a sequence of extraction within a three-


dimensional resource block model, block by block, in addition to a destination to each
one in compliance with the economic goal of achieving a maximized Net Present Value
of a mine project. This problem has been approached in various ways dating back to the
1960s, commonly as an integer program, in an intelligent manner to handle computing
limitation of its time through a stepwise procedure. During last decades, in light of the
technological development, there have been researches on how to cope with the
scheduling problem considering a holistic view, by using methods based on aggregation,
optimization-based heuristics, Lagrangian relaxation, among others, whether separately
or combined. The present set of case studies aims to understand current developments
of Direct Block Scheduling methodology, analyzing the behavior of the McLaughlin
Deposit and the implications on Net Present Value and Production of Material, over the
life of the mine, in view of adopting cumulative operational constraints by using a hybrid
method composed by MILP and heuristics. The parameterized scenarios comprise a
default case; a restrict mining case; and a restrict and force mining case. In addition, a
preliminary analysis of simplified scenarios using the stochastic methodology will be
applied considering financial uncertainty as a guide for further and deeper studies. The
commercial software used has proven its capacity to handle a dataset 2.1 million blocks
in reasonable times and respecting the constraints imposed.

Keywords: direct block scheduling, mixed integer linear programming, mine planning,
strategic mine planning, global asset optimization
LIST OF FIGURES

Figure 1. Porter’s value chain framework in the mining business. Taken from
Camus (2011)................................................................................................... 10
Figure 2: Direct Block Scheduling result of the Default Case showing Period 0.
......................................................................................................................... 15
Figure 3: Direct Block Scheduling result of the Default Case showing Period 5.
......................................................................................................................... 15
Figure 4: Direct Block Scheduling result of the Default Case showing Period 10.
......................................................................................................................... 16
Figure 5: Direct Block Scheduling result of the Default Case showing Period 15.
......................................................................................................................... 16
Figure 6: Default Case and its Production Process tonnages over the mine
lifespan. ............................................................................................................ 17
Figure 7: Default Case and its Production Dump tonnages over the mine lifespan.
......................................................................................................................... 18
Figure 8: Default Case, its Cumulative Stocked ore tonnages over the mine
lifespan and the stock-in and stock-out process. .............................................. 18
Figure 9: Default Case and its Cumulative NPV over the mine lifespan. .......... 19
Figure 10: Direct Block Scheduling result of the Restrict Mining Case along with
the restricting surface used. ............................................................................. 20
Figure 11: A rough contour (in red) of the restricted area, which has been applied
during the first five years of the mine project. ................................................... 20
Figure 12: Direct Block Scheduling result of the Restrict Mining Case showing
Period 5. ........................................................................................................... 21
Figure 13: Direct Block Scheduling result of the Restrict Mining Case showing
Period 10. ......................................................................................................... 22
Figure 14: Direct Block Scheduling result of the Restrict Mining Case showing
Period 15. ......................................................................................................... 22
Figure 15: Cumulative Production Process tonnages over the mine lifespan. .. 23
Figure 16: Cumulative Total Production tonnages over the mine lifespan. ....... 23
Figure 17: Default Case, Restrict Mining Case, and their Cumulative Stocked ore
tonnages over the mine lifespan....................................................................... 24
Figure 18: Default Case, Restrict Mining Case, and their Cumulative NPV over
the mine lifespan. ............................................................................................. 24
Figure 19: McLaughlin Deposit, mined areas during the entire life of mine from a
top view. ........................................................................................................... 26
Figure 20: McLaughlin Deposit, non-restricted areas that are available to be
mined during the first five years. ....................................................................... 26
Figure 21: Direct Block Scheduling result of the Force Mining Case along with the
forcing surface used and a rough contour of the forced area (in red). .............. 27
Figure 22: Direct Block Scheduling result of the Force Mining Case showing
Period 2. ........................................................................................................... 27
Figure 23: Direct Block Scheduling result of the Force Mining Case along with the
restricting surface used and a rough contour of the restricted area (in red). .... 28
Figure 24: Direct Block Scheduling result of the Force Mining Case showing
Period 5. ........................................................................................................... 28
Figure 25: Direct Block Scheduling result of the Force Mining Case showing
Period 10. ......................................................................................................... 28
Figure 26: Direct Block Scheduling result of the Force Mining Case showing
Period 15. ......................................................................................................... 29
Figure 27: Cumulative Production Process tonnages over the mine lifespan. .. 29
Figure 28: Cumulative Production Process tonnages over the mine lifespan. .. 30
Figure 29: Default Case, Restrict Mining Case, and their Cumulative Stocked ore
tonnages over the mine lifespan....................................................................... 30
Figure 30: Default Case, Restrict Mining Case, and their Cumulative NPV over
the mine lifespan. ............................................................................................. 31
Figure 31: Comparison of processing time spent by each scenario in percentage.
......................................................................................................................... 31
Figure 32: Aggregated single and independent scenarios with fixed ore price
changes and their cumulative NPV's. ............................................................... 33
Figure 33: Economic-Stochastic scenario based on fixed ore price changes. .. 34
CONTENTS

1 INTRODUCTION ......................................................................................... 8

2 OBJECTIVES AND RELEVANCE ............................................................. 11

3 THE MCLAUGHLIN DEPOSIT .................................................................. 12

4 CASE 01: DEFAULT CASE ....................................................................... 13

5 CASE 02: RESTRICT MINING .................................................................. 19

6 CASE 03: FORCING MINING .................................................................... 25

9 APPLYING ECONOMIC UNCERTAINTY: MAKING ROOM FOR FUTURE


SUDIES ............................................................................................................ 32

9.1 Case 04: Fixed Price Changes – Single-Independent Scenarios


32

9.2 Case 05: Fixed Price Changes – Stochastic Simulation ........ 33

13 CONCLUSIONS .......................................................................................... 35

REFERENCES ................................................................................................. 36
8

1 INTRODUCTION

Scheduling is the process that determines a sequence of extraction within


a three-dimensional resource block model, block by block, in addition to assigning
a destination for each one in compliance with the economic goal of achieving the
highest Net Present Value (NPV). In summary, scheduling defines which blocks
should be mined, when and to where they should be sent (Goycoolea, et al.,
2013). Consequently, this is a guide to estimate the capital investments required
to a given mine project. Hence, this process consists of a crucial step regarding
Mine Planning, applicable from preliminary studies of feasibility up to long-term
planning, as it is intrinsically related to economic evaluation and strategic mine
planning.

Aiming to realize a feasibility study, a deposit is represented through a


3D block model that simplifies the reality. King (2011) points out the importance
of the effort spent modeling in order to achieve accurate estimates of NPV, which
will lead to better decisions. A set of information is then assigned to each block,
such as position, volume, density, grade, and economic value. This fragmentation
of information into cells is a requirement due to computing limitations.

Historically, the scheduling problem has been approached in various


ways dating back to the 1960s, commonly as an integer program (Lambert &
Newman, 2013). Moreover, the limitation of computers in that time has made
necessary to handle the scheduling problem through a stepwise procedure,
which has been divided into three general steps. These steps have consisted of
pit optimization; design of pushbacks; and scheduling (Goycoolea, et al., 2013).

The pit optimization defines an optimal pit along with nested pits, which
are multiples and submultiples of the optimal solution and represent the optimal
pit size according to different ore prices. This step is commonly generated
through the well-known Lerchs-Grossmann (LG) algorithm. The design of
pushbacks consists of operationalizing the pit and dividing it into phases
(Goycoolea, et al., 2013) that use the nested pits, or groups of them, as a guide
to operationalize the pit. Finally, the third step is the scheduling, as explained
before.
9

In light of the technological development of computers and its processing


capacities, there have been researches on how to cope with the scheduling
problem through an approach that was not possible before, i.e., considering a
holistic view. Such view has been approached through methods based on
aggregation, optimization-based heuristics, Lagrangian relaxation, among
others, whether separately or combined (Newman, et al., 2010).

The need to stop using an intelligent approach that is already satisfactory


can be addressed by some of the following reasons. Strategic mine planning is a
cost-effective mean of increasing project value (King, 2011). In addition, such
global optimization allows for a wider range of variables being considered at once,
supporting decision-making at many levels (Whittle, 2004). Moreover, Whittle
(2004) points out the importance of performing different strategies in order to test
the merits of distinct scenarios on a trial and error basis. Such extensive
procedure, necessarily, requires to include many variables into the optimization
framework so that this task could be performed in reasonable time. Hence, a
holistic view will improve the quality of our assumptions, our decisions and, by
the end, our results.

Furthermore, mining engineers and geostatisticians deal with punctual


information from sampling to represent the entire deposit, which reduces
information accuracy. This lack of knowledge upon the orebody results in an
inherent financial risk to a Mining Project and a challenge to be overcome.
Ongoing strategies rely on Geostatistics and Kriging to estimate grades within
non-sampled regions. Carvalho (2016) explains the distinctions between
estimation and simulation: while estimation considers an average of a specified
set of unknown grades, according to a certain methodology adopted, simulation
techniques allow for a distribution of possible values in which each one has a
certain level of probability.

“Based on drill-hole data and their statistical properties,


conditional simulations generate several equally probable
models (or scenarios) of a deposit, each reproducing
available data and information, statistics and spatial
continuity, that is, the in-situ variability of the data. The
difference between the equally probable scenarios is a
quantitative measure/description of uncertainty.” (Benndorf
& Dimitrakopoulos, 2013)
10

Marinho and Martinez (2015) conducted a case study exemplifying the


benefits, aforementioned, of using a stochastic model as a strategy to reduce
geological uncertainty for long-term production scheduling. Their results have
shown a decrease in uncertainty from 9% to 8% with 80% of probability while
increasing NPV in 4%. Godoy and Dimitrakopoulos (2004) show an increase of
28% in NPV when comparing the stochastic simulation with the conventional
approach. Leite and Dimitrakopoulos (2014) have obtained an increase of 29%
in NPV by using Stochastic Integer Simulation against the traditional
methodology. Each one of these studies has used distinct considerations and
software but already corroborate with the emergent potential of this new
technology.

A study conducted by Juan Camus (2011) corroborates with this idea,


concluding that the upper portion of the value chain, i.e., resource-related
activities such as exploration, planning, and development, is the major
responsible for value creation in a Mining Business. Camus has used Michael
Porter’s (1985) value chain framework, as shown in Figure 1.

Figure 1. Porter’s value chain framework in the mining business. Taken from Camus (2011).
11

2 OBJECTIVES AND RELEVANCE

Mining still has a large impact on society’s and economic demands and
well-being, and it has been one of the oldest primary industries in the world. Due
to the increased consciousness of population regarding the environment and a
general sense of quality reduction of deposits, there is, more than ever, a
requirement for an improved process that explore the full potential of nature and
that are able to mitigate economic uncertainties. Those uncertainties are inherent
to every mining project due to the natural and geological heterogeneity of rock
formations, being unpredictable to a certain extent.

The impact that such uncertainties could have on a mining project are
substantially high to be underestimated so that it is needed to quantify precisely
all the risks involved within a mining business. Extensive researches have been
occurring during last decades in this field, attesting its relevance. Therefore, this
academic work intends to follow the same direction.

Herein, the objective is to explore two strands of thought that are related
to each other: a global mine planning scheduling that covers many constraints in
a single-step through Direct Block Scheduling and the possibility of quantifying
uncertainty via Stochastic Direct Block Scheduling. Consequently, this work aims
to verify the current developments of the technological industry by using a
commercial software known as SimSched Direct Block Scheduler, which uses a
hybrid method that combines Mixed Integer Linear Programming (MILP) and
heuristics.

Under the aforementioned purpose, three cumulative scenarios will be


tested in order to compare the behavior of the McLaughlin Deposit in light of the
method used. These cumulative scenarios comprise the use of operational
constraints that have not been possible to be incorporated in the traditional
methodology without manual manipulations. Finally, a preliminary analysis will be
done considering the potential of the stochastic methodology, guiding further
studies: a set of single scenarios simulating fixed price increase and decrease
will be compared with a single stochastic model that considers all the price
possibilities at once to find a single result.
12

3 THE MCLAUGHLIN DEPOSIT

Herein, the set of scenarios to be analyzed consist of a well-known


dataset, freely available in MineLib (Espinoza, et al., 2012): the McLaughlin
Deposit, situated in California, United States. “The McLaughlin deposit is situated
115 km north of San Francisco in the Coast Ranges geomorphic province of
Northern California.” (Enderlin, 2002). “The reserve captures much of the
geological diversity of the California Coast Ranges, which are internationally
famous for spectacular exposures of rocks deposited or modified by plate tectonic
processes over the past 140 million years.” (UC Davis Natural Reserve System,
2007).

Dean Enderlin has reported as a part of the Homestake's McLaughlin


mine closure plan that the McLaughlin deposit has been the largest gold
discovery of the 20th century in California. According to the information, the ore
reserve has exceeded 4 million ounces of contained gold, from which
approximately 3.3 million ounces were recovered (UC Davis Natural Reserve
System, 2005).

“The McLaughlin deposit is a world-class gold orebody, and


one of the world's finest examples of a hot springs-type
epithermal precious metals system. When discovered, the
deposit was remarkably well preserved, with an intact
siliceous sinter capping the central portion of the
mineralized outcrop. The subaerially deposited hot springs
sinter terraces marked the paleosurface of the Pleistocene
hot springs that formed the deposit. The presence of sinter
was especially important in demonstrating that this was a
fully preserved epithermal system from the top down… a
rarity in deposits such as this, and an opportunity for
scientists around the world to study this fascinating geologic
feature as it was mined.” (UC Davis Natural Reserve
System, 2005)

The dataset of the McLaughlin reserve, available in MineLib, is


discretized by 2,140,342 blocks of size 25x25x20 ft., that translates into
7.62x7.62x6.096 m.
13

4 CASE 01: DEFAULT CASE

As a default case, this study will adopt the parameters, available along
with the dataset within MineLib (Espinoza, et al., 2012), to calculate default block
values for each possible destination: the processing stream and the waste dump,
as shown in Table 1. Table 2 presents a complete set of parameters used to
perform the Direct Block Scheduling, including parameters commonly used by
the conventional scheduling such as default density, default slope angle, discount
rate, stockpiling costs and production limits.

Block Value Computation Case 01: Parameters


Selling Price $900/oz Default Density 3 t/m³
Mining Cost $1.32/ton Default Slope angles 45 degrees
Processing Cost $19/ton Discount rate 15 %
Recovery 90% Stockpiling
Table 1: Block Value Computation Parameters. Fixed mining cost 1.32 $/t
Rehandling cost 0.3 $/t
Operational Constraints
Minimum widths
Mining 50 m
Bottom 50 m
Vertical rate of advance
Preferred 20 m
Maximum 80 m
Destinations
Recovery
Process 0.9 %
Dump 0 %
Production Limits
Process
Period 1-2 1 Mt
Period 3-5 2 Mt
Period 6-end 4 Mt
Total
Period 1-2 5 Mt
Period 3-5 10 Mt
Period 6-end 20 Mt
Stockpile limit
Process 20 Mt
Table 2: McLaughlin Default Case: Parameters
14

The initial step consists of removing entirely the air blocks out of the
dataset. Differently from the majority of mining packages, SimSched does not
consider a topography surface to perform the scheduling. Thus, the way of
interpreting the topography limits within the model is by removing the air blocks
above it. Otherwise, those air blocks will be interpreted as waste material.

As SimSched Direct Block Scheduler (SimSched DBS) has been


designed to be a plugin, the developers decided to postpone the implementation
of tools that could facilitate the input and modifications of block values. According
to them, it has been a decision motivated by the priority they give to their clients
to guide the changes to be implemented. Besides, side tasks could be performed
through other programs and office suites during the first stages of the
development.

This way, the next step is the definition of economic values for each
destination, which is performed inside the block model file. The lack of a
calculator and the need to define economic values prior to the importation could
suggest an annoyance to perform different scenarios based on different
economic values and/or economic assumptions. However, the block model
allows adding as many columns of economic values as desired. This way, even
if a given mining project has one processing stream, it is possible to add many
routes, based on different assumptions, without the need to import the model
many times so that the user is able to select one processing stream at once.

Besides, SimSched decides to which destination the blocks should be


sent, based on economic value and any variables could be added to the block
model file to influence such values. This means it is not necessary to define,
arbitrarily, a cutoff grade to schedule, as this measure is not a static parameter
and could often need cutoff grade optimization. Consequently, this decision is
taken by considering the value of each block over time and the cutoff grade is a
consequence of a global scheduling.

Moreover, the software allows for operational constraints such as


minimum widths and vertical rate of advance, also known as sinking rate. These
constraints take into account the dimensions of entire blocks, rounding up to the
next integer as, currently, SimSched DBS does not allow mining them partially.
15

Figure 2, Figure 3, Figure 4, and Figure 5 show the visual evolution, every five
periods, of the mine project resulted by a Direct Block Scheduling using the set
of parameters defined by default.

Figure 2: Direct Block Scheduling result of the Default Case showing Period 0.

Although SimSched DBS intends to provide more operational solutions,


i.e., respecting parameters such as bottom and mining widths and vertical rate of
advance, it is possible to notice, especially in Figure 2, the challenge of
connecting areas of the same period. Of course, the superficial portion of the
deposit is the most inflexible due to its topography-constrained configurations,
and the software improves its responses, making such areas more connected, as
the mining achieve deeper regions and, consequently, more unconstrained – by
natural conditions – areas.

Figure 3: Direct Block Scheduling result of the Default Case showing Period 5.
16

Figure 4: Direct Block Scheduling result of the Default Case showing Period 10.

Figure 5: Direct Block Scheduling result of the Default Case showing Period 15.

From the point of view of the algorithm, and considering the surfaces
being used, which are defined as a grid of points, mining squares rather than
circular areas is quite easier to implement in order to connect regions of a given
period. Mining rounded areas consist in a challenge due to the difficulty to overlay
different circles without leaving blocks to the next period – or anticipating them
from a previous period –, which might result in disjointed areas over different
elevations.

Analyzing the quantifiable outcomes, Figure 6 shows that SimSched DBS


has respected the defined production constraints and additionally has achieved
almost the entire processing stream capacity. The software, excepting for the limit
curve that has been manually added, generates such charts automatically in
XLSX format, which is commonly used by well-known office suites. Respecting
the production limits is considered the “third” most important constraint to
SimSched DBS, even though the software does not have a clear-cut hierarchy
defined, as it considers each restriction at the same time. For the developers,
17

solutions that disrespect this parameter to achieve higher NPV are not considered
feasible as they might require a significant increase in the CAPEX (Capital
Expenditure) of the project. Geotechnical constraints are the “second” most
important ones as they are related to safety. Surfaces are the “most” important
constraints within the algorithm framework.

The importance given to surfaces regarding the mentioned “hierarchy”,


being used to force and/or restrict mining, provides the mine planners with the
power to manipulate the results. Such feature allows users to correct disjointed
areas, for instance, as well as operationalizing the final pit in a proper software
and importing the modified surface back to perform another scheduling. While
this possibility makes room to mine planners apply their knowledge and judgment
to take decisions, it also guarantees that such decisions will have more impact
than the algorithm choices. Besides, the operational manipulation to include
ramps and accesses will be postponed from an intermediate step – in the
conventional scheduling –, to a final step, which implies that more scenarios could
be fairly tested and compared in reasonable time before operationalizing the best
one.

Figure 6: Default Case and its Production Process tonnages over the mine lifespan.

Figure 7 shows that the Production Dump has not achieved the proposed
limits, which means the total amount of material has also not achieved the full
potential of the production and the chosen fleet. Achieving the full capacity is the
“secondary” objective after respecting the production limits, but if the software
18

does not find higher NPV while achieving the production limit, it decides to
disrespect such constraint, making room for engineers to define an alternative
limit. Figure 8 presents the stock process, the cumulative stocked ore and the
stock-in and stock-out process.

Figure 7: Default Case and its Production Dump tonnages over the mine lifespan.

Figure 8: Default Case, its Cumulative Stocked ore tonnages over the mine lifespan and the stock-in and
stock-out process.

The cumulative NPV of the Default Case for the Direct Block Scheduling
of the McLaughlin deposit has achieved 842.5 million monetary units (in this case
U.S. dollars have been used in the calculations of block values) as shown in
Figure 9. Regarding the processing time, the default case has spent 1.18 hour.
19

Figure 9: Default Case and its Cumulative NPV over the mine lifespan.

5 CASE 02: RESTRICT MINING

SimSched DBS uses surfaces to define slope angles and this


methodology has already shown its competence to eradicate geotechnical errors
in comparison with the blocks precedence methodology in prior studies (Baretta
& Marinho, 2014) (Baretta & Marinho, 2015). Following an approach studied by
Marinho (2013), a surface-constrained mine production scheduling has been
adopted, and the software allows using surfaces to force and restrict mining to
perform the mine scheduling.

Taking advantage of this feature, a second scenario has been prepared


supposing a restricted area due to a lack of concession rights during the first five
years of operation. The restricted area has been defined by calculating the
thickness of material from topography and surface of period five of the previous
scenario, which is the period in which the restricted area would be applied. By
filtering the thickness values greater than 50 meters, it has been possible to
determine the area that has been preferably mined during the first five periods of
Case 01. The region that has not been mined in expressive tonnages was
restricted. This arbitrary decision has been taken considering the intentions for
the Case 03, forcing a determined area for a supposed demand of placing a
secondary waste dump. The other parameters have remained the same. Figure
20

10 shows an overview of Direct Block Scheduling result of Case 02 along with


the restricting surface used.

Figure 10: Direct Block Scheduling result of the Restrict Mining Case along with the restricting surface
used.

SimSched DBS interprets restricting surfaces guaranteeing that regions,


in which the deposit is situated beneath the surface, or a portion of it, will not be
mined until the specified period. On the other hand, the uncovered region of the
model will be available to be mined. Figure 11 shows a rough contour (in red) of
the region that will be restricted during the first five years of the mine project.
Besides, any adaption of slope angle that might be required during the
scheduling, i.e., mining more or fewer blocks along the vertical direction on the
boundary of the surface, will occur outside the restricted area.

Figure 11: A rough contour (in red) of the restricted area, which has been applied during the first five years
of the mine project.
21

Figure 12: End of Period 2, showing that the forced area has been entirely exhausted.

Figure 12 presents the end of the second period, to which has been set
the limit of time to exhaust the forced region. Figure 13, Figure 14, and Figure 15
show the visual evolution, every five periods, of the mine project resulted by a
Direct Block Scheduling using the set of parameters defined as Case 02: Restrict
Mining. By comparing Figure 11 and Figure 13, it is possible to verify the
effectiveness of the restricting surface, as SimSched DBS has not mined anything
during the first five periods.

Figure 13: Direct Block Scheduling result of the Restrict Mining Case showing Period 5.
22

Figure 14: Direct Block Scheduling result of the Restrict Mining Case showing Period 10.

Figure 15: Direct Block Scheduling result of the Restrict Mining Case showing Period 15.

By comparing Figure 5 and Figure 15, it is possible to note visual


similarities regarding volumes and geometries during the fifteenth period of both,
the Default and the Restrict Mining cases. However, the software has found other
general solution within the sample space reduced by the new restriction. The
numerical analysis corroborates with that by showing a Life of Mine (LOM)
reduction of 1 year. Figure 16, Figure 17, Figure 18, and Figure 19 present
cumulative charts of the Default Case in addition to the Restrict Mining Case.

Figure 16 shows that the Restrict Mining Case has also respected the
production process limits, while Figure 17 shows another type of production
tonnages output reported by the software: the Total Production of material,
whether ore or waste. However, the Total Production does not allow reaching
relevant conclusions and the unpredicted oscillations seem to be a result of the
“NPV-oriented” process, always trying to achieve the highest cumulative Net
Present Value. Such variations might not be a problem, as the processing stream
has a much more constant production and is the major factor that will impact the
cash inflows and the economic performance of a given mine project. Moreover,
Figure 18 presents the Default Case, the Restrict Case and their cumulative
stocked material over the mine lifespan, and both limits being respected.
23

Figure 16: Cumulative Production Process tonnages over the mine lifespan.

Figure 17: Cumulative Total Production tonnages over the mine lifespan.
24

Figure 18: Default Case, Restrict Mining Case, and their Cumulative Stocked ore tonnages over the mine
lifespan.

Regarding the NPV, the Restrict Case has decreased 4.84% against the
Default Case, reducing from 842.5 million dollars to 801.6 million dollars. The
second case has an additional constraint in comparison with the previous
scenario, and it has produced a decrease in NPV. Regarding the processing time,
the second case has spent 1.07 hour, which represents a decrease of 9.3%.

Figure 19: Default Case, Restrict Mining Case, and their Cumulative NPV over the mine lifespan.

SimSched DBS contains its own heuristics, thus, it does not provide
necessarily deterministic results, which in turn could imply in an intuitive thought
that reduced sample spaces will produce poorer results. This way, more
25

constrained scenarios will not necessarily reduce the cumulative NPV of a given
mining project. Such mathematical perfection does not even exist currently for
the scheduling problem. Therefore, as constraints are added, there is just a
tendency to reduce the cumulative NPV. Due to those heuristics, the developers
have said the algorithm allows finding improved solutions for reduced sample
space scenarios because, this way, seeking higher NPV, the software details its
searches so that it is possible to achieve better results for scenarios with more
restrictions. Consequently, the mining professionals are able to verify if their
assumptions to obtain indirect gains by taking a certain decision will, and how
such assumptions will, affect the mining project and its valuation.

6 CASE 03: FORCING MINING

The third case has consisted of the same scenario to which has been
added another surface-constraint, intending to force mining. SimSched DBS
understands forcing-surfaces by mining every material that is situated above a
surface until reaching it within a specified range of time. The other regions of the
deposit will be beneath the surface, thus, they will be less preferred areas to be
mined due to the strength of the force-mining constraint. Some caution is required
to understand such preference: while the forced area is an obligation, the non-
forced area is a possibility. Therefore, despite the fact that non-forced regions are
less preferred to be mined, it does not imply that such areas will be mined only
after the forced ones.

The motivation behind this scenario has been the possibility of


considering the demand of exploiting material from a peripheral area aiming the
allocation of a secondary waste dump inside an exhausted region of the pit. By
using the topography surface and the Surface 18, which is the latest one
generated by Case 01, the thickness of the material being mined during the entire
life of mine has been determined by taking the difference between the topography
Z coordinates and the Surface 18 Z coordinates. Then, the thickness values have
been filtered, removing the zero values.

Figure 20 and Figure 21 show areas from top view perspective that are
slightly distorted regarding X and Y coordinates proportions but clarify the
procedure realized. While Figure 20 presents the full area being mined during the
26

entire life of mine in Case 01, Figure 21 presents the available area during the
first five periods, i.e., the non-restricted area. These steps could have been made
at once, but have been separated to a better understanding. Analyzing Figure 21,
the most reasonable area to place a waste dump has been the one in which Y
coordinate is greater than 1800 meters. The aforementioned procedure in
addition to the manipulation of the surface to change the elevation points resulted
in Figure 22.

Figure 20: McLaughlin Deposit, mined areas during the entire life of mine from a top view.

Figure 21: McLaughlin Deposit, non-restricted areas that are available to be mined during the first five years.
27

Figure 22: Direct Block Scheduling result of the Force Mining Case along with the forcing surface used and
a rough contour of the forced area (in red).

Figure 23: Direct Block Scheduling result of the Force Mining Case showing Period 2.

In addition, analyzing Figure 23 it is possible to note that at the end of the


second period the entire forced area has been mined, in comparison with Figure
22, in which appears two shades of dark blue. Moreover, Figure 24 shows the
results of Case 03 along with the restricted surface that has been used.
Furthermore, Figure 25 shows the fifth period of the mine project, which permit to
compare the restricting surface and conclude the effectiveness of the restriction.
28

Figure 24: Direct Block Scheduling result of the Force Mining Case along with the restricting surface used
and a rough contour of the restricted area (in red).

Again, Figure 25, Figure 26, and Figure 27 show the physical evolution
of the mine project, every five years.

Figure 25: Direct Block Scheduling result of the Force Mining Case showing Period 5.

Figure 26: Direct Block Scheduling result of the Force Mining Case showing Period 10.
29

Figure 27: Direct Block Scheduling result of the Force Mining Case showing Period 15.

Figure 28 presents a cumulative chart with the production process


tonnage of each scenario and the common limit. Similarly, the Case 03 has
reduced the LOM by one year in comparison with the Case 02.

Figure 28: Cumulative Production Process tonnages over the mine lifespan.
30

Figure 29: Cumulative Production Process tonnages over the mine lifespan.

Figure 30: Default Case, Restrict Mining Case, and their Cumulative Stocked ore tonnages over the mine
lifespan.

Figure 29 and Figure 30 show the total production tonnage and the stock
process in cumulative charts covering all cases. Despite the limits are being
respected, the results are still inconclusive.
31

Figure 31: Default Case, Restrict Mining Case, and their Cumulative NPV over the mine lifespan.

Finally, Figure 31 presents the cumulative NPV of all cases. The Restrict
and Force Case has achieved an NPV of 735.9 million dollars, representing a
decrease of 8.20% in comparison with the Restrict Case and a decrease of
12.65% in comparison with the Default Case. This case has performed the mine
scheduling within 1 hour, a decrease of 15.3% in comparison with the default
case. Figure 32 shows a comparison in percentage of the time taken to perform
a mine scheduling of each case. The default case has been set as 100%.

Figure 32: Comparison of processing time spent by each scenario in percentage.

In this analysis, the processing time has reduced as more constraints


have been considered. However, this is not a rule as extremely constrained
scenarios could make the software struggle to find a solution for longer times.
32

This benchmark, on the other hand shows that this base takes a reasonable time,
even though more complex datasets could take more time.

9 APPLYING ECONOMIC UNCERTAINTY: MAKING ROOM FOR FUTURE


SUDIES

Although not being sufficiently representative of reality, the following


scenarios will compare the adoption of fixed ore price changes, exemplifying
already implemented capacities of the software and making room for future
studies that could take advantage of such feature in order to perform deeper
analysis on economic uncertainty. Under this purpose, fixed increases of 30 and
50% and decreases of 30 and 50% in ore price have been stablished.

9.1 Case 04: Fixed Price Changes – Single-Independent


Scenarios

Initially, the resource block model has been manipulated in order to


create complementary information for processing streams based on different ore
prices. In spite of not having a proper interface to perform this task in an easy
way, it allows deciding easily which one will be used for a given scenario. These
manipulations could be easily done via well-known worksheet softwares and
even via mine planning softwares. Single and independent scenarios have been
run via Direct Block Scheduling, and have produced the chart shown in Figure
33. These scenarios have considered the default case, i.e., without operational
constraints.

The default case has achieved previously a NPV of 842.5 million dollars.
Considering a fixed increase in ore price of 30%, the NPV achieved has been
1157.7 million dollars, representing an increase in 37.4% in the cumulative NPV.
On the other hand, the fixed decrease of 30% in ore price has produced a NPV
of 470.4 million dollars, or a decrease of 44.2% in the cumulative NPV.

Considering the 50% fixed increase in ore price, the NPV achieved has
been 1368.9 million dollars, while the 50% decrease in ore price has produced a
cumulative NPV of 255.6 million dollars. These values represent an increase of
62.5% and a decrease of 69.7%, respectively, in comparison with the default
case.
33

Figure 33: Aggregated single and independent scenarios with fixed ore price changes and their cumulative
NPV's.

9.2 Case 05: Fixed Price Changes – Stochastic Simulation

Stochastic simulations requires equally probable models and the ongoing


studies around this methodology apply it, in general, considering uncertainties
related to geological aspects, such as grade and/or volume of ore. While single
scenarios of distinct models could be run separately, a stochastic scenario
consist of obeying all the single scenarios at once through an adapted resource
block model that contains equally probable values for a given set of variables, to
which there is a certain level of uncertainty. In this optimization approach, the
results come along with risk-profile indicators.

Herein, for lack of time, only a preliminary analysis will be considered,


just to exemplify other possibilities of exploring the stochastic optimization and its
current commercial potential. Using an adapted resource block model that
includes columns of economic values, based on the aforementioned fixed price
changes, for the same processing stream, a financial stochastic scenario has
been run. The results is shown in Figure 34 and P10 and P90 curves, which
represent the probabilities of occurrence of 10 and 90%, have been removed due
to their insignificance in light of the representativeness of the values adopted.
34

Figure 34: Economic-Stochastic scenario based on fixed ore price changes.

The maximum cumulative NPV obtained has been 1541.3 million dollars
and the minimum has been 235.6 million dollars. The expected life of mine has
reduced in 2 years considering the default case.

Comparing Cases 04 and 05, it is possible to notice that the results of


Case 04 are inside the minimum-maximum range generated by the “stochastic”
case. REASON. The expected NPV has achieved 888.4 million dollars, an
increase of 5.45% when comparing with the default case. REASON.
35

13 CONCLUSIONS

This study has focused on a novel and global approach to the mine
scheduling by analyzing a commercial software that performs Direct Block
Scheduling (DBS). The software used, SimSched Direct Block Scheduler, which
is still a beta version, uses Mixed Integer Linear Programming and heuristics.
Three cumulative and operational scenarios have been run in addition to two
preliminary studies comparing economic uncertainties through a simplification
that has intended to explore the stochastic potential of DBS.

This approach consists of more global optimization framework that allows


for much more variables being considered at once. Such possibility increases the
information available by considering a vast range of assumptions performed
easily and in reasonable time, which in turn support the decision-making at many
levels. Therefore, a holistic view has been addressed for many researches as a
fundamental part of strategic mine planning. Besides, the stochastic mine
planning allows for many equally probable models, respecting each once at once.
The stochastic scheduling has been proving its efficiency to handle uncertainties,
originally considering grades and volumes but also with a high potential yet to be
explored, i.e., considering uncertainties regarding other variables.

The three chief cases have shown the efficiency of the current state of
SimSched Direct Block Scheduler to handle a dataset of more than 2 million
blocks in reasonable times from one up to 1.18 hour. The Net Present value has
decreased as more restrictions have been added, although this is not a rule.

Moreover, the surface-constraints have shown their efficiency and a


potential to be explored in further studies due to the possibilities of taking into
account specific situations such as a need to preserve a determined area or even
postpone it because of a lack of concession rights. Those surfaces permit to force
mining in order to place an in-pit crusher or to induce a pre-stripping in a specific
region of the mine. Furthermore, the surface-constrained scenarios make room
for mine planners to use their knowledge and expertise to conduct the software
prior to the scheduling. Further studies could explore deeper the economic and
geotechnical uncertainties.
36

REFERENCES

Baretta, F. S. & Marinho, A., 2014. Impacts of Slope Angle Approximations on Pit
Optimization.. Belo Horizonte, IBRAM: Instituto Brasileiro de Mineração.

Baretta, F. S. & Marinho, A., 2015. The impacts of slope angle approximations
on open pit mining production. Fairbanks, Society for Mining, Metallurgy, and
Exploration.

Benndorf, J. & Dimitrakopoulos, R., 2013. Stochastic Long-Term Production


Scheduling of Iron Ore Deposits: Integrating Joint Multi-Element Geological
Uncertainty. Journal of Mining Science, 49(1), pp. 68-81.

Camus, J., 2011. Value Creation in the mining business. Mining Engineering,
63(3), pp. 43-52.

Carvalho, J. P. d., 2016. Simulação Estocástica e Sequenciamento de Blocos de


um Depósito de Cobre, Belo Horizonte: Departamento de Engenharia de Minas.

Enderlin, D., 2002. Geology of the McLaughlin Deposit. [Online]


Available at: http://nrs.ucdavis.edu/McL/natural/geology/region/region1.htm
[Acesso em 18 10 2016].

Espinoza, D., Goycoolea, M., Moreno, E. & Newman, A. N., 2012. MineLib: A
Library of Open Pit Mining Problems. Annals of Operations Research, 6
December, 206(1), pp. 91-114.

Godoy, M. & Dimitrakopoulos, R., 2004. Managing risk and waste mining in long-
term production scheduling of open-pit mines. SME Transactions, Volume 316,
pp. 43-50.

Goycoolea, M., Moreno, E. & Rivera, O., 2013. Direct Optimization of an open
cut Scheduling Policy. APCOM 2013 - Applications of Computers and Operations
Research in, pp. 424-432.

Guimarães, O. & Marinho, A., 2014. Sequenciamento Direto de Blocos. Belo


Horizonte, CBMINA: 8º Congresso Brasileiro de Mina a Céu Aberto.

King, B. M., 2011. Optimal Mining Practice In Strategic Planning. Journal of


Mining Science, 47(2).
37

Lambert, W. & Newman, A., 2013. Analyzing solutions of the open-pit block
sequencing problem obtained via Lagrangian techniques. Mining Engineering,
February, Volume 65, pp. 47-53.

Leite, A. & Dimitrakopoulos, R., 2014. Stochastic optimization of mine production


scheduling with uncertain. International Journal of Mining Science and
Technology, 24(6), p. 755–762.

Marinho, A., 2013. Surface constrained life-of-mine production scheduling,


Montréal: s.n.

Marinho, A. & Martinez, L., 2015. Traditional versus stochastic mine planning
under material type and grade. Fairbanks, Society for Mining, Metallurgy, and
Exploration.

Newman, A. M. et al., 2010. A Review of Operations Research in Mine Planning.


Interfaces, 7 April, 40(3), pp. 222-245.

Porter, M., 1985. Competitive Advantage: Creating and Sustaining Superior, New
York: Simon & Schuster.

UC Davis Natural Reserve System, 2005. Homestake Mining Company -


McLaughlin Mine: Geology. [Online]
Available at: http://nrs.ucdavis.edu/McL/natural/geology/geo.htm
[Acesso em 24 10 2016].

UC Davis Natural Reserve System, 2007. Geology. [Online]


Available at: http://nrs.ucdavis.edu/mcl/natural/geology/
[Acesso em 25 10 2016].

Whittle, G., 2004. Global Asset Optimisation. Perth, The Australasian Institute of
Mining and Metallurgy, Spectrum Series, pp. 361-366.

Potrebbero piacerti anche