Professional Documents
Culture Documents
Developing an algorithm is the first step when devising a mathematical model to optimise
mining operations. An algorithm is a systematic procedure to accomplish a specific task. It
is the idea behind any reasonable computer programme. Mining has reached a stage
where mine planners have embraced optimisation when planning their operations. Mine
planning for any operation (surface or underground) is aimed at maximising the value
realized from the ore body. Consequently, most mines use computer-programming tools
to optimise the ore extraction processes.
The challenge faced by researchers in the mining industry when solving mining-related
problems, is in the design of algorithms that solve problems accurately, quickly and
efficiently. These algorithms need to apply to real mining environments. Hence, this
paper provides a guideline to be employed when developing an algorithm for mining-
related problems. There is no one process to develop an algorithm due to the unique
quality of the ore body setting environments; however, the material presented in this
paper provides generic thought processes.
INTRODUCTION
An algorithm is a procedure that takes any of the possible input instances and transforms them to the
desired output. It must solve a general but well-specified problem. An algorithmic problem is specified
by describing the complete set of instances it must work on and of its output after running on one of
these instances. Determining the type of problem is a first step towards finding a solution.
Generally, mining-related problems involve exploration, mining, processing and refining. It is worth
noting that exploration has not been fully developed as a problem, however, it is imperative to develop
algorithms to determine how and where to explore. Currently, optimisation problems are categorised
into surface and underground mining problems.
Mining-related problems generate along the mining process, which has three major stages: mining,
processing and refining. Most optimisation problems in surface mining operations are found when
undertaking production scheduling and cut-off grade optimisation. The major hurdles encountered
when optimising underground mining operations is optimising development, stope outlining, and ore
production scheduling in a single run.
Mining companies must advocate and develop global strategic mine plans that optimise both surface
and underground operations to manage any changes that may affect the operations according to their
profiles of operations. In order for mine plans to incorporate such changes, optimisation methods need
to be adopted and applied (Musingwini, 2016).
Most professional programmers are not well prepared to tackle algorithm design problems related to
mining. This can be attributed to the complexity of mining processes. The techniques of algorithm
design form one of the core practical technologies of computer science. Designing correct, efficient,
and implementable algorithms for real-world problems requires access to two distinct bodies of
knowledge: resources and techniques (Skiena, 2008).
Good algorithm designers understand several fundamental algorithm design techniques such as
heuristics, sorting and searching, dynamic programming, data structures and linear programming. The
most important algorithmic design technique is modelling. It involves conceptualizing a messy real-
world application into a clean problem that is defined in an algorithm. It is prudent to first figure out
what is known about the particular problem, rather than labouring from scratch to produce a new
algorithm for every task. Existing algorithms serve as a starting point rather than re-implementation of
the algorithms. They provide sufficient source material to model most any application (Skiena, 2008).
Operation research
Operation research is a mathematical science commonly applied to optimisation problems that are
concerned with improving the performance of an original. It determines the maximum (of profit,
performance, or yield) or minimum (of loss, risk, or cost) of some real-world objective. When operation
research is undertaken, these steps are followed (Little, 2012):
The paper shows the major algorithms developed in both surface and underground mining operations
to show the importance of applying operation research on mining optimisation problems. The
combination of these existing models is important as it strategically and tactically solves the problems
on a global scale, generating optimal solution. For instance, underground scheduling is commonly
done in a sequential manner in that the output from one planning process is the input to the next. This
is mostly applicable when using manual methods that generate non-optimal NPV, thus the need to
optimize the whole process. For example, Little (2016) developed an integrated solution for
underground mining operations with two major areas of concerns: stope layout and boundary and
138
production scheduling. There are various studies which aim to optimise development and
infrastructure along with stope layout and the production scheduling.
There are studies focusing on combining these common algorithms to get an optimal solution for both
surface and underground mining operations. These combined approaches are applied in mining
software such as Whittle 4X, Surpac, and Maptek Vulcan.
Mine planners use the economic block model to select blocks that should be included in the optimum
extraction of the orebody with an objective function to maximise the value. There are various
techniques used to determine which blocks are to be included in the final layout, most common are the
operations research techniques (algorithms). There are numerous algorithms developed to optimise
layout for either surface or underground mining.
Algorithms developed in surface mining
The development and implementation of mathematical formulations to solve complex optimisation
problems in mine planning has been progressive, especially in surface mining. Algorithms developed in
surface mining involves production scheduling which embodies several major aspects that affect the
overall mine output. These include cut-off grade optimisation, block economic evaluation, slope design
and requirements, pit design and optimisation and pushback design.
The main objective in open pit production scheduling is to maximise the net present value of
future cash flows with slope, bench or block precedence, grade blending and production capacities
(mining and milling capacities) as the constraints. It identifies the sequence of extraction then
undertakes a pushback/cutback/phase design and block-by-block production schedule based on
the constraints.
A mining complex constitutes a mine, ore processing stream and metal refinery. In order to avoid
losses in a mining system, these areas should be optimally planned (Githiria et al., 2016). Modelling of
these optimisation problems related to mining operations follow the same thought-process. They
follow the same conventional mining layout as shown in Figure 2. The changes arising along the
process depends on the nature of the problem, if surface or underground mining related.
Cut-off grade is defined as the grade that is normally used to distinguish between ore and waste within
a given ore body. It also distinguishes between various ore types before processing takes place for
different metallurgical processing options. Lane (1964, 1988) came up with a 3D algorithm in
calculating cut-off grades. Lane’s process accounts for economic and geological parameters and
production capacities in the calculation of cut-off grades. Lane’s theory uses mathematical derivations
to get six cut- off grades, which are then sorted using a sorting algorithm to get an optimum cut-off
grade. Three of the cut-off grades are limiting cut-off grades, derived by assuming that the three stages;
mining, processing and refining are each independently constraining throughput. The limiting cut-off
grades are based upon capacities, costs, and price. The other three are balancing cut-off grades that are
determined by assuming that two of the three stages are operating in unison at their capacity limits.
They are based on the grade, and capacities of the production stages. The rate of production in a
mining system is determined by the production capacities (Githiria and Musingwini, 2018).
The optimal cut-off grade approach (Lane 1964, 1988) has been modified in several studies by
incorporating different mining scenarios. For instance, Githiria, Muriuki and Musingwini (2016)
focussed on optimising cut-off grade under deterministic variables and developed a computer-aided
application using Lane’s algorithm. The application is useful in mine planning when coming up with a
cut-off grade policy. The cut-off grades are dynamic since Lane’s approach accounts for the change in
the grade-tonnage curve. This ensures that the cut-off grades used in mine planning are proportional
to the rate of ore depletion.
Floating cone is one of the common pit limit calculation methods. It is developed by Kennecott Copper
Corporation during the 1960s that entailed a series of interlocking frustum shaped removal increments
(Carlson, et al., 1966). It is a computerized method and thus is easy to understand by engineers
coupled with a simple computational algorithm. It is a better method than the trial and error method
as it produces fairly good results for mine planning, however it misses combinations of profitable
blocks and extends or shrinks the ultimate pit beyond optimal limit. Wright (1999) and Elahi, et al.
(2011), further modified it.
Lerch-Grossman (L-G) algorithm (Lerch and Grossman, 1965) involves modelling of an ore body, by
pit planning, using the algorithm to plan for an open pit mine. It obtains the ultimate pit limits and
pushback regions, and then obtains scenarios of subsequent production scheduling that should be
followed to maximise the mine value with emphasis on maintaining consistent production. Mining
software programmes such as Whittle 4X have incorporated the algorithm.
L.G 2D algorithm is based on dynamic programming while the 3D version of the algorithm is based on
graph theory. The L-G 3D Algorithm, a graph closure solution, is implemented in Whittle 4X to
generate the ultimate pit from the orebody model. Nested pits are thereby created within the ultimate
pits by changing capacities of arcs between the nodes of the graph. The nested pits are combined and
finally a production schedule can be properly obtained. Any feasible pit outline has a cash value and to
compute it, a mining/ production schedule must be decided upon and then the pit mined out,
progressively accumulating the revenues (Whittle, 1990).
The LG algorithm is widely applied in the industry due to its efficiency. However, recently newer
algorithms for network flow have been developed theoretically more efficient than the LG algorithm.
examples of the network flow algorithms include push-relabel and pseudoflow algorithms. Hochbaum
(1998, 2000, 2002) generalised the LG algorithm to a pseudoflow network model. It proved to be
theoretically more efficient than LG algorithm since it did not allow selection of constraints in its
calculations (Muir, 2004).
Other methods that have been applied in generating ultimate pit limits include; maximum flow
algorithm (Johnson, 1968; Asad and Dimitrakopoulos, 2013), Zhao and Kim algorithm (Zhao and Kim,
1992), Roman’s method based on dynamic programming (Roman, 1974), variations of Lerch-
Grossman algorithm using graph theory, and other methods/models based on linear programming.
Milawa NPV, Milawa Balanced and Fixed lead algorithms are some of the algorithms applied in
production scheduling using Whittle 4X software. They implement mathematical models that combine
pit shells to generate a pushback design. Using Whittle 4X along with calculated mining parameters,
the periodic production and mine value can be predetermined thereby showing elimination of
guesswork and its subsequent effects on mining results (Whittle, 2000). Table I presents the summary of
the algorithms commonly applied in surface mining optimisation problems.
Table I. Summary of some of the algorithms applied in surface mining
Algorithms developed in underground mining
Unlike surface mining, extraction of an underground orebody presents mine planners with great
challenges, as there are numerous possible options. The optimisation problem for underground mining
can be divided into four sub-problems, that is, development and infrastructure placement, stope
layout, equipment selection and deployment, and production scheduling. The circular logic presents
difficulty in deciding which part of the underground optimisation subsets should be a starting point of
optimisation (Figure 3). However, Erdogan et al. (2016) argued that stope layout be considered as the
initial point. The preferred optimisation process will be to optimise the underground mining problem
simultaneously. However, there are numerous constraints to be considered, which will increase the
solution time exponentially and making the problem NP-hard. Numerous algorithms have been
developed for optimisation of infrastructure and development placement, and production scheduling.
This paper presents a summary of algorithms that have been developed to generate stope layouts
(Table II).
Algorithms are developed for a specific problem, thus cannot be generalised. This becomes a
constraint, for example, particle swarm optimisation (PSO) is used to solve a mechanical problem but
to optimise stope layout it will have to be modified accordingly.
The thought-process in solving problems in mining requires two aspects, strategy and tactics. Strategy
represents the quest for developing a framework around which the path to the goal is constructed while
tactics are used to win the minor challenges encountered in the process to generating a solution. When
articulating the reasoning as to why the application is not working, the first thing is to check is whether
you have maximised all possibilities in coming up with a solution.
It is crucial to have a global strategy plan when designing an algorithm in order to come up with tactics
along the solution path. The strategic question is what kind of algorithmic problem the application will
be modelled into. The problem will define the technique used in modelling. The tactical decisions made
along the algorithmic path are critical to the ultimate quality of the solution, but they depend on the
success of the global strategy. The most important questions or decisions made when designing an
algorithm are based on a detailed understanding of the problem and do not require specific expertise.
However, one needs the know-how to come up with an algorithm development procedure. The
thought-process on designing an algorithm involves the following step- by- step procedure:
The algorithmic process gears to solve the problems arising in any operation in a step-to-step format as
shown in Figure 4. The mining engineering related problems are first identified, whether open pit or
underground, then later the model to be applied is determined. Selecting the right jobs and reasoning
about correctness matters most when coming up with an algorithm. A good algorithm designer does
not necessarily reinvent the wheel. There is a catalogue of algorithms developed to solve problems
which can be used as a reference when solving the specific problem in question.
In summary, there is a great need for improvement in mine planning through integrated optimisation
of the whole mining process. Mathematical optimisation techniques have been used in the mining
industry to develop several optimisation software options to maximise profits. Examples of these
mathematical optimisation techniques used in implementing algorithms include:
Exploration drilling results are used to produce a geological block model which is delineated into
thousands in some cases millions of mining blocks in 3D space. Thereafter, a suitable mining method
is selected. Subsequently, mine planners start with generation of mining layouts that are aimed at
maximising the value realised from the orebody. Generating an optimum mining layout is of the
outmost importance as the layout serves as input into the production scheduling process. Thus,
inputting a non-optimal layout into the formulation of production schedules could result in unrealistic
expectations being incorporated into the schedules. Accordingly, having a negative influence on the life
of mine and value realised.
Numerous algorithms have been developed to aid in decision making with regards to mining layout.
However, underground mining has been lagging behind and no known algorithm can guarantee
optimal solution in the 3D space. The challenge faced by mine planners when solving mining related
problems is in the design of algorithms that solve problems accurately, fast and efficient. These
algorithms need to applicable in a real mining environment. The generic approach in solving mining
optimisation problems involves formulating a mathematical model in form of an algorithm. Linear,
quadratic functions or any other mathematical function of the mining parameters against time and/or
any other relevant parameter are used to achieve the algorithmic procedure. However, there is no
single process used in developing an algorithm due to the uniqueness of the mining environment.
REFERENCES
Carlson, T. R., Erickson, J. D., O’Brain D. T. and Pana, M. T. ( 1966). Computer techniques in mine
planning, Mining Engineering, Vol. 18, No. 5, pp. 53-56.
Elahi Zenyi E, Kakaie R, Yousefi A.( 2011). A new algorithm for optimum pit design: floating cone
method III, Journal of Mining Environment, Vol. 2, No. 2, pp.118–25.
Erdogan, G., Cigla, M., Topal, E., Yavuz, M. ( 2016). Implementation and comparison of four stope
boundary optimisation algorithms in an existing underground mine, 6th International
Conference on Computer Applications in the Minerals Industries, Istanbul, Turkey, 5–7
October 2016
Githiria, J., Muriuki, J. & Musingwini, C. ( 2016). Development of a computer-aided application using
Lane’s algorithm to optimise cut-off grade, Journal of the Southern African Institute of
Mining and Metallurgy, Vol. 116, No. 11, pp. 1027-1035.
Githiria, J., Musingwini, C. ( 2018). Comparison of cut-off grade models in mine planning for
improved value creation based on NPV, Proceedings of the 6th Regional Conference of the
Society of Mining Professors (SOMP), Johannesburg, South Africa, pp. 347–362.
Hochbaum, D. S. and Chen, A.( 2000). Performance analysis and best implementations of old and new
algorithms for the open-pit mining problem, Operations Research, Vol. 48, No. 6, pp. 894-
914.
Hochbaum, D. S. ( 1998). The pseudoflow algorithm and the pseudoflow-based simplex for the
maximum flow problem, Proceedings of 6th International Integer Programming and
Combinatorial Optimization Conference. Heidelberg: Springer; 1412:325–37.
http://doi.org/10.1007/3-540- 69346-7_25.
Hochbaum, D. S. ( 2001). A new-old algorithm for minimum cut and maximum-flow in closure graphs,
Networks, special 30th anniversary paper, Vol. 37, No. 4, pp. 171-193.
Hochbaum, D. S. ( 2002). The Pseudoflow algorithm: a new algorithm for the maximum flow problem,
UC Berkeley manuscript, December.
Johnson, T. B. ( 1968). Optimum open pit mine production scheduling. PhD Thesis, Department of
IEOR, University of California, Berkeley, CA.
Koenigsberg E. ( 1982). The optimum contours of an open pit mine: an application of dynamic
programming, Proceedings of 17th International Symposium Application of Computers and Operation
Research in the Mineral Industry. pp. 274–87.
Lane, K.F. (1964). Choosing the optimum cut-off grade, Colorado School of Mines Quarterly, Vol. 59,
No.
4, pp. 811–829.
Lane, K.F., (1988). The Economic Definition of Ore: Cut-off Grade in Theory and Practice. Mining
Journal Books, London.
Lerchs, H. and Grossmann, I. F. ( 1965). Optimum design of open pit Mine, CIM Bulletin, No. 58(633),
pp. 47 – 54.
Little, J. ( 2012). Simultaneous optimisation of stope layouts and production schedules for long-term
underground mine planning, PhD thesis, University of Queensland, Brisbane, Australia.
Muir, D. C. W. ( 2004). Pseudoflow, New Life for Lerchs-Grossmann Pit Optimisation, Orebody
Modelling and Strategic Mine Planning, Spectrum Series, Vol. 14, pp. 97-104.
Nhleko, A.S.; Tholana, T. and Neingo, P.N. ( 2018). A review of underground stope boundary
optimization algorithms. Resource Policy, Vol. 56, pp. 59 – 69.
Roman, R.J. ( 1974). The role of time value of money in determining an open pit mining sequence and
pit limits, Proceedings of the 12th International Symposium on Application of Computers and
Operation Research in the Minerals Industries (APCOM), pp. 72-85.
Skiena, S.S. (2008). The Algorithm Design Manual, 2nd ed., DOI: 10.1007/978-1-84800-070-4 10, Springer-
Verlag London Limited.
Whittle, D. ( 2000). Proteus environment: Sensitivity work made easy. Whittle North American Mine
Planning Conference, CO.
Whittle, J. ( 1990). Open Pit Optimization, Surface Mining, 2nd Edition, SME-Online Digital Library.
Wright, E. A. ( 1999). MOVING CONE II - A Simple Algorithm for Optimum Pit Limits Design,
Proceedings of the 28th Symposium on the Application of Computers and Operations Research in
the Mineral Industries (APCOM), pp. 367-374.
Zhao, H. and Kim, Y.C. ( 1992). A new optimum pit limit design algorithm, Proceedings of the 23rd
International Symposium on the Application of Computers and Operations Research in the
Mineral Industries, pp. 423-434.
Joseph Muchiri Githiria
PhD Student
University of the Witwatersrand
J. Githiria is currently doing his PhD in Mining Engineering at the University of the Witwatersrand. He
holds a Master of Engineering Science (Mining) in Curtin University (Western Australia School of
Mines) and a Bachelor of Science in Mining and Mineral Processing Engineering. His research
interests include mine planning and optimisation in surface and underground mining and mine design
with aid of computer and operations research applications. He is a Member of the Australian Institute
of Mining and Metallurgy (AusIMM) and the Southern African Institute of Mining and Metallurgy
(SAIMM).
150
View publication stats