VII-th International Scientific Conference SGEM2007, 2007-04-27, Sofia, Bulgaria

A STUDY ON THE OPTIMIZATION ALGORITHMS FOR DETERMINING

OPEN-PIT AND UNDERGROUND MINING LIMITS

Asso.Pro.K.Shahriar*, Pro.K.Oraee**, PhD Stud. E.Bakhtavar*

*
Department of Mining and Metallurgy. Eng, Amirkabir University of Technology, Iran
**
Department of Mining. Eng, Stirling University, UK

ABSTRACT
Determination of Mining Limits (ML) for each of Open-Pit (OP) and Underground
(Ug) is one of the most significant part impacts the economic aspects. Optimization process
for the mentioned objective in both of OP and Ug mining was done on the economic block
model. In this paper, the applied Optimization Algorithms (OA)s for ML determination
extensively were studied. Because of the major difference between the shape and form of
mining limits in two main exploitation methods, implementation of the algorithms in
searching and founding the mining limits for OP and Ug is also different.
Keywords: Optimization algorithm, open-pit, underground, mining limit, determination

INTRODUCTION
In the last 40 years or so that modern computers were invented various form of
techniques to judge and select one choice out of the many options of mine planning were
used. Optimization has now been such a taboo word in mine planning that managers and
financiers almost routinely ask of any plans submitted for approval, “Have they been
optimized?” [1]. Underground operations and its mining limits (boundaries) are less
amenable to optimization techniques. Optimization techniques frequently refer to economic
criteria which require each flow projections. Although, implementation of mining limits
OAs initially were due to pit limits optimization, but during the last years these algorithms
were developed as the optimization tools for determination of underground mining limits.
In mining, there are three principal kinds of models (ore body models, financial or
economical models, and mine models) commonly employed in developing long term mine
plans. Mine models are described as a form of established structures. Like underground
mine will have shafts or inclines, outwardly and downwardly expanding stope or panels,
drifts and cross-cut and funnels or chutes joining with the level drifts. The pit is usually
represented by a series of levels (easily conceived as benches) and a set of perimeter, one
on each level expanding upwards from one level to the next. In mining, financial or
economical models are commonly utilized for planning, budgeting and economic
evaluations. Up to now, the whole of presented mining limits OAs were based on the
financial blocks model (with economical and grade values).

OPTIMIZATION CONCEPT
Optimization in term of mine planning is known as descriptive of a set of techniques
that introduces analytical mathematical methods to arrive at an option out of multiple

1
VII-th International Scientific Conference SGEM2007, 2007-04-27, Sofia, Bulgaria

choice-thus bringing it into planning activities. The techniques embody a process of three
stages [2].
1. The creation of a mathematical model of the activity.
2. The adoption of criterion.
3. The development of an algorithm.

CLASSIFICATION OF OAs FOR MINING LIMITS DETERMINATION

In 1978 Kim described the various optimization techniques. His classification is
included rigorous (in the sense that the algorithm can be proved to find optimum solution)
and heuristic [3]. According to the two mentioned optimization techniques (rigorous and
heuristic), mining limits OAs based on economical block models are classified in two main
groups named open pit and underground. In this case, the most significant algorithms which
are studied in this paper for both of mining methods are summarized in two classes as
figure 1.

Mining Limits Optimization Algorithms

Open-Pit Underground
Rigorous Heuristic Rigorous Heuristic
1) Moving cone
1) Dynamic 1) Dynamic 1) Floating Stope
Programming (1965) 2) Network Flow Programming (1995)
Analysis (1968)
2) Graph 2) Maximum
3) Linear 2) Branch & Bound
Theory (1965) (1995) Value
Programming
Neighbourhood
3) Parametric 4) Korobov (1974) 3) Probable Stope
Analysis (1975) (2004)
5) Genetic
4) Stochastic Algorithm (1994)
Process (2006) 6) Artificial
Intelligence (1996)

Figure 1: Classification of mining limits optimization algorithms based on economical block models

INVESTIGATION OF PIT LIMITS OAs

A number of OAs to solve the basic open pit mining problem especially for pit limit
determination have been developed over the years. These algorithms include rigorous and
heuristic approaches. Almost all optimal open pit design algorithms, with the exception of
elementary methods that are used for some stratiform deposits, are applied to a regular,
fixed, three dimensional block model of orebody. The orebody is subdivided into regular
blocks and a value is estimated for each block. This value is almost always the net
(undiscounted) revenue that would be obtained by mining and treating the block and selling

2
VII-th International Scientific Conference SGEM2007, 2007-04-27, Sofia, Bulgaria

its contents. Some methods such as parameterization (Parametric Analysis), used grade
values in the block model [4]. The entire mentioned pit limit OAs are discussed in the next
sections.

PIT LIMITS OAs BASED ON THE RIGOROUS LOGIC

A) DYNAMIC PROGRAMMING (DP)

The optimum final pit limit is that combination of mined blocks which maximizes profit
within practical constrains, such as the stability of the pit slopes and ease of access for
machinery and trucks from one level to the next [5]. Two dimensional Lerchs and
Grossmann algorithm (1965) was the first pit limit optimization technique based on
dynamic programming (DP) algorithm [6]. In 1971, Johnson and Sharp developed a three
dimensional DP algorithm which is a direct extension of Lerchs and Grossmann. There was
basically two-step procedure, first repeating LG along all cross-sections and then applying
it to a longitudinal section [7]. This satisfies constraints on the individual cross-sections
well, but not in the longitudinal direction, and so some smoothing is still required. In 1982,
a new three dimensional DP algorithm which conducts a search, subject to the geometrical
constraints, through a fully three dimensional array of blocks was introduced by
Koenigsberg [8]. Wilke et al, (1984) were proposed a modification of Koenigsberg, s
formulation that will always yield a non-degenerate solution [9]. However, to achieve
reasonable computing times and storage requires a number of modifications that render the
algorithm sub-optimal [10].

B) GRAPH THEORY (GT)

Lerchs and Grossmann (1965) were represented a new pit limit OA based on rigorous
reasons named Graph Theory (GT) which makes use of property of a block model of an
open cut mine that it can be modeled as a weighted-directed graph in which the vertices
represent blocks and the arcs represent mining restrictions on other blocks. Early version of
the GT algorithm, as described by Zhao and Kim (1992) operates by applying a set of
transformations to the graph’s tree structures until there are no waste blocks preventing the
removal of desirable ore blocks [11], [12]. The GT algorithm is able to ensure the true
optimality of pit design.

C) PARAMETRIC ANALYSIS (PA)

One radically different approach to pit optimization is to parameterize the pit design as
a function of a number of variables. This approach was pioneered by Matheron [13] and
have been described in a number of places [14], [15], [16], [17]. This algorithm is to divide
the problem into two separate parts- technical and economical- and assume that the only pit
design of any interest is that which maximizes the quantity of metal. In the latest
improvement, Whittle (1988) introduced an economic parameter as the ratio of metal price
to the mining cost and developed the Whittle 4-D software, which generates incremental
pits by increasing this economic parameter [18].

3
VII-th International Scientific Conference SGEM2007, 2007-04-27, Sofia, Bulgaria

D) STOCHASTIC PROCESS (SP)

Stochastic Process (SP) algorithm as a pit limit optimization technique was initially
used and introduced by Jalali et al (2006) on the principles of SP and benefits from a
mathematical support based on the probability theory [19].

PIT LIMIT OAs BASED ON THE HEURISTIC LOGIC

Up till the early 1980s, heuristic algorithms were widely used in the mining industry
because they execute faster and are conceptually simpler than other optimizing algorithms.
The algorithms based on the heuristic techniques are not guaranteed to deliver optimal
solutions as well as optimization approaches [20].

A) MOVING CONE (MC)

Moving Cone (MC) algorithm which described by Pana (1965) is to search for cones in
which the total weight of all the blocks in the cone is positive [21]. These cones are added
to the already generated pit. The heuristic idea lies in the assumption that every cone in the
optimal pit is profitable, whereas in fact an optimal pit may consist of a collection of cones
share negative value blocks and have total weight which is positive. This algorithm is very
simple in concept, easy to program and reaches a solution in quite a short time. However,
the variants that have been published to date do not always yield a true optimum, and this
can usually be demonstrated by means of simple counter-examples [22], [23].

B) NETWORK FLOW ANALYSIS (NFA)

In 1968, the use of Network Flow Analysis (NFA) to solve the ultimate pit limit
determination problem for an open cut mine was first proposed by Johnson [24]. Picard
(1976) developed a formal proof for the concept [25].Giannini et al (1991) developed the
software package PITOPTIM, which uses maximum flow algorithm for computing the
optimal pit contour [26]. Yegulalp and Arias (1992) and Yegulalp et al (1993) conducted
computational experiments to compare the performance of a variant of the push-relabel
maximum flow algorithm, the excess-scaling algorithm of Ahuja and Orlin (1989), and
Lerchs and Grossmann algorithm [27], [28], [29].

C) LINEAR PROGRAMMING (LP)

Meyer (1969) was introduced Linear Programming (LP) technique to optimize pit limit.
However, the most significant disadvantage of the approach was its excessive need for
computer memory and computing time that made it impractical [30]. In 1994, Huttagosol
and Cameron formulated ultimate pit limit problem as a large scale transportation problem
with an LP solution [31]. The adapted simplex algorithm of LP is used to solve the dual
system. The required computing time and memory are the main constraints of the method
especially for a large scale problem.

D) KOROBOV (K)
The Korobov algorithm (1974) is a cone-based algorithm that allocates the values from
positive blocks against the negative or zero blocks that are contained in the extraction cones

4
VII-th International Scientific Conference SGEM2007, 2007-04-27, Sofia, Bulgaria

of the positive blocks. A flow chart for the algorithm has been presented else where [32],
[33]. A major improvement to the Korobov algorithm was suggested by Dowd and Onur
(1992), in which they propose joint testing of cones before it decided whether or not to
remove a cone [34].

E) GENETIC ALGORITHM (GA)

Firstly, Denby and Scholfield (1994) employed Genetic Algorithm (GA) as a
optimization tool in open pit design [35].This technique was established on the basis of a
self-training GA and concurrently supplies a combined pit limit and extraction schedule
that intends to maximize the net present value (NPV) of the production.

F) ARTIFICIAL INTELLIGENCE (AI)

Achireko and Frimpong (1996) applied Artificial Neural Networks (ANN)s in
combination with “Conditional Simulation” technique to optimize the pit limits [36]. They
referred the random field properties connected to the ore grades and reserves, which are
ignored in preceding algorithms and displayed them using the “Modified Conditional
Simulation” (MCS) based on the best linear unbiased estimation and local average
subdivision techniques. ANNs are used to classify the block into classes based on their
conditioned economic values.

INVESTIGATION OF UNDERGROUND LIMITS OAs

There are few algorithms available to optimize underground stope boundaries, such as:
Dynamic Programming (DP), Branch and Bound (B&B), Floating Stope (FS), Octree
Division (OD), Downstream Geostatistical (DG), Maximum Value Neighbourhood (MVN),
Probable Stope (PS). OD and DG techniques are implemented on the grade block model of
ore body. But the other mentioned algorithms were performed on the economic block
models. In this study, the algorithms based on economic block model were investigated.

UNDERGROUND LIMITS OAs BASED ON THE RIGOROUS LOGIC

A) DYNAMIC PROGRAMMING (DP)

Initially, the DP algorithm was used by Riddle (1977) to optimize stope layout of block
caving method [37]. This method was established due to modifying the Johnson and Sharp
(1971) approach [7]. The DP algorithm by Riddle is a multi-section two dimensional
solution for three dimensional problems. It means the approach is provided an optimum
stope in two-section, but it is failed to determine the actual optimal stope in three
dimensions. It is notable that this method is limited to the block caving mines and can not
be able in optimizing the layout of other underground stopping methods.

B) BRANCH AND BOUND (B&B)

Ovanic and young (1995) used a separable programming with Branch and Bound
(B&B) technique for economic optimization of stope boundary. An optimal economic
stoping boundary was developed by optimizing the starting and ending locations for mining

5
VII-th International Scientific Conference SGEM2007, 2007-04-27, Sofia, Bulgaria

within each row of blocks. To determine these locations, two piecewise linear, cumulative
functions are used for each row. The first function sums block values along the row for
inclusion within the stope boundary, while the second function sums block values for
exclusion. In this algorithm the stope boundary model is optimized by a mixed integer
approach known as "special ordered sets" [38],[39].

C) PROBABLE STOPE (PS)

Jalali and Ataee-pour (2004) established a technique (named Probable Stope) based on
Riddle DP algorithm (1977) to optimize the stope limits of the mining methods which are
feasible for vein deposits [40]. There is a main difference between PS technique and the
others. This algorithm is implemented basis on a particular economic block model
including the constraints of stope dimensions. Therefore, the most significant constraints
within the objective function are eliminated. It is resulted the algorithm be simple in
concept, easy to program and reaches a solution quickly.

UNDERGROUND LIMITS OAs BASED ON THE HEURISTIC LOGIC

A) FLOATING STOPE (FS)

Floatin Stope (FS) was implemented on a fixed economic block model of the orebody.
The FS is the tool developed by Datamine to define optimal limit for mineable ore or stope
envelope which can be economically extracted by underground stoping methods. The
general concept of the FS method was established by Alford (1995) [41].

C) MAXIMUM VALUE NEIGHBOURHOOD (MVN)

Ataee-pour (1997) proposed the MVN algorithm to optimize stope boundaries using a
three-dimensional fixed economic block model to locate the best neighbourhood of a block,
which guarantees the mine geometry constraints. The neighbourhood concept is based on
the number of mining blocks equivalent to the minimum step size. Since several
neighbourhoods are available for each block, the one that provides the maximum net value
is located for inclusion in the final stope. It can apply to any underground mining method,
although it does not guarantee the true ‘optimum’ stope layout [42].

CONCLUSIONS
There are main and basic differences between the specifications of an open-pit and
those of a stope, which influence mining technology to be employed. The actual optimal
solution is guaranteed for the pit limit optimization and several computer packages are
available. However, only few algorithms have been developed for optimizing final stope
limits in underground mining. Up to now, the all presented mining limits OAs were based
on the financial blocks model (with economical and grade values). These algorithms were
classified as rigorous and heuristic. It is notable that the problem of true optimally has not
been solved in the stope limit optimization yet except FS algorithm of Datamine.

6
VII-th International Scientific Conference SGEM2007, 2007-04-27, Sofia, Bulgaria

REFERENCE

1- Jayanta, B: Principles of Mine Planning; ALLIED Publishers PVT.LTD, 2003, pp. 23-36.
2- Lane, K.F: “Optimization: is it the best?” Strategic programming, Optimization with Whittle; Third
Biennial Conference, 1999, pp. 1-8.
3- Kim, Y.C: Ultimate pit design methodologies using computer models-the state of the art; Mining
Engineering, Colorado, 1978, pp. 133-137.
4- Dowd, P.A., Onur, A. H: Optimizing open pit design and sequencing; proceeding of 23th International
APCOM Symposium, 1992, pp. 411-422.
5- Shenggui, Z., Starfield, A.M: Dynamic programming with colour graphics smoothing for open-pit design
on a personal computer; International Journal of Mining Engineering, 1985, pp. 27-34.
6- Lerchs, H., Grossmann, I.F: Optimum Design of Open Pit Mines; Canadian Institute of Mining Bulletin,
58, 1965, pp. 17-24.
7- Johnson, T.B., Sharp, R.W: Three Dimensional Dynamic Progromming Method for Optimal Ultimate pit
Design; US bureau of Mines, Report of Investigation, No.7553, 1971.
8- Koenigsberg, E: The optimum contours of an open pit mine: an application of dynamic programming;
proceeding of 17th International APCOM Symposium, 1982.
9- Wilke, F.L., Mueller, K., Wright, E. A: Ultimate and Production Scheduling Optimization; proceeding of
18th International APCOM Symposium, 1984, pp. 29-39.
10- Wright, E. A: The use of Dynamic Programming for Open Pit Mine Design: some practical implications;
Mining Science and Technology, 4, 1987, pp. 97-104.
11- Zhao, Y., Kim, Y.C: A new optimum pit limit design algorithm; SME Preprint no. 90-9, SME Annual
Meeting, Salt lake City, Utah, USA, 1990.
12- Zhao, Y., Kim, Y.C: A new graph theory algorithm for optimal ultimate design; proceeding of 23th
International APCOM Symposium, 1992, pp. 423-434.
13- Matheron, G: Parametrage des contours optimaux; Note Geostatistique no.128, Center de Geostatistique
et de Morphologie Mathematique, Internal Report N-401., Fontainebleau, France, 1975, pp. 54
14- Coleou, T: Technical parameterization of reserves for open pit design and mine planning., proceeding of
21th International APCOM Symposium, 1989, pp. 485-494.
15- Francois-Bongarcon, D., Guibal, D: Algorithms for parameterizing reserves under different geometrical
constraints., Proceeding of 17th International APCOM Symposium, 1982, pp. 297-309.
16- Francois-Bongarcon, D., Marechal, A: A new method for open pit design- parameterization of the final
pit contour., Proceeding of 14th International APCOM Symposium, 1976, pp. 573-583.
17- Degdelen, K., Francois-Bongarcon, D: Towards the complete double parameterization of recovered
reserves in open pit mining., Proceeding of 17th International APCOM Symposium, 1982, pp. 288-296.
18- Whittle, J: Beyond Optimization in Open Pit Design; Computer Applications in the Mineral Industry,
Balkema, Rotterdam, 1988, pp. 331-337.
19- Jalali, S.E., Ataeepour. M., Shahriar. K: Pit Limits optimization using Stochastic Process; Canadian
Institute of Mining Bulletin, 58, 2006.
20- Hochbaum, D. S., Chen, A: Performance analysis and best implementations of old and new algorithms
for the open-pit mining problem; Journal of Operation Research, Vol. 48, No. 6, 2000, pp. 894-914.
21- Pana, M. T: the simulation approach to open pit design., Proceeding of 5th International APCOM
Symposium, 1965.
22- Lemieux, M: Moving cone optimizing algorithm., In computer methods for the 80s in the mineral
industry, New york., 1979, pp. 329-345
23- Robinson, R.H., Prenn, N. B: An optimum pit design model., Proceeding of 10th International APCOM
Symposium, 1973, pp. 155-163.
24- Johnson, T.B: Optimum Open Pit Mine production scheduling., Operations Research Center, University
of California, Berkeley., 1968, pp. 120.
25- Picard, J.C: Maximal Closure of a Graph and Applications to Combinatorial problems., Management
Science, vol. 22, No. 11., 1976, pp. 1268-1272.

7
VII-th International Scientific Conference SGEM2007, 2007-04-27, Sofia, Bulgaria

26- Giannini, L. m., Caccetta, L., Kelsey, P and Carras, S: A New High Speed Network Flow Technique for
Optimum Pit design Facilitating Rapid sensitivity Analysis., Australasian Institute of Mining and Metallurgy,
Melbourne, 1991, pp. 57-62.
27- Yegulalp, T. M., Arias, A. J: A fast algorithm to solve the ultimate pit limit problem., Proceeding of 23th
International APCOM Symposium, 1992, pp. 391-397.
28- Yegulalp, T. M., Arias, A. J., Muduli, P: New development in ultimate pit limit problem solution
methods. SME Trans, 1993, pp. 1853-1857.
29- Ahuja, R. K., Orlin, J. B: A fast simple algorithm for the maximum flow problem., Operation Research
(37-5), 1989, pp. 748-759.
30- Meyer, M: Applying linear programming to the design of ultimate pit limits., Management Science, 16,
No.2, 1969, pp. B121-135.
31- Huttagosol, P., Cameron, R. E: A computer design of ultimate pit limit by using transportation
algorithm., Proceeding of 23th International APCOM Symposium, 1992, pp. 443-460.
32- David, M., Dowd, P. A., Korobov, S: Forecasting departure from planning in open pit design and grade
control., proceeding of 12th International APCOM Symposium, 1974, pp. F131-142.
33- Korobov, S: Method for determining optimal open pit limits., Technical report (EP74-R-4), Ecole
Polytechnique de Montreal, 1974.
34- Dowd, P. A., Onur, A. H: Optimizing open pit design and sequencing., Proceeding of 23th International
APCOM Symposium, 1992, pp. 411-422.
35- Denby, B., Schofield, D: Open pit design and scheduling by use of Genetic Algorithms., The Institution
of Mining and Metallurgy, London, vol. 103, 1994, pp. A21-26.
36- Achireko, P., Frimpong, S: Open pit optimization using Artificial Neural Networks on conditionally
simulated blocks., Proceeding of 26th Int. APCOM Symposium, 1996, pp. 285-290.
37- Riddle, J. M:A dynamic programming solution of a block-caving mine layout., Proceeding of 14th Int.
APCOM Symposium, SME, Colorado., 1977, pp. 767-780.
38- Ovanic, J., Young, D. S: Economic optimization of stope boundary using separable programming with
special Branch and Bound techniques., proceeding of 3rd Canadian Conference on Computer Applications in
the Mineral Industry., 1995, pp. 129-135.
39- Ovanic, J., Young, D. S: Economic optimization of open stope geometry., Proceeding of 28th Int.
APCOM Symposium, 1999, pp. 285-290.
40- Jalali, S. M. A., Ataee-pour, M: A 2D Dynamic Programming Algorithm to Optimize Stope Boundary.,
Proceeding of Mine Planning and Equipment Selection (MPES)., 2004, pp. 45-52.
41- Alford, C: Optimization in underground mine design., Proceeding of 25th Int. APCOM Symposium,
1995.
42- Ataee-pour, M: A new heuristic algorithm to optimize stope boundaries., Proceeding of the 2th Regional
APCOM Symposium, 1997.