Documenti di Didattica
Documenti di Professioni
Documenti di Cultura
Resources Policy
journal homepage: www.elsevier.com/locate/resourpol
A review of cut-off grade policy models for open pit mining operations
Mohammad Waqar Ali Asad n, Muhammad Asim Qureshi, Hyongdoo Jang
Department of Mining Engineering and Metallurgical Engineering, WASM, Curtin University, Australia
art ic l e i nf o
a b s t r a c t
Article history:
Received 10 March 2016
Received in revised form
29 April 2016
Accepted 12 May 2016
In an open pit mining operation, the heterogeneity of the grade-tonnage distribution of the deposit
dictates that all available material within the boundaries of an open pit may not be processed. Given this
heterogeneity, it is imperative that the valuable (ore) and waste materials are clearly identied. In this
context, the cut-off grade denes the quantity of ore and waste, ensuring smooth supply of ore to the
processing streams. While, the earliest signicant effort towards the development of models for cut-off
grade policy dates back to 1960s, a number of contributions on this vital aspect of a mining operation
have been made since then. This paper presents a comprehensive overview of the available literature on
cut-off grade policy models and suggests possible areas of future research.
& 2016 Elsevier Ltd. All rights reserved.
Keywords:
Optimization
Modelling
Cut-off grade
Processing
Open pit mining
1. Background
Fig. 1 presents the layout of an ideal open pit mining system
that constitutes three essential components: a mine (open-pit), a
processing plant, and a renery. Mine produces ore and waste
materials, ore is then transported to the suitable processing
streams, and waste is hauled to the waste dumps. Processing plant
then upgrades the metal content in raw ore and produces concentrate, which is then fed to the renery for the production of
metal as the nal marketable product (McKee et al., 1995). As
material ows from one component of the mining system to the
next, overall economy of the operation is the driving force that
guides the decision making process on this material movement.
Cut-off grade is the most important economic criterion that
denes the supply of ore and waste material from the mine to
subsequent processing streams and waste dumps, respectively
(King 1999, 2001; Wooler, 2001). The material with metal content
greater than the cut-off grade is designated as ore because under
existing economic situation, not only this material will pay for the
cost of mining, processing, and rening but also it will generate
some prot. On the contrary, the material with grade less than the
cut-off grade is identied as waste, and if this material is required
to be mined, it will incur the cost of mining that includes excavation and haulage from the mine to the waste dumps.
Given this denition, a cut-off grade policy thus describes a
schedule or sequence of cut-off grades over the life of an
n
Corresponding author.
E-mail addresses: waqar.asad@curtin.edu.au (M.W.A. Asad),
m.m.qureshi@student.curtin.edu.au (M.A. Qureshi),
Hyongdoo.jang@curtin.edu.au (H. Jang).
http://dx.doi.org/10.1016/j.resourpol.2016.05.005
0301-4207/& 2016 Elsevier Ltd. All rights reserved.
Stockpiles
Mine
Waste
Processing Streams
Potential Ore
Ore
Metals or Final
Products
Run of Mine
Leach
Crushed Leach
Concentrates
Refinery
Flotation
Others
-
Waste Dumps
143
Tailings
0.00
0.20
0.00
0.00
0.15
0.13
0.00
0.26
0.14
0.00
0.29
0.35
0.27
0.00
0.25
0.21
0.41
0.25
0.28
0.18
0.00
0.29
0.46
0.43
0.00
2. Breakeven model
The traditional or breakeven model takes the metal selling
price ( s ), rening, market and/or sales cost ( r ), mining cost ( m),
processing cost ( p), metal content or grade ( g ), and metallurgical
recovery ( y ) as inputs, and derives the prot ( P ) per tonne of
material as follows (Henning, 1963; Taylor, 1972, 1984):
25
20
16
12
0
0.00
20
15
10
0
0.50
1.00
1.50
2.00
Grade (% Cu)
2.50
3.00
3.50
0.0-0.5
0.5-1.0
1.0-1.5
1.5-2.0
2.0-2.5
2.5-3.0
3.0-4.0
144
Table 1
Economic and operational parameters of a copper mining operation (Lane, 1964).
Description
Value
$550.00 per tonne
$50.00 per tonne
$0.50 per tonne
$0.60 per tonne
$4,000,000.00 per year
Copper price (s )
Rening, marketing or sales cost (r )
Mining cost (m)
Processing cost (p)
Fixed or period cost (f )
Metallurgical recovery (y )
Discount rate (d )
Mining capacity (M )
Processing capacity (C )
Rening capacity (R )
90%
15%
20,000,000 t per year
10,000,000 t per year
90,000 t per year
=
Table 2
Grade-tonnage distribution of copper mineralization (Lane, 1964).
Quantity q (tonnes)
gl
gu
0.0
0.15
0.20
0.25
0.30
0.35
0.40
0.45
0.50
0.55
0.60
0.65
0.70
0.15
0.20
0.25
0.30
0.35
0.40
0.45
0.50
0.55
0.60
0.65
0.70
1.56
14,400,000
4,600,000
4,400,000
4,300,000
4,200,000
4,100,000
3,900,000
3,800,000
3,700,000
3,600,000
3,400,000
3,300,000
42,300,000
P =( s r ) gy mp
(1)
m +p
( s r ) y
(2)
p
( s r ) y
(3)
3. Lane's model
Unlike breakeven model, the cut-off grade calculation in Lane's
Table 3
Cut-off grade policy derived from the breakeven model.
Year
Qm (tonnes)
Qc (tonnes)
Qr (tonnes)
CF ($)
NPV ($)
1
2
3
4
5
6
7
8
9
0.1333
0.1333
0.1333
0.1333
0.1333
0.1333
0.1333
0.1333
0.1333
0.7538
0.7538
0.7538
0.7538
0.7538
0.7538
0.7538
0.7538
0.7538
11,467,890
11,467,890
11,467,890
11,467,890
11,467,890
11,467,890
11,467,890
11,467,890
08,256,880
10,000,000
10,000,000
10,000,000
10,000,000
10,000,000
10,000,000
10,000,000
10,000,000
07,200,000
67,844
67,844
67,844
67,844
67,844
67,844
67,844
67,844
48,848
18,187,987
18,187,987
18,187,987
18,187,987
18,187,987
18,187,987
18,187,987
18,187,987
13,095,351
85,337,863
79,950,555
73,755,151
66,630,436
58,437,014
49,014,579
38,178,778
25,717,607
11,387,262
Pt
145
Qm
(4)
f +vd
QmpQc
vm=( s r ) Qr m+
(14)
Qmt M ,t
(5)
f +vd
Qc
vc =( s r ) QrmQm p+
(15)
QctC ,t
(6)
QrtR,t
(7)
f +vd
QrmQmpQc
vr = s r +
(16)
NPV =
t
t = 1 ( 1+d )
Subjected to
C; ifq > C
o
Qc =
qo; otherwise
(8)
q
Qm=Qc 1 + w
qo
(9)
Qr =Qc ( g y)
(10)
( 1 + d)
v=
( 1 + d)
Pt +W
( 1+d)t
(11)
(12)
(13)
Fig. 4. Presentation of the cash ows and the present value in Lane's model.
(17)
146
16
14
12
($ millions)
10
0
0.00
0.10
0.20
0.40
0.30
0.50
0.60
0.70
0.80
Grade (% Cu)
previous sections, a careful comparison of both Lane and breakeven models demonstrates the signicance of Lane's model, and
this became a reason for its large-scale acceptance in the mining
industry. Today, not only the Lane's model has been implemented
in the standard strategic mine planning software (Whittle and
Vassiliev, 1998), but also several extensions of the original Lane's
model are available.
In this context, Mol and Gillies (1984) suggest an improvement
into traditional cut-off models (breakeven and Lane), such that it is
relevant to the iron mining operations, where market driven
contracts dene the required grade specications, thus maximization of the marketable reserves is the priority, and material
blending to achieve required grade specications becomes
imminent.
Dagdelen (1992, 1993) present the steps of algorithm for implementation of the Lane's model and a case study that conrms
the benets of this model. Through examples, it demonstrates that
mining companies recognize the value of using higher cut-off
grades during early years of a mining operation, as it generates
higher cash ows, ensuring an early return of the capital investments. Also, Dagdelen (1993) provides more insights into
16
14
12
($ millions)
10
0
0.00
0.10
0.20
0.30
0.40
0.50
0.60
Grade (% Cu)
0.70
0.80
147
Table 4
Cut-off grade policy derived from the Lane model.
Year
Qm (tonnes)
Qc (tonnes)
Qr (tonnes)
CF ($)
NPV ($)
1
2
3
4
5
6
7
0.5036
0.5034
0.4590
0.4106
0.3581
0.3011
0.2396
0.9999
0.9998
0.9705
0.9389
0.9035
0.8655
0.8248
17,847,221
17,847,221
16,831,262
15,829,898
14,828,295
13,848,932
02,971,851
10,000,000
10,000,000
10,000,000
10,000,000
10,000,000
10,000,000
02,303,661
89,997
89,985
87,339
84,471
81,314
77,902
17,101
26,074,739
26,071,221
25,254,067
24,320,538
23,242,717
22,026,447
04,760,751
95,765,702
84,055,818
70,592,970
55,927,848
39,996,487
22,753,244
04,139,784
Table 5
Lane's model based cut-off grade policy with stockpiles.
Year
Material Source
Qm (tonnes)
Qc (tonnes)
Qr (tonnes)
CF ($)
NPV ($)
1
2
3
4
5
6
7
8
9
Mine
Mine
Mine
Mine
Mine
Mine
Mine
Stockpile
Stockpile
0.5036
0.5036
0.4670
0.4198
0.3686
0.3132
0.2535
0.2396
0.2396
0.8100
0.8100
0.7903
0.7651
0.7375
0.7076
0.6756
0.2813
0.2813
17,847,221
17,847,221
17,004,459
16,010,921
15,020,620
14,046,817
02,222,742
10,000,000
05,820,183
10,000,000
10,000,000
10,000,000
10,000,000
10,000,000
10,000,000
01,695,897
10,000,000
05,820,183
89,997
89,997
87,809
85,010
81,944
78,620
12,730
31,252
18,189
26,074,739
26,074,739
25,402,434
24,499,351
23,461,819
22,286,569
03,557,755
03,375,911
01,964,842
97,399,414
85,934,586
72,750,035
58,260,107
42,499,772
25,412,919
06,938,289
04,421,277
01,708,558
balancing cut-off grades, and shares linear interpolation as an alternative procedure to graphical approach in Lane (1964, 1988),
which is more suitable for algorithmic implementations.
Whittle and Vassiliev (1998) conrm that changes in processing costs, recoveries, and capacities impact the cut-off grade
calculation in Lane's model, and consequently, based on a stochastic liberation modelling technique, it provides a recovery
prediction system, which feeds variable predicted values for recoveries (as opposed to the constant and average values for recovery) as an input to the Lane's model. Whittle and Wooller
(1999) establish the relevance between the cut-off grade and the
subsequent milling time, apply the stochastic liberation modelling
technique in Whittle and Vassiliev (1998), and simultaneously
optimise the performance of the comminution circuit and processing cut-off grade in the Lane's model. Wooler (2001) also applies Whittle programming's Opti-Cut (a commercial implementation of the Lane's model) for dening the optimal strategy for a milling operation, and a simultaneous optimization of the
mine cut-off grade and mill throughput.
Nieto and Bascetin (2006) as well as Bascetin and Nieto (2007)
modify the opportunity cost concept in (Eqs. (14)16) with an
optimization factor, and realizing the non-linear nature of the
model implement the Generalized Reduced Gradient (GRG) algorithm to generate solution through their modied model. The
optimization factor is introduced to address the convergence of
NPV over a number of steps in the iterative process, which consequently leads to an improvement in overall NPV of the
operation.
Asad (2007) incorporates commodity price and operating cost
escalation into the basic Lane's model. Thus, the commodity price
and operating costs do not remain constant during the life of
operation, rather depending upon the dened escalation rates,
these economic parameters vary from one year to next. Therefore,
this study offers a relatively realistic cut-off grade policy. An application of the algorithm at a hypothetical copper deposit demonstrates the impact of these changes on overall NPV of the
operation. While a sensitivity analysis in this study delineates the
relative importance of escalation in economic parameters, Asad
and Topal (2011) complement stockpiling option in Asad (2007).
Asad and Topal (2011) present the mathematical formulation for
creation of stockpiles, describe the strategy to reclaim the stockpile material after exhaustion of in-pit reserves, compare the cutoff grade policies with and without stockpiling, and demonstrate
the benets in terms of increase in NPV and life of operation.
Osanloo et al. (2008) modify and improve the basic Lane's
model by incorporating the environmental issues specic to the
porphyry copper deposits. Apart from the traditional framework in
basic Lane's model, this modied model accounts for separate
waste dumps and tailing dams for acid and non-acid generating
wastes. Consequently, the mathematical formulation in (Eqs. (13)
17) includes the operating costs associated to the disposal of these
wastes to the waste dumps and nally to the tailing dams. An
application of this model not only reects an improvement in NPV
as compared to the basic Lane's model, but also it ensures the
environmental sustainability of the open pit mining operations.
Similarly, Narri and Osanloo (2015) suggest a sustainable mining
model through a reduction in the undesirable impacts of environmental issues associated to the open pit mining, and their
improved Lane's model that denes different types of wastes, not
only incorporates the environmental costs, but also considers the
possible revenues from waste rock reclamation.
He et al. (2009) apply a combination of the genetic algorithm
and neural networks nesting method for dynamic optimization of
cut-off grades. This study realizes the complexity and non-linearity of the basic model and offers a solution through evolutionary
approaches. More specically, a neural network establishes the
local link between the revenue factor ((Eqs. (14)16)) and chromosomes ( gl ), and genetic algorithm performs a search for the
optimal cut-off grade globally by satisfying (Eq. (17)). An application of the method at an actual iron mine suggests changes in the
present scheme of cut-off grades, which leads to substantial improvement in NPV.
Gholamnejad (2008, 2009) introduces waste dump rehabilitation cost into the prot function of the basic Lane's model, which
leads to the change in relationships for vm , vc , vr ((Eqs. (14)16)),
with a consequent shifting of the optimum point . A case study of
this modied model demonstrates that inclusion of the waste
dump rehabilitation cost leads to a reduction in the value of ,
coupled with a consequent decrease in the quantity of waste sent
to the waste dumps, which indirectly translates into the benet of
processing low-grade ores.
148
f +vd
QmpQc
vm=( s1r1) Qr1+( s2r2 ) Qr2 m+
(18)
f +vd
Qc
vc =( s1r1) Qr1+( s2r2 ) Qr2mQm p+
(19)
f +vd
Qr +( s r ) Qr mQmpQc
vr1= s1 r1+
1 2 2 2
R1
(20)
f +vd
Qr mQmpQc
vr2=( s1r1) Qr1+ s2 r2+
1
R2
(21)
Here, s1= selling price of metal 1, s2= selling price of metal 2, r1=
rening or marketing cost of metal 1, r2= rening or marketing
cost of metal 2, Qr1= quantity of metal 1 to be rened/marketed,
Qr2= quantity of metal 2 to be rened/marketed, R1= rening or
marketing capacity for metal 1, and R2= rening or marketing
capacity for metal 2.
This indicates that as opposed to a single optimum cut-off
grade ( ) dening ore and waste on the grade-tonnage curve, an
intercept between 1, 2 identies the ore and waste on the gradetonnage distribution in two minerals case. Fig. 7 shows this intercept on a surface representing the grade-tonnage distribution in
this scenario (Lane, 1988). This leads to an update in Eq. (17) as
follows:
(22)
149
150
grade strategies developed under selling price and grade uncertainties. More specically, as a rst step, given a set of equally
probable selling price and orebody realizations, the algorithm
applies Lane's model for the development of all feasible cut-off
grade strategies, where each strategy relates to a unique selling
price and grade-tonnage curve realizations, and then in second
step, ranking system selects the best among feasible strategies.
Also, the exibility to close the mine prematurely during early
years of operation is built into the system.
Thompson and Barr (2014) consider selling price uncertainty,
formulate the cut-off grades optimization problem as a system of
nonlinear partial differential equations, and solve this formulation
using a numerical approach. The model simultaneously generates
the maximum value based cut-off grade strategy and hedging
statistics for a set of equally probable selling price realizations. An
implementation of the model in a real life scenario establishes that
as compared to the cut-off grade values realized through conventional or deterministic (Lane's model and extensions) models,
the projected cut-off grades are far lower under market uncertainty regimes as well as long-term valuation horizons, conrming the results reected in Asad and Dimitrakopoulos (2013)
that marginal/low-grade orebodies would suffer the most in such
conditions.
6. Concluding remarks
While breakeven model provides a basic understanding into
the calculation of cut-off grades, the drawbacks of its application
to realistic, large-scale, and complex open pit mining operations
are obvious, as it is proven that keeping a constant cut-off grade
throughout the life of operation allows exposure to serious economic risks.
Lane's model not only overcomes the deciencies in breakeven
model but also apart from the importance of economic inputs, it
recognises the production based limitations of a mining operation
as well as the relevance of the grade-tonnage distribution. Thus, it
follows a systematic approach to dene an optimal cut-off grade
policy, which becomes a valid reason for its acceptability in both
mining industry and academia. Specially, the extensions in both
basic (single mineral) and multiple-minerals based Lane's models
are a proven evidence of its exibility or adaptation to implementation under varying inputs and solution approaches.
These extensions cover economic, technical, environmental, and
algorithmic (solution methods) variations, offering an expansion
in the value and applicability of the Lane's approach.
Moreover, apart from a few exceptions, a majority of the stochastic models also offers an extension into the basic Lane's model.
These stochastic models consider commodity price and/or grade
uncertainty and generate risk quantied cut-off grade policies, but
they work within the connes of the conceptual framework dened in basic Lane's approach. The stochastic models are realistic
and more relevant, but a very limited number of studies consider
stochastic nature of the inputs to cut-off grade models. Thus, in
future, there exists enormous scope of contribution through the
151
development and implementation of stochastic models for dening the schedule of cut-off grades in open pit mining operations.
Given the nature of Lane's models and their extensions, not
only their algorithmic implementations rely on a pre-dened
mining sequence but also being iterative generate solutions by
trial and error method (Dagdelen and Kawahata, 2008), and subsequently, become prone to offer a heuristic solution to the cut-off
grade problem. Alternatively, mathematical programming based
cut-off grade models derive an exact or true optimum solution;
however, to date, only few studies are available in literature. Thus,
the development and implementation of mathematical programming based cut-off grade models becomes another area of future
research.
Nevertheless, given the scale of open pit mining operations, the
size or number of decision variables within the mathematical
models is excessive, which leads to the computational complexity
of these models. Consequently, generating the solution to the
mathematical programming based cut-off grade models within
reasonable time becomes a challenge. Thus, the development of
new methods or algorithms to solve these models is another
emerging eld of future research.
So far, in the context of inputs to cut-off grade models, while
the nature of economic or operational inputs varies from one
model to the next, the structure of the geological input remains
same, i.e. the grade-tonnage curve of the deposit as the geological
input is the common factor among these models. Nevertheless, as
explained in previous sections, apart from known mining sequence, uniform distribution of grades in a given mineralization is
the basic assumption in conversion of the orebody model (Fig. 2)
into a grade-tonnage curve, and given the heterogeneity of grades,
this assumption is unrealistic and invalid. Therefore, a schedule of
cut-off grades that is derived using the grade-tonnage curve provides a strategic guidance only, because in practice, it lacks the
synchronization with the short-term operational plans. Consequently, during the exploitation of the reserves, in a given period,
it is possible that the available or accessible material for mining
may not satisfy the dened cut-off grade for that period.
This limitation of the available heuristic as well as mathematical
programming based models presents an opportunity on the development and implementation of mathematical models that take
orebody models (Fig. 2) as an input and endeavor to generate the
optimal mining sequence and cut-off grade policy simultaneously.
References
Abdel Sabour, S.A., Dimitrakopoulos, R., 2011. Incorporating geological and market
uncertainties and operational exibility into open pit mine design. J. Min. Sci.
47 (2), 191201.
Abdollahisharif, J., Bakhtavar, E., Anemangely, M., 2012. Optimal cut-off grade determination based on variable capacities in open-pit mining. J. South. Afr. Inst.
Min. Metall. 112 (12), 10651069.
Asad, M.W.A., 2005. Cutoff grade optimization algorithm with stockpiling option for
open pit mining operations of two economic minerals. Int. J. Surf. Min. Reclam.
Environ. 19 (3), 176187.
Asad, M.W.A., 2007. Optimum cut-off grade policy for open pit mining operations
through net present value algorithm considering metal price and cost escalation. Eng. Comput.: Int. J. Comput.-Aided Eng. Softw. 24 (7), 723736.
Asad, M.W.A., Topal, E., 2011. Net present value maximization model for optimum
cut-off grade policy of open pit mining operations. J. South. Afr. Inst. Min.
Metall. 111 (11), 741750.
Asad, M.W.A., Dimitrakopoulos, R., 2013. A heuristic approach to stochastic cut-off
grade optimization for open pit mining complexes with multiple processing
streams. Resour. Policy 38, 591597.
Ataei, M., Osanloo, M., 2003a. Determination of optimum cut-off grades of multiple
metal deposits by using the Golden Section search method. J. South. Afr. Inst.
Min. Metall. 103 (8), 493499.
Ataei, M., Osanloo, M., 2003b. Methods for calculation of optimal cut-off grades in
complex ore deposits. J. Min. Sci. 39 (5), 499507.
Ataei, M., Osanloo, M., 2004. Using a combination of genetic algorithm and the grid
search method to determine optimum cut-off grades of multiple metal
152