Resources Policy 38 (2013) 591–597

Contents lists available at ScienceDirect

Resources Policy
journal homepage: www.elsevier.com/locate/resourpol

A heuristic approach to stochastic cutoff grade optimization for open

pit mining complexes with multiple processing streams
Mohammad Waqar Ali Asad a,1,n, Roussos Dimitrakopoulos b
a
Department of Mining Engineering, Western Australian School of Mines, Faculty of Science and Engineering, Curtin University, Australia
b
COSMO Stochastic Mine Planning Laboratory, Department of Mining and Materials Engineering, McGill University, Canada

art ic l e i nf o a b s t r a c t

Article history: Cutoff grade specifies the available supply of metallic ore from an open pit mine to the multiple
Received 6 December 2012 processing streams of an open pit mining complex. An optimal cutoff grade strategy maximizes the net
Received in revised form present value (NPV) of an open pit mining operation subject to the mining, processing, and marketing/
11 September 2013
refining capacity constraints. Even though, the quantities of material flowing from the mine to the
Accepted 19 September 2013
market are influenced by the expected variation in the available metal content or inherent uncertainty in
Available online 18 October 2013
the supply of ore, the majority of cutoff grade optimization models not only disregard this aspect and
JEL classification: may lead to unrealistic cash flows, but also they are limited in application to an open pit mining
Q operation with single processing facility. The model proposed herein determines the optimal cutoff grade
Keywords: policy based on a stochastic framework that accounts for uncertainty in supply of ore to the multiple ore
Stochastic optimization processing streams. An application on a large-scale open pit mining operation develops a unique cutoff
Cutoff grade grade policy along with a portfolio of mining, processing, and marketing/refining rates. Owing to the
Heuristic algorithms geological uncertainty, the approach addresses risk by showing a difference of 14% between the
Open pit mine planning minimum and maximum production rates, cash flows and NPV.
& 2013 Elsevier Ltd. All rights reserved.

Introduction
In an open pit mining complex, a typical open pit produces
material of various categories, which is transported to the appro-
priate processing streams for recovering metal as a valuable
product. Knowing the ultimate pit limit, i.e., the size or extent of
an open pit mine, the supply of material in terms of grade–tonnage
curves is established (Hustrulid and Kuchta, 2006), that is the
available quality (grade categories) and quantity (tonnes within
these categories) of material within the pit limit. The grade–
tonnage curves along with the economic parameters such as price
of metal, operating costs, metallurgical recovery, and discount rate
become the basic input to determine the optimal cutoff grade
policy.
Cutoff grade policy defines the amount of ore and waste in a
given period during the life of an operation. While waste is hauled
to the waste dumps, ore being valuable material is sent to the
processing streams for crushing, grinding, and upgrading to
produce concentrate, which, in turn, may be upgraded further by
refining to produce a final marketable product (Lane, 1964; Taylor,
n
Corresponding author. Tel.: þ 61 8 9088 6139.
E-mail addresses: waqar.asad@curtin.edu.au, mwaasad@gmail.com
(M.W.A. Asad), roussos.dimitrakopoulos@mcgill.ca (R. Dimitrakopoulos).
1
Previously (12/2009 to 06/2012) Research Associate at COSMO Stochastic
Mine Planning Laboratory, McGill University, Canada.
1972). Cutoff grade policy maximizes NPV of an open pit mining
complex subject to the mining, processing, and marketing/refining
capacity constraints (Rendu, 2008; King, 2011). Given that multiple
ore processing options exist, cutoff grade for a particular proces-
sing option is defined from the known economic parameters and
grade–tonnage curves, and depending upon the quality of ores
described in grade–tonnage curves, it is sent to the most economic
processing stream (Dagdelen and Kawahata, 2008).
While, mine is a source of supplying ore to a number of processes
that convert raw ore to a profitable product, expected variation in
metal content throughout the extent of the orebody defines the
inherent uncertainty in supply of ore. Taking the limited exploratory
drill-hole information as an input, the geostatistical technique of
conditional stochastic simulation (Goovaerts, 1997; Boucher and
Dimitrakopoulos, 2012; Machuca-Mory and Deutsch, 2013) is used
to characterize this geological uncertainty, resulting in a set of
equally probable simulated realizations of the orebody, showing
the variation of the metal content and corresponding tonnages from
one realization to the next. Thus, as opposed to a traditional
procedure that calculates or derives a single grade–tonnage curve
from a single (or constant and known) estimated orebody realiza-
tion, the stochastic procedures derive multiple or equally probable
grade–tonnage curves from a set of conditionally simulated or
equally probable realizations of the orebody. As a result, each
simulated realization of the orebody is translated into grade cate-
gories and corresponding tonnages within these grade categories to

0301-4207/$ - see front matter & 2013 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.resourpol.2013.09.008
592 M.W.A. Asad, R. Dimitrakopoulos / Resources Policy 38 (2013) 591–597
create a grade–tonnage curve. Consequently, being derived from
simulated or equally probable realizations of the orebody, each
grade–tonnage curve in the set of simulated grade–tonnage curves
becomes equally probable, i.e., at the time of actual mining, the
probability of supplying material from (or corresponding to) a
particular grade–tonnage curve remains the same. Given this
fundamental difference between the traditional and stochastic
procedures, it is well-established that the geological uncertainty
impacts the ore and metal production targets (Elkington and
Durham, 2011). For instance, Baker and Giacomo (1998) validate
that owing to the discrepancies came from the poor description of
the orebody, out of 48 mining projects in Australasia, 13 revealed
20% more than projected reserves, and 9 realized 20% less than the
originally forecasted reserves. Similarly, a World Bank survey in
Canada and USA shows that 73% of mining projects were closed
prematurely due to problems in their ore reserve estimates, and led
to severe losses in capital investment (Vallee, 2000). Consequently,
recognizing the importance of incorporating geological uncertainty,
it is imperative to adopt a stochastic framework for making a
strategic decision on cutoff grade policy. Owing to the variation in
metal content (grade) and corresponding tonnages from one grade–
tonnage curve to the next, the stochastic framework honors the fact
that given a defined cutoff grade, there is a possibility that the same
material may be identified as ore in one grade–tonnage curve and
waste in another grade–tonnage curve (Godoy and Dimitrakopoulos,
2011). Thus, as opposed to the traditional approaches that ignore the
geological uncertainty by considering a single (or constant and
known) grade–tonnage curve, a stochastic framework jointly con-
siders multiple simulated or equally probable grade–tonnage curves
and addresses the risk of not having enough ore to feed the
processing streams under geological uncertainty (Satybaldiev and
Freidin, 2006; Dimitrakopoulos and Abdel Sabour, 2007; Elkington
and Durham, 2011).
Lane proposed a heuristic approach to cutoff grade optimization
(Lane, 1964, 1988) that not only considers a constant grade–tonnage
curve, but also limited in application to open pit mining operations
with mine feeding ore to a single processing plant. Lane's theory
has been modified in a number of traditional cutoff grade optimiza-
tion models (Dagdelen, 1992, 1993; King, 2001; Asad, 2002; Cetin
and Dowd, 2002; Ataei and Osanloo, 2004; Asad, 2005, 2007;
Bascetin and Nieto, 2007; Osanloo et al., 2008; He et al., 2009;
King, 2011). Dagdelen (1992, 1993) and Asad (2002) propose the
steps of the algorithm that implements Lane's approach. King
(2001) includes variations in ore type throughput in cutoff grade
optimization models for multi-element mineralization. Cetin and
Dowd (2002) suggest a genetic algorithm for multi-mineral cutoff
grade optimization. Ataei and Osanloo (2004) recommend a com-
bination of genetic algorithm and grid search technique to deter-
mine cutoff grades for multi-metal deposits. Asad (2005)
incorporates stockpiles into the cutoff grade optimization model
for multi-mineral deposits. Asad (2007) looks into the impact of
metal price and cost escalation on cutoff grade optimization.
Bascetin and Nieto (2007) suggest an NPV maximization model
through an optimization factor to determine cutoff grades. Osanloo
et al. (2008) incorporate environmental issues into cutoff grade
optimization model. He et al. (2009) share a genetic algorithm and
neural network based strategy to simulate a complex mining
system for calculating the optimal cutoff grade. King (2011) explains
the intricacies of Lane's approach giving details of various policies to
consider operating and administrative costs. Dagdelen and
Kawahata (2008) present a linear programming based cutoff grade
optimization model that considers constant grade–tonnage curve
with multiple processing options. Rendu (2008) covers the proce-
dural aspects of considering multiple processing options in a cutoff
grade optimization model. While constant grade–tonnage curve as
an input is the common feature of these studies, majority of them
do not attempt to look into the combined impact of considering a
set of simulated or equally probable grade–tonnage curves derived
from simulated or equally probable orebody realizations and multi-
ple processing options.
Unlike previous studies, realizing the importance of acknowl-
edging geological risk to open pit mine planning, we propose an
extension in Lane's theory herein, such that, it is applicable in a
stochastic framework that not only takes into account multiple
simulated or equally probable grade–tonnage curves generated
from a set of simulated or equally probable orebody realizations
(Boucher and Dimitrakopoulos, 2012; Horta and Amilcar, 2010),
but also considers supply of ore to multiple ore processing
destinations in an open pit mining complex. The proposed cutoff
grade optimization model (i) maximizes NPV of future cash flows;
(ii) satisfies the mining, processing, and marketing/refining capa-
cities constraints; (iii) develops a unique and optimal cutoff grade
policy for the life of operation by simultaneously utilizing all
available equally probable realizations of the grade–tonnage
curves; and (iv) addresses risk by developing a portfolio of
possible mining, processing, and marketing/refining rates that
correspond to the optimal cutoff grade policy.
In the following sections, we discuss the stochastic cutoff grade
optimization model, describe the heuristic procedure for calculat-
ing optimum cutoff grade, present the steps to implement the
heuristic approach, and demonstrate the benefits of the proposed
model in a case study followed by conclusions.
Stochastic cutoff grade optimization model
The stochastic optimization model is limited in application to
an open pit mining complex that consists of a single material
source (mine), multiple material destinations (processing streams
and waste dump), and a market/refinery receiving concentrate
from these processing streams. Also, it is assumed that an optimal
ultimate pit limit has been established and the available reserves
(in terms of multiple simulated or equally probable grade–tonnage
curves) within the pit limit are known (Asad and Dimitrakopoulos,
2013). However, the model jointly accounts for a set of simulated
or equally probable grade–tonnage curves and multiple processes
for defining the cutoff grade policy. The uniqueness of the model
lies in simultaneous utilization of a set of simulated or equally
probable grade–tonnage curves for developing an exclusive cutoff
grade policy. The following parameters (Lane, 1964; Hustrulid and
Kuchta, 2006; Asad and Dimitrakopoulos, 2012) facilitate descrip-
tion of the model:
t period (year) indicator;
T life of operation (years);
p process indicator;
ω grade–tonnage curve indicator;
n grade categories indicator;
N number of grade categories in a particular grade–
tonnage curve;
φωt cash flow for grade–tonnage curve ω during period t
($/year);
S selling price of metal ($/tonne of metal);
r marketing/refining cost ($/tonne of metal);
m mining cost ($/tonne of material);
cp processing cost of process p ($/tonne of ore);
f administrative or fixed cost ($/year);
M mining capacity (tonnes/year);
Cp processing capacity of process p (tonnes/year);
R marketing/refining capacity (tonnes/year);
Qmωt quantity of material mined for grade–tonnage curve ω
during period t (tonnes of material);
 M.W.A. Asad, R. Dimitrakopoulos / Resources Policy 38 (2013) 591–597 593

Qcωpt quantity of ore processed for grade–tonnage curve ω in then (Lane, 1964; Asad, 2007)
process p during period t (tonnes of ore); ′
ððS  rÞQr ω  ∑p cp Qcωp  mQmω  ft Þ þ V ′ω
Qr ωt quantity of marketable product or metal in concentrate Vω ¼ t′
ð5Þ
to be marketed/refined for grade–tonnage curve ω dur- ð1 þ dÞ
ing period t (tonnes of marketable product); Hence, the difference between V ω and V ′ω , i.e., vω , is the
yp metallurgical recovery of process p (%); increase in the present value achieved through mining next Qmω
d discount rate (%). of material. An expansion of the denominator in Eq. (5) gives
vω ¼ ððS  rÞQr ω  ∑cp Qcωp  mQmω ðf þ dV ω Þt′Þ ð6Þ
Given a set of equally probable grade–tonnage curves, the p
model develops the cutoff grade policy for life of operation, i.e.,
If g ω represents the average grade of ore, then
it determines the cutoff grade γ p for process p from periods 1 to T,
Qr ω ¼ ∑p Qcωp g ω yp updates Eq. (6) as
such that, the present value of future cash flows is maximized
subject to the mining, processing, and marketing/refining capacity vω ¼ ððS  rÞ∑Qcωp g ω yp  ∑cp Qcωp  mQmω ðf þ dV ω Þt′Þ ð7Þ
p p
constraints. The mathematical representation of the model is as
follows: Eq. (7) presents the basic value function that is in fact max-
imized by satisfying the limiting operational capacities and solving
T φωt
max ∑ t ð1Þ for the optimal value of cutoff grade. Therefore, it dictates the
t ¼ 1 ð1 þ dÞ derivation of mine, processes, and market/refinery limiting cutoff
subject to grades. The term ðf þ dV ω Þ is the opportunity cost of not mining
the remainder of the deposit due to limiting capacities of mining,
Qmωt r M; 8t ð2Þ processing, and marketing/refining stages. Therefore, depending
upon the limiting stage, the time t′ becomes Qmω =M, Qcωp =C p , or
Qcωpt rC p ; 8t ð3Þ Qr ω =R, if mine, processes, or market/refinery are limiting the
operation, respectively. Consequently, the opportunity cost
Qr ωt rR; 8t ð4Þ ðf þdV ω Þ may be distributed per tonne of material, per tonne of
ore, or per tonne of saleable product, if mine, processes, or market/
here φωt ¼ ðS rÞQr ωt  ∑p cp Qcωpt  mQmωt f is the cash flow
refinery are limiting the operation, respectively. This leads to the
realized from mining Qmωt quantity of material, processing Qcωpt
formulation for mine, processes, and refinery limiting cutoff
quantity of ore in process p, and marketing/refining Qr ωt quantity
grades, as follows (Lane, 1964; Asad, 2007):
of marketable product/concentrate during period t.
cp
The formulation (1)–(4) decides on the optimal cutoff grade γ p γ mp ¼ ð8Þ
such that the function in (1) is maximized relying on the fact that ðS  rÞyp
the constraints (2)–(4) may be limiting the mining operation
f þdV ω
either individually or in pairs. If constraints (2)–(4) limit the cp þ
Cp
operation individually, then the optimal cutoff grade corresponds γ cp ¼ ð9Þ
to mine limiting γ mp , process limiting γ pp , or market/refinery ðS  rÞyp
limiting γ r p cutoff grades. However, if constraints (2)–(4) limit the cp
operation in pair, then maximum available capacity of these γrp ¼
 ð10Þ
f þ dV ω
constraints may be utilized by balancing the output from mine, Sr  yp
R
process, or market/refinery stages, such that the optimal cutoff
grade corresponds to mine-process γ mp, mine-market γ mr, or The balancing cutoff grades are meant for harmonizing the
process-market γ pr balancing cutoff grades. As such, depending available material in ground with the production capacities. Hence,
upon the constraints (2)–(4), the optimal cutoff grade for a they are deduced from the grade–tonnage curves. Assume that
particular process is selected among the limiting and balancing a grade–tonnage curve ω consists of N grade categories
cutoff grades. ½½g ω ð1Þ; g ω ð2Þ; ½g ω ð2Þ; g ω ð3Þ; …; ½g ω ðN  1Þ; g ω ðNÞ and qωn quantity
(tonnes) of material in each grade category. If lower grade in
½g ω ðnÞ; g ω ðn þ 1Þ (let us identify this grade category as nn) is chosen
as a cutoff grade, i.e., γ ′ ¼ g ω ðnÞ, then the quantity of ore per unit of
Calculating optimum cutoff grades—a heuristic approach
material mined, recoverable metal in concentrate per unit of
material mined, and the recoverable metal in concentrate per unit
Formulation (1)–(4) is solved using a heuristic procedure that
of ore processed, represented as mpω ðγ ′Þ, mr ω ðγ ′Þ, and prω ðγ ′Þ,
maximizes the NPV of operation by keeping the quantities of
respectively, are calculated as follows (Lane, 1964; Asad and Topal,
material mined, processed, and marketed/refined within the
2011):
available limiting capacities in a given period t. The heuristic
framework allows the selection of the optimal cutoff grade γ p for a qoω ðγ ′Þ
mpω ðγ ′Þ ¼ ð11Þ
process p, such that it corresponds not only to the NPV of the qoω ðγ ′Þ þ qwω ðγ ′Þ
operation, but also to the operational capacities, leading to
qoω ðγ ′Þg ω ðγ ′Þy
dynamic values changing from one period to the next with the mr ω ðγ ′Þ ¼ ð12Þ
exhaustion of reserves. qoω ðγ ′Þ þ qwω ðγ ′Þ
Given a grade–tonnage curve ω, if total quantity of material in
pr ω ðγ ′Þ ¼ g ω ðγ ′Þy ð13Þ
ground is Q, and due to limited available operational capacities, if
it takes time t′ to mine next Qmω of material, then the present here qoω ðγ ′Þ, qwω ðγ ′Þ, g ω ðγ ′Þ, and y represent the quantity of ore,
value V ω of future cash flows at time zero (now) is equal to the quantity of waste, average grade, and average metal recovery
present value of cash flows generated by mining next Qmω of corresponding to a particular value of γ ′, respectively. Mine and
material at the end of time t′ and the present value of remainder processing stages, mine and market/refining stages, and processing
ðQ  Qmω Þ of the deposit V ′ω at time t′. If cash flow from mining and market/refining stages will be balanced if the quantity of ore per
′
next Qmω is represented as ððS  rÞQr ω  ∑p cp Qcωp  mQmω ft Þ, unit of material mined, quantity of metal in concentrate per unit of
594 M.W.A. Asad, R. Dimitrakopoulos / Resources Policy 38 (2013) 591–597

material mined, and the quantity of metal in concentrate per unit of 6. Compute:
ore processed are equal to the ratios ∑p C p =M, R=M, and R=∑p C p (a) Limiting economic cutoff grades γ mp , γ pp , γ r p using Eqs.
respectively. Applying linear approximation on the curves obtained (8)–(10), respectively.
from Eqs. (11) to (13), we get the formulation for balancing cutoff (b) γ p using Eq. (17).
grades as follows (Lane, 1964; Asad and Topal, 2011): (c) qoω ðγ p Þ, qwω ðmin γ p Þ, g ω ðγ p Þ using Eqs. (18)–(20),
2 3 respectively.
∑p C p
6  mp ω ð γ ′Þ 7 (d) Qmω , Qcωp , Qr ω as follows:
γ mp ¼ min6 M þ g ω ðnÞ7 ð14Þ Qmω ¼ M, Qcωp ¼ fQmω ∑p qoω ðγ p Þ=
ω 4mpω ðγ ′ þ 1Þ  mpω ðγ ′Þ 5 Set
g ω ðn þ 1Þ  g ω ðnÞ ð∑p qoω ðγ p Þ þ qwω ðmin γ p ÞÞgfC p =∑p C p g, Qr ω ¼ ∑p Qcωp γ ω yp .
If Qcωp 4 C p or Qr ω 4 R, then
2 3
R Qcωp ¼ C p , Qmω ¼ f∑p Qcωp gf1 þqwω ðmin γ p Þ=∑p qoω ðγ p Þg,
6  mr ω ðγ ′Þ 7 Qr ω ¼ ∑p Qcωp γ ω yp .
γ mr ¼ min6 M þ g ω ðnÞ7 ð15Þ
ω 4mrω ðγ ′þ 1Þ  mrω ðγ ′Þ 5 If Qr ω 4 R or Qmω 4 M, then
g ω ðn þ 1Þ  g ω ðnÞ Qr ω ¼ R, Qcωp ¼ ðQr ω C p =∑p C p Þ=γ ω yp ,
Qmω ¼ f∑p Qcωp gf1 þ qwω ðmin γ p Þ=∑p qoω ðγ p Þg.
2 3
R (e) v′ ¼ ∑t φωt =ð1 þ dÞt .
 pr ω ðγ ′Þ
6 ∑p C p 7 7. Compare V ω with v′ for NPV convergence (within some
γ pr ¼ min6 þ g ω ðnÞ7 ð16Þ
ω 4pr ðγ ′ þ 1Þ  pr ðγ ′Þ
ω ω
5 tolerance). If V ω is converged, then go to step 8, otherwise
g ω ðn þ 1Þ  g ω ðnÞ go to step 5.
8. Given γ p and the corresponding flow of ore and waste to
The optimum cutoff grade corresponding to a particular processing
respective processes and waste dump, adjust the grade–
stream is selected among the limiting economic ðγ mp ; γ pp ; γ r p Þ and
tonnage curve ω.
balancing ðγ mp; γ mr; γ prÞ cutoff grades. Eq. (8) reveals that the
9. Set t ¼ t þ 1, go to step 2.
optimum cutoff grade may never be less than mine limiting cutoff
10. Present the optimal cutoff grade policy from periods 1 to T,
grade, because it corresponds to the minimum (break-even) cutoff
showing quantity mined, cutoff grade for process p, quantity of
grade, and cutoff grade less than this value would result in sending
ore processed in process p, quantity of metal in concentrate to
waste to a particular processing stream. Similarly, Eq. (9) shows that
be marketed/refined, cash flow, and the respective NPV.
the optimum cutoff grade may never exceed the process limiting
cutoff grade, because, it would schedule some of the valuable
materials to the waste dumps. Hence, the optimum cutoff grade for
a processing stream p refers to the possible values between γ mp and
Application at a copper mining complex
γ pp , i.e., γ mp r γ p r γ pp . If m~ corresponds to the median value, then
the optimum cutoff grade is selected as
This section presents an application of the stochastic cutoff
γ p ¼ mð ~ γ mp ; γ pp ; γ mpÞ; mð
~ mð ~ γ mp ; γ r p ; γ mrÞ; mð
~ γ pp ; γ r p ; γ prÞÞ ð17Þ grade optimization model to an actual copper mining complex.
The open pit mining complex constitutes an open pit, four ore
processing facilities, and a waste dump. The open pit produces
sulphide and oxide ores along with waste material to uncover
Implementing the proposed heuristic
these ores. The ore may be processed in two flotation-mills, a bio-
leach pad, or an acid leaching plant, and the waste material is
The heuristic procedure simultaneously accounts for a set of
destined to the waste dump. Mining capacity stands at 182.5
simulated or equally probable grade–tonnage curves and deter-
million tonnes of ore and waste. The processing capacity of
mines a single as well as optimum cutoff grade policy. The
flotation-mills, bio-leaching pad, and acid-leaching plant is 43.8
procedure is computation-intensive and iterative, requiring imple-
million, 21.9 million, and unlimited tonnes of ore, respectively.
mentation of the algorithmic steps in a programming language.
Similarly, it is assumed that the operation has an unrestricted
Given a particular period t, the iterative steps continue finding the
potential to refine or market the annual copper production. Fig. 1
optimum values of γ p until the convergence of NPV subject to the
explains the flow of material from the mine to processing facilities,
quantities of material mined, ore processed, and metal in concen-
waste dump, and refinery.
trate marketed/refined within the operational limits. The steps of
Table 1 presents fifteen (15) simulated or equally probable
the algorithm are given as follows:
grade–tonnage curves derived from a set of fifteen (15) simulated
or equally probable orebody realizations. The grade categories and
1. Set t¼1.
tonnages (quantity of material) corresponding to these categories
2. Compute reserves Q ωt , if Q ωt ¼ 0, then go to step 10.
3. Compute balancing cutoff grades γ mp, γ mr, γ pr based on mpω ,
mr ω , and pr ω curves using Eqs. (11)–(13), respectively, such Bio-Leach Pad
that (21.9 Million)

qoω ðγ ′Þ ¼ ∑ qωn ð18Þ

n Z nn
Acid-Leach Plant
Mine (Unlimited Capacity)
qwω ðγ ′Þ ¼ ∑ qωn ð19Þ (182.5 Million)
non
n Market/Refinery

 Flotation Mill A
g ðnÞ þg ω ðn þ 1Þ
∑n Z nn qωn ω (43.8 Million)
2
g ω ðγ ′Þ ¼ ð20Þ Waste Dump
qoω ðγ ′Þ
Flotation Mill B
(43.8 Million)
4. Set v′ ¼ 0.
5. Set V ω ¼ v′. Fig. 1. Flow of material from mine to refinery in an open pit mining complex.
 M.W.A. Asad, R. Dimitrakopoulos / Resources Policy 38 (2013) 591–597 595

within equally probable grade–tonnage curves represent the

004.266
013.232

000.260

000.025
088.568

000.725
669.831
432.766

000.071
001.574

000.145
037.114
available reserves inside the optimal ultimate pit limit. Utilizing

15
these equally probable grade–tonnage curves, the optimal cutoff
grade policy is developed by maintaining the discount rate at 8%
and copper recoveries from flotation mill A, flotation mill B, bio-

033.295
012.046

001.862
000.784
000.263

000.025
090.949

003.730

000.073
467.816

000.215
637.519 leach pad, and acid leaching plant at 80%, 81%, 35%, and 55%,
14

respectively. The refined metal is expected to generate $4409.00

per tonne of copper at a marketing/refining cost of $550.00 per
tonne of copper, mining cost of $1.50 per tonne of material, and

002.048
416.487

000.280
088.023
685.635

034.961

005.691

000.710

000.144
014.311

000.170
000.119
processing cost of $6.00 per tonne of ore processed in flotation
13

mills A and B, $1.50 per tonne of ore processed in bio-leach pad,

and $4.00 per tonne of ore processed in acid leaching plant,
respectively.
005.844
678.686

033.954
012.864

001.294
000.306
087.475
425.106

002.156

000.163
000.518
000.189
Knowing the input information, the steps of the heuristic
12

approach given in the previous section are implemented to

develop the optimal cutoff grade policy along with a portfolio of
production rates and cash flows. Table 2 demonstrates this unique
000.429
000.094
452.679
096.367
034.568

005.283

000.732

000.073
000.072
001.813
642.351

014.117

cutoff grade policy generated by simultaneously utilizing all

11

available grade–tonnage curves. It also shows corresponding

Quantity (tonnes in millions) of material in simulated orebody realization

production rates and cash flows for an equally probable grade–

tonnage curve in realization 1.
000.824
000.594
659.883
436.952
094.058

004.925
035.873

001.729
013.315

000.143

000.165
000.117

As shown in Table 2, available reserves (1,248,577,642 tonnes)

10

in this equally probable grade–tonnage curve are exhausted in

fifteen (15) years. Based on the proposed heuristic stochastic
model, the optimal cutoff grade policy promises an NPV of
429.875

036.928

005.405

000.833

000.238
000.024
000.096
660.670

001.847
013.213

000.310
099.139

$3588.74 million. During year 1, acid-leaching plant processes

ore between 0.399% and 0.477%, flotation mills process ore
9

between 0.477% and 1.43%, and bio-leach pad processes ore

carrying a metal content of 1.43% or more. This dictates that
013.007
092.205

005.243

001.035
002.397

000.097
034.972

000.631
454.376

000.210
000.215
644.190

material below 0.399% is treated as waste and transported to the

waste dumps. As opposed to the flotation mills and the acid-
8

leaching plant, the bio-leaching process pays the highest oppor-

tunity cost due to the minimum available processing capacity
000.686
658.343

033.706

001.733
005.762
445.341

000.470
087.472

000.213
014.516

000.192
000.144

(21,900,000 tonnes). Owing to this highest opportunity cost

coupled with the least metal recovery (35%), bio-leach pad
7

processes the highest grade ore during fifteen (15) years of

production. Given the uncertainty in supply of material derived
000.468

000.069
448.988

002.678

000.308

000.239
032.279

005.270

001.401
013.051
090.711
653.115

from the equally probable grade–tonnage curves, the proposed

approach develops an optimal cutoff grade policy that remains
6

unique as shown in Table 2. However, it generates a portfolio of

possible cash flows and NPV, owing to the variation in mine,
091.638

001.699

000.094
481.067

005.039

000.679

000.072
619.419

013.418

000.186

000.167
035.101

process, and market/refinery rates. Fig. 2 presents a risk profile in

terms of minimum, maximum, and average cumulative quantity of
5

metal in concentrate to be marketed/refined over life of operation.

036.006
621.223

005.309
002.360
000.888
014.248

000.024
000.353
454.291

000.376
113.312

000.188

Conclusions
4

This paper presents an extension in Lane's heuristic approach

686.864

031.430

000.024
000.073
004.357
422.541

001.951

000.215
000.717
089.105

011.158

000.142

to cutoff grade optimization. The proposed modification enhances

the application of Lane's original model not only to open pit
3
Simulated realizations of the grade–tonnage curves.

mining operations with multiple ore processing facilities, but also

accounts for geological uncertainty. The model utilizes multiple
004.866
091.904

013.464
034.363

002.085
000.783

000.095
000.069
665.319
435.317

000.192

000.121

simulated or equally probable grade–tonnage curves derived from

equally probable simulated realizations of the orebody model.
2

However, irrespective of considering multiple grade–tonnage

curves, the heuristic approach suggests a unique cutoff grade
033.644

004.606
397.895

013.875

000.681
000.468
086.926

000.024
001.762
708.314

000.190
000.192

policy and looks into the risk of maintaining the cash flows to
maximize NPV based on the possible variations in production from
1

mine, processes, and market/refinery during the life of operation.

An application of the approach at a large copper mine that
Copper grade (%)

produces relatively uniform (less variable) and high-grade copper

5.50–4 5.50

ore shows a difference of 13.83% between the minimum and

1.50–2.00

2.50–3.00

3.50–4.00

4.50–5.00
0.00–0.55

2.00–2.50

3.00–3.50

4.00–4.50

5.00–5.50
0.50–1.00
1.00–1.50

maximum NPV generated through a set of fifteen (15) equally

Table 1

probable grade–tonnage curves. A similar trend may be observed

in Fig. 2 showing minimum, average, and maximum quantities of
596 M.W.A. Asad, R. Dimitrakopoulos / Resources Policy 38 (2013) 591–597

Table 2
The optimal cutoff grade policy and corresponding production rates, cash flows, and NPV for a grade–tonnage curve.

Year Qm (MT) Flotation mill A Flotation mill B Bio-leach pad Acid leaching plant Qr (MT) ϕ ($M) NPV ($M)

CoG (%) Qc (MT) CoG (%) Qc (MT) CoG (%) Qc (MT) CoG (%) Qc (MT)

1 157.41 0.48 22.19 0.48 8.46 1.43 11.10 0.40 44.39 0.36 772.55 3588.74
2 153.92 0.45 23.31 0.44 6.89 1.28 11.66 0.37 46.62 0.34 704.80 3041.48
3 151.00 0.42 24.47 0.41 5.46 1.12 12.23 0.35 48.94 0.33 643.42 2523.61
4 140.44 0.38 25.67 0.38 4.18 0.96 12.84 0.32 46.21 0.30 553.09 2030.61
5 112.49 0.34 26.87 0.34 3.11 0.80 13.44 0.30 31.12 0.24 413.11 1595.71
6 093.99 0.31 27.97 0.31 2.33 0.65 13.99 0.27 20.54 0.19 316.17 1277.21
7 081.56 0.28 28.96 0.28 1.73 0.51 14.48 0.25 13.07 0.16 246.67 1037.93
8 075.23 0.27 29.45 0.26 1.34 0.44 14.72 0.24 09.12 0.15 202.90 0854.56
9 070.97 0.26 29.76 0.25 1.02 0.40 14.88 0.23 06.42 0.13 174.20 0703.79
10 067.07 0.25 29.38 0.25 0.75 0.37 14.98 0.23 04.44 0.12 155.87 0571.97
11 049.86 0.25 18.50 0.25 0.50 0.36 15.02 0.23 02.92 0.09 133.45 0449.38
12 038.41 0.25 11.29 0.24 0.33 0.35 15.06 0.23 01.85 0.07 120.82 0341.21
13 030.29 0.24 06.20 0.24 0.19 0.34 15.10 0.23 01.06 0.06 112.73 0238.02
14 021.04 0.22 00.21 0.22 0.08 0.23 15.52 0.21 00.30 0.04 091.38 0135.31
15 004.90 0.20 03.79 0.20 0.01 0.14 00.31 0.19 00.03 0.02 047.45 0047.45

MT ¼ Millions of Tonnes; CoG ¼ Cutoff Grade; $M ¼ Millions of Dollars.

3000 the members of the COSMO Stochastic Mine Planning Laboratory—

Minimum Maximum Average
AngloGold Ashanti, Barrick, BHP Billiton, De Beers, Newmont,
2500 and Vale.
(Thousands of Tonnes)
Copper Production

2000

References
1500

Asad, M.W.A., 2002. Development of generalized cutoff grade optimization algo-

1000
500
0 Asad, M.W.A., 2007. Optimum cutoff grade policy for open pit mining operations
1 3 5 7 9 11
Production Period (Year)
Fig. 2. Minimum, maximum, and average copper production in the optimal cutoff
grade policy.
metal in concentrate to be marketed/refined. Therefore, ignoring
the geological uncertainty in open pit mining operations produ-
cing relatively low-grade and highly variable metallic ores may
lead to serious economic implications. The proposed approach
provides a tool for evaluating alternative policies as part of
feasibility studies at the long-term mine planning stage. Conse-
quently, the resultant cutoff grade policy (as presented in Table 2)
facilitates a risk-quantified decision making process involving
major investments for sustainable utilization of mineral resources.
Apart from considering multiple ore processing streams, the
proposed heuristic is limited to single mine or material source and
market/refinery (i.e., it sums the quantities of metal in concentrate
from individual processing streams). A mining complex with
multiple mines and limited capacity of individual market or
refining processes presents a very complex supply chain system
under geological uncertainty, and a heuristic procedure may not
capture the details of such a complex optimization problem. As
such, future studies require the development and implementation
of Stochastic Integer Programming (SIP) based mathematical
formulations to solve such a complex optimization problem.
Acknowledgments
The work in this paper was funded from Natural Sciences and
Engineering Research Council of Canada CRDPJ Grant 411270-10 and
rithm for open pit mining operations. Journal of Engineering and Applied
Sciences 21 (2), 119–127.
Asad, M.W.A., 2005. Cutoff grade optimization algorithm with stockpiling option for
open pit mining operations of two economic minerals. International Journal of
Surface Mining, Reclamation and Environment 19 (3), 176–187.
13 15 through net present value algorithm considering metal price and cost escala-
tion. Engineering Computations 24 (7), 723–736.
Asad, M.W.A., Dimitrakopoulos, R., 2012. Optimal production scale of open pit
mining operations with uncertain metal supply and long-term stockpiles.
Resources Policy 37 (1), 81–89.
Asad, M.W.A., Dimitrakopoulos, R., 2013. Implementing a parametric maximum
flow algorithm for optimal open pit mine design under uncertain supply and
demand. Journal of the Operational Research Society, 64 (2), 185–197, http://dx.
doi.org/10.1057/jors.2012.26.
Asad, M.W.A., Topal, E., 2011. Net present value maximization model for optimum
cutoff grade policy of open pit mining operations. Journal of the South African
Institute of Mining and Metallurgy 111 (11), 741–750.
Ataei, M., Osanloo, M., 2004. Using a combination of genetic algorithm and grid
search method to determine optimum cutoff grades of multiple metal deposits.
International Journal of Surface Mining, Reclamation and Environment 18 (1),
60–78.
Baker, C.K., Giacomo, S.M., 1998. Resource and reserves: their uses and abuses by
the equity markets. In: Ore Reserves and Finance: A Joint Seminar between
Australasian Institute of Mining and Metallurgy and ASX, Sydney.
Bascetin, A., Nieto, A., 2007. Determination of optimal cut-off grade policy to
optimize NPV using a new approach with optimization factor. Journal of the
South African Institute of Mining and Metallurgy 107 (2), 87–94.
Boucher, A., Dimitrakopoulos, R., 2012. Multivariate block-support simulation of the
Yandi iron ore deposit, Western Australia. Mathematical Geosciences 44 (4),
449–468.
Cetin, E., Dowd, P.A., 2002. The use of genetic algorithms for multiple cutoff grade
optimization. In: Proceedings of the 32nd International Symposium on Appli-
cation of Computers and Operations Research in the Mineral Industry, pp. 769–
779.
Dagdelen, K., 1992. Cutoff grade optimization. In: Proceedings of the 23rd Interna-
tional Symposium on Application of Computers and Operations Research in
Minerals Industry, pp. 157–165.
Dagdelen, K., 1993. An NPV optimization algorithm for open pit mine design. In:
Proceedings of the 24th International Symposium on Application of Computers
and Operations Research in Minerals Industry, pp. 257–263.
Dagdelen, K., Kawahata, K., 2008. Value creation through strategic mine planning
and cutoff grade optimization. Mining Engineering 60 (1), 39–45.
Dimitrakopoulos, R., Abdel Sabour, S.A., 2007. Evaluating mine plans under
uncertainty: can the real options make a difference?. Resources Policy 32 (3),
116–125.
Elkington, T., Durham, R., 2011. Integrated open pit pushback selection and
production capacity optimization. Journal of Mining Science 42 (2), 177–190.
 M.W.A. Asad, R. Dimitrakopoulos / Resources Policy 38 (2013) 591–597
Godoy, M.C., Dimitrakopoulos, R., 2011. A risk quantification framework for
strategic mine planning: method and application. Journal of Mining Science
47 (2), 235–246.
Goovaerts, P., 1997. Geostatistics for Natural Resources Evaluation. Oxford
University Press, New York.
He, Y., Zhu, K., Gao, S., Liu, T., Li, Y., 2009. Theory and method of genetic-neural
optimizing cutoff grade and grade of crude ore. Expert Systems with Applica-
tions 36 (4), 7617–7623.
Horta, A., Amilcar, S., 2010. Direct sequential co-simulation with joint probability
distributions. Mathematical Geosciences 42 (3), 269–292.
Hustrulid, W., Kuchta, M., 2006. Open Pit Mine Planning and Design, second edition
Taylor and Francis/Balkema, The Netherlands.
King, B., 2001. Optimal Mine Scheduling Policies (Ph.D. thesis). Royal School of
Mines, Imperial College, London University, UK.
King, B., 2011. Optimal mining practice in strategic planning. Journal of Mining
Science 47 (2), 247–253.
Lane, K.F., 1964. Choosing the optimum cutoff grade. Colorado School of Mines
Quarterly 59, 811–829.
597
Lane, K.F., 1988. The Economic Definition of Ore, Cutoff Grade in Theory and
Practice. Mining Journal Books, London.
Machuca-Mory, D.F., Deutsch, C.V., 2013. Non-stationary geostatistical modeling
based on distance weighted statistics and distributions. Mathematical Geo
sciences 45 (1), 1–30.
Osanloo, M., Rashidinejad, F., Rezai, B., 2008. Incorporating environmental issues
into optimum cutoff grades modelling at porphyry copper deposits. Resources
Policy 33 (4), 222–229.
Rendu, M., 2008. An Introduction to Cutoff Grade Estimation. Society of Mining,
Metallurgy, and Exploration, Inc. (SME), United States.
Satybaldiev, N.M., Freidin, A.M., 2006. Optimization of mineral mining geometry
and mine production capacity. Journal of Mining Science 42 (3), 245–250.
Taylor, H.K., 1972. General background theory of cutoff grades. Transactions of the
Institution of Mining and Metallurgy, A160–A179.
Vallee, M., 2000. Mineral resource þ engineering, economic and legal feasibility¼ ore
reserve. CIM Bulletin 93.