International Journal of Mining, Reclamation and

Environment

ISSN: 1748-0930 (Print) 1748-0949 (Online) Journal homepage: https://www.tandfonline.com/loi/nsme20

A mathematical programming model for optimal

cut-off grade policy in open pit mining operations
with multiple processing streams

Asif Khan & Mohammad Waqar Ali Asad

To cite this article: Asif Khan & Mohammad Waqar Ali Asad (2020) A mathematical programming
model for optimal cut-off grade policy in open pit mining operations with multiple processing
streams, International Journal of Mining, Reclamation and Environment, 34:3, 149-158, DOI:
10.1080/17480930.2018.1532865

To link to this article: https://doi.org/10.1080/17480930.2018.1532865

Published online: 05 Dec 2018.

Submit your article to this journal

Article views: 289

View related articles

View Crossmark data

Citing articles: 3 View citing articles

Full Terms & Conditions of access and use can be found at

https://www.tandfonline.com/action/journalInformation?journalCode=nsme20
INTERNATIONAL JOURNAL OF MINING, RECLAMATION AND ENVIRONMENT
2020, VOL. 34, NO. 3, 149–158
https://doi.org/10.1080/17480930.2018.1532865

A mathematical programming model for optimal cut-off grade

policy in open pit mining operations with multiple processing
streams
Asif Khana and Mohammad Waqar Ali Asadb,c
a
Department of Mining Engineering, Karakoram International University, Gilgit-Baltistan, Pakistan; bMining
Engineering and Metallurgical Engineering, Western Australian School of Mines, Kalgoorlie, Australia; cMinerals,
Energy and Chemical Engineering, Curtin University, Kalgoorlie, Australia

ABSTRACT ARTICLE HISTORY

Cut-off grade classifies the available supply of ore (valuable) and waste Received 20 April 2018
material within a mineralised deposit. Given the mining, processing and Accepted 3 October 2018
refining limitations of a mining operation, an optimal cut-off grade policy KEYWORDS
ensures that the flow of ore from the mine to the processing and refining Mining; modelling;
facilities is maintained at the maximum possible throughput. This policy optimisation; mixed integer
defines a schedule of cut-off grades along with corresponding quantities linear programming; cut-off
of mineralised material to be mined, processed and metal refined in each grade; open pit mining
period of the scheduling horizon. The criteria that controls the develop-
ment of cut-off grade policy aligns with the strategic objectives of an
operation in order to maximise the discounted value (net present value
or NPV) over the life of operation. This paper proposes a new mixed
integer linear programming (MILP) based model that maximises NPV
subject to the mining, processing, refining capacity constraints and
develops an optimal cut-off grade policy for an open pit mining opera-
tion with multiple processing streams. An implementation of the pro-
posed method on hypothetical and realistic data promises a relatively
higher NPV as compared to the traditional Lane’s approach practiced in
the mining industry.

1. Introduction
An ideal open pit mining operation comprises three fundamental components: a mine, a proces-
sing facility and a refinery. A mine produces mineralised material that is classified either as ore or
as waste, where ore contains the metal content greater than or equal to a predefined cut-off grade,
and waste with a metal content less than the cut-off grade is destined to the waste dumps [1]. The
processing facility applies physical (crushing and grinding) and chemical (flotation, leaching, etc.)
processes to remove or decrease the unwanted constituents locked into the ore, and thus converts
ore into a high-grade concentrate, which is then processed in the refinery for production of the
final marketable product i.e. refined metal. Cut-off grade is thus the most important economic
criterion that guides the flow of materials from one stage in the mineral or metal supply chain to
the next, which translates into a direct impact on the future cash flows [2–4]. The cut-off grade
policy identifies a schedule of cut-off grades and the associated flow of materials from the mine
(quantity of material to be mined), processing plant (quantity of ore to be processed) and the
refinery (quantity of metal to be refined) along with the relevant cash flows generated over the life
of operation.

CONTACT Mohammad Waqar Ali Asad [email protected]

© 2018 Informa UK Limited, trading as Taylor & Francis Group
150 A. KHAN AND M. W. A. ASAD

The development of cut-off grade policy requires economic, geological and technical or
operational inputs. Metal selling price, refining or marketing cost, mining cost, processing cost,
fixed or period cost and discount rate constitute the economic parameters. The primary geological
input constitutes an orebody model that describes the resource mineralisation in terms of the
qualitative (metal content) and quantitative (tonnes of material) data on a block-by-block basis.
However, this block-by-block information is then converted into a grade-tonnage curve or
distribution with a number of grade bins or increments and the corresponding quantity of
materials within each bin or increment [3]. The technical and operational parameters include
metallurgical recovery as well as mining, processing and refining capacities. This set of inputs is
valid for an ideal mining operation that comprises a mine, a processing plant and a refinery.
However, depending upon the quality and the chemical composition of the ore, if a mining
operation comprises multiple processing streams (such as leaching for low-grade ore and flotation
for relatively high-grade ore), the processing capacity as well as recovery of individual processing
streams would be required as an input [4,5].
Different approaches for determining cut-off grade policy for an open pit mining operation are
available in technical literature. In this context, the break-even cut-off grade model relies exclu-
sively on the economic parameters and incorporates processing cost (c; $ per tonne of ore),
refining or marketing cost (r; $ per tonne or gram or troy ounce of metal), metal price (s; $ per
tonne or gram or troy ounce of metal) and the metallurgical recovery (y; %). Equation (1) presents
the break-even model based on processing cut-off grade (gp ; % or grams per tonne or troy ounces
per tonne) that generates the cut-off grade policy [1]:

c
gc ¼ (1)
ðs  rÞy

The break-even model follows simple steps towards defining the cut-off grade policy [1,3].
However, given that it accounts for the economic parameters only, it has the fundamental flaw
of ignoring the grade-tonnage distribution of the resource mineralisation, the operational capa-
cities and the time value of money, which leads to a cut-off grade schedule that remains constant
over the life of a mining operation [3,6].
Lane [7,8] addressed these issues in the break-even cut-off grade policy and proposed a
pioneering formulation that maximises NPV subject to the mining, processing and refining
capacity constraints. Thus, the formulation relies not only on the economic parameters, but
also accounts for the geological (grade-tonnage distribution) and technical/operational inputs
(mining, processing and refinery capacities). Equation (2) in Lane’s model presents an increase in
the present value of future cash flows generated by mining Qm quantity (tonnes) of material and
processing Qo quantity (tonnes) of ore at an average grade
g (% or grams per tonne or troy ounces
per tonne) over time t. In addition, apart from the mining (m; $ per tonne of material), processing
(c; $ per tonne of ore) and refining (r; $ per tonne or gram or troy ounce of metal) costs, the term
ðf þ dV Þ is the opportunity cost that constitutes f as fixed or period cost ($ per unit time t (period
or year)), d as the discount rate (%) and V as the present value of the mining operation, where dV
is the possible interest that would have been generated by investing V in an alternative oppor-
tunity at a discount or interest rate d [1,3,8].

v ¼ ½ðs  rÞ
g y  cQo  mQm  ðf þ dV Þt (2)

However, given the mining (M; tonnes of material per year), processing (C; tonnes of ore per
year) and refining (R; tonnes or grams or troy ounces of metal per year) capacities, the time t


would be defined as QMm , QCc , or QRc g y if mine, processing plant, or refinery is limiting the operation,
respectively. Consequently, Equation (2) leads to Equations (3)-(5) representing the increase in
present value if mine, processing plant or refinery is limiting the operation and the opportunity
 INTERNATIONAL JOURNAL OF MINING, RECLAMATION AND ENVIRONMENT 151

cost ðf þ dV Þ is distributed per tonne of material mined, per tonne of ore processed or per unit
(tonne or gram or troy ounce) of metal refined, respectively [1,3,8].


f þ dV
vm ¼ ½ðs  rÞ
g y  cQo  m þ Qm (3)
M



f þ dV
vc ¼ ðs  rÞ
g y  c þ Qo  mQm (4)
C


 
f þ dV
vr ¼ s rþ
g y  c Qo  mQm (5)
R
Given Equations (3)-(5), Asad et al. [3] describe the procedure to determine the optimum cut-off
grade that corresponds to the grade that satisfies the function max½minðvm ; vc ; vr Þ. The original
procedure in Lane [7] presented above is applicable to the operations with a single processing
facility, while the later extensions in the Lane approach incorporate more complex operational
requirements [1,2,5,9–21].
Lane’s approach and its extensions cover several realistic aspects of open pit mining operations;
however, the fundamental formulation generates a heuristic solution due to the underlying
assumptions and the steps of the algorithm implementing this model, making it susceptible to
missing the optimum values for cut-off grades [3,4,11]. This creates an opportunity for the
development and implementation of mathematical programming models that offer the optimal
solution to this problem. In this context, Dagdelen and Kawahata [4,18] discuss the intricacies in
the development of optimal cut-off grade policy for open pit mining operations comprising
multiple mines, dumps, stockpiles and processing streams; however, it does not share the
mathematical formulation due to its implementation in a commercial software i.e. OptiMine®.
Ganguli et al. [19] present a cut-off grade formulation that maximises the NPV of a mining
operation subject to mine sequencing, reserve, production capacity and grade blending con-
straints. Moosavi et al. [20] share a model that provides a simultaneous solution to production
sequencing and cut-off grade optimisation problems under geological uncertainty. Yasrebi et al.
[21] propose a non-linear mathematical model for cut-off grade policy optimisation utilising the
conceptual framework and structure of inputs similar to the one described in the original Lane’s
model [7,8]; however, a comparison with Lane’s model [7,8] using hypothetical data given in
Hustrulid et al. [22] reflects no improvement in NPV.
While very few studies on the application of mathematical programming models for cut-off
grade policy are available [18–21], with one exception that relates to application in a commercial
software [18], a majority of the models are applicable on open pit operations with a single
processing facility [19–21]. Therefore, this paper contributes a new mixed integer linear program-
ming (MILP) based approach to cut-off grade policy optimisation for open pit mines with
multiple processing streams. The proposed method aims to maximise the NPV of a mining
operation while taking into account the grade-tonnage distribution of the resource mineralisation,
fixed or period costs, as well as mining, processing and refining capacity constraints. An
implementation of the model on hypothetical and realistic data confirms its flexibility, computa-
tional efficiency, as well as better performance as compared to Lane’s model. With this back-
ground, Section 2 provides the details of the proposed MILP formulation and Section 3 shares its
implementation as well as comparison with Lane’s approach followed by conclusions in Section 4.

2. Proposed MILP formulation

In addition to some of the parameters defined in Section (1), the following indices, parameters
and decision variables provide a basis for the proposed MILP based mathematical model:
152 A. KHAN AND M. W. A. ASAD

B ¼ the total number of grade bins;

T ¼ scheduling horizon or the total number of production periods or years;
where t ¼ 1; 2;    ; T;
D ¼ the total number of material destinations (processing streams and waste dumps);
b ¼ grade bin index, where b ¼ 1; 2;    ; B;
d ¼ material destination index, where d ¼ 1; 2;    ; D, and d ¼ 1 for waste dump, d ¼ 2 for
flotation circuit, d ¼ 3 for leach pad, etc.;
Mt ¼ mining capacity in tonnes of material during year t;
Ctd ¼ processing capacity in tonnes of ore at destination d during year t;
Rt ¼ refining capacity in tonnes or troy ounces or grams of metal during year t;
yd ¼ metallurgical recovery in % at destination d;
gLb ¼ lower grade in the grade bin b;
gUb ¼ upper grade in the grade bin b;
qb ¼ available quantity of material in grade bin (between gLb and gUb ) b;

gb ¼ average grade of material in % or troy ounce or grams of metal in grade bin b;
cd ¼ processing cost in $ per tonne of ore processed at material destination d;
i ¼ discount rate in %;
vbdt ¼ ð1þi
vbd
¼ discounted economic value in $ per tonne of material supplied from grade bin b
Þt n
ðs  rÞ
g b yd  m  cd ; if dÞ1
to destination d in year t; where vbd ¼
m; if d ¼ 1
ft ¼ fixed or period cost in $ per year if mining takes place in year t;
N ¼ A reasonably big number;

qbdt ¼ quantity of material supplied from grade bin b to destination d in year t; where this
linear variable remains within 0 
qbdt  qb
n
1; if material is mined in year t
xt ¼ ; i.e. this binary variable xt 2 f0; 1g
0; otherwise
Given these indices, parameters and decision variables, the mathematical formulation is as
follows:
" #
X
T X
B X
D
ft
Maximise z ¼ vbdt
qbdt  xt (6)
t¼1 b¼1 d¼1 ð1 þ iÞt

Subject to:

X
B X
D

qbdt  Mt ; "t (7)
b¼1 d¼1

X
B

qbdt  Ctd ; "d Þ 1 and "t (8)
b¼1

B X
X D

qbdt
gb yd  Rt ; "t (9)
b¼1 d > 1

X
B X
D

qbdt  Nxt ; "t (10)
b¼1 d¼1
 INTERNATIONAL JOURNAL OF MINING, RECLAMATION AND ENVIRONMENT 153

X
D X
T

qbdt  qb ; "b (11)
d¼1 t¼1

X
D  X
D
qb

qbdt 
qðbþ1Þdt ¼ 0; "t and "b Þ B (12)
d¼1
qðbþ1Þ d¼1

Equation (6) presents the objective function that maximises the discounted value or NPV of the mining
operation. Equations (7)-(9) satisfy mining, processing and refining capacities in a given year, respec-
tively. More specifically, Equation (7) requires that the total quantity of material supplied from all grade
bins to all destinations remains within the available mining capacity in a given year. Similarly, Equation
(8) maintains that the total quantity of material (ore) supplied to an individual processing stream remains
within the processing capacity. In addition, Equation (9) ensures that the total quantity of metal refined in
a given year remains within the available refining capacity. Equation (10) forms a link between the linear
(
qbdt ) and binary (xt ) variables, and provides a basis for incurring fixed cost only if material is supplied
from grade bin(s) to destination(s) in a given year. Equation (11) constitutes the grade bin reserve
constraint, as it requires that the total quantity of material supplied from a grade bin to all material
destinations over the scheduling horizon shall not exceed the total quantity of available reserves in that
grade bin. Equation (12) aligns with the algorithms that implement Lane’s formulation [1–3,10,12] as it
preserves the shape of the grade-tonnage distribution and maintains a balanced or proportionate supply
of material from each grade bin over the scheduling horizon, such that the operation processes a blend of
low and high-grade material throughout the life of operation.

2.1. The structure of the proposed formulation

Given that each grade bin b constitutes a lower grade (gLb ), an upper grade (gUb ) and the quantity
of material (qb ) between gLb and gUb , the grade-tonnage distribution of the deposit may be
presented as ½ðg1 ; g2 Þ; q1 ; ½ðg2 ; g3 Þ; q2 ; . . . . . . . . . . . . . . . . . . ½ðgB1 ; gB Þ; qb¼B . The proposed
formulation takes this grade-tonnage distribution as an input. In this context, the average grade
 

g b ¼ gLb þg
2
Ub
in each grade bin b derives the available and produced quantity of metal, thus
contributes in the objective function (Equation (9)) and the refinery capacity constraint (Equation
(6)), respectively. Similarly, the available quantity of material (qb ) in each bin b becomes input for
Equations (11)-(12), thus helps to schedule ore and waste production within the available reserves
ensuring a proportionate blend of low and high-grade materials at ore processing destinations.
Therefore, the proposed formulation is structured such that in a given period t, it schedules the
supply of ore from a range of grade bins to a processing destination d; logically, the optimal cut-
off grade (Gdt ) for a particular processing destination d in period t then corresponds to the grade
bin b that contributed ore at the minimum gLb . For example, if a processing destination d receives
ore from grade bins b ¼ 3 ððg3 ; g4 Þ; q3 Þ to b ¼ B ððgB ; gBþ1 Þ; qB Þ during period t, then Gdt ¼ g3 .

3. Case studies
This section demonstrates the functionality, robustness and the performance of the proposed
MILP formulation in three different scenarios:
Case A – takes hypothetical data (a copper mining operation) given in Lane [7], Asad et al. [3]
and Hustrulid et al. [22] as an input and considers a single processing stream (flotation).
Case B – takes realistic data (a gold mining operation) given in Dagdelen [1] and Dagdelen and
Kawahata [4] as an input and considers a single processing stream (flotation).
Case C – takes realistic data (a copper mining operation) given in Asad and Dimirakopoulos
[23] as an input and considers multiple processing streams (flotation and crushed leach).
154 A. KHAN AND M. W. A. ASAD

Given that the applications of Lane’s model for mining operations with single processing
streams has been reported in the earlier studies [1,3,4], a comparison of the proposed MILP
and Lane’s approaches in Cases A and B not only helps to validate the proposed method, but also
demonstrates the relevance and value of the proposed method. Once Cases A and B confirm the
significance of the proposed method, Case C demonstrates the application of the proposed MILP
formulation on a realistic copper mining operation with multiple processing streams.
Table 1 presents the applicable economic as well as operational parameters and Figure 1
demonstrates the shape of the grade-tonnage distributions for Cases A, B and C, respectively.
Given these inputs, an implementation of the Lane’s approach through algorithms in Asad et al.
[3] generates the cut-off grade policies for Cases A and B. Tables 2 and 3 present these cut-off
grade policies, respectively. In addition, for the implementation of the proposed model, C++ code
in Microsoft Visual Studio environment generates the relevant MILP formulations for Cases A, B
and C, then the academic version of the commercial optimisation solver ‘CPLEX’ [24] solves these
formulations. A simple post-optimisation analysis of the CPLEX solutions then derives the
optimal cut-off grade policies presented in Tables 4, 5 and 6.

Table 1. Economic and operational parameters for Cases A, B and C.

Case
A B C
Description (Copper) (Gold) (Copper)
Metal price ($ per tonne of Cu or $ per troy ounce of Au) 550.00 600.00 5000.00
Refining cost ($ per tonne of Cu or $ per troy ounce of Au) 50.00 5.00 200.00
Mining cost ($ per tonne of material) 0.50 1.20 1.05
Mill processing cost ($ per tonne of ore) 0.60 19.00 3.00
Crushed leach processing cost ($ per tonne of ore) - - 1.00
Fixed or period cost ($ per year) 4,000,000 8,350,000 4,000,000
Mill recovery (%) 90 90 90
Crushed leach recovery (%) - - 25
Discount rate (%) 15 15 15
Mining capacity (tonnes of material) 20,000,000 Unlimited 30,000,000
Mill processing capacity (tonnes of ore) 10,000,000 1,050,000 15,000,000
Crushed leach processing capacity (tonnes of ore) - - 3,000,000
Refining capacity (tonne of Cu or troy ounces of Au) 90,000 Unlimited 90,000

Figure 1. Grade-tonnage distribution of the mineralisation in Cases A, B and C [1,5,23].

 INTERNATIONAL JOURNAL OF MINING, RECLAMATION AND ENVIRONMENT 155

Table 2. Case A – cut-off grade policy based on lane model.

Processing plant – Flotation
Year Qm (tonne) Waste (tonnes) Gt (Cu %)
g(Cu %) Qc (tonnes) Qr (tonnes) Cash flow($)
1 17,651,413 7,651,413 0.501 1.000 10,000,000 90,000 26,174,065
2 17,651,413 7,651,413 0.501 1.000 10,000,000 90,000 26,174,065
3 16,738,894 6,738,894 0.459 0.973 10,000,000 87,580 25,420,614
4 15,745,903 5,745,903 0.411 0.941 10,000,000 84,707 24,480,352
5 14,746,456 4,746,456 0.357 0.906 10,000,000 81,521 23,387,342
6 13,767,826 3,767,826 0.299 0.867 10,000,000 78,070 22,150,943
7 04,598,093 1,011,113 0.236 0.826 3,586,980 26,668 04,882,984
NPV @ 15% = $96,302,398

Table 3. Case B – cut-off grade policy based on lane model.

Processing plant – Flotation
Year Qm (tonne) Waste (tonnes) Gt (oz. per tonne)
g(oz. per tonne) Qc (tonnes) Qr (oz.) Cash flow($)
1 17,956,369 16,906,369 0.160 0.259 1,050,000 244,899 95,867,547
2 17,181,554 16,131,554 0.151 0.259 1,050,000 240,686 94,290,312
3 16,400,229 15,350,229 0.142 0.250 1,050,000 236,034 92,459,842
4 15,617,394 15,567,394 0.131 0.244 1,050,000 230,906 90,347,951
5 14,835,239 13,785,239 0.119 0.238 1,050,000 225,241 87,916,175
6 14,063,400 13,013,400 0.106 0.232 1,050,000 219,033 85,148,838
7 11,798,909 10,748,909 0.091 0.210 1,050,000 198,432 75,608,137
8 09,494,627 08,444,627 0.077 0.185 1,050,000 175,134 64,511,143
9 08,167,280 07,117,280 0.064 0.159 1,050,000 150,670 51,547,999
NPV @ 15% = $411,798,622

Table 4. Case A – cut-off grade policy based on proposed MILP model.

Processing plant – Flotation
Year Qm (tonne) Waste (tonnes) Gt (Cu %)
g(Cu %) Qc (tonnes) Qr (tonnes) Cash flow($)
1 17,268,900 7,268,860 0.484 0.989 10,000,000 89,009 25,870,050
2 16,295,100 6,295,100 0.438 0.959 10,000,000 86,329 25,016,950
3 15,330,100 5,330,100 0.389 0.927 10,000,000 83,419 24,044,550
4 14,367,800 4,367,800 0.335 0.891 10,000,000 80,228 22,930,050
5 13,435,300 3,435,300 0.278 0.853 10,000,000 76,811 21,687,950
6 12,538,100 2,538,100 0.216 0.813 10,000,000 73,166 20,313,800
7 11,664,700 1,664,700 0.150 0.769 10,000,000 69,215 18,775,250
NPV @ 15% = $96,955,389

Table 5. Case B – cut-off grade policy based on proposed MILP model.

Processing plant – Flotation
Year Qm (tonne) Waste (tonnes) Gt (oz. per tonne)
g(oz. per tonne) Qc (tonnes) Qr (oz.) Cash flow($)
1 17,519,600 16,469,600 0.155 0.257 1,050,000 242,541 94,988,375
2 16,745,700 15,695,700 0.145 0.252 1,050,000 238,114 93,282,990
3 15,970,200 14,920,200 0.136 0.247 1,050,000 233,248 91,318,320
4 15,199,400 14,149,400 0.124 0.241 1,050,000 227,919 89,072,525
5 14,437,700 13,387,700 0.112 0.235 1,050,000 222,095 86,521,285
6 13,405,100 12,355,100 0.099 0.226 1,050,000 213,208 82,472,640
7 10,800,300 09,750,270 0.085 0.200 1,050,000 188,669 70,997,695
8 08,788,260 07,738,260 0.073 0.177 1,050,000 167,350 60,727,338
9 07,037,760 05,987,760 0.062 0.155 1,050,000 146,645 50,508,463
10 05,608,950 04,558,950 0.051 0.135 1,050,000 127,632 40,910,300
NPV @ 15% = $413,788,876

A comparison of the cut-off grade policies presented in Table 2–3 (Lane’s model) and 4–5
(proposed MILP model) confirms that the proposed MILP outperformed the traditional
Lane’s model in Cases A and B. More specifically, as indicated in Tables 2 and 4, the
156 A. KHAN AND M. W. A. ASAD

Table 6. Case C – cut-off grade policy based on proposed MILP model.

Processing – Flotation Processing – Crushed leach
Gt
g
Year Qm (tonne) Waste (tonnes) (Cu %) (Cu %) Qc (tonnes) Gt (Cu %)
g(Cu %) Qc (tonnes) Qr (tonnes) Cash flow ($)
1 30,000,000 12,000,000 0.208 0.414 15,000,000 0.166 0.187 3,000,000 57,356 191,809,136
2 30,000,000 12,000,000 0.208 0.414 15,000,000 0.166 0.187 3,000,000 57,356 191,809,136
3 30,000,000 12,000,000 0.208 0.414 15,000,000 0.166 0.187 3,000,000 57,356 191,809,136
4 30,000,000 12,000,000 0.208 0.414 15,000,000 0.166 0.187 3,000,000 57,356 191,809,136
5 30,000,000 12,000,000 0.208 0.414 15,000,000 0.166 0.187 3,000,000 57,356 191,809,136
6 30,000,000 12,000,000 0.208 0.414 15,000,000 0.166 0.187 3,000,000 57,356 191,809,136
7 28,608,800 10,608,800 0.198 0.405 15,000,000 0.154 0.176 3,000,000 55,946 186,501,944
8 25,961,300 07,961,330 0.175 0.384 15,000,000 0.127 0.151 3,000,000 53,036 175,315,067
9 23,680,800 05,680,790 0.152 0.365 15,000,000 0.100 0.126 3,000,000 50,228 164,231,211
10 21,241,700 04,261,740 0.122 0.342 15,000,000 0.083 0.102 1,979,940 46,622 150,503,329
11 18,787,400 03,769,340 0.084 0.314 15,000,000 0.083 0.084 18,100 42,394 134,747,666
NPV @ 15% = 966,172,130

MILP model contributes 0.6% higher NPV as compared to the Lane’s model. Here, Lane’s
model exploits the resource at relatively higher cut-off grades, where both processing and
refinery limit the operation during initial years and then processing plant exclusively becomes
a bottleneck for the remaining life of operation. This leads to a policy that generates higher
cash flows initially, but sufficiently low production of refined copper during the final year of
the operation affects the cash flow as well as the overall NPV of the operation. On the
contrary, the MILP model not only maintains a consistent supply of ore to the processing
plant throughout the life of operation, but also it produces higher quantity of refined copper
that leads to higher cash flows in the final year of the operation. However, owing to the
discounting over 7 years, this higher cash flow in the final year translates into a relatively less
affect (0.6% increase) on overall NPV of the operation. Again in Case B, Tables 3 and 5
reflect a similar pattern, where the MILP model based cut-off grade policy demonstrates the
production of relatively higher quantity (troy ounces) of gold over the life of operation, which
generates more cash flows, ultimately leading to 0.5% higher overall NPV as compared to the
Lane’s model.
Table 6 demonstrates the application of MILP model as it generates the optimal cut-off grade
policy for a copper mining operation considering flotation and leaching as processing streams.
The MILP model fully utilises the available mining and processing capacities and maintains a
consistent supply of relatively high-grade ore to the flotation and low-grade ore to the leaching
streams. A dynamic cut-off grade policy initially (years 1–6) ensures a balance among the mining
and processing stages at their maximum throughputs, then (years 7–11) responds to the limiting
capacities at the flotation and leaching streams. The optimal policy also reflects that leaching as an
additional processing facility helps to process ore that would otherwise have been destined to the
waste dumps, which ensures that the production of a relatively higher quantity of metal coupled
with the higher cash flows and overall NPV over the life of operation can be achieved.

4. Conclusions
This paper shares a new MILP based approach for defining the optimal cut-off grade policy for
open pit mining operations considering multiple processing destinations/streams. Aligning with
the industry accepted Lane’s model, the proposed MILP formulation maximises the NPV, satisfies
production capacity constraints and maintains a balanced and sustained supply of valuable
material over the life of operation. The applications in hypothetical and realistic studies confirm
the functionality, validity, relevance and value of the proposed model over the traditional cut-off
 INTERNATIONAL JOURNAL OF MINING, RECLAMATION AND ENVIRONMENT 157

grade optimisation models. As reflected in Cases A and B, the proposed model generates 0.6% and
0.5% higher NPV over the life of operation as compared to Lane’s model; this capability of the
proposed method promises a benefit for future feasibility studies, strategic decision-making and
investments in the mineral resources industry.

Disclosure statement
No potential conflict of interest was reported by the authors.

References
[1] K. Dagdelen, Cut-off grade optimization, Proceedings of the 23rd International Symposium on the
Application of Computers and Operations Research in the Minerals Industry, Tucson, Arizona, (1992),
pp. 157–165.
[2] M.W.A. Asad, Optimum cut-off grade policy for open pit mining operations through net present value
algorithm considering metal price and cost escalation, Eng Computations 24 (7) (2007), pp. 723–736.
doi:10.1108/02644400710817961.
[3] M.W.A. Asad, M.A. Qureshi, and H. Jang, A review of cut-off grade policy models for open pit mining
operations, Resour Policy 49 (2016), pp. 142–152. doi:10.1016/j.resourpol.2016.05.005.
[4] K. Dagdelen and K. Kawahata, Value creation through strategic mine planning and cut-off grade optimization,
Mining Eng 60 (2008), pp. 39–45.
[5] J.M. Rendu, An Introduction to Cut-Off Grade Estimation, 2nd, Society of Mining, Metallurgy, and
Exploration, SME, Englewood, Colorado, 2014. (ISBN: 978–0–87335–393–9).
[6] H.K. Taylor, General background theory of cut-off grades. Transactions of the Institute of Mining and
Metallurgy: Section A, (1972), pp. 160–179.
[7] K.F. Lane, Choosing the optimum cut-off grade, Colo Sch Mines Q 59 (1964), pp. 811–829.
[8] K.F. Lane, The Economic Definition of Ore – Cut-Off Grades in Theory and Practice, Brisbane, Australia,
2015. COMET Strategy, Brisbane, Australia.
[9] A. Bascetin and A. Nieto, Determination of optimal cut-off grade policy to optimize NPV using a new
approach with optimization factor, J South Afr Inst Mining Metallurgy 107 (2007), pp. 87–94.
[10] K. Dagdelen, An NPV optimization algorithm for open pit mine design, Proceedings of the 24th International
Symposium on the Application of Computers and Operations Research in the Minerals Industries,
Montreal, Quebec, (1993), pp. 257–263.
[11] M.W.A. Asad, Cut-off grade optimization algorithm with stockpiling option for open pit mining operations of
two economic minerals, Int J Surface Mining Reclamation Environ 19 (3) (2005), pp. 176–187. doi:10.1080/
13895260500258661.
[12] M.W.A. Asad and E. Topal, Net present value maximization model for optimum cut-off grade policy of open
pit mining operations, J South Afr Inst Mining Metallurgy 111 (2011), pp. 741–750.
[13] M. Osanloo, F. Rashidinejad, and B. Rezai, Incorporating environmental issues into optimum cut-off grades
modelling at porphyry copper deposits, Resour Policy 33 (4) (2008), pp. 222–229. doi:10.1016/j.
resourpol.2008.06.001.
[14] J. Githiria, C. Musingwini, and J. Muriuki, Development of a computer-aided application using Lane’s
algorithm to optimize cut-off grade, J South Afr Inst Mining Metallurgy 116 (11) (2016), pp. 1027–1035.
doi:10.1016/j.eswa.2008.09.018.
[15] Y. He, K. Zhu, S. Gao, T. Liu, and Y. Li, Theory and method of genetic-neural optimizing cut-off grade and
grade of crude ore, Expert Syst Appl 36 (4) (2009), pp. 7617–7623. doi:10.1016/j.eswa.2008.09.018.
[16] J. Gholamnejad, Incorporation of rehabilitation cost into the optimum cut-off grade determination, J South
Afr Inst Mining Metallurgy 109 (2009), pp. 89–94.
[17] B. King, Optimal mining principles, Proceedings of the Orebody Modelling and Strategic Mine Planning
Conference, Perth, Australia, (2009), pp. 1–6.
[18] K. Dagdelen and K. Kawahata, Cut-off grade optimization for large scale multi-mine, multi-process mining
operations. Proceedings of the International Symposium on Mine Planning and Equipment Selection,
Curran Associates, Bangkok, (2007), pp. 226–233.
[19] R. Ganguli, K. Dagdelen, and E. Grygiel, Mine Scheduling and Cut-Off Grade Optimization Using Mixed
Integer Linear Programming, 3rd, SME Mining Engineering Handbook, SME, Englewood, Colorado, 2011,
850–852. Chapter 9.10: Systems Engineering.
[20] E. Moosavi, J. Gholamnejad, M. Ataee-Pour, and E. Khorram, Optimal extraction sequence modelling for
open pit operation considering dynamic cut-off grade, Mineral Resour Manag 30 (2014), pp. 173–186.
158 A. KHAN AND M. W. A. ASAD

[21] A.B. Yasrebi, A. Wetherelt, P. Foster, G. Kennedy, D.K. Ahangaran, P. Afzal, and A. Asadi, Determination of
optimised cut-off grade utilising non-linear programming, Arabian Journal of Geosciences 8 (10) (2015), pp.
8963–8967. doi:10.1007/s12517-014-1756-5.
[22] W. Hustrulid, M. Kuchta, and R. Martin, Open Pit Mine Planning and Design, 3rd ed., CRC Press/ Balkema, Leiden,
The Netherlands, 2013.
[23] M.W.A. Asad and R. Dimirakopoulos, Optimal production scale of open pit mining operations with uncertain
metal supply and long-term stockpiles, Resour Policy 37 (2012), pp. 81–89. doi:10.1016/j.resourpol.2011.12.002.
[24] CPLEX Optimizer (IBM Analytics), Available at April 17, 2018. https://www.ibm.com/analytics/data-
science/prescriptive-analytics/cplex-optimizer