See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/303988381

Mining rate optimization considering the stockpiling: A theoretical economics

and real option model

Article  in  Resources Policy · March 2016

CITATIONS READS

0 272

2 authors, including:

Kuangyuan Zhang
Pennsylvania State University
15 PUBLICATIONS   187 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Environmental Risk and Policy View project

Indoor Air Quality View project

All content following this page was uploaded by Kuangyuan Zhang on 29 July 2018.

The user has requested enhancement of the downloaded file.

 Resources Policy 47 (2016) 87–94

Contents lists available at ScienceDirect

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

Mining rate optimization considering the stockpiling: A theoretical

economics and real option model
Kuangyuan Zhang a,n, Andrew N. Kleit b
a
Lecturer of Energy and Mineral Economics, Department of Energy and Mineral Engineering, Pennsylvania State University, 228 Hosler Building, University
Park, PA 16802, USA
b
Professor of Energy and Environmental Economics, Department of Energy and Mineral Engineering, Pennsylvania State University, 201 Hosler Building,
University Park, PA 16802, USA

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

Article history: In the extraction of metal from a mine, the intermediate ore between the economic and the breakeven
Received 12 November 2015 cutoff grade is usually stockpiled for future processing once the mine is depleted. This research estab-
Received in revised form lishes a theoretical two-stage economic model to derive the value of this stockpile and how it affects
28 December 2015
optimal mining rate. By deriving the optimal condition for objective profit function and parameterized
Accepted 6 January 2016
Available online 14 January 2016
analysis, this research finds the stockpiling option can significantly boost a mine's profit. Processing the
stockpiled material affects the optimal mining rate and cutoff grade strategy significantly compared to
Keywords: the case that it is not processed. The research also investigates the optimal mining rate's sensitivity to
Metal mining strategy input variables such as commodity price, discounting rate, capital cost, and processing capacity, etc. In
Stockpiling management
addition, the intrinsic advantages of this approach compared to the broadly used Lane's model (1988) are
Pseudo-elasticity
discussed.
Cutoff grade
& 2016 Elsevier Ltd. All rights reserved.

1. Introduction Based on a theoretical two-stage model, this research examines

In the extraction of metal from a mine, the intermediate ore be-
tween the economic and the breakeven cutoff grade is usually stock-
piled for future processing once the mine is depleted. As widely stu-
died and discussed, the economic cutoff grade is defined as the cutoff
grade strategy that maximizes the net present value of a mine by
considering the opportunity cost, which is usually higher than the
breakeven cutoff (Zhang et al., 2015). This research establishes a two-
stage mining model to derive the value of stockpiling and how it af-
fects mining decisions, i.e. the optimal mining rate. For decades, in-
dustry practitioners and academic researchers have been interested in
questions on stockpiling, such as that how valuable the stockpiled
material is and its impact on the optimal mining strategy (Young,
1991); when is the best time (or price threshold) to process the
stockpiled material (Dixit and Pindyck, 1994; Conrad, 2000; Zhang
et al., 2015); how price expectations will impact ore's quality selection
(Withagen, 1998; Lee et al., 2006); and how will the stockpiled ore’s
quality changes as it is processed (Mason, 2011). Nevertheless, there is
only limited theoretic work studying the stockpiled option and its
impacts on mining production rate.
n
Corresponding author.
E-mail addresses: kuangyuan.thu@gmail.com (K. Zhang),
ank1@psu.edu (A.N. Kleit).
the production of stockpiled material after depletion. The profit
functions in both stages and their first order conditions (FOC) are
formulated explicitly and the optimal mining rate is obtained.
Profit from the stockpiled material in our model accounts for more
than 15% of the mine’s total value. The model also examines op-
timal mining rate's sensitivity to input variables, such as the
commodity price, the capital cost and the processing capacity.
This paper is organized as following: Section 2 reviews the
previous relevant literature. Section 3 describes the two-stage
model considering the stockpiling of the intermediate grade ma-
terial, and derives the profit-maximizing conditions. Section 3 also
analyzes the two-stage production model, the profit function, as
well as the first order condition of the objective function. Section 4
discusses the results and compares it with the two-stage model
with Lane’s (1988) model. Conclusions are provided in Section 5.
2. Literature review
Lane’s (1988) model has been adopted in the mining industry
for decades to determine the optimal cutoff grade strategy for
metal mines. By maximizing the net present value of a mine, the
Lane model suggests a series of economic cutoff grades that are
usually higher than the minimum grade that is economic to be
processed, i.e. the breakeven or heuristic cutoff. As shown in Fig. 1,

http://dx.doi.org/10.1016/j.resourpol.2016.01.005
0301-4207/& 2016 Elsevier Ltd. All rights reserved.
88 K. Zhang, A.N. Kleit / Resources Policy 47 (2016) 87–94
Fig. 1. A two-stage production model with cutoff grades and processing the
stockpile material.
the portion of the materials with the highest quality (the part with
dark color in Fig. 1) is mined and processed early in production.
This strategy leaves a portion of ore not processed, which may still
generate positive cash flows. These intermediate grade materials
(the middle part in Fig. 1) are extracted during the production
time, and usually stockpiled beside the mine to be processed after
the mine is depleted. Although mined during the production, the
lowest grade portion of the deposit (the part with light color in
Fig. 1) is regarded as waste and never processed.
Theoretical work by Krautkraemer (1988) uses a cylinder
model and optimal control approach to explore the optimal
strategy of mining rate and how changes in mining cost and metal
price impact the strategy. It is found that, if the price increases
more quickly than the discount rate, the mine is depleted sooner
with a higher cutoff grade. Krautkraemer (1989) extends the cy-
linder model to study the impacts of variability in price on the
selection of ore's quality in the metal deposit. However, the
stockpiling question was not examined in this model.
Epaulard and Pommeret (2003) derive a closed-form solution
of optimal extraction strategy for non-renewable resource based
on a recursive utility approach. Both uncertainties in resource
stock and technical progress are considered. It is found that un-
certainty in metal price leads to a conservative use or stockpile of
the resource for future extraction. Cairns and Shinkuma (2003)
examine the relationship between commodity price and the
choice of cutoff grade, which also arises in the two-stage model
proposed in this research. A follow-up similar work by Cairns and
Davis (2007) addresses the optimal stopping rule for nonrenew-
able resource extraction against the resource’s quality and ex-
pectations on price, and extend the model to sequential develop-
ment and timing issues of natural resource projects.
Asad (2005) models two minerals' grades and tonnage as a
series of two-dimension contours. Each axis in diagram (as shown
in Fig. 2) represents the grade for one mineral. A straight line is
drawn on the area of contours, and the intercept of the line with
one axis represents the cutoff for one mineral in absence of the
other. Solving for the optimal cutoff policy for the two minerals is
equivalent to determining the pair of cutoff intercepts maximizing
the net present value of the mine. Asad solves for the economic
cutoff grade, and considers processing the intermediate grade
material after exhausting the mine by a numerical iterative
method. It is found that stockpiles, with the advantage of no ad-
ditional mining cost, can generate a significant amount of addi-
tional positive cash flows. Asad’s work can be improved by ex-
ploring the impacts of the stockpiled on the mining strategy, i.e.
the optimal mining rate. In this research, it is assumed that the
stockpiling is stocked just on ground besides the processing fa-
cility with no degrading, additional capital expenditure; the unit
cost of material re-handling is assumed to be the same as
Fig. 2. Two dimensional grade and tonnage distribution of deposit (Asad, 2005).
processing the ores from the mine.
There are also several pieces of literature that have explored
the issues of ore quality, cutoff grade, and breakeven grade in a
metal mine using numerical methods. Dimitrakopoulos et al.
(2007) use stochastically simulated reserve models for designing
open pit mines, and apply the approach in a gold deposit. It is
found that both uncertainties from ore quality and gold price will
lead to a reduction in mining production. Osanloo et al. (2008)
address the environmental concerns when determining the opti-
mal cutoff grade based on copper deposits, which can minimize
the adverse environmental impacts of the mining projects. They
find that an appropriate chosen cutoff strategy can help to mitigate
the adverse environmental impacts of mining activities and
treating the waste. Shafiee and Topal (2010) argue that the
stockpiled mineral products such as gold can mitigate the negative
impacts of uncertainty in market, and provide the mining com-
panies an option for hedging strategy. Nieto and Zhang (2013)
study the cutoff grade for a by-product mine based on a rare earth
case of Dysprosium and Neodymium. It is found that the changes
in the by-product’s price can lead to a significant shift in the pri-
mary product's cutoff. Franco-Sepúlveda and Velilla-Avilez (2014)
review the relevant work on a mine's cutoff grade using a hy-
pothetical gold mine to illustrate Lane’s algorithm that leaves the
intermediate grade material stockpiled. However, there is a gap in
literature addressing the interaction of stockpiled material and the
grade selection in mining operations.
3. A two-stage production model
3.1. The general model
In this section, a model of two-stage production of a metal
deposit with variety in quality is presented. There are two stages
of production (as shown in Fig. 1): the first stage production in-
cludes mining the material in deposit with a constant production
rate of M , processing the part of the mined material with quality
higher than cutoff x1 in the time before the depletion; the part of
mined material with quality lower than x1 (the intermediate grade
material) is stockpiled for the second stage production after the
 K. Zhang, A.N. Kleit / Resources Policy 47 (2016) 87–94 89

first stage; the second stage production processes the stockpiled Q

M
intermediate grade material until all such material is processed.
∫ {pyMF1(M )−AM − RC}*e−δt dt
Note that as the stockpiled material is already on the ground, there (7)
o
is no additional mining/hauling cost associated. For simplicity, it is
assumed in this research that the unit processing cost to be the The exponential fraction term in the above equation is needed
same for ore form the mine and the stockpiling. The breakeven or to discount future value to present value. There is a minimum
heuristic cutoff grade for the final period of the second stage quality in the reserve that is economic to produce (or to generate
processing is defined as x2, which is determined by the metal positive cash flow), which is denoted as x2. To solve for x2, consider
price, production cost and the recovery rate. The two-stage model that the profit function the second stage declines until it equals to
is advanced compared to the Lane model in the sense that, 1) the zero,
economic cutoff grade in the first stage production is determined pyC*x2 − RC = 0 (8)
by choosing an optimal mining rate; 2) The stockpiled inter-
mediate material is processed in the production strategy de- By manipulating (8) and solving for x2, the cutoff grade for the
final stage of production thus becomes
scribed, but ignored in Lane’s analysis. The processing capacity C is
assumed to be constant across all the years of production. The total R
x2 = ,
quantity of material is Q , here measured in tons. This section py (9)
describes a general model, while a deposit with uniform grade
which is the minimum quality in deposit that is economic to
distribution based on the general model is assessed in the next
produce.
section.
Let the second stage end at time T2. At any time t in the second
The material in the deposit is assumed to have homogeneous
stage ( T1≤t ≤T2), the quality of ore that goes to the processing plant
tonnage-grade distribution (Peattie and Dimitrakopoulos, 2013; is a function of t and M , such that, x = x(t , M ) and
Emery et al., 2006). Thus the tonnage-grade distribution of one
unit of ore is identical to another unit. Assume z represents the x2 = x( T2) ≤ x( t ) ≤ x( T1) = x1 (10)
metal grade in percent, which is in the range 0– G, and G is the
and x(t ) is decreasing as time goes,
highest grade in the deposit,
ẋ( t ) < 0 (11)
0≤z ≤ G<1 (1)
Thus, the profit function at a time point t in the second stage is
Assume f (z ) is the probability distribution function for z , such
that, pyC*x( t , M ) − RC (12)
G
The present value (the exponential fraction term in the fol-
∫0 f (z )dz=1
(2) lowing equation is needed to discount future value to present
with value) of the total profits in the second stage in continuous time is
therefore
0 < x2 < x1 < G. (3)
T2

For one ton of material that has been mined, given processing ∫ ⎡⎣ pyC*x( t , M ) − RC ⎤⎦*e−δt dt
cutoff grade x1, the metal content in the material for the first stage T1 (13)
processing is
G
3.2. Uniform distribution model
∫ f (z )zdz = F1( M ).
x1(M ) (4) Assume there is a uniform distribution of grade, such that

The average grade of the ore that goes to the first stage of 1
f ( z) =
production is, G (14)
G where the highest grade is G and the lowest grade is 0. In each
M M
C
∫ f (z )zdz =
C
F1(M ) time of the first stage production, the profit function is
x1(M ) (5)
G + x1( M ) 1
Let the price for per unit of metal be p; y the metallurgical
pyM*
2
− AM − RC =
2
( )
pyM G + x1( M ) − AM − RC
(15)
recover rate (the portion of metal content that can be recovered in
As the grade distribution is uniform, in the first stage the
ore, over the total metal content in ore); C is the processing ca-
portion of ore that is processed is (which is constrained by the
pacity; R is the processing cost for per unit of ore; A is the mining
mining capacity M and processing capacity C ),
cost for per unit of mined material; δ is the discount rate. In each
period of the first stage production, the profit function is G − x1 C
=
G M (16)
M G
pyC*
C
∫x (M) f (z)zdz − AM − RC = pyMF1(M ) − AM − RC. (6) This implies x1 equals,
1

Q ⎛ C⎞
The first stage of production will last T1 = M years, as all the ore x1 = G⎜ 1− ⎟
⎝ M⎠ (17)
will be mined. The profit function in continuous time for the first
stage production is, Eq. (15) therefore implies the profit at each time point in the
90 K. Zhang, A.N. Kleit / Resources Policy 47 (2016) 87–94

first stage equals T2

π2 = ∫T ⎡⎣ pyC*x( t , M ) − RC ⎤⎦*e−δt dt
1
⎛ GC ⎞ T2 T2
pyM ⎜ G − ⎟ − AM − R = pyCa ∫T te−δt dt + ( pyb − R)C ∫T e−δt dt
⎝ 2M ⎠ (18) 1 1

( δt + 1)e−δt |T2 − 1
The present value of the total profit in the first stage becomes, = − pyCa* T1 ( pyb − R)C*e−δt |TT12
δ2 δ
pyCa ⎡ − δT − δT ⎤
T1 ⎛ ⎛ GC ⎞ ⎞ =− ⎣ ( δT2 + 1)e 2 − ( δT1 + 1)e 1⎦
π1(M ) = ∫0 ⎜ pyM ⎜ G − ⎟ − AM − RC ⎟e−δt dt δ2
⎝ ⎝ 2M ⎠ ⎠ 1
1 ⎛ GC ⎞
− ( pyb − R)C e−δT2 − e−δT1
δ
( )
=
δ
{pyM ⎜ G −
⎝ 2M ⎠
(
⎟ − AM − RC} − e−δT1 + 1 ) pyCa ⎡ − δT − δT ⎤
⎛ pyG ⎞
= ⎣ ( δT1 + 1)e 1 − ( δT2 + 1)e 2⎦
1 δ2
= {( pyG − A)M − ⎜
δ ⎝ 2 ⎠ (
+ R⎟C} 1 − e−δT1 ) (19) 1
+ ( pyb − R)C e−δT1 − e−δT2
δ
( ) (26)
From the beginning of production to the end of the second
stage, all the tonnage of material with grade higher than x2 will be The present value of the total profit for the two stages π is
processed. The total processed tonnage is, therefore

G − x2 ⎛ x ⎞ 1⎧ ⎛ pyG ⎞ ⎫
Q*
G
= Q ⎜ 1− 2 ⎟
⎝ G⎠ (20)
π (M ) = π1 + π2 = ⎨ ( pyG − A)M − ⎜
δ⎩ ⎝ 2
+ R⎟C ⎬ 1 − e−δT1
⎠ ⎭ ( )
pyCa ⎡ − δT − δT ⎤
+ ⎣ ( δT1 + 1)e 1 − ( δT2 + 1)e 2⎦
With C units processed per period the total production time δ2
becomes 1
+ ( pyb − R)C e−δT1 − e−δT2
δ
( ) (27)
Q⎛ x ⎞
T2 = ⎜ 1− 2 ⎟
C⎝ G⎠ (21) The first order condition for the profit functions for both stages
will be derived below. It is known that the length of the first
Q
Since the grade distribution is uniform and processing capacity period is T1 = M
, so the first order derivative is:
is constant, x(t , M ) can be modeled as a linear function of t . This
implies that ∂T1 Q
=− 2
∂M M (28)
t − T1
x( t , M ) = x1 + (x2 − x1)
T2 − T1 (22) The derivative of the profit in the first stage respect to the
mining rate is (using formula (19) and (28)),
Based on Eqs. (9), (10) and (17), it is known that x1 > x2, i.e.,
C R ∂π1 1⎧ ∂T ⎡ ⎛ pyG ⎞ ⎤⎫
G(1− M )> py . To simplify the study of how x(t , M ) changes in the
∂M δ⎩
⎪

⎪ (∂M ⎣
)
= ⎨ ( pyG − A) 1 − e−δT1 + δe−δT1 1 ⎢ ( pyG − A)M − ⎜
⎝ 2
+ R⎟C ⎥⎬
⎠ ⎦⎭
⎪

second stage, denote in (22),

1⎧ Q ⎡ ⎛ pyG ⎞ ⎤⎫
=
δ⎩
⎪

⎪ (
⎨ ( pyG − A) 1 − e−δT1 − δe−δT1
M 2⎣ )
⎢ ( pyG − A)M − ⎜
⎝ 2
+ R⎟C ⎥⎬
⎠ ⎦⎭ (29)
⎪

x − x1
a( M ) = 2
T2 − T1 (23)
In formula (26) for the profit in the second stage, there are two
∂a Q
x1T2 − x2T1 terms. The FOC for the first term in (26) is (using ∂M
=0 and T1 = M
),
b( M ) =
T2 − T1 (24)
pyCa ⎡ − δT ⎤

Then (22) can be written as,

∂{ δ2
− δT
⎣ ( δT1 + 1)e 1 − ( δT2 + 1)e 2⎦ }
∂M
pyCa ⎧ ∂T1 −δT1 ∂T ⎫
t − T1 = ⎨δ
2 ⎩
e + ( δT1 + 1)( − δ )e−δT1 1 ⎬
x( t , M ) = x1 + ( x2 − x1) δ ∂M ∂M ⎭
T2 − T1
pyCa ∂T1 −δT1
=
x2 − x1 x T − x2T1
t+ 12
=
δ ∂M
e {
− δT1e−δT1 − e−δT1 }
T2 − T1 T2 − T1
pyCa ⎛ Q ⎞
= at + b (25) =
δ ⎝
⎜ − 2 ⎟ − δT1e−δT1
M ⎠
( )
pyCaQ −δT1
It can be shown mathematically that the grade function in the = T1e
M2
second stage x(t )=at + b is not a function of the mining rate M , i.e.
pyCaQ2 − δQ
∂a
=0 and
∂b
=0. The details of the derivation can be found in = e M
∂M ∂M M3 (30)
Appendix.
With denotations in formulas (23) and (24), the present value The first order condition for the second term in (26) is
of the second stage in (13) can be written as,
 K. Zhang, A.N. Kleit / Resources Policy 47 (2016) 87–94 91

9
∂ { C
δ
( pyb − R)( e−δT1 − e−δT2)}
∂M 8

Cutoff Grade for 1st Stage (gram/ton)

C ⎛ ∂T ⎞ 7
= ( pyb − R)⎜ − δe−δT1 1 ⎟
δ ⎝ ∂M ⎠
6
⎛ ⎛ Q ⎞⎞
= C ( pyb − R)⎜ − e−δT1⎜ − 2 ⎟⎟
⎝ ⎝ M ⎠⎠ 5
CQ ( pyb − R) δQ 4
= e− M
M2 (31)
3
By summing (30) and (31), the derivative for the profit function
2
in the second stage respect to mining rate becomes

∂π2 ⎡ pyCaQ2 CQ ( pyb − R) ⎤ − δQ

1
=⎢ + ⎥e M
∂M ⎢⎣ M3 M2 ⎥⎦ (32) 0
0 200,000 400,000 600,000 800,000
The sign (positive or negative) of FOC for the first stage in Mining Rate (tons/year)
formula (29) is indeterminate. However, in the FOC for the second
Fig. 3. Cutoff grade for the first stage production ( x1) by differentiated mining rates.
stage in formula (32), if the price p is high enough, then
pyb − R > 0, and the formula (32) is positive. So the present value
of the profit in the second stage increases to the mining rate in the 25
first stage. This is because if there is an increase in M , the mine will
be depleted in the first stage more quickly. This leaves more ore to
20
be stockpiled and processed in the second stage, and increases the
profit from the second stage accordingly.
15
Year

4. Parameterization and analysis

10
4.1. Parameterization

5
The economic parameters of a gold mine are adopted to ex-
amine how the changes in the mining rate in the first stage will
affect the mine’s total profit based on the two-stage model. Fol- 0
lowing to a hypothetic gold mine’s information and data from 0 100,000 200,000 300,000 400,000 500,000 600,000 700,000 800,000
Batten et al. (2014), an Fanacial Analysis (2009) and Goldcorp’s Mining Rate (tons/year)
2014 Annual Report, assume there are 1 million tons of material Length of 1st Stage Length of 2nd Stage
underground that can be mined; the gold price is $1200/Oz.,
Fig. 4. Length of the first ( T1) and second ( T2 − T1) stages by differentiated mining
($42.34/gram); the metallurgical recovery rate is 70%; the highest rates.
grade in the deposit is 8 gram/ton; discount rate is 3%; the cost for
mining one ton of underground material is $100; the cost for
$140
processing one ton of ore is $100. A processing plant is built such
that the processing capacity is 25,000 tons of ore each year. The $120
summary of the economic parameters for the gold mine is given in
Table 1. $100
Assuming the above parameters, the cutoff grade by the end of
the second stage x2 is calculated to be 3.37 gram/ton from (9). Eq.
Millions

$80
(21) implies the total production time length is 23 years. From Eq.
(25), the value of the function of quality's change in the second $60
stage, a is calculated to be
0.2; b is calculated to be 8.
Using on Eq. (17), the graph for the cutoff grade in the first $40
production stage is calculated (shown in Fig. 3). The cutoff grade in
the first stage increases as the mining rate expands (note that this $20

Table 1 $0
0 200,000 400,000 600,000 800,000
Economic input parameters of the gold mine (Batten et al., 2014; Goldcorp Annual
Mining Rate (tons/year)
Report, 2014).
profit 1st stage profit 2nd stage
Total material quantity ( Q ) 1,000,000 Tons
Fig. 5. Present value of the profit in the 1st stage ( π1) and 2nd stage ( π2 ) by dif-
Gold price ( p ) 42.34 $/gram
ferentiated mining rate.
Recovery rate ( y ) 70% %
Highest grade in deposit ( G ) 8 Gram/ton
Processing cost ( R ) 100 $/ton
does not mean the mining rate will change during the years that
Processing capacity ( C ) 25,000 Tons/year
Discount rate ( δ ) 3 % mine is in operation), but the increase rate diminishes as the cutoff
Mining cost ( A ) 100 $/ton grade approaches the highest grade in the deposit – 8 gram/ton. In
Capital cost per unit of mining rate ( ω ) 200 $/ton
the first stage, as the total material quantity Q are mined with a
92 K. Zhang, A.N. Kleit / Resources Policy 47 (2016) 87–94

$90 Table 2
Pseudo-elasticities for input variables on optimal mining rate M *.
$80
Each input variable's change
$70

10%
5%
1% 1% 5% 10%
$60
Total material quantity ( Q ) 0.61 0.61 0.60 0.60 0.60 0.59
Millions

$50 Gold price (p) 0.60 0.59 0.58 0.58 0.57 0.57
Recovery rate ( y ) 0.60 0.59 0.58 0.58 0.57 0.57
$40 Highest grade in deposit ( G ) 0.60 0.59 0.58 0.58 0.57 0.57
Processing capacity (C ) 0.41 0.40 0.40 0.40 0.39 0.39
$30 Discount rate ( δ ) 0.03 0.03 0.03 0.03 0.03 0.03
Processing cost ( R ) 0 0 0 0 0 0
$20
0.01
0.01
0.01
0.01
0.01
0.01
Mining cost (A )
Capital cost per unit of
0.62
0.59
0.58
0.57
0.55
0.53
$10
mining rate ( ω )

$0
0 100,000 200,000 300,000 400,000 500,000
Mining Rate (tons/year)
Fig. 6. Profit of the mine with average capital cost ($200 for 1 ton of mining ca-
pacity) to determine the optimal mining rate.
mining rate of M , the time length of first stage is T1 =
Q
Based on T1 = M , the length of the first stage can be obtained
based on the parameters in Table 1 and is shown in Fig. 4 against
differentiated mining rate. Fig. 4 also shows the length of the
second stage, which can be calculated by T2 − T1. As all the material
above the breakeven cutoff x2 are processed at full processing
capacity, the total processing quantity and capacity are fixed. Thus,
the production time T2 is fixed, which can be calculated by Eq. (21).
The total production time T2 is calculated to be 23 years, and there
is a trade-off between the lengths of the first stage and second
stage. The deposit is depleted sooner as the mining rate increases,
but this effect diminishes as the mining rate increases.
Fig. 5 shows the present values for the profits in first stage and
second stage based on Eqs. (19) and (26). Note that these results
do not consider the initial investment to build the mining opera-
tion, and this issue will be examined in later paragraphs. It is
found that both profit curves show an increasing trend respect to
the mining rate, but the increase trend is diminishing as the
mining rate expands.
Fig. 5 shows that the production of the stockpiled material in
the second stage can provide at most 15% of the mine's total profit
using Batten et al. (2014) and Financial Sense (2009)’s parameters.
As the mining rate increases, the share of the first stage's profit
decreases as the share of the second stage's profit increases.
From Fig. 5, it can be observed that the profits from both stages
increase as the mining rate M increases. However, the result of
profits in this graph does not consider initial investment to build
up the mining operation Assume the capital cost increases as the
mining rate M increases linearly. According to Financial Sense
(2009)’s report on gold mining expenditures, an average of $465 is
required for per ounce of produced gold as capital expenditure to
build up a gold mine. Using a grade of 8 gram/ton (0.28 ounce/ton)
and a recovery rate of 0.7, the average cost of building up 1 ton of
mining capacity is $465*0.28/0.7 ¼$186. Here assumes a higher
average capital cost of $200 to build up 1 ton of mining capacity, to
reflect the increase in capital cost from 2009 to 2014. Fig. 6 shows
that the optimal mining rate M * is 211,900 tons per year. Based on
the optimal mining rate, the cutoff grade for the first stage is
7.06 gram/ton; the first stage lasts 5 years, and the second stage
lasts 18 years; About 15% of the mine's profit comes from pro-
cessing the stockpiled material; The maximum profit of produc-
tion by considering the capital cost is $79.7 million.
To examine how responsive the optimal mining rate M * is to a
change of input variable, this section assess the pseudo-elasticity
600,000 700,000
for an input variable v ,
%ofChangeinM * ∂M * v
ϵv = =
%ofchangeinv ∂v M * (33)
Q
M
. To estimate the pseudo-elasticities of input variables, six sce-
narios of change in each input variable are considered. The cor-
responding percent of change in M * can be calculated. Table 2
shows the pseudo-elasticities for the input variables.
From Table 2, it can be observed that total material quantity,
gold price, recovery rate, highest grade in deposit, processing ca-
pacity, and discount rate have positive pseudo-elasticities This
means that an increase in each of these input variables will lead to
an increase in the optimal mining rate. Higher elasticity means
that the optimal mining rate is more responsive to a change in the
variable in question.
Total material quantity has the highest value of pseudo-elasti-
city, and the elasticities are almost unchanging across the relevant
levels (around 0.6); Gold price, recovery rate, and highest grade in
deposit all have same pseudo-elasticity (around 0.58); and pro-
cessing capacity has a pseudo-elasticity around 0.4.
Processing capacity has a positive pseudo-elasticity on the
mining rate. The pseudo-elasticity of discount rate is positive and
close to zero (0.03). Higher discount rates will make the mine's
owner set a higher mining rate to deplete the mine faster, and
partially avoiding the discounting effect for the future cash flows.
This effect, however, is relatively small.
The optimal production rate is perfectly inelastic to processing
cost, as processing cost’s elasticity is zero. This means the pro-
cessing cost will not affect the optimal choice of the production
rate, which is for the reason that, the two-stage model assumes
that the processing plant always operates in full capacity until the
depletion, no matter how the processing cost changes. The capital
cost per ton of mining rate has a negative pseudo-elasticity, so an
increase in capital cost per unit of mining rate decreases the choice
of optimal mining rate. The elasticity of mining cost is negative
and close to zero (
0.01). Higher mining cost for each ton of
mined material (due to it is more difficult to mine) promotes the
mine’s owner to use a lower mining rate.
4.2. Comparison to the Lane's model
Lane (1988) establishes an approach to calculate the cutoff
grade strategy across the life of a metal mine based on numerical
techniques. The Lane model and the model introduced here ad-
dress the similar problem about a metal mining operation with
variability in ore quality and constraints in processing capacity.
This research provides several insights that cannot be drawn from
previous literature.
 K. Zhang, A.N. Kleit / Resources Policy 47 (2016) 87–94
Lane’s approach calculates a cutoff grade strategy, which is
usually higher than the breakeven cutoff, the lowest grade in the
deposit that is economic to process. This leaves the intermediated
grade material, or the stockpiled ore that still may generate posi-
tive cash flows, but does not consider this portion of the deposit
when making decision on the optimal cutoff grade strategy and
mining rate. The two-stage model in this research covers proces-
sing the stockpiled, and determines the optimal mining rate and
cutoff grade strategy by considering the profits from the first stage
production as well as from processing the stockpiled.
Additionally, there is a gap in previous literature incorporating
the capital cost of building up the mine and optimal mining ca-
pacity. The model introduced in this research explicitly formulates
the profit function for production and its first order condition, and
solving for the optimal mining capacity by modeling capital cost.
As the mine is constrained by processing capacity, choosing an
optimal mining rate strategy is equivalent to choosing an optimal
cutoff grade strategy.
In addition to determining the optimal mining rate, this re-
search measures how responsive the optimal mining rate is to
changes in input variables by calculating the pseudo-elasticities for
input variables. Variables from the most positively responsive to
least responsive for the optimal mining rate, are total material
quantity, gold price, recovery rate, highest grade in deposit, and
processing capacity; optimal rate is the most negatively responsive
to per unit capital cost, and is much less negative responsive to the
mining cost; processing cost proves to be perfectly inelastic with
the optimal mining rate; the effects of several input variables (total
material quantity, gold price, recovery rate, highest grade in de-
posit, processing capacity, and per unit capital cost) are found to
be diminishing. The Lane model, at best, can be helpful on de-
termine the effects of input variables on the mine's profit, but not
on the optimal mining rate, or whether their effects are
diminishing.
There are also some intrinsic advantages of a theoretical model
compared to the Lane's approach, which can be used to explain the
results intuitively. It is found that the processing cost is perfectly
inelastic with the optimal mining rate. This can be clearly ex-
plained by the theoretical model, as the terms of the processing
cost are canceled out in the first order conditions of the profits in
the first and second stages. However, it is fairly difficult to reach
this conclusion based on methods in previous literature.
This research finds the grade function in the second stage
∂a ∂b
x(t )=at + b is not a function of the mining rate, i.e. ∂M =0 and ∂M =0.
Although not straightforward intuitively, this conclusion can be
reached and proved by the theoretical two-stage model in this
research. The previous literature fails to provide insights about
processing the stockpiled, or how the stockpiled material’s grade
changes in the second stage.
5. Conclusion
In mining practice, the Lane model has been used for decades
as a standard tool to determine a metal mine's cutoff grade policy.
The Lane's algorithm proposes a higher cutoff grade than the
breakeven cutoff, which leaves part of the deposit with inter-
mediate grade not processed. These materials are usually stock-
piled, and processed to contribute to the mine's value. The stock-
piling has been discussed and interested extensively by the in-
dustry and academic researchers, but there is limited literature
addressing this issue by incorporating the mining strategy of the
original deposit. This research has established a theoretic two-
stage model, examining the processing of the stockpiled, and its
impacts on the optimal mining rate, cutoff grade strategy, and
93
mine's profit. Profit from the stockpiled material cannot be ig-
nored, which may account as much as 15% of the mine’s total
profit.
The mine is assumed to have variability in ore's quality and is
constrained by the processing capacity. It is assumed that the
processing plant is always in full capacity across all the production
years. By deriving explicitly the mine’s profit function and its first
order condition, the optimal mining rate strategy is determined by
the two-stage model with parameterized analysis.
This research also measures how responsive the optimal
mining rate is to the changes in the input variables, with the cal-
culations of their pseudo-elasticities. Total material quantity, metal
price, recovery rate, highest grade in deposit, processing capacity
and discount rate prove to positively impact the optimal mining
rate; as the mine’s processing plant is always operating in full
capacity during the production years; the processing cost is per-
fectly inelastic with the optimal mining rate; mining cost and the
capital cost of the mining equipment negatively impact the opti-
mal mining rate; the effects of several variables are diminishing as
their percent of change increases.
The two-stage model in this research has intrinsic advantages
over previous numerical work. By a theoretical model, it is
straightforward to explain that the processing cost is perfectly
inelastic to the optimal mining rate, as well as why the change of
ore' quality is not a function of mining rate when processing the
stockpiled.
This research implies that, by choosing an optimal mining
strategy, the stockpiled material contributes to the mine’s value by
a great part, and should not be ignored. In addition, the con-
sideration of the stockpiled material significantly impacts the
mining strategy (mining rate and cutoff grade) of the deposit. The
two-stage model is also beneficial to policy makers and mineral
property owners, when assessing the impacts of a policy pro-
moting the stockpile management, as well as making mining
strategic decisions.
Last but not the least, this theoretical research has its limita-
tions by the nature of model, primarily to balance for the mathe-
matical simplicity and consistency. These limitations will be the
future research directions. 1) a constant marginal capital cost is
assumed in the model, which does not cover the effect of “eco-
nomics of scale”; 2) it should be noted that the stockpiling is not
necessarily to be processed after depletion, which is examined in
this research and can be processed any time before depletion; 3)
Some ores such as sulfuric ones cannot be stockpiled for a long
time due to degrading and oxidizing, and which can be depicted
by multiplying a degrading factor to the profit function in the
second stage; 4) the location of stockpile is very important, and
will affect economic feasibility for the mine; for simplicity, this
research assumes that the stockpiling is stocked just on ground
besides the processing facility with no degrading or additional
capital expenditure; 5) the economic cutoff grade considering the
opportunity cost is usually decreasing as the mine exploration
goes on; however, this research assumes that the economic cutoff
grade maintains constants during the mine’s life, which will be
also improved in the future research; 6) the optimal condition for
processing capacity C would be also another interesting future
research topic.
Acknowledgments
The authors wish to thank Dr. Antonio Nieto for helpful dis-
cussion on the topic. Support from Energy and Environmental
Economics and Policy Initiative in Pennsylvania State University is
gratefully acknowledged.
94 K. Zhang, A.N. Kleit / Resources Policy 47 (2016) 87–94

Q
Appendix. : Derivation of grade changes in the second stage and plug formulas (35), (34), (17), (21) and T1 = ,
M

This appendix proves that the function of grade change in the x1′T2 − x1′T1 − x2T1′ + x1T1′
second stage is not x(t )=at + b is not a function of the mining rate ⎛
GC Q ⎛⎜ x ⎞ GC Q Q C⎞Q
∂a ∂b
M , i.e. ∂M =0 and ∂M =0. Note that T1 = M and x1=G(1− M ) are func-
Q C = 1 − 2⎟ − 2 + x2 2 − G⎜ 1 − ⎟ 2
2
M C ⎝ G ⎠ M M M ⎝ M⎠M
tions of M , and T2 and x2 are not, following derivatives can be
obtained, GQ Qx GCQ Qx GQ GCQ
= − 22 − + 22 − 2 + =0
∂T1 Q M2 M M3 M M M3 (41)
=− 2
∂M M (34) Thus, by formulas (40) and (41), the derivative of a and b re-
spect to M is zero,
∂x1 GC
= 2 ∂a
∂M M (35) =0
∂M (42)

∂T2
=0 ∂b
∂M (36) =0
∂M (43)

δx2 It has been shown that the grade change in the second stage is
=0
δM (37) not a function of the mining rate M .

Based on formula (23) and (24), the derivatives of a(M ) and

b(M ) respect to M are,
References
∂a
=
(
∂
x2 − x1
T2 − T1 ) Asad, M.W.A., 2005. Cutoff grade optimization algorithm with stockpiling option for
∂M ∂M
open pit mining operations of two economic minerals. Int. J. Surf. Min. Reclam.
′ ′

=
( x2 − x1) ( T2 − T1) − ( x2 − x1)( T1 − T2) Environ. 19 (3), 176–187.
2 Batten, Jonathan A., Ciner, Cetin, Lucey, Brian M., 2014. On the economic de-
( T2 − T1) terminants of the gold–inflation relation. Resour. Policy 41, 101–108.
Cairns, Robert D., Shinkuma, Takayoshi, 2003. The choice of the cutoff grade in
GC ( x − x )Q
− 2 ( T2 − T1) − 2 21 mining. Resour. Policy 29 (3), 75–81.
= M 2
M Cairns, Robert D., Davis, Graham A., 2007. Strike when the force is with you: optimal
( 2 1)
T − T (38) stopping with application to resource equilibria. Am. J. Agric. Econ. 89 (2), 461–472.
Conrad, Jon M., 2000. Wilderness: options to preserve, extract, or develop. Resour.
Energy Econ. 22 (3), 205–219.

∂b
=
∂ ( x1T2 − x 2T1
T2 − T1 ) Dimitrakopoulos, R., Martinez, L., Ramazan., S., 2007. A maximum upside/minimum
downside approach to the traditional optimization of open pit mine design. J.
∂M ∂M Min. Sci. 43 (1), 73–82.
′ ′ Dixit, Avinash K., Pindyck, Robert S., 1994. Investment Under Uncertainty. Princeton
=
( x1T2 − x2T1) ( T2 − T1) − ( x1T2 − x2T1)( T2 − T1) University Press, Princeton, N.J., United States.
2
( T2 − T1) Emery, X., Ortiz, J.M., Rodríguez, J.J., 2006. Quantifying uncertainty in mineral re-
sources by use of classification schemes and conditional simulations. Math.

=
( x1′T2 + x1T2′ − x2′ T1 − x2T1′)( T2 − T1) + ( x1T2 − x2T1)T1′ Geol. 38 (4), 445–464.
2 Epaulard, Anne, Pommeret, Aude, 2003. Optimally eating a stochastic cake: a re-
( T2 − T1) cursive utility approach. Resour. Energy Econ. 25 (2), 129–139.
Fanacial Analysis, 2009. Capital Expenditure Clouds Gold Mining Profitability.
=
( x1T2 − x2T1)( T2 − T1) + ( x1T2 − x2T1)T1′
′ ′
〈http://www.financialsense.com/node/631〉.
2
( T2 − T1) Franco-Sepúlveda, Giovanni, Velilla-Avilez, Danilo Arturo, 2014. Mining planning as
a function of changes in the law of cutting mine planning review of variation as
x1′T22 − x2T1′T2 − x1′T2T1 + x2T1′T1 + x1T2T1′ − x2T1T1′ a function cutoff grade (in Spanish). Earth Sci. Bull. (35), 25–30.
=
( T2 − T1)
2 Goldcorp Annual Report, 2014. Link: 〈http://s1.q4cdn.com/038672619/files/
ar2014/_doc/Goldcorp_AR14_full.pdf〉.

=
(
T2 x1′T2 − x1′T1 − x2T1′ + x1T1′ ) Krautkraemer, Jeffrey A., 1989. Price expectations, ore quality selection, and the
2 supply of a nonrenewable resource. J. Environ. Econ. Manag. 16 (3), 253–267.
( T2 − T1) (39) Krautkraemer, Jeffrey A., 1988. The cutoff grade and the theory of extraction. Can. J.
Econ. 21 (1), 146–160.
Look at the numerator of the formula (38), and plug in x1, x2, T2 Lane, Kenneth F., 1988. The Economic Definition of Ore: Cut-off Grades in Theory
Q and Practice. Mining Journal Books, London.
from formulas (17), (9), (21), and T1 = ,
M Lee, Junsoo, List, John A., Strazicich, Mark C., 2006. Non-renewable resource prices:
GC (x2 − x1)Q deterministic or stochastic trends? J. Environ. Econ. Manag. 51 (3), 354–370.
− (T2 − T1) − Mason, Charles F., 2011. On stockpiling natural resources. Resour. Energy Econ. 33
M2 M2 (2), 398–409.
GC Q x Q (x − x )Q Nieto, A., Zhang, Kuangyuan, 2013. Cutoff grade economic strategy for by-product mi-
=− [ (1 − 2 ) − ] − 2 21 neral commodity operation: rare earth case study. Min. Technol. 122 (3), 166–171.
M2 C G M M Osanloo, M., Rashidinejad, F., Rezai., B., 2008. Incorporating environmental issues
GQ Qx2 GCQ Qx2 Qx1 into optimum cut-off grades modeling at porphyry copper deposits. Resour.
=− + + − + Policy 33 (4), 222–229.
M2 M2 M3 M2 M2 Peattie, R., Dimitrakopoulos, R., 2013. Forecasting recoverable ore reserves and their
GQ GCQ Qx1 uncertainty at Morila gold deposit, mali: an efficient simulation approach and
=− + + future grade control drilling. Math. Geosci. 45 (8), 1005–1020.
M2 M3 M2 Shafiee, Shahriar, Topal, Erkan, 2010. An overview of global gold market and gold
C price forecasting. Resour. Policy 35 (3), 178–189.
GQ GCQ G(1 − )Q
=− + + M Withagen, Cees, 1998. Untested hypotheses in non-renewable resource economics.
2 3 2 Environ. Resour. Econ. 11 (3), 623–634.
M M M
Young, Denise, 1991. Productivity and metal mining: evidence from copper-mining
GQ GCQ GQ GCQ firms. Appl. Econ. 23 (12), 1853–1859.
=− + + − =0
M2 M3 M2 M3 (40) Zhang, Kuangyuan, Nieto, Antonio, Kleit, Andrew N., 2015. The real option value of
mining operations using mean-reverting commodity prices. Miner. Econ. 28 (1),
Look at the numerator the formula (39), and divide that by T2, 11–22.

View publication stats