Cut off Grade Optimizer Integrating Contaminant Constraints

Sebastián H. Troncoso
Universidad de Chile
Patricio A. Prieto
Universidad de Chile
Enrique A. Rubio
Universidad de Chile

Abstract

Open pit mine planning is normally based on the selection of an economic envelope that has been
generated by the Lerchs & Grossman algorithm, which requires individual economic valuation of each
block of a geological block model. The selected envelope, usually called Final Pit, is later discretized in
pushbacks and benches, which correspond finally to the fractions that the production schedule will
allocate in time. This planning methodology has been widely used by the mining industry in countless
projects to exploit open-pit mines, however, in many of the deposits can be found some undesirable
elements such as arsenic, phosphates, ash among others that tend to reduce the value of elements of
economic interest, either by disrupting the processes of concentration or as deemed impurities that affect
the quality of the final product. Thus, a fair amount of blending would be required to capture the real value
of the ore body. The traditional approach under estimates a block value by penalizing the block value by
the amount of contaminant even when at the mine a set of blocks is amine at a time. This paper proposes a
methodology for computing cut off grades over time that would maximize net present value allowing
blending that should fulfill a set of metallurgical constraints. Preliminary results show significant
increases in value and reserves quantity, questioning the use of fixed cut-off grades in generating a
production schedule when blending applies to fulfill a set of geometallurgical constraints. Finally, the
model allows the authors to show the importance of having adequate planning tools that will lead to the
ability to look at the potential economic value of a mineral deposit.

Introduction

Certainly in the current days mineral deposits are far away from being monometallic, each time
technology advances found new uses from elements that in the past were not considered, nevertheless, this
new desirable elements are normally found together with non desirable elements which reduces the value
of the firsts either by disrupting the processes of concentration or as deemed impurities that affect the
quality of the final product. Such is the case of the deposit schematized by the following grade-tonnage
curve:
9,000 3 9,000 3,000
8,000 8,000
2.5 2,500
7,000 7,000
Grade (%, ppm)

Grade (%, ppm)

Tonnage (Mt)

Tonnage (Mt)

6,000 2 6,000 2,000

5,000 5,000
1.5 1,500
4,000 4,000
3,000 1 3,000 1,000
2,000 2,000
0.5 500
1,000 1,000
0 0 0 0
0 0.5 1 1.5 2 2.5 0 0.5 1 1.5 2 2.5
Tonnage Cu grade (%) Au grade (ppm) Cu cut-off grade (%) Tonnage As grade (ppm) Cu cut-off grade (%)

1
 Figure 1. Grade tonnage curve of high arsenic content deposit.

If only the left graph of Figure 1 is look, the amounts of copper and gold lead to think in a very interesting
mining project, nevertheless, when the right part is observed, the amounts of arsenic (nowadays
considered as an impurity) reduces the previous feelings about the economic feasibility of the project.
Moreover, the project looks worst if it is considered the mineralogical association between copper and
arsenic and its response to the actual concentrations processes, shown in Figure 2, and also the following
penalties for the presence of arsenic in concentrate:

 2.5 US$/t of concentrate for each 1,000ppm above 1,000ppm of As in concentrate.

 3.5 US$/t of concentrate for each 1,000ppm above 25,000ppm of As in concentrate.

40,000
35,000
As in concentrate (ppm)

30,000
25,000
20,000
15,000
10,000
5,000
0
0 500 1000 1500 2000
As head grade (ppm)
Figure 2. Arsenic grade in concentrate as function of arsenic head grade.

The previous issues motivate the generation of an economic production schedule that does not exceed a
certain amount of arsenic in the material sent to processing plant to make the product attractive to the
market which has strict rules regarding arsenic concentrations in the concentrate. In the case study is
considered a 150ppm limit for arsenic in head ore, constraint that must not be overpass under any
circumstance.

The first approximation to the problem is to compute a production schedule with standard open pit
planning tools, without considering the arsenic constraint. The results are shown in the following figure:
120 1.2 120 1000

100 1 100 800

Tonnage (Mt/year)

Tonnage (Mt/year)

80 0.8 80
Cu (%), Au (ppm)

600 As (ppm)
60 0.6 60
400
40 0.4 40

20 0.2 20 200
NPV: 840.5 MUS$ NPV: 840.5 MUS$
0 0 0 0
1 2 3 4 5 1 2 3 4 5
Ore Waste Cu Au Period (years) Ore Waste As As Limit Period (years)
Figure 3. Production Schedule without consider arsenic constraint (tonnages, grades and Net Present Value).

Naturally, due to the high arsenic content in the deposit, the arsenic limit of 150ppm is overpass in every
period of the schedule, becoming this productive program unfeasible, despite its economic value. A
second approach is to try to reduce the total amount of arsenic in the final pit envelope by considering a
fixed cutoff grade as an upper bound for this element, so those blocks in the model which an arsenic
content higher than the considered cutoff grade, are immediately treated as waste material, despite its

2
economic value. By considering an arsenic cutoff grade of 300ppm, the resulting production schedule is
the following:
120 1.2 120 160
140
100 1 100
120
Tonnage (Mt/year)

Tonnage (Mt/year)
Cu (%), Au (ppm)
80 0.8 80
100

As (ppm)
60 0.6 60 80
60
40 0.4 40
40
20 0.2 20
20
NPV: 203.2 MUS$ NPV: 203.2 MUS$
0 0 0 0
1 2 3 4 5 1 2 3 4 5
Ore Waste Cu Au Period (years) Ore Waste As As Limit Period (years)
Figure 4. Production Schedule considering arsenic constraint through cutoff grade of 300ppm in final pit construction
(grades, tonnages and Net Present Value).

Although the arsenic limit is not overpass in any period, becoming this schedule feasible, the net present
value of the productive program was strongly reduced, generating a not economically attractive project. If
it is considered a higher arsenic cutoff grade (400ppm), which allows a greater arsenic content in the
economic envelope and hopefully, a higher economic value (due to mineralogical association between
arsenic, copper and gold), the project seems in the following way:
120 1.2 120 250

100 1 100 200

Cu (%), Au (ppm)
Tonnage (Mt/year)

Tonnage (Mt/year)

80 0.8 80
150

As (ppm)
60 0.6 60
100
40 0.4 40

20 0.2 20 50
NPV: 505.1 MUS$ NPV: 505.1 MUS$
0 0 0 0
1 2 3 4 5 1 2 3 4 5
Ore Waste Cu Au Period (years) Ore Waste As As Limit Period (years)
Figure 5. Production Schedule considering arsenic constraint through cutoff grade of 400ppm in final pit construction
(grades, tonnages and Net Present Value).

Despite the more interesting economic value of 505.1MUS$, the schedule becomes unfeasible since
periods 1 and 4 exceed the arsenic limit. Naturally, with enough time and effort, considering variable
arsenic cutoff grades and other alternatives of sequencing and pushbacks selection, could be found a
feasible production schedule with an acceptable economic value, nevertheless, this approximation is not
different than trial and error and does not consider an important issue: block blending for economic
valuation, which in this case, due to the presence of a undesirable element which can be diluted, becomes
an interesting alternative to look at the economic value deposit.

A mixed integer linear program model is proposed to tackle the cut off grade optimization for open pit
mine planning, beginning in the final pit envelope selection until production scheduling process, that
based on a MILP model allows to generate an economically interesting project for the described deposit,
without overpass the arsenic constraints in any period of the schedule. The methodology presented in this
paper is as follows:

 Economic envelop selection

 Pushback selection

3
  Pushback/ Bench grade range discretization
 Optimization model
 Optimization results

Economic Envelope Selection (Final Pit)

The first step towards applying the proposed methodology is to compute the final pit. This final pit would
be used as a hard boundary to limit the search for optimal blending blocks. It is assumed that this envelope
contains a production schedule with specific constraints (arsenic limit by period). Several economic
envelopes were generated using the Lerchs & Grossmann algorithm, considering different block valuation
scenarios, show in Table 1.

Table 1. Scenarios for economic envelope selection.

Scenario As cutoff (upper bound for Ore – Waste differentiation, ppm) Penalties in block valuation
1 - 1
2 - 0
3 200 0
4 300 0
5 400 0
6 500 0

For each one of the previous scenarios, were generated several nested pits varying the copper and gold
prices in fixed proportions (“Revenue Factors”). Then, for different processing plant capacities, were
selected those envelopes that in each case could sustain the firsts 5 years of production. Each one of the
generated scenarios was economically evaluated assuming a constant processing plant feeding with
constant grades in all periods (the total copper and gold fines were not modified in each case), which is a
biased approximation but allows to look rapidly for an economic envelope which potentially could sustain
a 5 years production schedule satisfying the considered arsenic constraints, with an acceptable economic
return. The results are shown in the following figure:
2,000
1,800
1,600
NPV (MUS$)

1,400
1,200
1,000
800
600
400
20 40 60 80 100
Scn. 1 Scn. 2 Scn. 3 Plant Capacity (Kt/d)
Scn. 4 Scn. 5 Scn. 6

Figure 6. Productive scenarios economic analisys.

Figure 6, the production scenarios at 100,000 tonnes per day seems to be the more interesting ones, from
an economic perspective, so, this is a preliminary processing plant capacity to be considered in the
scheduling process. Looking now at the arsenic content of the envelopes selected for each case, the results
were the following:

4
 800 210
As grade in envelope (ppm)

As grade in envelope (ppm)

700 190
600 170
500 150
400 130
300 110
200 90
100 70
0
50
20 40 60 80 100
20 40 60 80 100
Scn. 1 Scn. 2 Scn. 3 Plant capacity (Kt/d) Scn. 3 Scn. 4 Plant capacity (Kt/d)
Scn. 4 Scn. 5 Scn. 6 Scn. 5 Scn. 6

Figure 7. Arsenic content in analysed envelopes.

Since the final pit envelope must potentially be able to sustain a production schedule with the 150ppm
arsenic constraint, scenarios 1 and 2 are discarded. Scenario 6 is also discarded to be considered right on
the limit, leaving little chances to the scheduler. The selected envelope (scenario 5) is specified in the
following table:

Table 2. Selected final pit envelope.

150 Economic envelope
Cu Au As
Scenario Plant capacity (t/d) Total rock (Mt) Ore (Mt) Ore/Waste
(in ore, %) (in ore, ppm) (in ore, ppm)
5 100,000 515.9 184.9 1.79 1.05 0.43 125.0

Additionally to the previous final pit selection, an interesting observation could be made from Figure 6
and Figure 7. The only difference between scenarios 1 and 2 is the consideration of arsenic in concentrate
penalties in independent block valuation for nested pits generation. It can be seen that by only considering
the arsenic in concentrate penalties in block valuation, the Lerchs & Grossmann algorithm is unable to
generate an envelope that could reduce significatively the amount of arsenic in the generated envelopes,
reducing the possibility to construct a feasible production schedule.

Economic Envelope Discretization

The selected final pit was first discretized in five pushbacks, described in Table 3, without considering this
time an Ore – Waste differentiation, since this is going to be a decision made by the MILP model.

Table 3. Considered Pushbacks.

Pushback Total Rock (Mt) Cu (%) Au (ppm) As (ppm)
1 93.5 0.63 0.42 353.33
2 115.0 0.55 0.28 157.07
3 84.8 0.59 0.27 203.26
4 135.0 0.52 0.24 158.39
5 87.7 0.48 0.25 123.65

Each one of the five pushbacks was divided into 26 benches of 12m high each (corresponding to the high
of one block of the original block model). Then, each pushback – bench intersection was divided in seven
copper grade ranges, so each pushback – bench – range (PBR) is characterized by total tonnage, copper
grade, gold grade and arsenic concentration, and correspond to the portions that ad-hoc production
schedule optimizer must allocate in time, this is the typical bin definition to classify the ore tonnage in a
volume. Considered ranges are shown in Table 4.

5
Table 4. Considered copper grade ranges.
Range Cu min (%) Cu max (%)
1 0 0.4
2 0.4 0.8
3 0.8 1.2
4 1.2 1.6
5 1.6 2.0
6 2.0 2.4
7 2.4 2.8

Cut off grade Optimization Model

Parameters

Parameters of the proposed optimization model are the following:

 T: total periods for the schedule.

 I: total number of pushbacks.
 B: total number of benches.
 R: total number of grade ranges.

 𝑡𝑟𝑏𝑖𝑟 : tonnage in range r of bench b of pushback i (t).

 𝑙𝑎𝑠𝑟𝑏𝑖𝑟 : arsenic grade in range r of bench b of pushback i (ppm).
 𝑣𝑎𝑙𝑟𝑏𝑖𝑟 : independent economic value of range r of bench b of pushback i without mining cost
(US$/t).

 𝑑𝑒𝑠𝑐𝑡 : discount factor for cash flow in period t.

 𝑐𝑚: mining cost (US$/t).

 𝑙𝑎𝑠𝑚𝑎𝑥𝑡 : maximum admissible arsenic grade in period t (ppm).

 𝑟𝑒𝑚𝑎𝑥: maximum admissible waste/ore ratio for each period of the schedule.
 𝑟𝑒𝑚𝑖𝑛: minimum admissible waste/ore ratio for each period of the schedule.
 𝑟𝑢: ramp up (proportion of mine capacity for first period).

 𝐼𝑝 : processing plant unitary investment (US$/tpd)

 𝐼𝑚 : mine unitary investment US$/tpd)

Decision Variables

Decision variables of the model are the following:

 𝑡_𝑚𝑜𝑣𝑡 : total tonnage mined in period t.

 𝑡_𝑝𝑟𝑐𝑡 : total tonnage processed in period t.
 𝑡_𝑒𝑠𝑡𝑡 : total waste tonnage mined in period t.

 𝑥𝑏𝑖𝑡 : 1 if bench b of pushback i in open by period t, 0 otherwise.

 𝑦𝑏𝑖𝑡 : 1 if bench b of pushback i is closed by period t, 0 otherwise.
 𝑤𝑏𝑖𝑡 : 1 if bench b of pushback i is open in period t, 0 otherwise.

 𝑝𝑏𝑏𝑖𝑡𝑟 : proportion of range r in bench b of pushback i that is extracted and processed in period t.

6
  𝑛𝑝𝑏𝑏𝑖𝑡𝑟 : proportion of range r in bench b of pushback i that is extracted but not processed in
period t.

Objective Function

The objective function of the model is to maximize the NPV of the project and it looks in the following
way modeled as a function of parameters and decision variables:

𝑀𝑎𝑥𝑖𝑚𝑖𝑧𝑒
𝑇 𝑇
𝑡_𝑝𝑟𝑐𝑡 𝑡_𝑚𝑜𝑣𝑡
− {(𝐼𝑝 ∗ ∑ ) + (𝐼𝑚 ∗ ∑ )}
𝑇 𝑇 [1]
𝑡=1 𝑡=1
𝑇 𝐵 𝐼 𝑅

+ ∑ 𝑑𝑒𝑠𝑐𝑡 ∗ (∑ ∑ ∑ 𝑝𝑏𝑏𝑖𝑡𝑟 ∗ 𝑡𝑟𝑏𝑖𝑟 ∗ (𝑣𝑎𝑙𝑟𝑏𝑖𝑟 − 𝑐𝑚) − 𝑛𝑝𝑏𝑏𝑖𝑡𝑟 ∗ 𝑡𝑟𝑏𝑖𝑟 ∗ 𝑐𝑚)

𝑡=1 𝑏=1 𝑖=1 𝑟=1

Productive Constraints

a) Arsenic constraint:
𝐵 𝐼 𝑅

∑ ∑ ∑ 𝑡𝑟𝑏𝑖𝑟 ∗ 𝑝𝑏𝒃𝒊𝒕𝒓 ∗ 𝑙𝑎𝑠𝑟𝑏𝑖𝑟 − 𝑙𝑎𝑠𝑚𝑎𝑥𝑡 ∗ 𝑡𝑟𝑏𝑖𝑟 ∗ 𝑝𝑏𝑏𝑖𝑡𝑟 ≤ 0 , ∀𝑡 ∈ {1, … , 𝑇} [2]

𝑏=1 𝑖=1 𝑟=1

b) Ramp up in period 1:

𝑡_𝑝𝑟𝑐1 = 𝑟𝑢 ∗ 𝑡_𝑝𝑟𝑐2 , ∀𝑡 ∈ {1, … , 𝑇 − 1} [3]

c) Constant plant feeding since period 2:

𝑡_𝑝𝑟𝑐𝑡 = 𝑡_𝑝𝑟𝑐𝑡+1 , ∀𝑡 ∈ {2, … , 𝑇 − 1} [4]

d) Waste/ore ratio:

𝑟𝑒𝑚𝑎𝑥 ∗ 𝑡_𝑝𝑟𝑐𝑡 ≥ 𝑡_𝑒𝑠𝑡𝑡 , ∀𝑡 ∈ {1, … , 𝑇} [5]

𝑟𝑒𝑚𝑖𝑛 ∗ 𝑡_𝑝𝑟𝑐𝑡 ≤ 𝑡_𝑒𝑠𝑡𝑡 , ∀𝑡 ∈ {1, … , 𝑇} [6]

e) Total processed tonnage:

𝐵 𝐼 𝑅

∑ ∑ ∑ 𝑡𝑟𝑏𝑖𝑟 ∗ 𝑝𝑏𝑏𝑖𝑡𝑟 = 𝑡_𝑝𝑟𝑐𝑡 , ∀𝑡 ∈ {1, … , 𝑇} [7]

𝑏=1 𝑖=1 𝑟=1

f) Total mined tonnage:

𝐵 𝐼 𝑅

∑ ∑ ∑ 𝑡𝑟𝑏𝑖𝑟 ∗ (𝑝𝑏𝑏𝑖𝑡𝑟 + 𝑛𝑝𝑏𝑏𝑖𝑡𝑟 ) = 𝑡_𝑚𝑜𝑣𝑡 , ∀𝑡 ∈ {1, … , 𝑇} [8]

𝑏=1 𝑖=1 𝑟=1

g) Total mined waste tonnage:

𝐵 𝐼 𝑅

∑ ∑ ∑ 𝑡𝑟𝑏𝑖𝑟 ∗ 𝑛𝑝𝑏𝑏𝑖𝑡𝑟 = 𝑡_𝑒𝑠𝑡𝑡 , ∀𝑡 ∈ {1, … , 𝑇} [9]

𝑏=1 𝑖=1 𝑟=1

7
Geometric and logic constraints

a) Mass balance between processed and not processed tonnage:

𝑇

∑ 𝑝𝑏𝑏𝑖𝑡𝑟 + 𝑛𝑝𝑏𝑏𝑖𝑡𝑟 = 1, ∀𝑏 ∈ {1, … , 𝐵}, 𝑖 ∈ {1, … , 𝐼}, 𝑟 ∈ {1, … , 𝑅} [10]

𝑡=1

b) Indicates if benches were opened:

𝑥𝑏𝑖𝑡 ≤ 𝑥𝑏𝑖(𝑡+1) , ∀𝑏 ∈ {1, … , 𝐵}, 𝑖 ∈ {1, … , 𝐼}, 𝑡 ∈ {1, … , 𝑇 − 1} [11]

c) Indicates if benches were closed:

𝑦𝑏𝑖𝑡 ≤ 𝑦𝑏𝑖(𝑡+1) , ∀𝑏 ∈ {1, … , 𝐵}, 𝑖 ∈ {1, … , 𝐼}, 𝑡 ∈ {1, … , 𝑇 − 1} [12]

d) Do not close benches before opening:

𝑤𝑏𝑖𝑡 = 𝑥𝑏𝑖𝑡 − 𝑦𝑏𝑖𝑡 , ∀𝑏 ∈ {1, … , 𝐵}, 𝑖 ∈ {1, … , 𝐼}, 𝑡 ∈ {1, … , 𝑇} [13]

𝑇 𝑇

∑ 𝑥𝑏𝑖𝑡 ≥ ∑ 𝑦𝑏𝑖𝑡 , ∀𝑏 ∈ {1, … , 𝐵}, 𝑖 ∈ {1, … , 𝐼} [14]

𝑡=1 𝑡=1

e) Ranges in each bench can be extracted only if the bench is open:

𝑤𝑏𝑖𝑡 ≥ 𝑝𝑏𝑏𝑖𝑡𝑟 + 𝑛𝑝𝑏𝑏𝑖𝑡𝑟 , ∀𝑏 ∈ {1, … , 𝐵}, 𝑖 ∈ {1, … , 𝐼}, 𝑡 ∈ {1, … , 𝑇}, 𝑟 ∈ {1, … , 𝑅} [15]

f) Bench b+1 of pushback i can be opened only if bench b of pushback i was extracted:
𝑇 𝑇

∑ 𝑥𝑏𝑖𝑡 ≥ ∑ 𝑥(𝑏+1)𝑖𝑡 , ∀𝑏 ∈ {1, … , 𝐵 − 1}, 𝑖 ∈ {1, … , 𝐼} [16]

𝑡=1 𝑡=1

g) To close benches if they are completely extracted:

𝑡

𝑤𝑏𝑖(𝑡+1) ≤ 1 − ∑ 𝑝𝑏𝑏𝑖𝑘𝑟 + 𝑛𝑝𝑏𝑏𝑖𝑘𝑟 , ∀𝑏 ∈ {1, … , 𝐵}, 𝑖 ∈ {1, … , 𝐼}, 𝑡 ∈ {1, … , 𝑇 − 1}, 𝑟 ∈ {1, … , 𝑅} [17]
𝑘=1

h) Bench b+1 cannot be extracted if bench b is not completely extracted:

𝑡

𝑤(𝑏+1)𝑖𝑡 ≤ ∑ 𝑝𝑏𝑏𝑖𝑘𝑟 + 𝑛𝑝𝑏𝑏𝑖𝑘𝑟 , ∀𝑏 ∈ {1, … , 𝐵 − 1}, 𝑖 ∈ {1, … , 𝐼}, 𝑡 ∈ {1, … , 𝑇}, 𝑟 ∈ {1, … , 𝑅} [18]
𝑘=1

i) Bench b of pushback i can be open only if bench b of pushback i-1 is already open:
𝐵 𝐵

∑ 𝑥𝑏𝑖𝑡 ≤ ∑ 𝑥𝑏(𝑖−1)𝑡 , ∀𝑖 ∈ {2, … , 𝐼}, 𝑡 ∈ {1, … , 𝑇} [19]

𝑏=1 𝑏=1

8
Production Schedule and Results analysis

The model was implemented in AMPL and solved using CPLEX Parallel 10.2 (8 threads). Solution time
was 14,432.03 seconds in an 8 processors machine (2 x IntelXeon Cuadcore [email protected], 4Gb
RAM). The obtained solution is analyzed in the following titles.

Production Schedule

The solution found with the production schedule ad-hoc optimizer is shown in the following figures:
140 1.2 140 160

120 120 140

1
100 120
100
Tonnage (Mt/year)

Tonnage (Mt/year)
0.8

Cu (%), Au (ppm)
100
80 80

As (ppm)
0.6 80
60 60
60
0.4
40 40
40
20 0.2 20 20
NPV: 714.9 MUS$ NPV: 714.9 MUS$
0 0 0 0
1 2 3 4 5 1 2 3 4 5
Mineral Waste Cu Au Period (years) Mineral Waste Period (years)
Figure 8. Resulting production Schedule from ad-hoc optimizer (tonnages, grades and Net Present Value).

It can be observed that solution does not exceed the arsenic limit in any period of the schedule, becoming
this solution feasible. The NPV increases to 714.9MUS$, value that captures 85% of the schedule that
does not considers arsenic constraints in its construction (Figure 3), which can be considered as an upper
limit for the value of the deposit.

Mining sequence

Although the pushbacks considered to build the parameters for the ad-hoc production schedule optimizer
were selected manually, fixing with this the sequence of the extraction, the model was able to extract, in
the same period, material from more than 1 pushback, which allows the model to look for materials in
different locations of the deposit in order to satisfy the imposed productive constraints for each period of
the schedule. The found solution it shown in the following figure:

Table 5. Active pushbacks by period (1: active, 0: inactive).

Period 1 Period 2 Period 3 Period 4 Period 5
Pushback 1 1 0 0 0 0
Pushback 2 1 1 0 0 0
Pushback 3 1 1 1 0 0
Pushback 4 0 0 1 1 1
Pushback 5 0 0 0 0 1

From Table 5 it can be observed that in period 1 material is extracted from pushbacks 1, 2 and 3, in period
2 from pushbacks 2 and 3 and so on, showing that the scheduler is following the predefined pushback
sequencing. The following table shows the proportions of ore and waste mined each period from each
pushback:

9
Table 6. Ore and waste tonnage by pushback and period.
Pushback 1 Pushback 2 Pushback 3
Ore (Mt) Waste (Mt) Ore (Mt) Waste (Mt) Ore (Mt) Waste (Mt)
Period 1 24.1 69.4 0.5 1.3 7.8 4.9
Period 2 - - 42.7 70.5 0.5 5.4
Period 3 - - - - 32.9 33.3
Period 4 - - - - - -
Period 5 - - - - - -
TOTAL 24.1 69.4 43.2 71.8 41.2 43.6

Pushback 4 Pushback 5 TOTAL

Ore (Mt) Waste (Mt) Ore (Mt) Waste (Mt) Ore (Mt) Waste (Mt)
Period 1 - - - - 32.4 75.6
Period 2 - - - - 43.2 76.0
Period 3 10.4 30.9 - - 43.2 64.2
Period 4 43.2 50.3 - - 43.2 50.3
Period 5 0.2 - 43.0 44.7 43.2 44.7
TOTAL 53.8 81.2 43.0 44.7 205.2 310.7

It is important to note that the differentiation between ore and waste, or more specifically, material send to
processing and material send to waste dump is not an input for the scheduler (as in many actual
commercial open pit mine planning tools) if not is a result from the scheduling process.

Material selection for processing

A relevant aspect related with the obtained solution is the processing of grade ranges with high
independent economic values (associated to high Cu and Au grades) but arsenic content over than
150ppm, which is cause of the economic valuation of the blend of materials sent to processing plant each
period of the schedule. Despite the final pit was generated considering a 400ppm cutoff grade as an upper
limit to reduce the total amount of arsenic in the envelope, many of the blocks that were considered as
waste in the preliminary valuation for execute Lerchs & Grossmann algorithm, because they exceed the
arsenic cutoff grade, were processed by the optimization model, since in the model, the preliminary
difference between ore and waste does not exist. In the same way, the model processed materials with
negative independent economic value in order to reduce the arsenic grades in other materials of higher
economic value but high arsenic content, which were treated as waste in the traditional methodology. Take
into consideration the following table:

Table 7. Processed tonnage by pushback over 150ppm As, 400ppm As and with non positive independent economic value.
Pushback - Bench - Range (Mt)
As >= 150ppm As >= 400ppm Independent economic value <= 0
Pushback 1 15.1 0.3 0.0
Pushback 2 16.4 1.7 4.0
Pushback 3 15.1 1.2 4.8
Pushback 4 15.3 2.9 1.0
Pushback 5 7.2 0.4 0.0
TOTAL 69.1 6.5 9.8

The second column of the Table 7 shows the total amount by pushback of processed materials with an
arsenic content over 150ppm. This tonnage corresponds to a 33% of the total processed material in the
first 5 productive years and it is clearly a relevant proportion. The third column shows the total amount by
pushback of processed materials with an arsenic content over 400ppm, despite this amount is naturally
lower than the previous, it interesting to note that this material were discarded from processing when the
final pit was chosen.

10
The fourth column of Table 7 shows the total amount, by pushback, of processed materials with negative
independent economic valuation. Certainly, due the optimization model ensures that the found solution is
optimal it can be inferred that the processing of these blocks allows to dilute the arsenic concentrations in
other materials, generating that the overall material has a positive economic value and an arsenic grade
lower or equal than 150ppm. This shows the potential of blending of the tool and how blending becomes
part of the production scheduler.

3 1800
1600
Cu grade (%), Au grade (ppm)

2.5
1400

As grade (ppm)
2 1200
1000
1.5
800
1 600
400
0.5
200
0 0
1 2 3 4 5 1 2 3 4 5
min Cu max Cu min Au max Au Period (years) min As max As Period (years)

Figure 9. Minimum and maximum copper, gold and arsenic grades processed each period in a FBR basis.

Figure 9 shows the minimum and maximum copper, gold and arsenic grades processed each period in a
PBR basis. It can be seen that each period, the minimum processed copper and gold grades where zero,
but it does not means that any PBR with zero copper or gold grade where processed, revealing a very
interesting issue: which was the cutoff grade? Moreover, there is a cutoff grade? Something, similar
happens for arsenic, it can be seen that maximum processed arsenic grade is in each period higher that
arsenic limit, but in the overall, the arsenic constraint is not overpassed.

Conclusions

From the case study analyzed on this paper, the production schedule obtained from the proposed
methodology shows an NPV of 714.9MUS$ for the first five years of production, which correspond to an
improvement of 251% in the feasible scenario found with actual commercial mine planning tools.

The optimal solution found, which considers ore blending before economic evaluation, considers the
processing of 9.8Mt of material which its isolated economic valuation is negative. This material, with low
arsenic content dilutes the high arsenic grades, which are highly correlated with copper and gold in the
deposit, resulting in a blend that does not exceed the arsenic constraints in any period of the schedule.

Mining sequence and material selection for processing is not conditioned by a cutoff grade; they should be
determined in order to maximize an objective function with certain productive constraints. In other words,
the destiny of each mined part of the deposit is a consequence of its location (pushback-bench) and the
constitutive characteristics (grades, grindability, metallurgical recovery and others) of itself and its
neighbors in space and time; to mine and process a specific block could be economic depending on the
blocks which are extracted in the current and other periods.

Finally, the results of this analysis show the importance of having adequate planning tools that lead to the
ability to look at the potential economic value of a mineral deposit.

Acknowledgments

To the University of Chile and Delphos Mine Planning Laboratory, to support the publication of the
results associated to the research presented in this paper

11
References

KING, B., 1999. Cash Flow Grades – Scheduling Rocks with Different Throughput Characteristics.
Optimizing with Whittle, Third Biennial Conference, 7p 103-109.

WHITTLE, J., WHARTON, C., 1995. Optimizing Cut-Off Grades. Optimizing with Whittle, 11p 159-
169.

ESFANDIARI, B., ARYANEZHAD, B.M., ABRISHAMIFAR, A., 2004. Open pit optimization including
mineral dressing criteria using 0-1 non-linear goal programming, Mining Technology (Trans. Inst. Min.
Metall. A), Vol. 113, 13p A3-A15.

RAMAZAN, S., DIMITRAKOPOULUS, R., 2004. Recent applications of operations research and
efficient MIP formulations in open pit mining, Transactions of the Society for Mining, Metallurgy, and
Exploration, Vol. 316, 6p 73-78.

12