You are on page 1of 36

Accepted Manuscript

Title: Optimal design and planning of heap leaching process. application to copper oxide leaching Author: Jorcy Y. Trujillo Luis A. Cisternas Edelmira D. G alvez Mario E. Mellado PII: DOI: Reference: To appear in: Received date: Revised date: Accepted date: 18-7-2012 9-6-2013 25-7-2013 S0263-8762(13)00323-7 http://dx.doi.org/doi:10.1016/j.cherd.2013.07.027 CHERD 1328

Please cite this article as: Trujillo, J.Y., Cisternas, L.A., G alvez, E.D., Mellado, M.E., OPTIMAL DESIGN AND PLANNING OF HEAP LEACHING PROCESS. APPLICATION TO COPPER OXIDE LEACHING, Chemical Engineering Research and Design (2013), http://dx.doi.org/10.1016/j.cherd.2013.07.027 This is a PDF le of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its nal form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

OPTIMAL DESIGN AND PLANNING OF HEAP LEACHING PROCESS. APPLICATION TO COPPER OXIDE LEACHING
Jorcy Y. Trujillo3, Luis A. Cisternas2,31, Edelmira D. Glvez1,2, Mario E. Mellado2
1

Submitted to: Chemical Engineering Research and Design Date: July 12, 2012

ABSTRACT

Although the process of heap leaching is an established technology for treating minerals, such as copper, gold, silver, uranium and saltpeter, as well as remediating soil, no studies to date have investigated process optimization. This work presents a methodology for the design and planning of heap leaching systems to optimize the process. This methodology consists of the creation of a superstructure that represents a set of alternatives to search for the optimal solution; from this superstructure, a mixed integer nonlinear programming (MINLP) model was generated, and a BARON-GAMS solver was used to find the optimal solution. This method was applied to the extraction of copper from systems with one, two and three heaps, and the effects of copper price, ore grade and other variables were analyzed for each system. From the results, it can be concluded that this methodology can be used to optimize heap leaching processes, including planning and design issues.

Corresponding author: L.A.Cisternas, e-mail: lcisternas@uantof.cl

Ac ce

pt

ed
1

an

Department of Metallurgical Engineering, Universidad Catlica del Norte, Antofagasta, Chile 2 Centro de Investigacin Cientfico Tecnolgico para la Minera (CICITEM), Antofagasta, Chile 3 Department of Chemical Engineering, Universidad de Antofagasta, Antofagasta, Chile

us

cr

ip t
Page 1 of 35

Keywords: Heap leaching, process design, process optimization

NOMENCLATURE
process unit}. , heap leaching unit}. , unit of solvent extraction}. , species to be extracted} Variables and Parameters Area of heap. [m2] Recovery constants for disjunction model.

Costs. [MUS$]

Cost of building a heap . [MUS$/cycle] Total fixed cost of heap j. [MUS$/cycle] Acid consumption at infinite time. [kg/ton] Acid consumption of heap j. [kg/ton]

Ac ce

Initial acid consumption. [kg/ton] Acid consumption of heap j. [kg/cycle] Total acid consumption. [kg/cycle] Variable cost of heap j. [MUS$/ton] Operational cost before leaching of heap j. [MUS$/ton] Operational cost after leaching. [MUS$/ton] Linear availability of species Planning time horizon. [days]
2

pt

ed

Acid consumption constants for disjunction model.

in heap . [ton/m]

an

us
Page 2 of 35

cr

ip t

Sets

Income. [MUS$] Mass flow from process unit to of species . [ton/cycle]

Input mass flow to process unit of species . [ton/cycle] Output mass flow from process unit of species . [ton/cycle]

Molecular weight of species k. [ton/ton-mol] Mass of mineral loaded on heap of species . [ton] Constant in the big M method. Number of cycles. Production of species . [ton/cycle] Price of species . [MUS$/ton] Price of acid. [MUS$/kg]

Output volumetric flow from the SX unit. [m3/cycle]

Recovery of species

Cycle time of the heap system. [days]

Ac ce

End time of leaching of heap . [days] Profits. [MUS$] Weight factor for income I. Weight factor for cost C. Concentration of stream from process unit to of species . [ton/m3]

Concentration of output stream from the SX unit of species . [ton/m3] Binary variable of the cycle times. Disjunctive binary variable. Height of heap. [m]
3

pt

Recovery at infinite time. [%]

ed

Recovery from heap of species . [%] in disjunctive model. [%]

Volumetric flow from process unit

an

to . [m3/cycle]

us
Page 3 of 35

cr

ip t

Molecular weight of acid. [ton/ton-mol]

Greek letters Recovery constants for Mellado et al. (2011) model. Acid consumption constant for the analytical model. Ore density. [ton/m3] Grade of species k in the ore. [%] Superscript Lower bound . Upper bound. Subscript Disjunctive.

Ac ce

pt

ed
4

an
Page 4 of 35

us

cr

ip t

1. INTRODUCTION
Heap leaching is a hydrometallurgical process that is widely used in the mining industry, in which low-grade metals, such as copper, gold, silver and uranium, are extracted; the process also has new applications in the treatment of non-metallic minerals, such as studies have been performed with the goal of improving the leaching process; these works include searching for the ideal operational conditions to achieve better extraction of metals. However, most modeling-based research has focused on the search for models based on first principles (Dixon and Hendrix, 1993a; Dixon and Hendrix, 1993b) and semi-empirical models (Mellado et al., 2009; Mellado et al., 2011). An analysis of the relationship between the operational time and costs and the identification of an optimal operation time and heap operations, no previous studies have analyzed the optimization of the entire system. Heap leaching is a mineral processing technology in which piles of ore (crushed or run-ofmine rock) are leached with various chemical solutions to extract valuable minerals. Large tonnages are involved, and metal is recovered over a long period. As shown in Figure 1, a the extraction of metal begins. Usually an aqueous cyanide solution is used for precious metals, and an aqueous sulfuric acid solution is used for copper ores. The metal is recovered from the pregnant solution by a recovery process that depends on the metal to be extracted. For example, solvent extraction (SX) followed by electrowinning is used for copper, and activated carbon or precipitation with zinc are used for gold (Gupta and Mukherjee, 1990). In this work, an SX process was considered as the recovery process. The barren solution from the recovery process is recirculated to the heap system. A heap system can include several heaps, and several irrigation networks can be used. Figure 1: Hydrometallurgical process based on heap leaching The objective of this research was to develop a methodology that allows design and planning to simultaneously find the optimal operational conditions, such as recovery and leach time, and the optimal design parameters, such as heap height. The methodology is
5

Ac ce

pt

ed

barren solution is pumped to the pile surface and sprayed, and as it percolates downward,

an

size have been performed (Padilla et al., 2008); however, despite the importance of heap

us

cr

ip t

saltpeter (Valencia et al., 2008), and soil remediation (Carlsson et al., 2005). Several

Page 5 of 35

based on a superstructure that represents the flowsheet alternatives and a mathematical model that is solved through a mixed integer nonlinear programming (MINLP) approach. The development of the model with its superstructure and corresponding equations is provided in section 2; the application of the model to systems with one, two and three heaps is provided in section 3, with the corresponding characteristics and sensitivity analysis. Finally, the last section includes the conclusions.

The design and planning of heap leaching systems is a coupled problem. Padilla et al. (2008) studied the economic optimization of this process by analyzing variables such as the leaching time and heap height in a copper mineral treatment plant and found that the design economic standpoint because these variables affect both the recovery and operational capacity of the leaching process. The purpose of this work is to develop a methodology to design and plan heap leach systems. To accomplish this goal, a superstructure representing a set of flowsheet mathematical model was developed using mass balances and their corresponding operational conditions to obtain a model that can determine the values of variables and parameters affecting the process. The resulting model corresponds to an MINLP. The superstructure was built based on heap leaching and solvent extraction units. The alternatives to look for the optimum solution was constructed, based on which a (heap height) and operational planning (leaching time) are coupled problems from an

Ac ce

pt

ed
6

schematic in Figure 2 represents these units, where

superstructure, a mixer is used at the input of the unit (square) and a divider is used at the output of the unit (triangle) to represent the set of connection possibilities between the different units. The rectangle in the middle of the figure represents the unit itself, either a heap leaching or SX unit. This type of representation has been used in other problems, e.g., water networks (Castro and Teles, 2013) and reactor networks (Silva and Salcedo, 2011). Figure 2: Process units used in the modeling

an

represents a process unit. In this

us

cr

2. MATHEMATICAL MODEL

ip t
Page 6 of 35

To develop the mathematical model, the following sets are defined. First, the set of units is defined as leaching unit} and species to be extracted is represented as is a process unit}, following which the subsets J , heap

, solvent extraction unit} are defined. The set of .

evaporative water losses are compensated for with fresh water so that the water losses and fresh water flow rates are null. The water balance and the cost of water losses were not considered because these balances and costs depend on weather conditions, heap irrigation technology, and evaporation losses/evaporation mitigation technologies in solution pools. b) The heaps are considered to be in a series, where heap heap

. c) During solvent extraction, the output concentration of the valuable species is . This concentration is assumed constant and known

assumed to be known and constant,

because it depends on downstream operations, in this case from the electrowinning stage. solution or loss of valuable species. d) A constant density is assumed in the dissolutions.

2.1 Mass Balances Mass balance in j heaps

Equation 1 shows the mass balance of species

pt

ed
in heap j in the liquid phase as a function

Ac ce

of the input and output mass flow rates and the recovery in the heap:

Additionally, it is assumed that this stage is ideal and does not represent entrainment of

an

us

has been operated longer than

cr

where

is the output mass flow of species

in heap , [ton/cycle], and

the recovery of the valuable species

in the leaching heap process, which depends on the , along with other variables. The term

leaching time and the height of the heap

, [%/cycle] is included in equation 1 because the heaps operate in series, and the recovery of the heap j, , begins where the recovery of the heap j-1,
7

ip t
(1)

To perform the corresponding balances, the following assumptions were considered. a) The

represents

, ends.

Page 7 of 35

can be calculated using an adequate model, for example, the model proposed by Lizama et al. (2005) for bioleaching, Mellado et al. (2011) for low grade ores, or Glvez et al. (2012 ) for Caliche ores. Note that input solution of heap , the superscript o ; therefore, the superscript represents the and

represents the output solution of heap in heap , given by the following:

area of the heap grade .

multiplied by the density of the ore

The balance of acid consumption is given by equation (3), where the first term represents the consumption of acid in heap j, and the second term is the credit for the recovery of acid in the electrowinning stage.

ed

an

us

In equation 2,

is the linear availability of species , [ton/m] which corresponds to the with a corresponding ore

cr

where

is the acid consumption of heap j by cycle [kg/ton of ore cycle] and MWk is the

The total acid consumption is the sum of the acid consumption for each heap, that is, (4)

Mass balance in the mixers and dividers in unit For the mixers and dividers, the mass balance per component was performed based on the input and output mass flows rates of the different process units. There is a mixer and a divider on each process unit.

Ac ce

molecular weight of species k.

pt

ip t
(2) (3)
Page 8 of 35

represents the mass of the valuable species

(5)

(6)

on the production of the valuable species

[ton/cycle], and the input and output mass

an

flows to the process unit, as shown in the following equation:

us

The mass balance for species

in the solvent extraction process unit was performed based

cr

Mass balance in SX

2.2 Planning and operation constraints

To design and plan the heap leaching process, certain restrictions must be considered. First, the number of cycles throughout the planned time horizon must be obtained as follows:

pt

ed

where

Ac ce

is the planned time horizon [days],

is the number of cycles and represents the

cycle time. A cycle consists of all operation phases of a heap, e.g., if a plant uses two heaps, the cycle consists of the operation of both heaps. The time horizon is the length of time used to evaluate the alternatives over the same period of time. The cycle time is the period required to complete one cycle of heap operation. In this work it is assumed that overlapping operation is used, this is, simultaneous heaps can be operated. Then, the cycle time is the maximal operation time between all heap units, calculated as follows:

ip t
(7) (8)
Page 9 of 35

(9)

Note that if a non-overlapping operation is used, then the cycle time is the sum of the operation times of all heap units. Heap starts and finishes the operation at times be written as a disjunction expression: and , respectively. Equation (9) can

cr
the big M method (Biegler et al. 1997) as follows:

In this work, equation (10), the disjunctive expression for the cycle time, is represented by

an
.

us ed Ac ce pt
where operation time. Here, the constraint in equation (12) only allows one choice of the inequalities redundant if and enforced if
10

, is a binary variable that represents the selection of the heap with the maximal . M is a

large parameter introduced in the right side of equations (11), (13) and (14), which renders

ip t
(10)
(11) (12) (13) (14)
Page 10 of 35

The following operational bounds must be considered to achieve satisfactory results. These restrictions include lower and upper bounds in the mass flows, the height of the heap, the recovery, the number of cycles and the cycle time as follows:

us an M ed Ac ce pt
where (14) is tUP. 2.3 Specifications
11

cr

. Then, based on equation (17), a good value of M in equations (11), (13) and

ip t

(15)

(16)

(17)

(18)

(19)

(20)

(21)

Page 11 of 35

In the following discussion, the expressions to calculate some bounds and parameters are given. First, the model uses the mass flow rates of species k; however, in industry, concentration and volumetric flow rates are used. The relationship between these variables is as follows:

stream moving from process unit to unit from process unit to

, and

represents the volumetric flow rate is the maximal flow of species , this

[m3/cycle]. Because

parameter can be calculated as the sum of the output flows of all of the SX units as follows:

an
the mass flows is given by

us

cr

where

represents the concentration of species

in mass per volume [ton/m3] in the

In addition, from equation (22), the balance for the solvent extraction unit as a function of

ed

Ac ce

where

corresponds to the output volumetric flow rate from the SX unit and at the output of the SX unit.

pt

corresponds to the concentration of species

2.4 Objective Function

The expression to be maximized is an economic optimization that must be simultaneously maximized in terms of income and minimized in terms of costs. The income is represented as the total production multiplied by the price, and the cost is the number of cycles produced multiplied by the cost of each heap over the time horizon.

12

ip t
(22) (23) (24)

Page 12 of 35

A possible expression is represented in equation (25), where the costs; and are weight factors; and

is the income;

represents .

represents the profits when

Income, cost and profits can give different optimal solutions (Novak Pintari and Kravanja, 2006), and therefore different weight factor values can be analyzed. Maximize

The income,

[MUS$], can be represented in terms of production, the number of cycles,

[MUS$/ton of metal], which includes the cost of the purification process (e.g., solvent extraction or ion exchange) and the product recovery process (e.g., electrowinning or crystallization), and represents the cost after the leaching is performed. The income is given by

an

us

the price of the species to be extracted

and the post-leaching cost

cr

In addition, the cost,

[MUS$], is given by the number of cycles, the cost of building and

operating heap , and acid consumption, as shown in the following equation:

pt

ed

where

Ac ce

is the acid price in [MUS$/kg]. [MUS$/cycle], is defined as a function

Then, the cost of building and operating heap

of the different costs of the heap leaching system according to the following expression:

13

ip t
(25) (26) (27) (28)
Page 13 of 35

where

represents the total fixed cost, including the cost per square meter and per cycle is the variable cost per ton of ore and per cycle [MUS$/ton of ore cycle]

in the process;

in the process that includes the power, water consumption, land preparation, maintenance and labor; and is the cost before leaching per ton of ore and per cycle in the

process [MUS$/ton of ore cycle], including costs such as mine operation, crushing and agglomeration.

2.5 Recovery Expressions

In this work, two methods are used to express the recovery of copper, but different models expression is used to approximate the recovery using straight lines, and an analytical model following manner:

ed

developed by Mellado et al. (2011) is then used. The disjunctive expression is written in the

an

can be used depending on the leaching technology and the leached ore. First, a disjunctive

us

cr
(29) (30) (31)
14
Page 14 of 35

The disjunctive expression of recovery in equation (29) is expressed using the Convex Hull method (Biegler et al. 1997):

Ac ce

The recovery is expressed by straight lines for different time ranges and heap heights.

pt

ip t

(32)

(33)

cr
where equation, is the binary variable, and and the operation time for each disjunction, respectively. Mellado et al. (2011) can be written as follows:

are constants used to approximate the recovery from the straight line

time and size kinetics scales and includes variables such as particle radius, heap height, as variables, keeping other variables constant. Then, the analytical expression proposed by

pt

ed

irrigation rate and porosities. Here, only the heap height and leaching time are considered

Mellado et al. (2011) developed an empirical knowledge-based model, which considers two

an

indicate the lower and upper bounds of

us

Where heap

Ac ce

represents the recovery at t = infinity, which is a function of the height of the

as follows:

The parameters of equations (29) and (37) are shown in Table 1.

15

ip t
(36)
(37)

(34) (35)

Page 15 of 35

Table 1: Parameters considered in the analytical model by Mellado et al. (2011) for copper recovery and the analytical model (eq. 39) for acid consumption.

Figure 3 shows the graphic representation of equations (27) and (34) for the disjunctive Similarly as for copper recovery, two ways of expressing the acid consumption were used. First, a disjunction expression based on straight lines for different time ranges and heap written as follows:

us
Convex Hull method (Biegler et al. 1997):

cr

heights was used, and then an analytical model was used. The disjunctive expression is

The disjunctive expression for acid consumption in equation (38) is expressed using the

ed

an
(39)

pt

where

The analytical model is

Ac ce

are constants.

The parameters of equations 39 and 41 are given in Table 1. The disjunctive model of acid consumption (Eqs. 39 and 40) is used together with the disjunctive model of copper recovery (Eqs. 30 to 35). In the same way, the analytical model
16

ip t
(40) (41)

model and the model by Mellado et al. (2011).

(38)

Page 16 of 35

of acid consumption, Eq. 41, is used together with the analytical model of Mellado et al. (2011), Eq. 36.

3. CASE STUDIES
systems of one, two and three heaps to analyze the effects of variables such as copper price, heap of 200,000 m2 with an ore grade of 0.9% copper. The copper price is assumed to be leaching system. Other ore characteristics and heap costs are indicated in Table 2, and they are shared by all examples. The costs used in this work are those normally observed in the context of copper production in Chile and were updated from the work of Schmidt (2001). Table 2: Parameters considered in the case studies.

We must consider that the variable cost, the pre-leaching cost and the total cost in systems with two and three heaps apply only to the first heap. In other words, the costs of heaps two and three are zero because the heaps operate in series and because their costs have been two and three can be neglected. considered in the first heap; furthermore, it is assumed that the operational cost of heaps In the following sections, the three studied cases are shown, which were implemented in GAMS and solved by BARON using an AMD Athlon II Dual-Core M300 2.00 GHz processor. In all studied cases only one parameter was changed, while all other parameters were kept constant and equal to the values in table 2. Figure 3: Recovery as a function of time with heap heights of 6 and 9 m.

3.1 One-heap leaching system

Ac ce

pt

ed

M
17

an

us

7,700 US$/ton. A time horizon of 360 days was used to analyze, design, and plan the

cr

heap size, ore grade, acid price and variable cost on the process. The case studies consider a

ip t

This section illustrates the application of the model for the copper heap leaching process in

Page 17 of 35

The one-heap leaching case corresponds to the application of the model to a heap leaching system with the corresponding variables, parameters (Table 2), limits and equations. For this system, the effects of the copper price, ore grade, heap area, acid price and variable cost were studied using both the disjunctive model and the model by Mellado et al. (2011). Copper prices significantly affect the profits of the process; higher copper prices result in a larger income and therefore greater profits. However, variables such as recovery, cycle time, and heap height did not change. The second and third columns in Table 3 show the model by Mellado et al. (2011). effect of the copper price on the profit calculated using both the disjunctive model and the

Table 3: Profit fluctuations based on the copper price and different leaching systems using the disjunctive model (DM) and Mellado et al. (2011) model.

The heap area does not affect the height of the heap, number of cycles, recovery or cycle time; however, it does affect profits, as the production of copper increases with heap size. When the effect of the ore grade on the system is analyzed, a strong impact on the cycle grade increases. Nevertheless, no changes in the heap height are observed (9 m for the disjunctive model and 9.5 m for the Mellado et al. model, for all values of ore grade). The second and third columns in Tables 4 and 5 show the effect of the ore grade on the number of cycles and the recovery, respectively. Figure 4 shows the effect of the ore grade on profits, using both the disjunctive model and the model by Mellado et al. (2011). In the disjunctive model, cycle times and recoveries are constant for ore grades of 0.5 and 0.9%, whereas in Mellado et al. (2011), model cycle times and recoveries decrease with ore grade. This difference between the models can be explained by the underestimation of recoveries in the disjunctive model, as seen in Figure 3. As the feed grade increases, higher copper production is achieved because lower leaching time decreases recovery but allows more leaching cycles throughout the time horizon. time and recovery is observed, along with an effect on the profits, which increase as the ore

Ac ce

Figure 4: Profit fluctuations for ore grades of 0.5, 0.9 and 1.3%.
18

pt

ed

an

us

cr

ip t

Page 18 of 35

Table 4: Cycle time fluctuations based on the ore grade and different leaching systems using the disjunctive model (DM) and Mellado et al. (2011) model. Table 5: Recovery fluctuations as a function of the ore grade for the different leaching systems using the disjunctive model (DM) and Mellado et al. (2011) model (M). NC=Does not converge.

The effect of the variable cost (assuming constant expenses for acid consumption), including expenses for power, water consumption, ore transportation, maintenance and labor, is significant. If the variable cost is increased, the cycle time and recovery increase proportionally, and profits and income decrease. Figure 5 shows the effect of the variable disjunctive model and the Mellado et al. (2011) model. Additionally, the second and third columns in Table 6 show the profits for both models. cost on the cycle time (Figure 5a) and recovery (Figure 5b), calculated for both the

Figure 5 Fluctuations in cycle times (a) and recovery (b) at 0.005, 0.01 and 0.02 kUS$/ton of ore. Table 6: Profit fluctuations based on the variable cost and different leaching systems using the disjunctive model (DM) and Mellado et al. (2011) model. The effect of the acid price was studied separately from other variable costs. When the observed: cost increases and profit decreases as the acid price increases. Nevertheless, no change in the height of the heap is observed (9 m for the disjunctive model and 9.5 m for the Mellado et al. model, for all values of acid price). For cycle time and recoveries, no changes are observed for disjunctive model. On the other hand, cycle time and recoveries increase as acid price increases for the Mellado et al. model. More details are given in table 7. Table 7: Effect of acid price for the Disjunction and Mellado et al. models.

Ac ce

effect of the acid price on the system is analyzed, an impact on the cost and profit is

pt

ed

M
19

an

us

cr

ip t
Page 19 of 35

3.2 Two- and three-heap leaching systems The models are applied to a system formed by two- and three-heap leaching units to compare its behavior to the single heap system and to observe the effect of parameters such as copper price, heap area, ore grade, acid price and variable cost. The results obtained in these systems show a trend similar to that obtained for the single heap leaching system. For example, when the copper price increases, the profits increase in systems with one, two and effect of the copper price, using the disjunctive model and the model by Mellado et al. three heaps. Table 3 shows some of the results obtained for the three systems based on the (2011).

characterize. Table 6 shows some of the results obtained for the three systems based on the effect of the variable cost for the disjunctive and Mellado et al. (2011) models; the results heap; this is true for most cases, except for the Mellado model with three heaps. therefore, the number of cycles increases; the only difference in these cases is the shorter leaching time. With more heap leaching units in a system, the leaching time in each heap is using both the disjunctive model and the Mellado et al. (2011) model. As shown in Tables Table 4 shows that for a two-heap system and an ore grade of 0.5%, the cycle time for each heap is 40 and 49 days, and the entire two-heap system operates for approximately 80 and 98 days, for the disjunctive model and the Mellado et al. model, respectively. These values are similar to the values for a one-heap system. In addition, the recovery decreases as the ore grade increases in the different systems. Table 5 shows the results obtained for the three systems based on the ore grade for the disjunctive model and the model by Mellado et al. (2011). As shown in Table 5, in all systems, the recovery is low. For example, the recovery in heap two for the two-heap system with an ore grade of 0.9% is 44 and 42% for the disjunctive
20

Ac ce

3, 4 and 5, there are convergence problems in both models.

pt

shorter. Table 4 shows the results obtained for the three systems based on the cycle times

ed

As in the single heap system, if the ore grade is increased the cycle time decreases, and,

show that in a system with more than one heap, the profits are higher with each additional

an

In addition to the copper price, the variation of profits with other parameters is important to

us

cr

ip t

Page 20 of 35

model and the Mellado et al. model, respectively. To analyze this situation, a minimal recovery value of 70% to 90% was assigned to the model. The results of this test indicate that the model obtains high recoveries; however, there are variations in the cycle time and profits. The cycle times increase significantly, and the profits decrease compared to the results shown in Table 5. Therefore, in some cases the maximal recovery is not necessarily the best measure of operational efficiency, based on economic considerations. As in the single heap system, if the acid price is increased the cost increases, and therefore disjunctive model and the Mellado et al. (2011) model. It can be observed that the number
both models, leaching times and recoveries are approximately constant with the acid price.

Table 8. Effect of acid price for the disjunction model and Mellado et al. (2011) model. NC=Does not converge.

In all of the systems, the heap height calculated using the disjunctive model was 9 m, the in the model was obtained, which was not affected by parameters such as the copper price and ore grade. However, a test showed that if the upper bound of the height is increased in the Mellado et al. (2011) model, the optimal height obtained is below the upper bound

pt

ed

maximal height allowed. With the Mellado et al. (2011) model, the maximal height allowed

because if the heap height

is increased, the recovery

M
21

Finally, Figure 6 shows the effect of the weight factors on the income and costs calculated with the Mellado et al. (2011) model. The values in the graph indicate the weight factors . The point indicates the minimal cost, and as a result, the indicates the maximal income, and as a

minimal income. Additionally,

result, the maximal cost. The income increases significantly as the income weight factor increases until . For an income weight between 0.5 and 0.6, the cost increases but between 0.6 and 0.7, the income increases as

the income is almost constant. Then, for

the income weight factor increases; however, the cost also increases.

Ac ce

an

us
decreases.

of heaps has a greater effect than the acid price on the total leaching time and recoveries. In

cr

the profit decreases. Table 8 shows the results obtained for the three systems using both the

ip t

Page 21 of 35

As shown in these studies, parameters such as copper price, ore grade and variable cost have an important effect on the profit as well as on some process variables. In addition, the results of the disjunctive and Mellado et al. (2011) models are generally similar. The results of the Mellado et al. (2011) model are slightly better than those of the disjunctive model because the disjunctive model is an approximation to the Mellado model, and the disjunctive model underestimates the recovery. However, the Mellado et al. (2011) model has more convergence problems than does the disjunctive model. Convergence problems and local optima can be attributed to the fact that the model is a mixed integer nonlinear programming problem; it is nonconvex due to exponential terms equations 1 and 2). To solve these problems, linearization (Quesada and Grossmann, 1995) disjunctive model presents fewer convergence problems compared to the Mellado model. The disjunctive model is a linearization of the latter. For bilinear expressions, the discretization approach can be used. (recovery in the Mellado et al., 2011 model) and bilinear terms in the mass balance (e.g., and discretization (Pham et. al, 2009) techniques can be used. This may explain why the

M
and

an

us

cr

4. CONCLUSIONS

In this work, a method has been developed to plan and design a heap leaching system. Two on the model by Mellado et al. (2011). The developed mathematical model is a MINLP. As determined in the copper leaching cases that were studied, one of the primary variables affecting the profit of the process is the copper price, as expected. Additionally, certain other variables are shown to be important, such as the ore grade, acid price and the variable cost, as they significantly affect the operation planning and the profit. The cycle time and recovery decrease, but the profit increases with increasing ore grade. In addition, the cycle time and recovery increase, but the profit decreases with increases in variable costs. The cycle time decreases in systems with more than one heap. Moreover, the use of more than
22

Ac ce

models were used to calculate the recovery, one based on disjunction and the other based

pt

ed

Figure 6 Income and cost fluctuations based on the weight functions

ip t
.

Page 22 of 35

one heap allows better control of the irrigation rate in each heap. These results show that as the number of heaps increases, there is more flexibility in planning and a greater production capacity. This study has shown that the design (heap height) and planning of a heap leaching operation are coupled problems from an economic perspective, and therefore, these variables must be included in the model. This coupling occurs because these are interactive heap leaching process. factors, and these variables affect both the recovery and the capacity of the operation in the

In addition, the results of the analytical model by Mellado et al. (2011) show higher that the trend in the results is similar. Additionally, the CPU time required to obtain results is less for the Mellado et al. (2011) model than for the disjunctive model. Because the algorithm. system converges to local optima, there is room for improvement in the computing

for the design and planning of heap leaching systems, and future studies must include the effect of other variables, such as the particle size, irrigation rate and acid concentration, to predict the optimal operational conditions.

Acknowledgments and 1090592.

The authors wish to thank CONICYT for its support through the Fondecyt Project 1090406

REFERENCES
Biegler, L. T., Grossmann, I. E., & Westerberg, A. W., (1997). Systematic methods of chemical process design. NJ: Prentice Hall.
23

Ac ce

pt

ed

Finally, this study should be considered as a new step in the development of methodologies

an

precision and convergence than those obtained using the disjunctive model, despite the fact

us

cr

ip t

Page 23 of 35

Carlsson, E., Buchel, G., (2005). Screening of residual contamination at a former uranium heap leaching site. Thuringia.,Germany Chemie der Erde, 65S1, 7595. Castro, P.M., Teles, J.P. (2013). Comparison of global optimization algorithms for the design of water-using networks. Computers & Chemical Engineering 52, 249 261

more solid reactants from porous ore pellets. Metallurgical Transactions, 24B, 157-168. Dixon, D.G. and Hendrix, J.L., (1993b). General model for leaching of one or more solid

Glvez, E.D., Moreno, L., Mellado, M.E., Ordez, J.I., Cisternas, L.A. (2012). Heap comparisons, Minerals Engineering, 33, 46-53

Gupta C.K., Mukherjee T.K. (1990). Hydrometallurgy in Extraction Processes, Vol I, CRC Press.

Lizama, H.M., Harlamovs, J.R., McKay, D.J., Dai, Z. (2005). Heap leaching kinetics are 630.

Mellado, M.E., Casanova, M.P., Cisternas, L.A., Glvez E.D. (2011). On scalable analytical models for heap leaching. Computers and Chemical Engineering, 35, 220-225. Mellado, M.E., Cisternas, L.A., Glvez E.D. (2009). An analytical model approach to heap leaching. Hydrometallurgy. 95, 33-38.

Ac ce

proportional to the irrigation rate divided by heap height. Minerals Engineering, 18, 623-

pt

ed

M
24

leaching of caliche minerals: Phenomenological and analytical models Some

an

reactants from porous ore pellets. Metallurgical Transactions, 24B, 157-168.

us

cr

Dixon, D.G. and Hendrix, J.L., (1993a). A Mathematical model for heap leaching of one or

ip t

Page 24 of 35

Novak Pintari Z., Kravanja, Z. (2006). Selection of the economic objective function for the optimization of process flow sheets. Ind. Eng. Chem. Res. 45, 4222-4232. Padilla, G.A., Cisternas, L.A., Cueto, J.Y.(2008). On the optimization of heap leaching. Minerals Engineering. 21, 673-678. Pham V., Laird C., El-Halwagi M.M. 2009. Convex hull discretization approach to the

multicomponent flows. Computer and Chemical Engineering, 19, 1219-1242. Schmidt, P. (2001). Criterios de diseo de un proceso hidrometalurgico para minerales de cobre oxidados. Trabajo de titulacin, Universidad de Chile. Chile. Silva, H.G., Salcedo, R.R., (2011). SIMOP: Application to global MINLP stochastic

Valencia J.A., Mendez. DA., Cueto J.Y., Cisternas L.A., (2008). Saltpeter extraction and

Ac ce

pt

modeling of caliche mineral heap leaching. Hydrometallurgy. 90, 103-114.

ed

optimization. Chemical Engineering Science 66, 13061321.

M
25

an

us

Quesada, I., Grossmann I.E. (1995). Global optimization of bilinear process networks with

cr

global optimization of pooling problems. Ind. Eng. Chem. Res. 48, 1973-1979.

ip t

Page 25 of 35

Table 1: Parameters considered in the analytical model of Mellado et al. (2011) for copper recovery and analytical model (eq. 39) for acid consumption. Parameters Copper recovery model (Eq. 36) Recovery Constant Recovery Constant Infinite Recovery Constant Infinite Recovery Constant Infinite Recovery Constant Acid consumption model (Eq. 39) Acid consumption constant Acid consumption constant Infinite acid consumption Initial acid consumption

Table 2: Parameters considered in the case studies. Parameters Apparent ore density Ore grade Time horizon Price of copper Area of heap Price of acid Post leaching cost, Pre leaching cost, Variable cost, Total fixed cost, Copper mass flow rate at the output of SX

ed

M Ac ce pt
Values 0.162 [kUS$/ton of acid] 0.16675 [MUS$/ton of copper] 0.006074 [MUS$/ton of ore] 0.0028554[MUS$/ton of ore] 0.00095 [MUS$] 3,500
26

an

us

0.0125 1 70.96 kg/ton ore 25 kg/ton ore

cr
Page 26 of 35

1 0.03

ip t

Table 3: Variation of profits based on the copper price and the different leaching systems using the disjunctive model (DM) and Mellado et al. (2011) model. NC=Does not converge.

Price[MUS$/ton] 0.0055 0.0077 0.0099

One Heap DM 174.8 344.0 513.2 Mellado 181.6 369.2 566.5

Two Heaps DM 603.0 907.6 1,212.3 Mellado 663.7 1,001.9 1,340.6

DM

Ore grade[%]

One Heap DM
80 80 40 97 65 55

M
DM
40 40 31 27

Cycle time[Days]

an
Two Heaps
49 31 28

Table 4: Variation of the cycle times based on the ore grade and different leaching systems using the disjunctive model (DM) and Mellado et al. (2011) model. NC=Does not converge.

us
Mellado

0.5 0.9 1.3

ed

Mellado

cr
DM
NC 27 NC

NC 1,032.0 1,539.5

Table 5: Variation of recovery as a function of the ore grade for the different leaching systems using the disjunctive model (DM) and Mellado et al. (2011) model (M). NC=Does not converge. Recovery % Heap 1 DM M 22 NC 22 NC 13 NC Three Heaps Heap 2 DM M 46 NC 46 NC 38 NC Heap 3 DM M 62 NC 62 NC 50 NC

Ore grade[%] 0.5 0.9 1.3

Ac ce

One Heap Heap 1 DM M 62 68 62 55 38 49

pt

Two Heaps Heap 1 Heap 2 DM M DM M 20 22 44 45 20 19 44 42 20 19 44 42

ip t
Three Heaps Mellado NC NC NC Three Heaps Mellado
42 22 NC
Page 27 of 35

Copper

Profits[MUS$]

Table 6: Variation of profits based on the variable cost and different leaching systems using the disjunctive model (DM) and Mellado et al. (2011) model. Profits[MUS$] Variable Cost One Heap Two Heaps Three Heaps [kUS$/ton] DM Mellado DM Mellado DM Mellado

Heap height Leaching cycles Profits [MUS $] Recovery [%] Cycle time [days] Income [MUS$]

Acid Price [kUS$/ton of acid] 0.11 0.162 0.21 0.11 0.162 0.21 Mellado et al. Model Disjunction Model 9.00 9.00 9.00 9.50 9.50 9.50 4.50 4.50 4.50 5.81 5.53 5.31 379,458 344,004 311,278 411,925 369,118 330,635 62 62 62 53 55 56 80.00 80.00 80.00 61.99 65.09 67.81 579,241 579,241 579,241 673,520 662,199 652,159

Table 8. Effect of acid price for the disjunction model and Mellado et al. (2011) model. NC=Does not converge.

DM Mellado et al. Acid Price One Heap Two Heaps Three Heaps One Heap Two Heaps Three Heaps [kUS$/ton] Profits[MUS$] 0.11 379.5 926.0 1,138.4 411.9 1,022.8 NC 0.162 344.0 907.6 1,032.0 369.1 1,002.0 NC 0.210 311.3 890.7 933.8 330.6 982.8 NC Total Leaching Time [Days] 0.11 80.0 50.8 80.0 62.0 45.5 NC 0.162 80.0 50.8 80 65.1 45.5 NC 0.210 80.0 50.8 80.0 67.8 45.5 NC Total Recovery [%] 0.11 62 44 62 53 42 NC 0.162 62 44 62 55 42 NC 0.210 62 44 62 56 42 NC

Ac ce

pt

ed

M
28

an

us

Table 7: Effect of acid price for Disjunction and Mellado et al. models.

cr
Page 28 of 35

ip t

0.005 0.01 0.02

400.1 330.2 192.5

446.2 351.5 203.3

799.3 479.4 385.1

892.4 703.0 406.5

1,198.3 990.7 462.7

1,198.3 1.054.5 303.2

Mine

Ore Spent Ore

Metal

Figure 1: Hydrometallurgy process based on heap leaching

Ac ce

pt

ed
29

an

us
Page 29 of 35

Pregnant solution

Recovery Process

Barren solution

cr

ip t

Heaps

Ac ce

pt

ed
30

an
Page 30 of 35

us

cr

Figure 2: Process units used in the modeling

ip t

Figure 3: Recovery as a function of time with heap heights of 6 and 9 m.

Ac ce

pt

ed
31

an

us
Page 31 of 35

cr

ip t

Figure 4: Variation of profits for ore grades of 0.5, 0.9 and 1.3%.

Ac ce

pt

ed
32

an

us
Page 32 of 35

cr

ip t

a)

Ac ce

pt

Figure 5: Variation of cycle times (a) and recovery (b) at 0.005, 0.01 and 0.02 kUS$/ton of ore.

ed
b)
33

an
Page 33 of 35

us

cr

ip t

Figure 6: Variation of income and costs based on the weight functions

an

us
and .

Ac ce

pt

ed
34

cr
Page 34 of 35

ip t

This work presents a methodology for the design and planning of heap leaching systems. The method used a superstructure that represents a set of alternatives to search for the solution. A MINLP model was generated and a BARON-GAMS solver was used to find the optimal solution It can be concluded that this method can be used to optimize heap leaching planning and design.

Ac ce

pt

ed
35

an
Page 35 of 35

us

cr

ip t

You might also like