Comparative study of approaches based on the Taguchi and ANOVA for optimising

the leaching of copper–cobalt flotation tailings

Bienvenu I. Mbuya1, Méschac–Bill Kime2, Alexis M.D. Tshimombo3

1
Departement de Métallurgie, Faculté Polytechnique, Université de Likasi, Likasi,
République Démocratique du Congo, 2Department of Metallurgy, University of
Johannesburg, Auckland Park, South Africa, 3Departement de Métallurgie, Faculté
Polytechnique, Université de Lubumbashi, Lubumbashi, République Démocratique du
Congo

Corresponding author: Méschac–Bill Kime, E-mail: meschackime@gmail.com

Abstract

Approaches based on Taguchi and analysis of variance (ANOVA) methods have been

proposed for studying the leaching of copper–cobalt flotation tailings. Taguchi method

was used to determine the optimum leaching conditions as well the most influential

operating parameters. ANOVA was used to determine the relationship between the

experimental conditions and yield levels, as well as to define the significances of

parameters on the leaching yields. An L25 (55) orthogonal array experimental plan was

used to assess the effects of initial acidity (25, 50 , 75, 100 and 125 g/L), leaching time

(30, 60, 90, 120 and 150 min.), temperature (25, 35, 45, 55 and 65°C), pulp density

(S/L=10, 15, 20, 25 and 30 solid percent ) and ferrous iron concentration (0.0, 1.0, 2.5,

5.0, and 7.5 g/L) on individual dissolution of Cu and Co. Under the optimum conditions

(50 g/L, 60 min, 45 ° C, 15%, 2.5 g/L for Cu and 100 g/L, 60 min, 65 ° C, 15% and 2.5

g/L for Co), leaching yields were 95.98% Cu and 97.74% Co. pH and Fe2+ were found to

be key influential parameters of Cu and Co, respectively.

1
KEYWORDS: Taguchi, ANOVA, copper, cobalt, tailings, flotation, leaching

1. INTRODUCTION

Owing to their unique physical, chemical and mechanical properties, copper and cobalt

are extensively used in various engineering applications. Copper is a ductile metal with

very high thermal and electrical conductivity. Its main usages include electrical

equipment such as wiring and motors, electromagnets, and gold alloys. Cobalt is a hard

and lustrous metal and is mainly obtained as a by–product of copper and nickel mining

activities. Cobalt is used in the manufacturing of aircraft’s engine parts, permanent

magnets, glasses and ceramic pigments. The Katanga Province in the Democratic

Republic of Congo hosts the northwestern part of the Central African Copperbelt. This

region abounds with large and rich deposits of sulphide and oxidised copper–cobalt ores

(Kime et al., 2016). Due to the depletion of high–grade copper–cobalt mineralisation of

the Central African Copperbelt and processing environmental considerations, slag and

tailings materials are the subject of many reclamation studies for the recovering of metals

such as copper and cobalt (Ramakokovhu et al., 2012; Lutundula et al., 2013). This is

mostly aimed at minimising waste disposal and at the same time generating revenues.

Reclamation of tails from old metallurgical extraction processes has also been identified

as key to the lifespan extension of mines (Ramakokovhu et al., 2012). The interest in

exploiting man–made tailings dams is not only getting the minerals back to production, it

is also encouraged by reduced operational costs (Antonijevi´c et al., 2008) since they

consist of mined and comminuted metal bearing materials (Lutundula et al., 2013). The

Katanga province of the Democratic Republic of Congo has been the stronghold of

2
intensive copper and cobalt mining operations in the last decades. These operations

generated great amount of tailings piles scattered throughout the province that were often

improperly disposed or simply abandoned by lack of a compelling mining legislation.

These tailings represent a real threat to the ecosystem and human health as they are

subjected to weathering and erosion through exposure to the rainfall and winds that lead

to the contamination of surrounding areas and the aquifer (Lutundula et al., 2013;

Andrews et al., 2013). It is, therefore, important to quell this threat through the

continuous research and development towards a cost effective processing of the tails.

The Kambove concentrator is a business unit of Gecamines that processes about 120,000

t/month of both copper and cobalt sulphide and oxidised ores (Progogine, 1972; Kime et

al., 2015; Mulenga and Mwashi, 2016). The plant receives three different types of

copper–cobalt ores mainly from three mines: Shangolowe, grading about 2.90% Cu and

0.12% Co (Kime et al., 2015); Kamfundwa, grading an average of 2.5% Cu and 0.3% Co;

as well as Kamoya Sud II at 2.7% Cu and 0.5% Co (Kanda, 2013). The industrial

flowsheet involves obtaining a bulk concentrate of copper and cobalt by flotation (Fig. 1)

that is either sold or processed by hydrometallurgy or pyrometallurgy.

The concentrator adopts the regime of the following reagents: sodium silicate, Na2SiO3

(66%SiO2, 250g/t) as the dispersant; Senfroth, G41 (10 g/t) as the frother; sodium

carbonate, Na2CO3 (99.99%, 50g/t) as the pH modifier; sodium hydrosulfide, NaHS (70–

72%, 3500 g/t) as the sulphidisation agent. Potassium amylxanthate, PAX (95%

minimum, 350 g/t) is the primary collector and the mixture (150g/t) of gasoil (93–94%)–

3
rinkalore (6–7%) is the secondary collector. The surface sulphidisation mechanism is

done through the hydrolysis of sodium hydrosulfide as depicted in the following

reactions.

NaHS + H2O →NaOH + H2S (1)

NaOH → Na+ + OH- K = 10-7 (2)

H2S H+ + HS- K = 10-11 (3)

HS- H+ + S2- K = 10-34 (4)

As can be seen from Eqs. (2–4), the dissociation constants of H2S and HS- are very low

compared to the dissociation constant of NaOH. This means that an alkaline pulp will be

obtained, where OH-, HS- and S2- will react with valuable minerals. The general Equation

for minerals sulphidisation can be given as follows:

MO(s) + S2-(aq) + 2H+(aq) MS(s) + H2O (5)

Eq. (5) yields a bulk concentrate of copper and cobalt and a tail that is stockpiled over

the years pending further processing.In general, the copper–cobalt tailings generated have

complex mineralogy. In fact, they combine different gangue minerals such as cupriferous

micas, calcites, siliceous, feldspar and iron bearing oxide gangue materials. These

materials are known for increasing acid consumption which is accompanied by impurity

co–extraction and operational cost incline (Whyte et al., 2001). The dissolution of Cu and

Co in H2SO4 aqueous solution is carried out according to the following Equations

(Habashi, 1970; Philibert et al., 2002).

CuO + H2SO4 CuSO4 + H2O (6)

4
CuCO3.Cu(OH)2 + 2H2SO4 2CuSO4 + CO2 + 3H2O (7)

CuSiO3.2H2O + H2SO4 CuSO4 + SiO2 + 3H2O (8)

CoO + H2SO4 CoSO4 + H2O (9)

Co2O3 + 2FeSO4 + 3H2SO4 2CoSO4 + Fe2(SO4)3 + 3H2O (10)

The leaching of Cu is effectively achieved in H2SO4 aqueous solution whereas the

leaching of Co can only successfully be achieved in the presence of reducing agents like

FeSO4.7H2O. In fact, cobalt high oxides such as Co2O3 require the addition of a reducing

agent to the leaching medium to enhance their dissolution (Philibert et al., 2002; Kime

and Makgoale, 2016). Considering the more or less refractory nature of tailings from

copper–cobalt flotation operations, it is important to evaluate the operating conditions

under which the dissolution of copper and cobalt is optimised. Many parameters can

influence the behaviour of the leaching of tailings. The most important are chemical

composition, leaching solution, solid/liquid (S/L) ratio, leaching time and pH (Lutandula

et al., 2013). Taguchi method has been applied in several previous works to determine

the optimum leaching conditions (Çalban et al., 2005; Safarzadeh et al., 2007; Guo et al.,

2010; llyas et al., 2010; Nkulu et al., 2013; Popescu et al., 2015). Taguchi method’s

advantages lie in that it helps optimising parameters in reducing the number of trials, a

time gain and a minimum cost (Phadke, 1988). ANOVA statistical method can be used in

synergy with Taguchi method to determine the relative importance of each parameter

(Hvalec et al., 2004). Taguchi and ANOVA are the two statistical methods that will be

used in this work. They will be applied to the leaching of the copper–cobalt flotation

tailing considering an L25(55) orthogonal experimental design. The current study used the

5
statistical software package Statistica v13, compatible with Microsoft Windows 8

operating system.

2. MATERIALS AND METHODS

2.1. Sample Preparation

The copper–cobalt flotation tailings from the Kambove concentrator were used for this

study. Particle size distribution (PSD) analysis of the tailings materials was performed

using a series of ASTM sieves between 300 and 38 µm. Given the fineness of the grains,

wet sieving was preferred to dry sieving. Nine fractions were hence obtained. The

chemical analysis of the size fractions was carried out using both Atomic Absorption

Spectrometry (AAS) and Atomic Emission Spectrometry using Inductively Coupled

Plasma (ICP–AES).

2.2. Leaching Testwork

Leaching tests were carried out in 600 mL glass beakers that were carefully cleaned with

distilled water. H2SO4 aqueous solutions were prepared by mixing distilled water with

selected amounts of H2SO4 (98%, p.a., Merck®) and FeSO4.7H2O (98%, p.a., Merck®)

as the reducing agent. A known amount of the sample was then poured into the beaker

containing the aqueous solution. The assembly was placed on a hot plate equipped with a

mechanical stirring device. The temperature was monitored using a thermometer that was

placed permanently in the solution. After completion of the leaching time, the leachates

and residues were analysed using AAS and ICP–AES.

6
2.3. Experimental Plan

a) Taguchi method

Following parameters were considered for the leaching experimental testwork: initial

acidity (A), Time (B), Temperature (C), Pulp density (D) and Concentration of Fe2+ (E).

Table 1 lists the leaching parameters and their levels.

The fractional experimental plan based on the L25(55) orthogonal array was used in order

to conduct the testwork. The L25(55) orthogonal array experimental plan comprised 5

parameters associated each to 5 variation levels. Possible interactions between factors in

the orthogonal matrix were not considered. The order of tests was obtained by inserting

the parameter values in the L25(55) orthogonal array experimental plan. The L25(55)

orthogonal array experimental plan is shown in Table 2. This array displays in the

columns, integer values that represent the levels of factors and in the rows, the values

representing a set of specific levels for each factor (Çalban et al., 2005; Safarzadeh et al.,

2007; Guo et al., 2010; llyas et al., 2010; Nkulu et al., 2013; Popescu et al., 2015).

The philosophy of the robust Taguchi design method lies in minimising the impact of

confounding factors by identifying the optimal combination of controllable parameters in

order to make the process more robust. This amounts to jointly optimise the mean results

and dispersion results around the calculated mean (Droesbeke et al., 1997). Taguchi

method takes then into account the dispersion of results by integrating the analysis of

additional information like the signal–to–noise ratio (Madhav and Phadke, 2010). It also

uses a quality loss function to measure the variability in the performance characteristics

7
around the targeted robustness value. The value of the quality loss function is

transformed into the Signal–Noise ratio (S/N), which is a performance indicator

(Bernard, 2000). The value of the performance indicator can be obtained directly from

the quality loss function as follows (Safarzadeh et al., 2007; Goupy, 1996):

S/N=10log (L) (11)

where L is the quality loss and SN is expressed in dB (decibel)

Generally, there are 3 categories characterising the performance in the analysis of the

ratio S/N: the minimiser, the maximiser and the target value (Phadke, 1988). The largest

S/N ratio corresponds to the best performance. Therefore, the optimum level of a

parameter is the one with the largest S/N ratio (Demirci et al., 2011). The performance

characteristics are evaluated using following expressions (Droesbeke et al., 1997):

The higher, the better:

1 1
SN L 10log (12)
n Yi2

The lower, the better:

1
SN S 10log Yi2 (13)
n

where SNL and SNS are performance characteristics, n is the number of repetitions of the

performance for the experimental combination and Yi is the value of the performance of

the ith experience.

The value of the experiment performance can be predicted using the following prediction

function (Safarzadeh et al., 2007).

8
 T T T
YOpt Ai Bi ... (14)
n n n

where n is the total number of tests, T is the sum of all the test responses and Ai, Bj,… are

average responses of levels i, j,…

b). Analysis of variance (ANOVA)

ANOVA is the second statistical approach that was used. Unlike what its name suggests,

ANOVA does not consider differences in variances amongst populations but mean

differences (Vivier, 2002). This method takes its name from its use of variance measures

to judge the significance of test results. In other words, the significance of the average

differences measured amongst populations (Rabier, 2007). Generally, the principle of

ANOVA consists in subdividing the total variation in a factorial component relative to

the regression equation or model and a residual component. The first component is tested

with respect to the second one. The factorial and residual components are mathematically

represented by mean squares (Bernard, 2005). Ultimately, the interest in using ANOVA

is to help testing in an absolute manner the influence of factor variations of a given

response. Equations resulting from ANOVA are well presented in Mohd et al. (2009);

Laffly (2007); and Hvalec et al. (2004). ANOVA will be useful to determine statistically:

the most significant parameters, the contribution and the degree of contribution of each

factor on the metal dissolutions.

3. RESULTS

Chemical analysis results of the sample are shown in Table 3. It can be seen that about

73.66% Cu oxide and 86.67 % Co contained in the tailings sample was in the oxide form,

9
the remaining percentages were in sulphide form. As can be seen from Fig. 2, the

chemical assay of the size fractions showed that well above 50.00% of the sample was

less than 38 μm. The chemical analysis revealed that this fraction had a high content of

Cu and Co and contained about 0.55 % Cu and 0.51% Co.

The mineralogical examination results revealed that Cu was mostly in the form of

malachite (Cu2CO3(OH)2). Copper was also found in low proportions in the forms of

chrysocolla (CuSiO3.2H2O), chalcocite (Cu2S) and pseudomalachite (Cu5-

(PO4)2(OH).H2O). Cobalt was mainly found in the form heterogenite (aCoO.bCo2O3).

The gangue was constituted of limonite (Fe2O3.H2O), dolomite (CaMg(CO3)2) and silica

(SiO2).

Results from the test conducted according to the L25 (55) orthogonal array experimental

plan are shown in Table 4.

Table 5 provides the average S/N ratio of factors for each variation level.

Since higher S/N ratio is better in Cu–Co stirred leaching (Safarzadeh et al., 2007; Nkulu

et al., 2013), the leaching is optimised when responses are as large as possible. Fig. 3

shows the numerical value of the maximum point corresponds to the optimised value of

the extraction of Cu according to the experimental design. It can be seen that the

optimum for Cu corresponds to the levels A2B2C3D2E3 (letters represent the parameters

and indexes represent the levels). The optimum leaching conditions obtained for Cu were

10
50 g/L; 60 min; 45 °C; 15 solid% and 2.5 g/L and correspond to the initial acidity (A),

time (B), temperature (C), pulp density (D) and amount of ferrous iron (E), respectively.

The optimum leaching conditions for Co are given in Fig. 4. One can notice that the

parameterisation that provides the optimum cobalt recovery is A4B2C5D2E3 and it

corresponds to the followings: 100 g/L; 60 min; 65 °C; 15 % solids et 2.5 g/L for the

initial acidity, time, temperature, pulp density and the amount of ferrous iron,

respectively.

In order to examine the validity of the results obtained using Statistica v13, an

experiment was performed under the optimum leaching conditions. The results are shown

in Table 6. It can be seen that the leaching yields of Cu and Co were 95.98% and 97.74%,

respectively, and their respective total acid consumptions were 116.71 kg/dmtu (dry

metric tonne unit) and 123.20 kg/dmtu. The predictive model (Eq. 14) applied under

optimum conditions for the leaching of Cu and Co gave the leaching yields of 96.05% Cu

and 97.98% Co.

Taguchi oriented practitioners often use ANOVA to determine the factors that influence

the average response and signal–to–noise ratio. So the statistical analysis of variance

(ANOVA) was performed to see whether the process parameters were statistically

significant. The ANOVA results are given in Tables 7 and 8, which shows the

quantitative contribution of each factor on the quality characteristics. To better illustrate

the contribution of each parameter on the performance of Cu and Co leaching process,

11
the percentage contributions of all factors to the leaching yields of copper and cobalt are

plotted as Pareto charts in Figs. 5 and 6.

ANOVA results of Cu leaching yields are shown in Table 7. It can be seen that all the

Fexp obtained were lower than F0.05;4;4 = 6.3882 read in the Ficher–Senedor tables. In other

words, no parameter was significant. The initial acidity contributed with about 20.83% on

the leaching efficiency of Cu, compared to 16.78% and 11.86% of the contributions of

temperature and pulp density. The remaining percentage (38.02%) was the contribution

of the residue on the leaching yield of cobalt. The residue term or error represents the

contribution of uncontrollable factors and factors that were not considered in this study.

However, the robustness of the method of Taguchi helps minimising the influence of this

term.

ANOVA results on Co leaching yields are shown in Table 8. It can be seen that the

concentration of Fe2+ had a Fexp = 76.7183 much higher than the F0.05;4;4 =6.3882 read in

Ficher–Senedor tables. This means that the concentration of Fe2+ was the most important

factor for the dissolution of Co. The value of the critical probability of the concentration

of Fe2+, related to F was 0.000492. This corresponded to the smallest value obtained and

therefore Fe2+ was the most significant factor. Its contribution to the leaching efficiency

of Co was 67.87%, against only 2.81% of the initial acidity.

4. DISCUSSIONS

12
Two major differences were noticed in the dissolutions of Cu and Co. Values of 50 g/L

for the initial acidity and 45 °C for the temperature were obtained for Cu, whereas values

of 100 g/L for the initial acidity and 65 °C for the temperature were obtained for Co. This

hypothetically suggested that the Cu and Co leaching processes were governed by

different phenomena. This hypothesis was confirmed with results in Table 5 where

acidity (∆ = 0.20882) and concentration of ferrous ions (∆= 3.81655) were the most

influential parameters for Cu and Co, respectively. Moreover, Fig. 6 obtained using the

metric values of the SN functions (Table 5) shows that the presence of outliers for acidity

and temperature. A re–parameterisation of these two parameters was considered in order

to eliminate the aberrant points but the results did not impact on the dissolution yields.

Therefore the atypical and extreme points obtained were statistically consistent with the

results obtained. During the validation process, the L25 (55) orthogonal array experimental

plan considered helped to achieve a prediction within 0.07% (Cu) and 0.24% (Co) of the

actual test results. The small discrepancies could be related to certain uncontrollable

variables not taken into account in this work. This emphasises the importance of

expanding the number of variables in order to further improve the results.

5. CONCLUSION

This study was intended to optimise the leaching of Cu–Co flotation tailings in H2SO4

aqueous solution using the Taguchi and ANOVA statistical approaches. Using the

Taguchi L25(55) orthogonal experimental design and considering the initial acidity,

leaching time, temperature, pulp density and ferrous iron concentration as controllable

parameters, the optimized conditions for the leaching of a copper–cobalt flotation tailings

13
were calculated as 50 g/L, 60 min, 45 ° C, 15%, 2.5 g/L for Cu and 100 g/L, 60 min, 65 °

C, 15% and 2.5 g/L for Co. The leaching yields were 97.74% Cu and 94.66% Co.

Moreover, using ANOVA, it was verified that the most effective parameters for the

dissolution of Cu and Co from the copper–cobalt flotation tailings were the pH and

concentration of Fe2+, respectively. These two factors contributed to about 20.83% and

67.87% in the leaching yields of Cu and Co, respectively. These results clearly showed

that the phenomena involved in the dissolution of Cu and Co from the tailings were

different. It is therefore recommended that a kinetic study is further undertaken.

However, the findings of this work are gratifying due to improved Cu and Co extraction

from the flotation tailings.

REFERENCES

1. Andrews, J.W., Moreno, C.J.G., and Nairn, R.W. (2013). Potential recovery of

aluminum, titani–um, lead, and zinc from tailings in the abandoned Picher mining district

of Oklahoma. Miner. Econ., 26(1):61–69.

2. Antonijevi´c, M.M., Dimitrijevi´c, M.D., Stevanovi´c, Z.O., Serbula, S.M. and

Bogdanovic, G. (2008). Investigation of the possibility of copper recovery from the

flotation tailings by acid leaching. Journal of Hazardous Materials, 158(1):23–34.

3. Bernard, C. (2000). Conception robuste : Les concepts et la méthodologie de

Taguchi dans le design de produits. École Polytechnique de Montréal et Génistat

Conseils Inc.

4. Bernard, C. (2005). Modèles d’analyses de la variance avec Statistica, Génistat

Conseils Inc.

14
5. Çalban, T., Çolak, S. and Yeilyurt, M. (2005). Optimization of leaching of copper

from oxidized copper ore in NH3–(NH4)2SO4 medium. Chemical Engineering

Communications,192(11):1515 — 1524

6. Demirci, M.T., Samanci, A., Tarakçioglu, N., Asiltirk, I. (2011). Optimisation of

fatigue life parameters with Taguchi method. 6th International Advanced Technologies

Symposium (IATS’11), 16–18, Elazig, Turkey, pp. 453– 456.

7. Droesbeke, J.J., Fine, J., Saporta, G. 1997. Plans d’expériences, Application à

l’entreprise”, Technip, Ed Paris.

8. Goupy, J. (1996). La méthode des plans d’expériences – Optimisation du choix

des essais et de l’interprétation des résultats, Ed. Dunod, Paris.

9. Guo, D.C., Li, K.Y., Li, Y.L., Tan, H., and Wang, L.Y. (2013). SYNTHESIS

AND CHARACTERIZATION OF CARBAZOLE–BASED AZO ACYLHYDRAZONE

SCHIFF BASES. Chemical Engineering Communications, 200(11):1503–1512.

10. Habashi, F. (1970). Principles of extractive metallurgy. Volume 1 et 2.

11. Hvalec M., Gorsek A. and Glavic P. (2004). Experimental design of

crystallization processes using Taguchi method. Acta Chim. Solvenia, 51:245– 256.

12. Ilyas, S., Bhatti, H.N., Bhatti, I.A., Sheikh, M.A., and Ghauri, M.A. (2010).

Bioleaching of metal ions from low grade sulfide ore: process optimization by using

orthogonal experiment array design. African Journal of biotechnology, 9(19):2801 –

2810.

13. Kanda, J.N. (2013). Étude de la flottabilité de la malachite à l’aide de

l’amylxanthate de potassium et des acides gras – Cas d’étude : Flottation du minerai

15
oxydé de Kamfundwa au Katanga en RD Congo (in French). PhD thesis, Université de

Liège, Belgium.

14. Kime, M.B. and Makgoale, D. (2016). Characterization of copper–cobalt ores and

quantification of Cu2+, Co2+, Co3+, and Fe3+ in aqueous leachates using UV–visible

spectrophotometry. Chemical Engineering Communications, 203(12):1648–1655.

http://dx.doi.org/10.1080/00986445.2016.1230102

15. Kime, M.B., Ntambwe, J., and Mwamba, J. (2015). Laboratory evaluation of the

flotation response of a copper cobalt oxide ore to gasoil–rinkalore mixtures. International

Journal of Environmental, Chemical, Ecological, Geological and Geophysical

Engineering, 9(3):252–258.

16. Laffly, D. (2007). Analyse bivariée de variables qualitative et quantitative :

Analyse de Variance (ANOVA), in French. Laboratoire Société Environnement

Territoire LSEV.

17. Lutandula, M.S. and Maloba, B. (2013). Recovery of cobalt and copper through

reprocessing of tailings from flotation of oxidised ores. Journal of Environmental

Chemical Engineering, 1(1) :1085–1090.

18. Madhav, S., and Phadke, M.S. (2010). Introduction to Robust Design (Taguchi

Method), Publication scientifique. Accessed from

https://www.isixsigma.com/methodology/robust–design–taguchi–method/introduction–

robust–design–taguchi–method/

19. Mohd, A.L., Mohd, R. and Nurul, A. (2009). The implementation of Taguchi

method on EDM process Tungsten Carbide. European Journal of Scientific Research, 26

(4):609 – 617.

16
20. Mulenga, F.K. and Mwashi, A.M.M., (2016). Preliminary assessment of the

comminution circuit of the Kambove concentrator, Democratic Republic of Congo. New

technology and innovation in the Minerals Industry Colloquium Emperors Palace, 9–10

June 2016. Southern African Institute of Mining and Metallurgy

21. Nkulu, G., Gaydardzhiev, S., and Mwema, E. (2013). Statistical analysis of

bioleaching copper, cobalt and nickel from polymetalic concentrate originating from

Kamoya deposit in the Democratic Republic of Congo. Minerals Engineering 48:77–

85 ·

22. Phadke, M. S. (1988). Quality Engineering Using Design of Experiment,

Quality Control, Robust Design and Taguchi Method. Wardsworth and Books,

California.

23. Philibert, J., Vignes, A., Brecet, Y., and Cambrage P. (2002) .Métallurgie du

minerai au Matériau (cours et exercices corrigés, in French), Dunod, Paris.

24. Popescu, I.A., Varga, T., Fogarasi, S., Imre–Lucaci, Á., and Ilea, P. (2015).

Statistical Evaluation of Factors Affecting the Leaching Process of Waste Electrical and

Electronic Equipment using Sodium Persulfate. Chemical Engineering Communications,

203(3):414–423.

25. Progogine, A. (1972). Accroissement de la production du cuivre dans la

République du Zaïre: Rôle joué par les concentrateurs (in French). Académie royale des

Sciences d’Outre–Mer, Classe des Sciences Techniques, N.S., XVII–4, Brussels.

26. Rabier, F. (2007). Modélisation par la méthode des plans d’expériences du

comportement dynamique d’un module IGBT utilisé en traction ferroviaire (in French).

Thèse de doctorat, Institut National Polytechnique de Toulouse, Paris.

17
27. Ramakokovhu, M.M., Kasaini, H., Mbaya, R.K.K. (2012). Leaching

Characteristics of Upgraded Copper Flotation Tailings. International Journal of

Chemical, Molecular, Nuclear, Materials and Metallurgical Engineering, 6(8):853–857.

28. Safarzadeh, M.S., Moradkhani, D., Ilkhchi, M.O., Golshan, N.H. (2007).

Determination of the optimum conditions for the leaching of Cd–Ni residues from

electrolytic zinc plant using statistical design of experiments. Separation and Purification

Technology, 58(3):367–376

29. Vivier, S. (2002). Stratégie d’optimisation par la méthode des plans d’expériences

et application aux dispositifs électrotechniques modélisés par éléments finis (in French).

Thèse de doctorat, Université des sciences et technologie de Lille.

30. Whyte, R.M., Schoeman, N. and Bowes, K.G. (2001). Processing of Konkola

copper concentrates and Chingola refractory ore in a fully integrated hydrometallurgical

pilot plant circuit. The Journal of the South African Institute of Mining and Metallurgy,

101(8):427–436.

18
Table 1 Experimental parameters and quantitative values for each level

Parameters Levels

1 2 3 4 5

Initial acidity(g/L) - A 25 50 75 100 125

Time (min) - B 30 60 90 120 150

Temperature (°C) - C 25 35 45 55 65

Pulp density(%) - D 10 15 20 25 30

Concentration of Fe II(g/L) - E 0 1 2.5 5 7.5

19
Table 2 L25(55) orthogonal array experimental plan applied to the leaching

N° of experiment Parameters and their levels

A B C D E

1 1 1 1 1 1

2 1 2 2 2 2

3 1 3 3 3 3

4 1 4 4 4 4

5 1 5 5 5 5

6 2 1 2 3 4

7 2 2 3 4 5

8 2 3 4 5 1

9 2 4 5 1 2

10 2 5 1 2 3

11 3 1 3 5 2

12 3 2 4 1 3

13 3 3 5 2 4

14 3 4 1 3 5

15 3 5 2 4 1

16 4 1 4 2 5

17 4 2 5 3 1

18 4 3 1 4 2

19 4 4 2 5 3

20 4 5 3 1 4

20
21 5 1 5 4 3

22 5 2 1 5 4

23 5 3 2 1 5

24 5 4 3 2 1

25 5 5 4 3 2

21
Table 3 Chemical composition of the tailing samples

Sample 1 Sample 2 Average

Cutotal (%) 1.56 1.55 1.56

Cuoxide (%) 1.09 1.14 1.21

Cototal (%) 0.33 0.25 0.29

Cooxide (%) 0.27 0.23 0.25

Fetotal (%) 2.51 2.36 2.44

Mn (%) 0.18 0.14 0.16

Ni (%) 0.01 0.01 0.01

Zn (%) 0.62 0.80 0.71

SiO2 (%) 64.97 62.45 63.71

MgO (%) 3.38 6.81 5.1

Al2O3 (%) 3.06 3.84 3.45

22
Table 4 Experimental plan and leaching experimental results

No. CONTROLLABLE FACTORS RESPONSE S/NL

Acidity Time T° Density Conc. Cu Co Cu Co

(g/L) (min) (°C) (% Fe2+ (%) (%) (dB) (dB)

Solids)

1 25 30 25 10 0 91..53 49..04 39..23 33..81

2 25 60 35 15 1 96..52 87..53 39..69 38..84

3 25 90 45 20 2..5 95..89 87..39 39..64 38..83

4 25 120 55 25 5 94.72 87.40 39.53 38.83

5 25 150 65 30 7.5 92.79 87.10 39.35 38.80

6 50 30 35 20 5 95.94 87.54 39.64 38.84

7 50 60 45 25 7.5 97.17 90.28 39.75 39.11

8 50 90 55 30 0 93.64 57.74 39.43 35.23

9 50 120 65 10 1 97.83 93.00 39.81 39.37

10 50 150 25 15 2.5 98.29 95.11 39.85 39.56

11 75 30 45 30 1 95.98 87.67 39.64 38.86

12 75 60 55 10 2.5 95.62 90.58 39.61 39.14

13 75 90 65 15 5 96..54 92..56 39..69 39..33

14 75 120 25 20 7..5 95.00 88.06 39.55 38.90

15 75 150 35 25 0 97.12 57.88 39.75 35.25

16 100 30 55 15 7.5 96.07 90.35 39.65 39.12

17 100 60 65 20 0 97.75 68.55 39.80 36.72

18 100 90 25 25 1 95.10 90.65 39.56 39.15

23
19 100 120 35 30 2.5 95.19 92.26 39.57 39.30

20 100 150 45 10 5 96.59 90.24 39.70 39.11

21 125 30 65 25 2.5 96.57 92.62 39.70 39.33

22 125 60 25 30 5 95.33 87.45 39.58 38.83

23 125 90 35 10 7.5 96.12 92.86 39.66 39.36

24 125 120 45 15 0 96.05 63.63 39.65 36.07

25 125 150 55 20 1 95.36 92.52 39.59 39.32

24
Table 5 Marginal average as function of factor levels for the leaching of Cu and Co

Levels Parameters for Cu

A B C D E

1 39.48766 39.57276 39.55686 39.60135 39.57166

2 39.69588 39.68800 39.66141 39.70765 39.65929

3 39.64983 39.59583 39.67588 39.64396 39.67307

4 39.65783 39.62296 39.56153 39.65721 39.62926

5 39.63492 39.64658 39.67044 39.51595 39.59284

S/N ratio A2=50 g/L B2=60 min C3=45 °C D2=15 % E3=2.5 g/L

expected Solid

under

optimum

conditions

Influence 0.20822 0.11524 0.11902 0.1917 0.10141

of

parameter

Interaction 0.03x + 0.01x + - 0.01x+39.5 0.0012x+39.6

between 39.55 39.6 0.02x+39.69 9 2

parameter R² = 0.26 R² = 0.09 R² = 0.24 R² = 0.11 R² = 0.0021

s Linear

Fit

25
Levels Parameters for Co

A B C D E

1 37.82276 37.99332 38.05078 38.15734 35.41713

2 38.42387 38.53017 38.31892 38.58548 39.10833

3 38.29471 38.37832 38.39585 38.52307 39.23368

4 38.67892 38.49359 38.32885 38.33469 38.98922

5 38.58472 38.40958 38.71058 38.2044 39.05663

S/N ratio A4=100 g/L B2=60 min C5=65 °C D2=15 % E3=2.5 g/L

expected Solid

under

optimum

conditions

Influence 0.85616 0.53685 0.6598 0.42814 3.81655

of

parameter

Interactio 0.178x+37.82 0.080x+38.12 0.133x+37.9 - 0.716x+

n between 7 2 6 0.016x+38.4 36.2

parameter R² = 0.705 R² = 0.344 R² = 0.795 R² = 0.017 R² = 0.472

s Linear

Fit

26
Table 6 Results of confirmatory tests under optimum conditions

DESIGN Cu Co Acid consumption Residue

ATION Sol. Yi Bala Sol. Yi Bala Tota Cu Co Gan weight

eld nce eld nce l gue

kg/d % % kg/d % % kg/d kg/t kg/t kg/d kg/dmtu

mtu mtu mtu -Cu -Co mtu

Confirm. 14.9 95. 101. 2.94 89. 109. 116. 22.9 4.48 90.4 893.87

Cu 7 98 45 15 84 71 2 4

Confirm. 15.2 94. 100. 3.12 97. 84.1 123. 23.2 5.19 49.3 895.20

Co 5 66 66 74 3 20 0 1

27
Table 7 ANOVA of Cu leaching yield

Parameters df Sum of Variance F-Exp. p-value Contribution

Squares (%)

Acidity 4 0.128462 0.032115 1.5994 0.330115 20.83

Time 4 0.040241 0.010060 0.5010 0.740130 6.52

Temperature 4 0.073129 0.018282 0.9105 0.535108 11.86

Pulp density 4 0.103521 0.025880 1.2889 0.405837 16.78

Conc. Fe II 4 0.036903 0.009226 0.4595 0.765073 5.98

Residue 4 0.080317 0.020079 1.0000 0.330115 38.02

Total 24 0.462573 0.115643 100

F(4.4) =6.3882 (at 95% confidence level).

28
Table 8 ANOVA of Co leaching yield

Parameters df Sum of Variance F-Exp. p-value Contribution

Squares (%)

Acidity 4 2.24581 0.56145 3.1711 0.144872 2.81

Time 4 0.92186 0.23047 1.3017 0.402258 1.15

Temperature 4 1.11237 0.27809 1.5707 0.336236 1.39

Pulp density 4 0.71668 0.17917 1.0120 0.495541 0.90

Conc. Fe II 4 54.33231 13.58308 76.7183 0.000492 67.87

Residue 4 0.70821 0.17705 1.0000 0.144872 25.88

Total 24 60.037237 15.009312 100

F(4.4) =6.3882 (at 95% confidence level).

29
Figure 1. Simplified flowsheet of Kambove concentrator.

30
Figure 2. Particle size distribution of the oxidised copper-cobalt tailing.

31
Figure 3. Effects of controllable factors associated with their levels on the statistical
performance (SN) for the leaching of Cu.

32
Figure 4. Effects of controllable factors associated with their levels on the statistical
performance (SN) for the leaching of Co.

33
Figure 5. Pareto chart showing the percentage contribution of parameters on the leaching
of Cu.

34
Figure 6. Pareto chart showing the percentage contribution of parameters on the leaching
of Co.

35
Figure 7. Box plot for Cu leaching optimisation results according to the L25(55)

experimental plan.

36