Fuel 85 (2006) 289–297

www.fuelfirst.com

Application of the Box-Wilson experimental design method

for the spherical oil agglomeration of coal
Yakup Cebecia, İbrahim Sönmezb,*
a
Mining Engineering Department, Cumhuriyet University, 58140 Sivas, Turkey
b
Metallurgical and Materials Engineering Department, Cumhuriyet University, 58140 Sivas, Turkey
Received 17 March 2005; received in revised form 28 June 2005; accepted 27 July 2005
Available online 22 August 2005

Abstract
In this study, the Box-Wilson statistical experimental design method was employed to evaluate the effects of important variables such as
bridging liquid (oil) concentration, salt (CaCl2$2H2O) concentration and stirring speed on the agglomeration of bituminous coal. Response
function coefficients were determined by the regression analysis of experimental data and the predictions were found to be in good agreement
with the experimental results. The optimum kerosene concentration, CaCl2$2H2O concentration and stirring speed were determined as
30 wt%, 1 M and 1683 rpm, respectively, when considering combustible recovery and ash content.
In addition, contact angle and solution surface tension measurements were carried out to evaluate of agglomeration success with the
contact angle values and surface tension values. The surface tension of CaCl2 2H2O solutions and the average contact angle increased with
increasing CaCl2$2H2O concentration.
q 2005 Published by Elsevier Ltd.

Keywords: Coal; Box-Wilson design; Oil agglomeration; Kerosene; CaCl2$2H2O

1. Introduction beneficiation with high recovery. In this method, a bridging

liquid (immiscible oil) is added to a coal–water mixture, and
Modern mining methods and cleaning processes increase the hydrophobic particles are preferentially coated by the
the fine coal particle concentration continuously, which oil, whereas the hydrophilic mineral matter remains
create several problems in dewatering, drying, handling dispersed in the aqueous phase, due to the difference
transportation and storage. Significant quantities of these between the coal particles and mineral matter. Upon
fine coals are lost in the waste stream, which results in a agitation, the oil-coated particles collide with each other
substantial loss of valuable energy and causes considerable and form large agglomerates, which can be separated as a
environmental problems [1–3]. To making use of the clean coal [6,7]. The agglomerate can be separated from the
beneficial properties of fine coals, the mineral matter from suspension by floating, skimming and screening [5,8]. Basic
these is required to be removed. Conventional coal principles of oil agglomeration and the effects of various
beneficiation techniques are fairly unproductive in fine operating parameters have been investigated by many
coal processing. As an alternative, flotation, flocculation and researchers [5,8–10].
oil agglomeration are used in fine coal processing. Oil It is known that the electrokinetic behavior of the
agglomeration is accepted as the most effective method [4– particles affects the process recovery and the selectivity in
6] for cleaning of a material having very fine sized particle, the concentration process; based on the surface properties,
because of the simplicity of the process and mineral such as froth flotation, flocculation and oil agglomeration.
Coal recovery and separation efficiency, which are
achieved by oil agglomeration, depend on a number of
* Corresponding author. Tel.: C90 346 219 1010; fax: C90 346 219 factors including the surface properties of the solid, the
1173. nature of the agglomerates and certain properties of the
E-mail address: [email protected] (I. Sönmez). aqueous medium [11,12]. It has been observed that ionic
0016-2361/$ - see front matter q 2005 Published by Elsevier Ltd. strength of the aqueous medium affect the agglomeration
doi:10.1016/j.fuel.2005.07.017 performance [13,14]. The presence of electrolytes also
290 Y. Cebeci, I. Sönmez / Fuel 85 (2006) 289–297

de-stabilizes the wetting films on hydrophobic solids and When a particle is placed in an oil–water interface, it may
helps to thin the film during the process of establishing a be drawn into the aqueous phase (q!908) or remain
contact angle of coal particles [15]. The positive effect of concentrated at the interface (qZ908) or drawn into the oil
ionic strength on the agglomeration of the more oleophilic phase (qO908). Oil agglomeration preferentially occurs
coals appears to be due to the compression a the electrical when condition (qO908) is satisfied. In this situation, the
double layer surrounding individual particles, whereas the mineral/coal particles concentrating at the oil–water inter-
effect on the less oleophilic coals seems to be due to the face, migrate into the oil phase allowing maximum
adsorption of hydrated cations [12,13,16]. By compressing absorption of solids by the oil [22,23].
the electrical double layer around the negatively charged In this study, the Box-Wilson experimental design
coal particles and oil droplets, the repulsive electrostatic method was used in order to investigate the effects of
forces between particles and droplets seem to be largely important variables on combustible recovery on agglomera-
overcome [11]. tion of bituminous coal sample taken from Zonguldak. This
Another aspect of the aqueous medium that deserves experimental design is a response surface methodology used
special consideration is the presence of multivalent cations, for evaluation of a dependent variable as functions of
which form metal hydroxy complex ions in intermediate pH independent variables. The oil concentration, CaCl2$2H2O
(6–8) ranges and colloidal precipitates of metal hydroxides concentration and stirring speed were considered as
in higher pH (O9) ranges [11,17]. These ions are cations of independent variables and, combustible recovery and ash
metals such as calcium(II), magnesium(II), iron(II, III) and content were considered as dependent variables in the Box-
aluminum(III), which are frequently part of the inorganic Wilson statistical experimental design method. Also,
ash-forming mineral matter associated with coal [11,18]. contact angle measurements were carried out for declare
Wettability characteristics of minerals have substantial of agglomeration success with the contact angle values.
influence over the efficiency of beneficiation and processing
methods such as agglomeration, aggregation, flotation, dust 2. Materials and methods
controlling and solid–liquid separation [19,20]. The solid or
mineral surface is completely wetted by the liquid if the
2.1. Materials
surface tension of liquid is equal to or below the critical
surface tension of wetting value of solid or mineral surface
The coal sample, which was bituminous, was
(gC). For a good contact angle (qO0) at solid–liquid–air
obtained from TTK Zonguldak coal washery and ground
interface, the surface tension of liquid must be greater than
to !106 mm in a rod mill for the oil agglomeration tests.
the critical surface tension of wetting value of solid (gC)
The result of proximate analysis of the sample and the
[21]. The absorption mechanism of particle by the bridging
particle size distribution of the ground sample determined
liquid (oil) is given in Fig. 1, schematically. The position of
by wet screening are given in Table 1 and Fig. 2,
the solid particle at the interface between oil and water is
respectively.
governed by the relative values of the interfacial tensions.
At equilibrium
2.2. Chemicals
gSO Z gSW C gOW cos q (1) In the experiments, kerosene used as bridging liquid was
characterized as having a density of 0.780 g cmK3 and a
where viscosity 1.17 mPa s at 20 8C [24]. About 5 M stock solution
gSO is the solid–oil surface tension (mN mK1), of CaCl2$2H2O (147.02 g molK1) (Merck) salt was pre-
gSW is the solid–water surface tension (mN mK1), pared and required amounts of this solution was added into
gOW is the oil–water surface tension (mN mK1). the suspension to obtain the final desired concentration to
investigate the effect on agglomeration process. All these
γSO chemicals were analytical grade.
Table 1
Proximate analyses results of the coal sample
Oil
Components As received Air dried Dry basis
Ash (wt%) 14.00 14.00 14.08
γ OW θ Solid Volatile matter 26.59 26.60 26.75
(wt%)
Water
Fixed carbon 58.82 58.85 59.17
(wt%)
Calorific value (kCal/kg)
γSW
Net 6696 6699 6740
Gross 6947 6950 6988
Fig. 1. The situation of particle at an oil–water interface [22,23].
 Y. Cebeci, I. Sönmez / Fuel 85 (2006) 289–297 291

100
0.7 cm
Cumulative Undersize, %

Pulp Level

11.0 cm
10
10 100 1000
ParticleSize, ∝m
0.6 cm 7.4 cm
Fig. 2. Particle size distribution of the ground sample.
0.6 cm
2.3. Oil agglomeration studies

The experiments were performed in a 400 ml beaker 2.5 cm

having four baffles at the border to create homogeneity. 2.5 cm
IKA-WERK RW 20 model mechanical stirrer was used for
effective mixing. The agitation was provided by a centrally 7.4 cm
located flat blade turbine impeller (consisting of four blades)
at a fixed distance from the bottom of the vessel. The
experimental setup is shown schematically in Fig. 3. Mono-
distilled water (pHw6.5) was used in the experiments. The
solids concentration was kept constant at 5.0% by weight for
each test and the volume of suspension subjected to
agglomeration was 285 mL. The mixture of coal–water
was conditioned for 3 min before the salt solution addition.
Salt solution was added and the suspension was conditioned
for 3 min before bridging liquid addition. Suspension was
conditioned with bridging liquid for 3 min as agglomeration
time. The bridging liquid concentration was based on the
mass ratio of kerosene to the coal on dry basis. After the Fig. 3. Schematic illustration of the experimental setup.
completion of the experiment, agglomerates were separated
from suspension by using 106 mm test sieve and the 100 NRL Model Goniometer. During contact angle
agglomerates were washed with acetone to remove measurement tests, suspension was mixed using the same
the oils, dried and weighed. The ash contents of the conditions of oil agglomeration, and particles were removed
agglomerated product were determined. The results of from the suspension before the bridging liquid (oil)
the agglomeration process were evaluated by percentage of addition. Then, coal samples were dried in oven and
combustible recovery. The combustible recovery was prepared in the forms of pressed pellets in 12 mm diameter.
calculated using Eq. (2) The surface tension of kerosene used as a bridging liquid
was 25.95 mN mK1 (determined by drop weight method).
Mc ð1KAc Þ
Combustible Recovery ðCR; %Þ Z !100 (2)
Mf ð1KAf Þ
2.5. Box-Wilson experimental design
where
Ac ash content of clean coal, A Box-Wilson statistical experimental design method
Af ash content of feed, was used to determine the effects of major operating
Mc mass of clean coal, parameters on combustible recovery and ash content. More
Mf mass of feed. information about this method is given in related literature
[23,26]. The Box-Wilson design is a response surface
methodology, an empirical modeling technique, devoted to
2.4. Contact angle measurement studies the evaluation of the relationship of a set of controlled
experimental factors and observed results. This optimiz-
Contact angle (q) measurements were carried out by the ation process involves three major steps, performing the
method developed by Zisman [25] using a Rame-Hart, Inc. statistical design experiments, estimating the coefficients in
292 Y. Cebeci, I. Sönmez / Fuel 85 (2006) 289–297

Table 2
Experimental conditions according to a Box-Wilson statistical design

Axial points Factorial points

No. KC (%) SC (M) SS (rpm) No. KC (%) SC (M) SS (rpm)
A1 40.00 0.5 1250 F1 32.62 0.789 1683
A2 5.00 0.5 1250 F2 32.62 0.789 816
A3 22.50 1 1250 F3 32.62 0.211 1683
A4 22.50 0.0 1250 F4 32.62 0.211 816
A5 22.50 0.5 2000 F5 12.38 0.789 1683
A6 22.50 0.5 500 F6 12.38 0.789 816
Centre point F7 12.38 0.211 1683
A 22.5 0.5 1250 F8 12.38 0.211 816

KC, kerosene concentration; SC, salt concentration; SS, stirring speed.

a mathematical model, and predicting the response and good agreement with the observed values as shown in
checking the adequacy of the model. Three operating Table 3.
parameters: bridging liquid (kerosene) concentration (X1), Experimental results were modeled using a Statistica 5.0
salt concentration (X2) and stirring speed (X3) were chosen regression analysis program to determine the coefficients of
as the most important independent variables. The bridging the response function (Eq. (3)). The calculated coefficients
liquid concentration (X1) changed between 5 and 40 wt%, are listed in Table 4 and were used in calculating predicted
salt concentration (X2) varied between 0 and 1 M and the values of combustible recoveries and ash contents. The
stirring speed (X3) between 500 and 2000 rpm. Experimen- determination coefficient between the observed and pre-
tal conditions determined by the Box-Wilson statistical dicted values was (R2) 0.9703 and 0.7495 for combustible
design are presented in Table 2. The experimental consist of recovery and ash content, respectively, indicating a good
six axial (A), eight factorial (F) and centre (C) points. The agreement between the observed and predicted values.
center point was repeated three times. The effects of operating variables on the combustible
The combustible recovery or ash content (Y) was recovery and ash content were determined by obtaining
correlated with the other independent parameters (X1, X2, projections of the response functions on certain planes of
X3) using the Eq. (3). A Statistica 5.0 computer program was known parameter values.
used for determination of the coefficients of Eq. (3) by
Variation of the surface tension of CaCl2$2H2O solutions
regression analysis of the experimental data.
(determined by drop weight method) and the average
Y Z b0 C b1 X1 C b2 X2 C b3 X3 C b12 X1 X2 C b13 X1 X3 contact angles measured on the coal pellets are given in
Fig. 4.
C b23 X2 X3 C b11 X12 C b22 X22 C b33 X32 (3) It is shown in Fig. 4 that the surface tension of
CaCl2$2H2O solutions and the average contact angle
where
increase with increasing CaCl2$2H2O concentration. The
Y the predicted response function (combustible minimum contact angle was achieved for mono-distilled
recovery or ash content),
b0 constant, Table 3
b1, b2, b3 linear coefficients, Observed and predicted combustible recovery and ash content
b12, b13, b23 cross-product coefficients, Experiment Combustible recovery Ash content
b11, b22, b33 quadratic coefficients. no.
Observed Predicted Observed Predicted
A1 98.05 94.54 9.59 6.67
A2 57.27 61.87 5.97 6.31
A3 99.02 98.51 10.50 9.96
3. Results and discussions A4 99.13 98.87 10.06 10.32
A5 97.17 96.85 10.04 9.99
The combustible recoveries and ash contents obtained A6 87.78 87.47 12.37 12.12
F1 96.92 100.00 9.20 9.85
from the experiments that consisted of six axial (A), eight
F2 98.28 98.37 10.80 11.14
factorial (F) and centre points are summarized in Table 3 in F3 96.68 97.08 8.81 8.72
which a comparison of the experimental and predicted F4 98.39 100.00 11.41 11.79
values for combustible recovery and ash content is F5 90.02 87.28 8.99 8.82
presented. The observed combustible recoveries varied F6 71.65 72.54 7.90 8.21
F7 84.84 85.11 8.72 8.59
between 57.27 and 99.13%, while the observed ash contents
F8 78.59 75.89 10.18 9.76
were varying between 5.97 and 12.37%. The predicted
C (Ave) 97.97 98.01 10.25 10.25
values of combustible recovery and ash content were in a
 Y. Cebeci, I. Sönmez / Fuel 85 (2006) 289–297 293

K1.04!10K5
78 120

1.44!10K6
Surface Tension

Surface Tension (γ), mN.m–1

Contact Angle
76

Contact Angle, ˚
B33
110

74

K4.37!10K1
100
72
2.74
b22

70 90
K2

K8.53!10K3

0.0 0.2 0.4 0.6 0.8 1.0 1.2

K6.50!10

CaCl2.H2O Concentration, M
b11

Fig. 4. The surface tension of CaCl2$2H2O solutions and contact angle.

water as 938, and the average contact angle increased up to

K1.09!10K4
K2

109.58 for CaCl2$2H2O concentration of 1 M. The oil

1.10!10

agglomeration occurs successfully because the average

contact angle was higher than 908 at all CaCl2$2H2O
b23

concentrations.
K4

K1.09!10K4

3.1. The effect of stirring speed

K7.47!10

In order to determine the effect of stirring speed on

b13

combustible recovery and ash content at a CaCl2$2H2O

concentration of 0.5 M, Fig. 5 was developed for different
values of kerosene concentrations.
K2

7.69!10K2

As shown in Fig. 5, the combustible recovery increased

6.50!10

with the increasing stirring speed and kerosene concen-

b12

tration up to 30 wt%. Also, the combustible recovery

reached a maximum value at a stirring speed of 1250 rpm
for kerosene concentration of 30 wt%. Kerosene concen-
K4.35!10K3

tration of 40 wt%, the combustible recovery decreased with

K2

increasing stirring speed. At lower stirring speed and

4.35!10

kerosene concentrations, the combustible recovery was

low due to the insufficient contact between particle–particle,
b3

particle–microagglomerate and/or microagglomerate–

microagglomerate. Also, the insufficient wetting of hydro-
K18.33

phobic coal particles by kerosene could be another reason.

K6.09

In this case, most of the voids between particles were filled

b2

100 14
Com. Rec.
5.57!10K1

Kerosene, wt%
12
Combustible Recovery, %

5
80 10
10
Ash Content, %
4.74

20
b1

30
60 40 8
Coefficients of the response function

Ash Content

40
Kerosene, wt% 6
7.70
8.35

5
b0

10 4
20
20 30
2
Combustible recovery

40

0 0
0 500 1000 1500 2000 2500
Ash content
Coefficient

Stirring Speed, rpm

Table 4

Fig. 5. Effect of stirring speed on combustible recovery and ash content at

different kerosene concentration (CaCl2$2H2O: 0.5 M).
294 Y. Cebeci, I. Sönmez / Fuel 85 (2006) 289–297

with aqueous medium. As the concentration of the bridging attributed to the decreasing of the electrostatic repulsive
liquid was increased the voids between particles, filled with forces between coal particles and oil droplets. The
aqueous medium, were coated with an increasing concen- combustible recovery reached to maximum value with
tration of kerosene and also, since particle–particle collision 1 M CaCl2$2H2O concentration at higher stirring speeds.
energy and collision speed increased with increasing stirring This situation was based on the decreasing of adsorption of
speed, agglomerates in tighter structure were formed. ClK anions over the mineral matter and oil droplets and
Therefore, agglomerate size and combustible recovery also, it could be based on the increasing of the surface
increased. At a higher stirring speed and kerosene tension of the solution. Because, the hydrophobic coal
concentration, the amount !106 mm increased and the particle was not wetted completely by the solution and the
combustible recovery decreased because coal particles were kerosene attached to the coal surfaces sufficiently.
dispersed in kerosene. In addition, the spherical shape of As can be seen in Fig. 6, the ash contents of agglomerated
agglomerates turned into pasty lumps and the agglomerates product decreased with increasing the stirring speed at every
deviated from spherical shape. Also, the partial comminu- CaCl2$2H2O concentration. This behaviour was due to the
tion of the agglomerates caused by the collisions of the high dispersion of the mineral matter and the coal particles.
agglomerates with cell walls and with each other can be Also, this could be due to the electrostatic repulsive forces
another reason. between ash-forming mineral matter and kerosene droplets
As seen in Fig. 5, the ash contents of agglomerates because of the adsorption of ClK anions over the mineral
generally decreased with the increasing of stirring speed at matter particles and oil droplets. In the relevant literature, it
all kerosene concentrations. This behaviour can be based on was also stated that the excessive ClK anions increased the
the high dispersion of the mineral matter particles and the magnitude of negative electrokinetic potential value of the
coal particles and also, the agglomeration of the more kerosene droplets [27]. As it was expressed in initial studies,
hydrophobic feed particles. salts such as Al2(SO4)3, CaCl2, NaCl and FeSO4 affected the
Fig. 6 depicts the variation combustible recovery and ash zeta potential of oil droplets [24,27].
content as a function of stirring speed in different CaCl2$2- And also, the ash content of the product decreased with
H2O concentration, but at a constant kerosene concentration increasing the surface tension of solution. It could be based
of 22.5 wt%. on the insufficient attaching of kerosene to the mineral
As can be seen in Fig. 6, that the combustible recovery matter surfaces that had poor hydrophobicity at these
increased with the increasing of stirring speed especially at solution surface tensions.
high CaCl2$2H2O concentrations. This can be attributed to
the increasing of the contact between particle–particle, 3.2. The effect of kerosene concentration
particle–microagglomerate and/or microagglomerate–
microagglomerate and decreasing of the electrostatic A similar graph representing variation of combustible
repulsive forces between coal particles and oil droplets. recovery and ash content with kerosene concentration at
The combustible recovery decreased for stirring speed constant CaCl2$2H2O concentration (0.5 M), but different
above 1750 rpm and at low CaCl2$2H2O concentrations. stirring speed levels is given in Fig. 7.
This is based on the partial comminution of the agglomer- It can be shown in Fig. 7, that the combustible recovery
ates caused by the collisions of the agglomerates with cell reached its maximum value at 30 wt% kerosene concen-
walls and with each other. At low stirring speeds, the tration with all stirring speeds. At low bridging liquid
combustible recovery reached a maximum value at 0.25 M concentrations and stirring speeds, the combustible recov-
CaCl2 $2H 2O concentration. This behaviour can be ery was low due to the insufficient wetting of hydrophobic

100 14 100 14
Com. Rec. Com. Rec.
CaCl2.H2O, M Stirring Speed, rpm
12 12
Combustible Recovery, %
Combustible Recovery, %

0.00 80 500
816
0.25
10 10
Ash Content, %

1250
Ash Content, %

0.50 1683
0.75 60 2000
8
1.00 8
90 Ash Content
Stirring Speed, rpm
Ash Content
CaCl2.H2O, M 6 40 500
6
0.00 816

0.25 4 1250 4
1683
0.50 20 2000
0.75 2 2
1.00

80 0 0 0
0 500 1000 1500 2000 2500 0 10 20 30 40 50
Stirring Speed, rpm Kerosene Concentration, wt%

Fig. 6. Effect of stirring speed on combustible recovery and ash content at Fig. 7. Effect of kerosene concentration on combustible recovery ash
different CaCl2 2H2O concentration (Kerosene: 22.5 wt%). content at different stirring speeds (CaCl2$2H2O: 0.5 M).
 Y. Cebeci, I. Sönmez / Fuel 85 (2006) 289–297 295

coal particles by kerosene and also, the insufficient contact concentration, similarly. This is based on the reduction in
between particle–particle, particle–microagglomerate the thickness of the electrical double layer surrounding each
and/or microagglomerate–microagglomerate. In this case, coal particle and oil droplets. Above 25 wt% kerosene
most of the voids between particles were filled with aqueous concentration, the ash contents of agglomerated products
medium. As the concentration of the bridging liquid and decreased with the increasing kerosene concentration. This
stirring speed was increased, the voids between particles situation was attributed to the electrostatic repulsive forces
filled with aqueous medium were coated with an increasing between ash-forming mineral matter and kerosene droplets
concentration of kerosene and the particle–particle collision because of the adsorption of ClK anions over the mineral
energy and collision speed increased. Consequently, matter particles and oil droplets. Also, the ash content of the
compact and spherical agglomerates were formed and the product decreased with increasing the surface tension of the
combustible recovery increased. For the higher concen- solution. This situation could be based on the kerosene did
trations of kerosene (over 30 wt%), coal particles were not attach sufficiently to the mineral matter surfaces which
dispersed in kerosene and the spherical shape of agglom- had poor hydrophobicity at these solution surface tensions.
erates turned into pasty lumps; the agglomerates deviated
from spherical shape. Also, the combustible recovery 3.3. The effect of CaCl2$2H2O concentration
increased with increasing the stirring speed at kerosene
concentration of 40 wt%, because of the partial comminu- In an electrokinetic study performed for Zonguldak
tion of the agglomerates caused by the collisions of the coals, the isoelectric point (iep) value of coal was found to
agglomerates with cell walls and with each other. be pH 4. In the same study, it was also found that the zeta
As can be seen in Fig. 7, the ash content of agglomerates potential values of coal were K50 and C15 mV in high and
tended to decrease with increasing kerosene concentration low pH values, respectively [28]. Another electrokinetic
at all stirring speed. This could be attributed to the high study related to the kerosene reported that zeta potential of
dispersion of the mineral matter and the coal particles. kerosene droplets decreased up to the K20 mV at low pH
Variations of combustible recovery and ash content with values and increased up to the K90 mV at high pH values,
kerosene concentration at different levels of CaCl2$2H2O especially at the values over pH 8 [29].
concentrations, but a constant stirring speed of 1250 rpm is The effect of CaCl 2 $2H 2O concentration on the
shown in Fig. 8. The combustible recovery increased with combustible recovery and ash content at different stirring
the increasing of kerosene concentration up to 30 wt% and it speed, but at constant kerosene concentrations is given in
decreased partially afterward at the all CaCl2$2H2O Fig. 9.
concentrations studied. At low kerosene concentration, the As shown in Fig. 9, the combustible recovery decreased
combustible recovery was low due to the insufficient with the increasing of salt concentration at low stirring
wetting of hydrophobic coal particles by kerosene. Above speed (500 and 816 rpm) at kerosene concentration of
30 wt% kerosene concentration, the combustible recovery 22.5 wt%. This can be attributed to not overcome
decreased because of dispersion of the coal particles in electrostatic repulsive forces between coal particle and oil
kerosene and the spherical shape of agglomerates turned droplets. The combustible recovery did not change with
into pasty lumps and the agglomerates deviated from increasing salt concentration a stirring speed of 1250 rpm.
spherical shape. However, at higher stirring speed (1683 and 2000 rpm), the
It can be seen from Fig. 8 that as the kerosene combustible recovery increased with the increasing salt
concentration was increased up to 25 wt%, the ash contents concentration. This was due to the overcoming of
of the agglomerated products increased for all CaCl2$2H2O electrostatic repulsive forces between coal particles–coal

100 14 100 14
Com. Rec. Com. Rec.
CaCl2.H2O, M Stirring Speed, rpm
12 12
Combustible Recovery, %
Combustible Recovery, %

0.00
80 500
0.25 816
10 10
Ash Content, %
Ash Content, %

0.50 1250
0.75 1683
60 1.00 8 2000 8
90
Ash Content Ash Content
CaCl2.H2O, M 6 Stirring Speed, rpm 6
40 0.00 500
0.25
4 816 4
0.50 1250
20 0.75
1.00
2 1683
2
2000

0 0 80 0
0 10 20 30 40 50 0.0 0.2 0.4 0.6 0.8 1.0 1.2
Kerosene Concentration,wt% CaCl2 .H2 O Concentration, M

Fig. 8. Effect of kerosene concentration on combustible recovery and ash Fig. 9. Effect of CaCl2$2H2O concentration on combustible recovery and
content at different CaCl2$2H2O concentration (stirring speed: 1250 rpm). ash content at different stirring speeds (Kerosene: 22.5 wt%).
296 Y. Cebeci, I. Sönmez / Fuel 85 (2006) 289–297

14 2. Response function predictions determined by regression

100 Com. Rec.
Kerosene, wt% analysis were in good agreement with the experimental
12
Combustible Recovery, %

5
results.
80 10
10 3. The surface tension of CaCl2$2H2O solutions and the

Ash Content, %
20
30
60 40 8 average contact angle increase with increasing CaCl2$2-
Ash Content
H2O concentration. The oil agglomeration occurred
Kerosene, wt% 6
40 5
successfully with and without CaCl2$2H2O addition
10
4 because the average contact angles were higher than 908
20
20 30 at all experiments.
40
2
4. The combustible recovery increased with an increase in
0 0 stirring speed up to 1750 rpm and the ash content of the
0.0 0.2 0.4 0.6 0.8 1.0 1.2 product decreased with an increase in stirring speed
CaCl2 .H2O Concentration, M
depending on kerosene and CaCl2$2H2O concentrations.
Fig. 10. Effect of CaCl2$2H2O concentration on combustible recovery and 5. The combustible recovery increased with the increasing
ash content at different kerosene concentrations (stirring speed: 1250 rpm). kerosene concentration up to 30 wt% and the ash content
of the product decreased with the increasing kerosene
particles and/or coal particles–kerosene droplets because of concentration depending on stirring speed. Both the
adsorption of CaC2 cations on the coal particles. And also, it combustible recovery and ash content increased with
could be based on the increasing of the surface tension of kerosene concentration up to 30 wt% depending on
solution. In this case, the hydrophobic coal particles are CaCl2$2H2O concentration.
wetted by the solution to a lesser extent with respect to the 6. The combustible recovery decreased with the increasing
water and the kerosene droplets can sufficiently attaches to salt concentration at low stirring speeds (500 and
the coal surfaces. 816 rpm), and it increased with the increasing salt
As observed in Fig. 9, when the stirring speed was concentration at higher stirring speed (1683 and
increased, the ash contents of agglomerated product 2000 rpm). The ash contents of agglomerated product
decreased at all CaCl2$2H2O concentrations. This beha- decreased with the increasing CaCl2$2H2O concen-
viour was due to the high dispersion of the mineral matter tration depending on stirring speed. Also, the combus-
and coal. Also, the ash content of the product decreased with tible recovery did not change and the ash content of
increasing the surface tension of the solution. These results
agglomerated product decreased with the increasing salt
showed that the kerosene did not attach sufficiently to the
concentration depending on kerosene concentration.
mineral matter surfaces at these solution surface tensions.
7. Considering combustible recovery and ash content, the
The combustible recovery and ash content as a function
optimum kerosene concentration, CaCl2$2H2O concen-
of CaCl2$2H2O concentration at different kerosene concen-
tration and stirring speed were determined as 30 wt%,
tration and a constant stirring speed of 1250 rpm are
1 M and 1683 rpm, respectively.
presented in Fig. 10.
Fig. 10 proposed that the combustible recovery did not
change depending on the salt concentration at all kerosene
concentration studied and also, the combustible recovery
increased with the increasing of kerosene concentration. References
Also, Fig. 10 shows that when the CaCl2 $2H 2O
concentration was increased, the ash contents of agglomer- [1] Pawlak, W., Turak, A., Ignasiak, B., (1985). Selective agglomeration
ated product decreased at all kerosene concentration. This of low rank bituminous and subbituminous cretaceous coals, in:
behaviour was due to the high dispersion of the mineral Proceedings 4th International Symposium on Agglomeration, Tor-
matter and coal. Also, the ash content of the product onto, 907-915.
decreased with increasing the surface tension of the solution. [2] Cebeci, Y., (1996). The investigation of effects of some
parameters on the performance of selective agglomeration of
lignites, in: VIth International Mineral Processing Symposium,
Kuşadası, Turkey, 445-459.
4. Conclusions [3] Cebeci Y, Ulusoy U, Şimşek S. Investigation of the effect of
agglomeration time, pH and various salts on the cleaning of
The conclusions obtained from this study are as follows: Zonguldak bituminous coal by oil agglomeration. Fuel 2002;81(9):
1131–7.
1. The Box-Wilson statistical experimental design pro- [4] Capes C, Germain RJ. Selective oil agglomeration in fine coal
beneficiation. In: Liu YA, editor. Physical cleaning of coal present and
cedure was seen to be applicable in modeling to evaluate developing methods. NewYork: Marcel Dekker; 1982. p. 293–300.
the effects of important variables on the combustible [5] Mehrotra VP, Sastry KVS, Money BW. Review of oil agglomeration
recovery and ash content of agglomerated product in oil techniques for processing of fine coals. Int J Miner Process 1983;11:
agglomeration of bituminous coal. 175–201.
 Y. Cebeci, I. Sönmez / Fuel 85 (2006) 289–297 297

[6] Kim SS, Morsi BI, Chiang SH. Statistical analysis of the performance [17] Çelik MS, Somasundaran P. The effect of multivalent ions on the
of a selective agglomeration process using n-heptane as agglomerant. flotation of coal. Sep Sci Technol 1986;21(4):393–402.
Coal Preparation 1994;15:51–70. [18] Botsaris GD, Glazman YM. Interfacial Phenomena in Coal
[7] Kim SS, Morsi BI, Araujo G, Chiang S-H, Blachere J, Sharkey A. Technology. New York: Marcel Dekker; 1989 p. 141–53.
Effect of grinding conditions on the performance of a selective [19] Fuerstenau DW, Williams MC. A new method for characterization of
agglomeration process for physical coal cleaning. Coal Preparation the surface energy of hydrophobic particles. Particle Charact 1987;4:
1991;9(3-4):141–53. 7–13.
[8] Capes CE, Darcovich K. A survey of oil agglomeration in wet fine [20] Crawford RJ, Mainwaring DE. The influence of surfactant adsorption
coal processing. Powder Technol 1984;40:43–52. on the surface characterization of Australian coals. Fuel 2001;80:
[9] Capes, Capes CE. Principals and application of size enlargement in 313–20.
liquid system. In: Somasundaran P, editor. Fine Particle Processing. [21] Laskowski JS. The relationship between floatability and hydropho-
New York: AIME; 1980. p. 1442–62. Chapter 73. bicity Advances in Mineral Processing. Littleton, CO, USA: SME;
[10] Osborne DG. Flotation, agglomeration and selective flocculation, 1986 p. 189–208.
Coal Preparation Technology, vol. 1. London: Graham and Trotman [22] Puddington IE, Sparks BD. Spherical agglomeration process. Miner
Ltd; 1988. p. 460. Chapter 9. Sci Eng 1975;7(3):282–8.
[11] Fan CW, Markuszewski R, Wheelock TD. Effect of multivalent [23] Crozier RD, Crozier, R.D.. Flotation.: Pergamon Press; 1992.
cations on oil agglomeration of coal and pyrite. Coal Preparation [24] Cebeci Y, Sönmez İ. The investigation of coal-pyrite/lignite
1992;11:167–75. concentration and their separation in the artificial mixture by oil
[12] Yang GCC, Markuszewski R, Wheelock TD. Oil agglomeration of agglomeration. Fuel 2002;81:1139–46.
coal in inorganic salt solutions. Coal Preparation 1988;5(3-4):133–46. [25] Zisman WA. Relation of equilibrium contact angle to liquid and solid
[13] Fan CW, Markuszewski R, Wheelock TD. Coal and pyrite separation constitution in contact angle, wettability and adhesion, Advances in
by oil agglomeration in salt solutions. Fizykochemiczne Problemy Chemistry Series, No. Am Chem Soc 1964;43:1–51.
Mineralurgii 1987;19:17–26. [26] Davies OL, Davies, O.L. The design and analysis of industrial
[14] Sadowski Z, Venkatadri R, Druding JM, Markuszewski R, experiments. New York: Hafner Publishing Co.; 1956.
Wheelock TD. Behavior of oxidized coal during oil agglomeration. [27] Campbell JAL, Sun SC. Anthracite coal electronics. Soc Min Engng,
Coal Preparation 1988;6:17–34. AIME, Trans 1970;247:120–2.
[15] Blake TD, Kitchene JA. Stability of aqueous films on hydrophobic [28] Özbayoğlu G. In: Yarar B, Doğan Z, editors. Coal Flotation Mineral
methylated silica. J Chem Soc-Faraday Trans I 1972;68(8):1435. Processing Design, NATO Series. The Hague: Martinus Nijhoff;
[16] Fan CW, Markuszewski R, Wheelock TD. Role of induction time and 1987. p. 76–105.
other properties on the recovery of coal from aqueous suspensions by [29] Wen WW, Sun SC. An electrokinetic study on the oil flotation of
oil agglomeration with heptane. Energy Fuels 1989;3:376–81. oxidized coal. Sep Sci Technol 1981;16:1491–521.