Separation and Purification Technology 122 (2014) 87–95

Contents lists available at ScienceDirect

Separation and Purification Technology

journal homepage: www.elsevier.com/locate/seppur

Effect of settling test procedure on sizing thickeners

Gh.A. Parsapour, M. Hossininasab, M. Yahyaei, S. Banisi ⇑
Mineral Processing Group, Mining Engineering Department, Shahid Bahonar University of Kerman, P.O. Box 76175-133, Kerman, I.R. Iran

a r t i c l e i n f o a b s t r a c t

Article history: The main portion of the water in mineral processing plants is recovered in thickeners. The required cross-
Received 1 September 2013 sectional area for thickeners is generally calculated using the Coe–Clevenger, and the Talmage–Fitch
Received in revised form 1 November 2013 methods. Over the years, changes which were made in the original settling tests procedure led to inac-
Accepted 1 November 2013
curate results owing to floc structure variation especially for the flocculated suspensions. In this study,
Available online 11 November 2013
the effect of flocculation history, defined as the initial solids concentration where the flocs first were
formed, on their settling velocity was investigated using batch settling tests. The suspensions used for
Keywords:
settling tests were prepared by two different methods to arrive at equal solids fractions. The methods
Flocculation
Dewatering
were the conventional procedure of adding solids to a known amount of liquid and decantation. Results
Thickening of all settling tests (with and without flocculant) on various ores, and coal suspensions showed that the
Coal dewatering settling velocity of particles in any solids fraction of suspension in the mass settling region could be
Tailings disposal affected (up to 5-fold) by solids content of the suspension at which flocs were first formed. The decanta-
tion method provided lower settling velocity compared to that of the conventional procedure. This was
attributed to different flocs structure and size which were indirectly verified by the settling tests. In the
case of Interkarbon coal preparation plant, this translated to 27% underestimation of the thickener capac-
ity. It was then concluded that the settling tests performed to determine the required cross-sectional area
for thickening should be carried out either on a single sample with the solids concentration of the thick-
ener feed or various concentrations from the feed to the underflow concentrations provided that the sam-
ples obtained by decantation of the feed sample.
Ó 2013 Elsevier B.V. All rights reserved.

1. Introduction determination methods [2–18]. The Coe–Clevenger and Talmage–

Fitch approaches (to be described below) are the most widely used
1.1. Thickener cross-sectional area determination methods methods of thickener sizing in the mineral processing industry.

Prior to 1916, thickening was an art [1]. In 1916, Coe and Cle-
venger recommended a method to determine the required area
for thickeners based on a settling model [2]. They presented Eq.
(1) to determine thickener unit area (U.A.; needed area for thicken-
ing of one ton solids per hour) using liquid (water) balances:
 
1 1 1 concentration which determines the required unit area. Thus a ser-
U:A: ¼  ð1Þ
C C  u R  ql
where C is a solids fraction between the feed and the underflow, Cu
is the underflow solids fraction, ql is the liquid density (t/m3), and R
is the settling rate (m/h) of a suspension with a solids fraction of C.
Batch-scale settling tests were used to measure the rate of fall
of the interface between the settling solids and the clarified liquor
under gravity in a straight-walled vertical vessel. Over the years,
there have been many studies on the sedimentation behavior,
flocculation properties, and thickener cross-sectional area
⇑ Corresponding author. Tel./fax: +98 341 2112764.
E-mail address: banisi@mail.uk.ac.ir (S. Banisi).
1.1.1. The Coe–Clevenger method
Coe and Clevenger implicitly assumed that the settling rate of
solids is only a function of solids concentration [2]. This method
is based on the hypothesis that there is a critical flux rate at a solids
concentration between the actual feed and the desired underflow
ies of tests using concentrations ranging from the feed concentra-
tion to the underflow concentration are performed and the
resultant unit areas are plotted against these concentrations and
the maximum unit area is chosen as the design thickener unit area
[2–5,10,19–26]. The initial tangents of the settling curve (plot of
height of interface between the settling solids and the clarified li-
quor versus time) is used as the settling velocity of particles in the
Coe–Clevenger method [22].
1.1.2. Procedure of the Coe–Clevenger settling tests
The original procedure outlined by Coe and Clevenger [2] for the
settling tests was as follows:

1383-5866/$ - see front matter Ó 2013 Elsevier B.V. All rights reserved.
http://dx.doi.org/10.1016/j.seppur.2013.11.001
88 Gh.A. Parsapour et al. / Separation and Purification Technology 122 (2014) 87–95
(a) 1000 cm3 of feed suspension is taken and settling test is per-
formed by recording the height of interface in the cylinder
every 2 min.
(b) After completing (a) the suspension is thoroughly mixed by
shaking. Then, 25 cm3 of suspension is removed from the
cylinder and replaced by 25 cm3 of discharge suspension
(provided from small-scale continuous settling tests). After
mixing, step (a) is repeated. In this fashion, the solids con-
centration of suspension is gradually increased by replacing
a part of the suspension with the equal volume of thickened
underflow.
(c) Step (b) is repeated by removing 30, 45, 75, 100, 160 and
260 cm3 of previous suspension and replacing by an equal
volume of the discharge suspension. Upon completion of
tests, eight settling curves starting from the feed solids con-
centration to some higher values are obtained.
Coe and Clevenger [2] also provided an alternative approach
where a sufficient quantity of feed suspension was prepared and
suspensions with different solids concentrations were made ready
by decantation. The settling tests were then performed on each
sample having a certain solids fraction.
Up until 1964, there was no further mention of using decanta-
tion to prepare samples for the settling tests [3,19,20]. In 1964,
Moncrieff [4] stated that re-mixing of suspension (in the original
method of the Coe and Clevenger) could alter the suspension char-
acteristics – particularly if the suspension is flocculated – and agi-
tation may shear the flocs. Furthermore, he concluded that residual
turbulence after agitation can reduce the initial settling rate. To
overcome these shortcomings, he proposed to perform settling
tests on samples individually prepared to arrive at certain solids
concentration [4]. It is believed that by replacing agitation/re-mix-
ing (for flocs formed at a single given concentration) by floc forma-
tion at various different initial concentrations, Moncrieff may have
removed one source of error but introduced another (the initial
solids concentration at which the flocs were formed also affect floc
structure and therefore settling velocity). After 1964, the proce-
dure introduced by Moncrieff became the conventional method
of determining mud line settling velocity and loosely known as
the Coe–Clevenger method [7,10,22,23,27,28]. The results of the
tests are shown by drawing mud line height versus time and tan-
gents of the linear part of these curves are taken as settling
velocities.
1.1.3. The Talmage–Fitch method
In 1952, Kynch proposed a mathematical relationship between
the settling velocity and the solids concentration which made pos-
sible to determine the settling velocity of various solids concentra-
tion by a single settling test [29,30]. The Kynch model was based
on the free settling regime where flocs settle without receiving
any impacts from other flocs. Kynch assumed that settling velocity
of particles (V) in this region is only function of local solids concen-
tration (C) or V = f(C) [3,32].
Talmage and Fitch applied Kynch’s mathematical model to the
problem of thickener design. They calculated settling velocities of
various solids concentrations by a single batch settling test. They
showed that by constructing a tangent to the settling curve at
any point, the settling velocity and local concentration can be
determined from the gradient and y-intercept of the tangent
[24,25]. Thus, a set of data of solids concentration and settling
velocities can be obtained from a single batch-settling curve and
using Eq. (1) unit areas can be calculated and the maximum unit
area is chosen as the design thickener unit area.
Due to inaccuracy of the sizing methods, scale-up or safety fac-
tors up to 2 have been suggested [11,24,25]. It is worth to note that
when the original Coe–Clevenger method was introduced
flocculants were not in use. Since in the Talmage–Fitch method
the sample is usually prepared at the feed solids concentration it
seems that it is similar to real thickening operation where flocs
are formed at low solids concentrations. This feature made the Tal-
mage–Fitch method more suitable than the Coe–Clevenger method
in the case of flocculated suspensions [5,11,25,31,49] where the
floc structure is of prime importance. Given the significant role of
flocculation in suspension settling, it is important to describe and
understand the flocculation process.
1.2. Formation of flocs
Flocculants are added to gravity thickeners to enhance solid–li-
quid separation performance. It is a key step in many processes,
being essential to economic and effective solid–liquid separation
by gravity thickening. Addition of flocculant (long-chain polymers)
in a dilute stable suspension, bridges particles together [33–37],
leading to the formation of highly porous and irregularly shaped
aggregates, known as flocs [38,39]. Flocs have complex shapes,
low density, rapid initial settling velocity, and trap some liquid in
their structures, that is highly dependent upon floc formation pro-
cess [25,38,49]. The nature of flocs depends on various factors such
as nature of solids (surface chemistry, size, size distribution, shape,
density), nature of liquid (viscosity, dielectric constant), nature of
suspension (electric charge, pH, ion strength, temperature), and
nature of flocculant (chemical properties of main part and chains,
molecular weight, molecular weight distribution, electric charge,
density of electric charge) [27]. The properties of flocs, such as floc
size, density, structure, and strength significantly influence the
performance of solid–liquid separation processes [40–43].
It has been known that the size and structure of the flocs de-
pend on the solids concentration at which they are formed
[11,52]. At high solids concentration (above 10% solids volume
fraction), flocs form a wide network structure which in turn de-
creases settling velocity [37,50]. Talmage and Fitch [3] illustrated
that the settling velocity of a floc may be assumed to be a function
of the structure of the floc, as well as the suspension solids concen-
tration. They observed that the settled solids concentration is a
function of initial solids concentration and showed that the floc
structure was apparently affected by initial solid concentration,
and hence it could be assumed that the settling rate would also
be affected.
1.3. Effect of compression forces on final bed solids concentration
As flocs settle in the flocculated suspension, they form a bed at
the bottom of the suspension. This bed may consolidate further un-
der its own weight. When the individual flocs contact the consoli-
dated bed, the attractive forces between particles due to the
bridging flocculant cause the floc to stick to the bed in the position
of first contact. The floc is not free to roll around the top of the bed
to seek a lower gravitational potential energy. Subsequent colli-
sions of other flocs will tend to create an open network of floccu-
lated particles at average volume fraction much lower than can
be achieved by a stable sedimenting suspension that is free to seek
its lowest gravitational energy configuration [50]. Network consol-
idation is the process of packing flocs together without necessarily
changing the structure of individual flocs. Consolidation occurs un-
der the action of compressive forces (both external and internal).
Compressive forces are consisted of external and internal forces.
Push on suspension with a piston or increase the gravitational
forces in a centrifuge are examples of external forces. Weight of
material and water have been considered as internal forces; only
the weight of material has been taken into account by some
researchers [26,44,51]. In the great majority of metallurgical appli-
cation the thickener sizing is determined by clarification or mass
 Gh.A. Parsapour et al. / Separation and Purification Technology 122 (2014) 87–95
settling zones. However, if the solids must be retained in the com-
pression zone in the batch tests for long time (usually over 4 h) in
order to reach the desired underflow concentration, solids com-
pressibility must be controlled [52].
The volume fraction at which the suspension forms a network
structure is define as gel point and at the steady state, solids con-
centration at the top of bed is equal to the gel point [13,53,54]. Be-
low the gel point, the suspension is not networked and in this case
classical analysis of batch settling tests (Kynch theory) are valid
[12,26]. Gel point in some studies is assumed equal or larger than
10% solids volume fraction [13,37,50,53,54].
Compressive stresses on the suspension can be transmitted via
the network throughout the system and the structure then pos-
sesses the ability to support itself [50]. The network structure will
resist compression until compression forces become so strong that
the structure begins to deform irreversibly. Floc densification
which is the process of changing the structure of individual flocs
via rupture and reformation of bonds occurs and a denser (stron-
ger) structure forms. It has been shown that floc size and structure
is a function of compressive forces [50,51]. Pearse [6] investigated
the effect of suspension height on the final solids concentration by
testing same suspensions in the cylinders of constant diameter and
different heights. He observed that in the free settling regime the
height of the suspension column is not important (initial tangents
of settling curves are equal). But the final solids concentration de-
pends on the initial suspension column height. Yalcin [8] also
showed that initial suspension concentration had a significant ef-
fect on the final suspension concentration.
Densification of aggregates is a time dependent phenomenon
[26]. In the start of densification of aggregates that solids fraction
is low, flow around flocs is most resistance factor in densification
and by increasing time as the solids fraction is increased, effect
of flow around flocs is decreased and effect of flow through them
is increased [37]. At large time a steady-state aggregate diameter
is reached, which is the final extent of densification, beyond which
no change is exhibited [26]. Decreasing in the final suspension
height (increasing in the final solids fraction) is a result of densifi-
cation of aggregates.
1.4. The Interkarbon coal preparation plant
The Interkarbon coal preparation plant is located southeast of
Iran in Zarand. This plant processes the old tailings of the Zarand
coal washing plant. The feed rate of the plant is 100 t/h with the
ash content of 51%. Banisi and Yahyaei [45] designed a thickener
for this plant with a diameter of 22 m based on the dilution of feed
from solids concentration of 10% (w/w) to 4% and the target under-
flow solids concentration of 26%. The size distribution of the feed to
the thickener showed that 91% of particles were smaller than
38 lm and only 0.6% were coarser than 75 lm. Thickener unit area
was determined to be 9.6 m2/(t/h) using the Talmage–Fitch meth-
od. For the feedrate of 25 t/h (dry solids) to the thickener and using
a safety factor of 1.25, the diameter of the thickener was calculated
to be 21.9 m [45].
In this study, the effect of solids concentration at which flocs
were first formed on their settling velocities was investigated using
batch settling tests. The effect of floc size and structure on the sed-
imentation behavior, settling velocity and final bed solids concen-
tration were also indirectly studied. The term ‘‘flocculation
history’’, defined as the initial solids concentration where the flocs
were first formed, was introduced as a criterion influencing the
settling velocity. In order to keep the floc size and structure con-
stant when settling tests are performed at various solids fractions,
a procedure to perform settling tests was proposed and tested. The
effect of settling test procedure on the calculated cross-sectional
89
area required for thickening was also studied for the case of the
Interkarbon coal preparation plant.
2. Materials and methods
2.1. Materials
Size distribution and density of the ores used for the settling
tests are shown in Table 1. Settling tests were performed in neutral
pH and polyacrylamid type flocculants (known as A25, and A65 at
the Gol-E-Gohar iron ore company, and Interkarbon coal washing
plant, respectively).
2.2. Settling test procedure
Proper addition and mixing of a flocculant with the suspension
is one of the essential steps in the flocculation process [46].
Amount of mixing is very important because insufficient mixing
does not provide sufficient particles and flocculant molecules con-
tacts and excessive mixing breaks long flocculant molecules
[8,27,45]. Poor mixing of flocculant throughout slurry may also
lead to non-homogenous distribution of the polymer on the solid
particles [47]. It has been suggested that best mixing results by
reversing the test cylinder five times [6,48,49].
In this study, settling tests were performed in a one-liter clear-
scaled glass cylinder. Diameter and height of the cylinder were
6.07 and 34.5 cm, respectively. It was filled with specific weight
of sample and water and flocculants were also added when needed.
The cylinder containing the suspension was then mixed by revers-
ing the cylinder five times and finally mud line height was re-
corded against time. Flocculant stock solution of 0.05% was
prepared based on ISO standard (ISO 1086) by dissolving solid
flocculants in water.
2.3. Method of investigating effect of feed solids concentration on
settling velocity
Iron ore tailing, copper ore, lead and zinc ore and coal tailing
were tested in order to investigate the effect of initial solids con-
centration on the settling velocity under the conditions shown in
Table 2. To calculate the error of settling velocity measurements,
coal tailing tests of 2%, 4% and 8% (w/w) solids were repeated three
times. Note that all %’s are by weight unless otherwise mentioned.
2.4. Method of investigating effect of flocculation history on settling
velocity
In order to study the effect of flocculation history on the settling
velocity, suspensions used for the settling tests were prepared by
two different methods. The first method was the conventional pro-
cedure in which a specific amount of solids and water mixed into a
glass cylinder and flocculant added. In the second method, decan-
tation was used; in this method several cylinders were first filled
with a dilute suspension and after adding flocculant mixed as de-
scribed earlier. The suspension was then given a sufficient time
to settle. The clear water section of the cylinders was then si-
phoned and weighed; finally, suspensions with different solids
concentrations were prepared by careful mixing of the settled
materials.
Copper ore, iron ore tailing, and coal tailing suspensions were
prepared by the aforementioned methods. Settling tests of copper
ore and iron ore tailing were performed with solids concentration
of 5%, 10%, 15%, and 20% (w/w) using 5 g/t of flocculant; for coal
tailings, settling tests were performed at 2%, 4%, 8%, 15%, and 20%
(w/w) solids using 25 g/t of flocculant.
90 Gh.A. Parsapour et al. / Separation and Purification Technology 122 (2014) 87–95
Table 1
Size distribution and density of ores used in settling tests.
Ore 80% passing size; F80 (lm) 50% passing size; F50 (lm)
Copper 80 39
Lead and zinc 55 36
Iron ore tailing 70 49
Coal tailing 36 30
Table 2
Settling tests conditions.
Ore Feed weighting solids concentration (%)
Iron ore tailing 5 10 15
Copper 5 10 15
Lead and zinc 5 10 15
Coal tailing 2 4 6 8 10
3. Results and discussions
3.1. Effect of feed solids concentration on settling velocity of particles
Based on the Coe–Clevenger, and Kynch theories the settling
velocity of particles is a sole function of local solids concentration
around the particles [3,29]. In a settling test cylinder with a certain
feed solids there are different ranges of solids fractions. In other
words, from the top of the cylinder to the bottom, solids concentra-
tion gradually increases and at the bottom it reaches a maximum.
Moreover the solids fraction immediately below the clear liquor
interface increases with time. If settling tests are performed with
various feed solids concentrations it is therefore possible to find
the settling velocity of a certain given solids fraction using the tan-
gents to settling curves at specific points (based on desired solids
fraction). The question is if the settling velocities of a certain given
solids fraction (e.g., 15% solids fraction) in various suspensions
with different initial solids concentration are determined how do
they compare? If the initial solids concentration, at which flocs
were first formed, did not play any role, based on the aforemen-
tioned theory, all calculated settling velocities (using the
Talmage–Fitch method) should have been equal within an experi-
mental error. Stated differently, the settling velocity of any solids
fraction of suspension should be independent of the initial feed
solids concentration. Graphically, the settling velocity of a certain
given solids fraction versus feed solids concentration curve should
be a straight line parallel to the x-axis. In the following sections,
theses hypotheses will be examined through carefully designed
settling tests under different conditions.
3.1.1. Settling velocity of coal tailing (with flocculant)
Fig. 1 shows settling velocities of a particular solids fraction of
coal tailing containing 10.5% (which is equivalent to 4.2% by vol-
ume) when the feed solids concentration varied from 2% to 10%
(flocculant dosage: 35 g/t). Settling velocities were calculated by
the Talmage–Fitch method. Interestingly, the settling velocity of
this certain solids fraction (i.e., 10.5%) varied from 1.01 to
2.87 cm/min. The settling velocity of the same solids fraction
(10.5%) decreased as the feed solids concentration increased. On
the other hand, the initial settling velocities of lower solids concen-
trations were higher; for instance, in the suspension containing 2%
solids fraction and 35 g/t flocculant initial settling velocity was
40.4 cm/min. The results clearly indicates the effect of solids con-
centration at which the flocs were formed on their size and struc-
ture. To check the significance level of differences in velocity
values, the standard deviations obtained from the repeat tests
were also calculated (shown by error bars in Fig. 1; relative
20% passing size; F20 (lm) Density (g/cm3)
27 2.7
26 3.7
35 3.6
23 2.6
Flocculant dosage (g/t of dry solids)
20 25 0, 5, 10, 20, 30, 40
20 0, 5
20 25 0, 5
12 15 20 15, 25, 35
Fig. 1. Settling velocities of particles obtained by the Talmage–Fitch method for a
particular solids fraction containing 10.5 w/w% coal tailing (error bars show one
standard deviation based on repeat measurements); flocculant dosage: 35 g/t.
standard deviation was 0.13). Analysis of the results indicated sta-
tistically meaningful differences in velocity with an error of only
5%. A decreasing trend was observed for the settling rate rather
than having a constant value for a specific solids fraction.
3.1.2. Settling velocities of copper ore and lead and zinc ore (without
flocculant)
Settling velocities of a solids fraction containing 16.5% of copper
ore in three settling tests (starting from 5% to 15% solids) are
shown in Fig. 2 (without flocculant). Note that all velocities were
calculated by constructing a tangent to the settling curve at a spe-
cific point. It is also interesting to remind that, the settling velocity
of the same solids fraction varied from 0.46 to 1.14 cm/min in tests
with different feed solids concentrations. The velocity measure-
ments along with their standard deviations (error bars in Fig. 2)
also indicated that the velocity was not constant over the tested
range of initial solids fractions.
In order to explore the effect of ore density on the settling
velocity, the settling tests were also repeated for the lead and zinc
ore. Fig. 3 shows settling velocity of a solids fraction containing
16.5% for the lead and zinc ore at various feed solids concentration
(without flocculant). Note that all feed solid fractions which were
tested in this project were lower that of a networked suspension.
The same trend was observed as in Fig. 2 and the velocity of the
same solids fraction varied from 0.23 to 1.13 cm/min. The
differences in velocity for various initial solids concentration
 Gh.A. Parsapour et al. / Separation and Purification Technology 122 (2014) 87–95
Fig. 2. Settling velocities of particles obtained by the Talmage–Fitch method for a
16.5% solids fraction of copper ore (error bars show one standard deviation based
on repeat measurements); without flocculant.
Fig. 3. Settling velocities of particles obtained by the Talmage–Fitch method for a
16.5% solids fraction of the lead and zinc ore (error bars show one standard
deviation based on repeat measurements); without flocculant.
considering the error of measurements were significant at 95%
confidence level. Comparison of Figs. 2 and 3 indicated that ore
density also contributed to variation of settling velocity for any gi-
ven suspension.
3.1.3. Settling velocity of iron ore tailing (with and without flocculant)
Settling velocities of a 22% solids fraction for the iron ore tailing
(which is equivalent to 7.3% by volume) with and without
Fig. 4. Settling velocities of particles obtained by the Talmage–Fitch method for a 22% solids fraction of the iron ore tailing; (a) without flocculant; (b) flocculant dosage: 40 g/
t (error bars show one standard deviation based on repeat measurements).
91
flocculant are shown in Fig. 4. When no flocculant used the settling
velocities varied from 0.26 to 0.67 cm/min and with 40 g/t floccu-
lant settling velocities varied between 3.3 and 5.1 cm/min. Given
the error of measurements the differences found significant at
95% confidence level. The results reconfirmed that in all cases the
initial solids concentration had an effect on the settling velocity
of a given solids fraction for suspensions with and without
flocculant.
3.2. Comparison of settling velocities obtained using Coe–Clevenger
and Talmage–Fitch methods
In the Coe–Clevenger method the settling velocity of any given
suspension is measured by preparation of a suspension with the
desired solids concentration. In other words, for any feed solids
concentration a settling test must be performed. On other hand,
the Talmage–Fitch method requires only one settling test and the
settling velocity for any solids concentration is obtained from a sin-
gle settling curve. This implies that if the settling velocity of any
solids fraction is constant, regardless of the initial solids concentra-
tion where the flocs are formed, the settling velocities determined
by the Coe–Clevenger method and the Talmage–Fitch method
must be equal. Fig. 5 compares settling velocities of a suspension
containing 10% of iron ore tailing solids fraction for various floccu-
lant dosages determined by the two methods. Inspection of the re-
sults showed that the settling velocities obtained from the two
methods were not equal; the difference increased by increasing
the amount of flocculant dosages used for the tests. Since the
Coe–Clevenger method was originally proposed for suspensions
without any flocculant, where the structure of the flocs was not
complex, the difference increased as the amount of flocculant dos-
age was increased. This indicated the importance of the flocs struc-
ture on settling velocity which became marked in suspensions
containing flocculant.
3.3. Effect of feed solids fraction on settling time and final bed solids
concentration
In order to investigate the effect of feed solids fraction, two
points from the settling curves were chosen for further study. This
provided an opportunity to obtain some indirect information about
the structure of flocs formed at different feed solids concentrations.
The final solids concentration for any settling test was defined as
the solids concentration of settling particles after a long period
(24 h). This point indicated how the structure of flocs had an effect
on the amount of entrapped water. In other words, the higher final
solids concentration was indicative of a structure from which the
contained water could be released under the compression
92 Gh.A. Parsapour et al. / Separation and Purification Technology 122 (2014) 87–95
Fig. 5. Settling velocities of a 10% solids fraction of the iron ore determined by the
Coe–Clevenger (conventional method) and the Talmage–Fitch (initial solids
concentration was 5%) methods for various flocculant dosages (error bars show
one standard deviation based on repeat measurements).
originated from the weight of upper layers. The first point to study
on the settling curve was chosen to be 55% of the final solids con-
centration; the point was very close to the compression point
where the linear part of the curve ended. The second point was
chosen to be 90% of the final solids concentration. The choice of
these points was arbitrary and the idea was to study the behavior
of the settling suspension at two different regions: mass settling
and compression. Table 3 shows the effect of feed percent solids
on the final percent solids and on the required time to reach 55%
and 90% of the final percent solids (15 g/t flocculant). For suspen-
sions without flocculant the results are presented in Table 4. As
the feed percent solids increased the final bed solids fraction also
increased. The required time to form a dense suspension, defined
as about 90% of the final solids concentration, was significantly
lower for the dilute suspensions. For example, the required times
for coal tailing with 2% and 10% solids were found to 4.7 and
313.7 min, respectively. The very large difference indicated fast
settling suspension when the feed solids concentration was low
(2%). Almost the same trend in the required times was observed
to form a suspension with 55% of the final bed percent solids
(3.6 min versus 179 min).
The required time to reach 90% of the final percent solids from
55%, could be an indication of flocs structure, because at the high
solids fractions, only flow through flocs resists to densification of
aggregates [37]. The required time to reach 90% of the final solids
fraction from 50%, for coal tailing with 2% and 10% solids fractions
were found to be 1.1 min (4.7–3.6) and 134.7 min (313.7–179),
respectively. It was believed that the flocs structure when the flocs
first were formed at 10% solids was very packed and the required
time to release water from the pores was considerably high.
For the settling tests in suspensions without flocculant, the re-
quired time to reach 90% of the final percent solids for copper ore
with 5% and 20% solids were measured to be 24, and 106.7 min,
respectively. This observation also indicated the effect of flocs
structure even in suspensions without flocculant, but the effect
was more pronounced when flocculant used, judged by the re-
quired time difference to reach 90% of the final percent solids.
Table 3
Effect of feed solids fraction on the final bed solids fraction and the time required to reach two different (spatially-averaged) solids fractions (coal tailing; flocculant dosage: 15 g/
t).
Feed percent solids Final percent solids Time required to reach 55% of final percent solids (min)
2 37.46 3.62
4 38.75 14.64
8 42.29 110.85
10 43.28 179.03
The final bed percent solids obtained for suspensions without floc-
culant were significantly higher than that of reached by suspen-
sions with flocculant which showed that flocs in suspensions
having flocculant entrapped more water.
To further illustrate the effect of flocs structure on the bed sol-
ids concentrations, the trend of spatially-averaged suspension per-
cent solids for two feed containing 2% and 10% solids, over a period
of 1400 min for the coal tailing, is shown in Fig. 6 (15 g/t floccu-
lant). The rate of change in spatially-averaged suspension solids
concentration was significantly higher for the feed containing 2%
solids fraction, but the final percent solids was lower. Fig. 7 shows
the trend of spatially-averaged suspension solids concentration for
the copper ore when no flocculant used. It was observed that the
rate of change depended on the feed solids concentration. For in-
stance, the average rate of increase in spatially-averaged suspen-
sion percent solids for feed containing 2% and 10% coal tailing
were 0.07 and 0.03 percent per minute, respectively. It was con-
cluded that the feed solids (initial suspension) concentration
played an essential role in the settling velocity.
3.4. Effect of flocculation history on settling velocity of particles
Fig. 8 shows settling velocities of particles at various solids con-
centration of coal tailing suspensions which were prepared with
the conventional and decantation methods (25 g/t flocculant). It
was observed that particles settling velocities when suspension
prepared with conventional method were greater than that of
the suspension prepared by the decantation method. For instance,
for 4% solids concentration the difference in velocities between the
two methods was found to be 10 cm/min. The difference decreased
when the feed solids concentration increased such that for 8% sol-
ids fraction it became 0.86 cm/min. The main reason postulated to
be the difference in the shape and structure of the flocs which were
formed at different initial solids concentration. Since the coal tail-
ing sample was very fine (90% smaller than 38 lm) at high solids
concentrations the settling velocity was very low and the differ-
ence was difficult to detect.
In order to study similarity of the settling velocities obtained
using decantation and the Talmage–Fitch methods, a series of set-
tling tests were performed. To obtain the necessary samples a large
quantity of 2% solids suspension was prepared. By decantation,
samples with solids concentrations of 4%, 8%, 14.6%, and 19% were
obtained. Note that when preparing suspensions by decantation
method it is difficult to arrive exactly at a specific percent solids.
This explains why values such as 14.6% for solids concentration
were obtained. The standard settling test was performed on four
samples with solids concentration between 4% and 19% (25 g/t
flocculant). The settling velocities of the samples were measured
(Fig. 9). In accordance with the Talmage–Fitch procedure a single
settling test also was carried out with a 2% solids content sample.
Using the settling velocity curve for the 2% solids fraction sample,
by the gradient of the tangents drawn from the points correspond-
ing to 4%, 8%, 14.6%, and 19%, the settling velocities were calculated
(Fig. 9). With this way of experimentation, the flocculation history
was taken into account in determination of the settling velocity. It
is worth nothing that the single settling test was performed on the
2% solids fraction sample because it was the initial feed solids
Time to reach 90% of final percent solids (min)
4.69
24.27
190.67
313.68
 Gh.A. Parsapour et al. / Separation and Purification Technology 122 (2014) 87–95 93

Table 4
Effect of feed solids concentration on the final bed percent solids and the required time to reach two different (spatially-averaged) final percent solids (copper ore; without
flocculant).

Feed percent solids Final percent solids Time required to reach 55% of final percent solids (min) Time required to reach 90% of final percent solids (min)
5 54.04 14.23 24.05
20 58.80 23.05 106.74

Fig. 6. The change in the average solids concentration of the suspension over a
period of 1400 min (coal tailing suspension; flocculant dosage: 15 g/t). Fig. 9. Settling velocities of coal tailing suspension determined by the Talmage–
Fitch and the Coe–Clevenger (suspension prepared by decantation) methods (initial
feed solids concentration: 2%; flocculant dosage: 25 g/t).

concentration in which the flocs first were formed. The comparison

of the settling velocities obtained from the two methods (Fig. 9)
indicated that there was no significant difference between the re-
sults. The error bars shown in Fig. 9 for the measurements over-
lapped which indicated there was no statistically meaningful
difference between the results. The practical outcome of this find-
ing was that the settling tests performed to determine the required
cross-sectional area for thickening should be carried out either on a
single sample with the solids concentration of the thickener feed
(after dilution if it is diluted) or on the decanted samples. The con-
ventional procedure of measuring settling velocities in various per-
cent solids between the feed and the desired underflow
underestimates the required cross-sectional area needed for
Fig. 7. The change in the average solids concentration of the suspension over a
thickening.
period of 1400 min (copper tailing; without flocculant).

3.5. Recalculation of the diameter of the Interkarbon coal preparation

plant thickener

The Talmage–Fitch and the Coe–Clevenger methods were used

to re-size the Interkarbon coal preparation plant thickener. Both
decantation and conventional suspension preparation procedures
were used for the Coe–Clevenger method.

3.5.1. Required cross-sectional area estimated by the Talmage–Fitch

Fig. 8. Settling velocities of coal tailing suspension prepared with the conventional
(Coe–Clevenger) and decantation methods (flocculant dosage: 25 g/t).
method
In the Talmage–Fitch method, the settling curve for a solids
concentration of 2% was used to calculate the settling velocities
for various feed solids concentrations. For any given feed solids
concentration corresponding settling velocity was calculated by
constructing a tangent to the settling curve at any concentration.
Table 5 shows the results of such calculations for selected solids
concentration. The desired underflow solids concentration was
26% by weight. The limiting unit area was the one with the largest
value which was 8.9 m2/(t/h). For the feed rate of 25 t/h (dry solids)
and considering a safety factor of 1.25, the diameter of the required
thickener was found to be 18.8 m.
94 Gh.A. Parsapour et al. / Separation and Purification Technology 122 (2014) 87–95
Table 5
Unit areas calculated for various solids concentration by the Talmage–Fitch method.
Solids concentration (%) 2 4 8 14.6
2
U.A. (m /(t/h)) 1.0 2.5 4.0 6.5
Table 6
Unit areas calculated for various solids concentration by the Coe–Clevenger method
using conventional and decantation suspension preparation procedures.
Solids concentration (%) 2 4 8 14.6
U.A. (m2/(t/h))
Decantation method 1.0 2.4 3.2 8.1
Conventional method 1.0 1.1 1.9 5.6
3.5.2. Required cross-sectional area estimated by the Coe–Clevenger
method
Both conventional and decantation procedures were used to
prepare two series of suspensions at solids concentrations of 2%,
4%, 8%, 14.6% and 19%. To obtain the necessary samples by the
decantation a large quantity of 2% solids suspension was prepared
and in subsequent stages removing water provided higher solids
concentration suspensions. Initial tangents of any settling curve
was determined as the settling velocity of that solids concentra-
tion. Table 6 summaries the results of unit area calculation using
the Coe–Clevenger method when the desired underflow solids con-
centration was 26%. The limiting unit areas were 8.9 and 7.0 m2/(t/
h) when the decantation and conventional procedures were used,
respectively. Based on these limiting unit areas and the feed rate
of 25 t/h (dry solids) to the thickener and considering a safety fac-
tor of 1.25, the diameters of the required thickeners were calcu-
lated to be 18.8 and 16.7 m for the decantation and conventional
procedures, respectively. Note that the limiting unit area was
determined here solely by looking at a very restricted set of solids
fractions (corresponding to those used in the earlier decantation
study). For the Talmage–Fitch method however there is no need
to restrict consideration just to those values. But, in order to com-
pare the results, unit areas by both methods were calculated for
certain solids fractions.
Comparison of the results of the Coe–Clevenger and the Tal-
mage–Fitch methods showed that when the decantation procedure
was used the limiting unit areas of the Coe–Clevenger and the Tal-
mage–Fitch methods were equal. It is important to note that the
conventional procedure using the Coe–Clevenger method underes-
timated the required cross-sectional area for thickening.
The diameter of the Interkarbon coal preparation plant thick-
ener is 22 m and thickener feed rate is 25 t/h (day solids). The lar-
ger diameter is because the feed solids fraction in practice was 4%
instead of 2% which was the base to arrive at a diameter of 18.8 m.
It is interesting to observe that, based on the conventional settling
test procedure using the Coe–Clevenger method, a thickener of
16.7 m in diameter would be proposed. This implied that even after
diluting the feed to 2% solids fraction to increase the flocculation
performance the thickener capacity could only reach 20 t/h.
4. Conclusions
– Settling velocity of particles determined for a 22% solids fraction
of iron ore tailing varied from 0.26 to 0.67 cm/min when per-
forming settling test with different feed solids concentration
in suspensions without flocculant. When 40 g/t of flocculant
used the range of settling velocities were found to be from 3.3
to 5.1 cm/min. It was then concluded that the settling velocity
of particles for any given solids fraction depended on the initial
solids concentration.
– It was determined that for low feed solids concentration (2–4%)
the rate of suspension percent solids with time is high. This has
19 been the main driving force of feed dilution in new generation
8.9 of thickeners. It was found that the rate of increase in suspen-
sion percent solids with time for feed containing 2% and 10%
coal tailing were 0.07 and 0.03 percent per minute, respectively.
– The effect of flocculation history, defined as the feed solids con-
centration where the flocs first formed, on their settling velocity
was investigated using batch settling tests. Comparisons of the
settling velocities obtained using conventional feed preparation
19
and decantation procedures showed that flocculation history
highly affected the settling velocity of particles. For coal tailing
8.9
suspensions (with 25 g/t of flocculant) containing 4% and 8%
7.0
solids fractions the differences between settling velocities of
particles for the two methods were found to be 10 and
0.86 cm/min, respectively.
– It was concluded that the settling tests performed to determine
the required cross-sectional area for thickening should be car-
ried out either on a single sample with the solids concentration
of the thickener feed (after dilution if it is diluted) or on the
decanted samples. The conventional procedure of measuring
settling velocities in various percent solids between the feed
and the desired underflow underestimated the required cross-
sectional area for thickening.
– The required thickener diameter for the Interkarbon coal
preparation plant was calculated to be 18.8 m using the Coe–
Clevenger method when settling tests were performed on the
decanted samples; the Talmage–Fitch method also gave the
same result.
– The use of decantation and conventional settling test proce-
dures in the Coe–Clevenger method to determine the unit area
of the Interkarbon coal preparation was resulted in different
values (i.e., 8.9 and 7.0 m2/(t/h)). This translated to 27% differ-
ence in the thickener capacity.
Acknowledgements
The authors would like to thank Gol-E-Gohar Mining and
Industrial Company and Interkarbon Company supporting and
implementing the outcome of this research and also permission
to publish the article. Special gratitude also extended to the
reviewers of the paper who contributed much in the logical trans-
mission of ideas by providing constructive criticisms, comments
and questions.
References
[1] E.J. Roberts, Thickening – art or science?, Mining Transactions (1949) 61–64
March.
[2] H.S. Coe, G.H. Clevenger, Methods for determining the capacities of slime
settling tanks, Trans, AIME 55 (1916) 356–384.
[3] W.P. Talmage, E.B. Fitch, Determining thickener unit areas, Industrial
Engineering Chemistry 47 (1955) 38–41.
[4] A.G. Moncrieff, Theory of thickener design based on batch sedimentation tests,
Transactions of the Institution of Mining and Metallurgy 73 (1964) 729–759.
[5] K.J. Scott, Affecting settling rate of solids in flocculated suspensions, Mineral
Processing and Extractive Metallurgy 77 (1968) 85–97.
[6] M.J. Pearse, Factors affecting the laboratory sizing of thickeners, in: P.
Somasundaran (Ed.), Fine Particle Processing, AIME, 1980, pp. 1619–1642.
Chapter 81.
[7] J.H. Wilhelm, Y. Naide, Sizing and operating continuous thickeners, Mining
Engineering 33 (1981) 1710–1718.
[8] T. Yalcin, Sedimentation characteristics of Cu–Ni mill tailings and thickener
size estimation, CIM Bulletin 81 (1988) 69–75.
[9] F. Jones, J.B. Farrow, W.V. Bronswijk, Effect of caustic and carbonate on the
flocculation of haematite in synthetic Bayer liquors, Colloids and Surfaces A:
Physicochemical and Engineering Aspects 142 (1998) 65–73.
[10] T. Laros, S. Slottee, F. Baczek, Testing, sizing, and specifying sedimentation
equipment, in: A.L. Mular, D.N. Halbe, D.J. Barratt (Eds.), Mineral Processing
Plant Design, Practice, and Control, vol. 2, SME, 2002, pp. 1295–1312.
 Gh.A. Parsapour et al. / Separation and Purification Technology 122 (2014) 87–95
[11] D.A. Dahlstrom, Liquid-solid separation, in: M.C. Fuerstenau, N.H. Kenneth
(Eds.), Principles of Mineral Processing, SME, 2003, pp. 307–362.
[12] D.R. Lester, S.P. Usher, P.J. Scales, Estimation of the hindered settling function
R(u) from batch-settling tests, AIChE Journal 51 (2005) 1158–1168.
[13] S.P. Usher, P.J. Scales, Steady state thickener modelling from the compressive
yield stress and hindered settling function, Chemical Engineering Journal 111
(2005) 253–261.
[14] S. Diehl, Estimation of the batch-settling flux function for an ideal suspension
from only two experiments, Chemical Engineering Science 62 (2007) 4589–
4601.
[15] P. Grassia, S.P. Usher, P.J. Scales, A simplified parameter extraction technique
using batch settling data to estimate suspension material properties in
dewatering applications, Chemical Engineering Science 63 (2008) 1971–1986.
[16] A.D. Stickland, C. Burgess, D.R. Dixon, P.J. Harbour, P.J. Scales, L.J. Studer, S.P.
Usher, Fundamental dewatering properties of wastewater treatment sludges
from filtration and sedimentation testing, Chemical Engineering Science 63
(2008) 5283–5290.
[17] A.T. Owen, T.V. Nguyen, P.D. Fawell, The effect of flocculant solution transport
and addition conditions on feedwell performance in gravity thickeners,
International Journal of Mineral Processing 93 (2009) 115–127.
[18] P. Grassia, S.P. Usher, P.J. Scales, Closed-form solutions for batch settling height
from model settling flux functions, Chemical Engineering Science 66 (2011)
964–972.
[19] R.A. Couche, L.H. Goldney, The design of continuous thickeners for flocculated
materials, Proceedings the AUS.I.M.M. 191 (1959) 117–139.
[20] B. Fitch, Sedimentation process fundamentals, Society of Mining Engineers,
Trans, AIME 223 (1962) 129–137.
[21] K.J. Scott, J.L. Alderton, Maximum solids handling capacity of continuous
thickeners, Mineral Processing and Extractive Metallurgy 77 (1966) 201–210.
[22] B. Fitch, Batch tests predict thickener performance, Chemical Engineering 23
(1971) 83–88.
[23] M. Keane, Laboratory testing for design of thickener circuits, in: A.L. Mular,
M.A. Anderson (Eds.), Design and Installation of Concentration and
Dewatering Circuits, SME, 1986, pp. 498–505. Chapter 32.
[24] A. Gupta, D.S. Yan, Introduction to Mineral Processing Design and Operation,
Elsevier Science & Technology Books Publisher, Australia, 2006.
[25] B.A. Wills, T.J. Napier-Munn, Wills’ Introduction to Mineral Processing
Technology, seventh ed., Elsevier Science & Technology Books Publisher, 2006.
[26] B.V. Deventer, S.P. Usher, A. Kumar, M. Rudman, P.J. Scales, Aggregate
densification and batch settling, Chemical Engineering Journal 171 (2011)
141–151.
[27] B.M. Moudgil, B.D. Shah, Selection of flocculants for solid–liquid separation
processes, in: H.S., Muralidihara (Ed.), Advances in Solid–Liquid Separation,
Ohio, 1986, pp. 191–202.
[28] D.A. Dahlstrom, Selection of flocculants of solid–liquid separation equipment,
in: H.S., Muralidihara (Ed.), Advances in Solid–Liquid Separation, Ohio, 1986,
pp. 205–239.
[29] G.J. Kynch, A theory of sedimentation, Transactions of the Faraday society 48
(1952) 166–176.
[30] B. Fitch, Current theory and thickener design, Industrial and Engineering
Chemistry 58 (1966) 18–28.
[31] A.G. Waters, K.P. Galvin, Theory and application of thickener design, Filtration
& Separation 28 (1991) 110–116.
[32] C. Eswaraiah, S.K. Biswal, B.K. Mishra, Settling characteristics of ultrafine iron
ore slimes, International Journal of Minerals, Metallurgical and Materials 19
(2) (2012) 95–99.
[33] Y.A. Attia, Flocculation, in: J.S. Laskowski, J. Ralston (Eds.), Colloid Chemistry in
Mineral Processing, Elsevier Science & Technology Books Publisher, New York,
1992, pp. 277–308.
95
[34] R. Hogg, Flocculation and dewatering, International Journal of Mineral
Processing 58 (2000) 223–236.
[35] M.J. Pearse, Historical use and future development of chemicals for solid–
liquid separation in the mineral processing industry, Minerals Engineering 16
(2003) 103–108.
[36] I.G. Droppo, K. Exall, K. Stafford, Effects of chemical amendments on aquatic
floc structure, settling and strength, Water Research 42 (2008) 169–179.
[37] S.P. Usher, R. Spehar, P.J. Scales, Theoretical analysis of aggregate densification:
Impact on thickener performance, Chemical Engineering Journal 151 (2009)
202–208.
[38] P. Jarvis, B. Jefferson, J. Gregory, S.A. Parsons, A review of floc strength and
breakage, Water Research 39 (2005) 3121–3137.
[39] W. He, J. Nan, H. Li, S. Li, Characteristic analysis on temporal evolution of floc
size and structure in low-shear flow, Water Research 46 (2012) 509–520.
[40] O. Larue, E. Vorobiev, Floc size estimation in iron induced electrocoagulation
and coagulation using sedimentation data, International Journal of Mineral
Processing 71 (2003) 1–15.
[41] T. Li, Z. Zhu, D. Wang, C. Yao, H. Tang, The strength and fractal dimension
characteristics of alum–kaolin flocs, International Journal of Mineral
Processing 82 (2007) 23–29.
[42] Y. Wang, B.Y. Gao, X.M. Xu, W.Y. Xu, G.Y. Xu, Characterization of floc size,
strength and structure in various aluminum coagulants treatment, Journal of
Colloid and Interface Science 332 (2009) 354–359.
[43] M.S. Nasser, F.A. Twaiq, S.A. Onaizi, An experimental study of the relationship
between eroded flocs and cohesive beds in flocculated suspensions, Minerals
Engineering 30 (2012) 67–74.
[44] J.L. Chandler, Thickening, in: H.S. Muralidihara (Ed.), Advances in Solid–Liquid
Separation, Ohio, 1986, pp. 79–105.
[45] S. Banisi, M. Yahyaei, Feed dilution based design of a thickener for refuse slurry
of a coal preparation plant, International Journal of Coal Preparation and
Utilization 28 (2008) 201–223.
[46] F. Concha, N.N. Rulyov, J.S. Laskowski, Settling velocities of particulate systems
18: solid flux density determination by ultra-flocculation, International
Journal of Mineral Processing 104–105 (2012) 53–57.
[47] A.T. Owen, P.D. Fawell, J.D. Swift, D.M. Labbett, F.A. Benn, J.B. Farrow, Using
turbulent pipe flow to study the factors affecting polymer-bridging
flocculation of mineral systems, International Journal of Mineral Processing
87 (2008) 90–99.
[48] J.B. Farrow, J.D. Swift, A new procedure for assessing the performance of
flocculants, International Journal of Mineral Processing 46 (1996) 263–275.
[49] R. Hogg, P. Bunnaul, Sediment compressibility in thickening of flocculated
suspensions, Minerals and Metallurgical Processing, AIME Transactions 292
(1992) 184–188.
[50] K.L. Landman, L.R. White, R. Buscall, The continuous-flow gravity thickener:
steady state behavior, AIChE Journal 34 (1988) 239–252.
[51] I. Howells, K.A. Landman, A. Panjkon, L.R. White, Time-dependent batch
settling of flocculated suspensions, Applied Mathematical Modelling 14 (1990)
77–86.
[52] D.A. Dahlstrom, E.B. Fitch, Thickening, SME Mineral Processing Handbook, vol.
1, AIME, 1985, pp. 2–5. Chapter 9.
[53] Y. Zhang, A. Martin, P. Grassia, Mathematical modelling of time-dependent
densified thickeners, Chemical Engineering Science 99 (2013) 103–112.
[54] Y. Zhang, A. Martin, P. Grassia, Prediction of thickener performance with
aggregate densification, Chemical Engineering Science 101 (2013) 346–358.