University
stringclasses 19
values | Text
stringlengths 458
20.7k
|
|---|---|
Virginia Tech | flotation, which has been attributed to the exposure of asphaltene on a bitumen surface. As has
already been noted, the polar groups of asphaltene include carboxyl, amide, thiol, and hydroxyl
groups, which are more readily ionized at higher pHs, giving rise to an increased electrostatic
repulsive force. At the same time, the thickness of the asphaltene layer tend to increase with
increasing pH, causing stronger steric repulsion. Both of these forces contribute to the increased
stability of the wetting films formed on bitumen and hence to the slower kinetics of film thinning
at higher pHs.
4.3.4 Effect of Calcium Ions
Figure 4.5 shows the changes in the thickness of wetting films between the bitumen surface
and air bubble in 1 mM KCl solution at different CaCl concentrations. When the calcium
2
concentration was increased from 0 to 1 mM, the kinetics of film thinning slightly increased. As
the concentration was further increased to 10 mM, however, the kinetics decreased. The initial
increase in kinetics may be attributed to the likelihood that the Ca2+ ions react with the carboxylate
300
Ca2+ (mM)
0
0.1
1
200
10
)
m
n
(
h
100
0
0 3 6 9
t (s)
Figure 4.5 Effect of Ca2+ ions on the thinning kinetics of the wetting films formed
on bitumen in a 1 mM KCl solution at pH 8.4 to 7.5.
111 |
Virginia Tech | groups (RCOO−) of the asphaltene and, thereby, reduce the negative surface charge and hence the
positive disjoining pressure. At 10 mM calcium concentration, it is possible that calcium hydroxyl
or carbonate species adsorb on the surface, rendering the bitumen surface hydrophilic. In general,
the critical rupture thickness decreased with increasing calcium concentration, as shown in Figure
4.5, which may be attributed to the likelihood that in the presence of Ca2+ ions the wetting films
become thinner, subjecting the film to stronger hydrophobic forces. Overall, the changes in the
film thinning kinetics and the critical rupture thickness at different calcium concentrations were
not significant compared to other factors.
4.3.5 Effect of Magnesium Ions
Figure 4.6 shows the effect of MgCl concentration on the thinning kinetics of the wetting
2
films formed on bitumen in 1 mM KCl solution at pH 8.4-7.5. As shown, the addition of Mg2+ ions
had a relatively small effect on the kinetics of film thinning but a relatively large effect on rupture
300
Mg2+ (mM)
0
0.1
1
200
10
)
m
n
(
h
100
0
0 3 6 9
t (s)
Figure 4.6 Effect of Mg2+ ions on the thinning kinetics of the wetting films formed
on bitumen in a 1 mM KCl solution at pH 8.4 to 7.5.
112 |
Virginia Tech | time. The rupture time decreased substantially with increasing concentration of Mg2+ ions in
solution, which may also be attributed to the decrease in the electrostatic repulsion.
Masliyah, et al.3 studied the effect of Mg2+ and Ca2+ ions on laboratory flotation, and
showed that the effect was marginal. Thus, the results obtained in the present work is consistent
with the work of Masliyah et al. A detrimental impact was observed, however, with some ores in
the presence of montmorillonite clay.
4.3.6 Effect of Approaching Speed on Film Thinning
Much of the study conducted in the present work was to study the surface chemistry
variables. However, the interaction force between bubble and bitumen should also be affected by
the hydrodynamic force. To better understand the role of the hydrodynamic force, the velocity of
the bubble approaching a flat bitumen surface was varied during the film thickness measurement
using the BAP apparatus (Figure 4.1). Figure 4.7 compares the temporal and spatial profiles of the
wetting films formed at approach velocities of 0.2 and 4.6 µm/s in a 1 mM KCl solution with
natural pH of about 5.7. At the lower approaching velocity, the film thinned slowly and ruptured
at the critical film thickness (h ) of 40 nm in 2.78 s. At the higher approaching speed, the film
c
(a) 900 (b) 900
0.2 m/s 4.6 m/s
600 600 0 s
m) m)
n 0 s n
h
(
h
( 0.27
0.67
300 300 0.7
1.5
3.37
2.78
0 0
-50 -25 0 25 50 -80 -40 0 40 80
r (m) r (m)
Figure 4.7 Temporal and spatial profiles of the wetting films formed bitumen in a 1 mM KCl
solution at two different approach speeds: a) 0.2 and b) 4.6 m/s.
113 |
Virginia Tech | became flat initially and then began to form a dimple after 0.27 s. At longer thinning times, the
dimple became increasingly convex, forming a well-defined barrier rim at r = 73 µm. At 3.37 s,
the film ruptured catastrophically at the rim, with a rupture thickness of 70 nm. Similar
observations were made on hydrophobic surfaces by Pan et al.32
In Figure 4.8, the closest separation distances between bubble and bitumen have been
plotted as a function of time at the two different approach speeds employed in the present work,
i.e., 0.2 and 4.6 µm/s. Also shown in the figure is a plot of film thicknesses at the center of the
dimpled film. Initially, the dimpled film thinned faster than the non-dimpled film. After 1.7 s, the
non-dimpled film began to thin faster and ruptured before the dimpled film did. Does this mean
that less turbulence or weaker hydrodynamic force is better for bitumen aeration? It is difficult to
answer this question on the basis of a single set of experiments. In principle, dimpled films should
114
3
2
1
0
0
0
0
0
0
0
0 1 2 3
)
m
n
(
h
t ( s )
Figure 4.8 Comparison of the thinning kinetics obtained for dimpled and non-dimpled
wetting films in a 1 mM KCl solution. Black triangles represent the minimum film
thickness of the dimpled film, i.e., at the rim; black circles represent the film thicknesses
at the center of the dimpled films; and red circles, the minimum film thicknesses of the
non-dimpled films, i.e., at the center. |
Virginia Tech | thin more slowly because they have larger film radii than non-dimpled films, as shown in Figure
4.7. On the other hand, the film thinning kinetics should also depend on the magnitude of the
hydrodynamic force applied. Thus, there may be an optimum hydrodynamic force that can give
the fastest film thinning kinetics. Further investigation is needed to find optimum hydrodynamic
conditions for bitumen aeration.
4.4 Summary
In this study, the effects of solution chemistry on the thinning of wetting films formed
between bitumen and air bubble were studied. It was found that the kinetics of film thinning
decreased with increasing immersion time. In the presence of an electrolyte (KCl), the kinetics
increased due to double-layer compression. In the presence of KCl in the solution, the addition of
divalent cations such as Ca2+ and Mg2+ ions did not change the thinning kinetics significantly.
In addition, the effect of approach speed of air bubble to bitumen surface was investigated.
It was found that dimples were formed at high approach speeds, which is commonly observed with
other wetting and foam films. In general, film thinning rate is increased due to increased
hydrodynamic pressure; however, its effect is more significant on the rim of a film where the
curvature (or Laplace) pressure is higher than at the center of the film and hence the dimple
formation.
References
1. Long, J., Z. Xu, and J. Masliyah, On the role of temperature in oil sands processing,
Energy & fuels. 19(4), p. 1440-1446, 2005.
115 |
Virginia Tech | Chapter 5
Summary and Future Work
5.1 Research Revisits
Flotation is the major separation process used for the recovery of bitumen from oil sands.
The process involves two steps, bitumen liberation and aeration. The aeration step involves bubble
attachment on bitumen, followed by engulfment of the bubble by the bitumen drop. In the present
work, the kinetics of aeration (or bubble-bitumen interaction) have been studied. Most of the work
focused on the measurement of disjoining pressures of the water films of thickness h formed on
bitumen and on its hydrophobic and hydrophilic components, i.e., maltene and asphaltene. This
was accomplished using a modified thin film pressure balance (TFPB) and a new bubble against
plate (BAP) apparatus developed in the present work. Both were equipped with an optical
microscope, microinterferometry, and a high-speed video camera. The nano-scale images provided
information on the kinetics of film thinning in vertical directions, i.e., 𝜕ℎ⁄𝜕𝑡 and spatial variation
in radial directions, i.e.,𝜕ℎ⁄𝜕𝑟, which are then used to determine the disjoining pressures () in
TLFs.
It is believed that the results of the present study have improved the understanding of the
complex mechanisms involved in bubble-bitumen interactions, which is very different from the
bubble-mineral and bubble-coal interactions. With improved understanding, it is now possible to
develop a bitumen flotation model and a computer simulator that can have short-term industrial
applications.
The major research conclusions and suggested future work are presented below.
120 |
Virginia Tech | 5.2 Conclusions
It has been found that the film thinning rates increased in the order of asphaltene, bitumen,
and maltene, suggesting that the hydrophobicity of the substrates increases in the same order.
The experimental work has been carried out at temperatures ranging from 22 to 80 °C to
better understand the mechanisms involved in film thinning in the hot-, warm-, and cold-water
processing of bitumen. In general, the kinetics of film thinning increased with temperature,
indicating that the higher the temperature, the faster the film thinning and, hence, faster the aeration
kinetics. The results show also that the film thinning kinetics is sufficiently high at temperatures
as low as 35 °C. At lower temperatures, however, the film thinning and aeration kinetics are too
slow to warrant efficient flotation recovery. These results are consistent with the industrial
experience as reported in the literature.
The kinetic information on film thinning has then been used to determine the disjoining
pressure () in the wetting films. The results show that < 0 on maltene and bitumen, while >
0 on asphaltene. These results suggest that asphaltene may serve as a kinetic barrier for bubble-
bitumen interactions. In this regard, controlling asphaltene chemistry is important for the success
of bitumen flotation under various operating conditions.
The disjoining pressure data obtained in the present work have been analyzed in view of
the extended DLVO theory, which includes contributions from the electrical double-layer force,
van der Waals force, hydrophobic force, and steric force that are present in the wetting films
formed between air bubble and bitumen. It has been found that the negative disjoining pressure (
< 0) due to the hydrophobic force increases with increasing temperature, which agrees well with
the contact angles measured on thick films of bitumen. These results are also in agreement with
121 |
Virginia Tech | the industrial flotation practice. The disjoining pressure isotherms, established using the extended
DLVO theory, show also that steric force increases with increasing temperature.
The dynamic contact angle measurements conducted on bitumen show that, initially,
contact angles are small but subsequently increase above 90o, obviously due to the changes in the
interfacial tensions involved. It appears that the changes in interfacial tensions are brought about
by the diffusion of maltene from underneath the asphaltene layer formed on the surface of bitumen,
causing an increase in the surface hydrophobicity of bitumen. When the asphaltene layer is
saturated with maltene, an oily substance begins to coat the bubble surface. Simple thermodynamic
calculations based on the acid-base theory suggest that the asphaltene saturated with maltene,
rather than maltene, is more likely to coat the bubble surface. In effect, asphaltene is acting as a
spreading agent for maltene. Based on the data obtained in the present work, a model for the
bubble-bitumen interaction and subsequent engulfment has been proposed. The information
obtained in the present work will be useful for developing a bitumen flotation model and a
computer simulator that may be useful for industrial application near term.
5.3 Original Contribution
1. The present work represents the first measurement of disjoining pressure in the wetting
films of water formed on bitumen.
2. It has been established that the kinetics of the thinning of the wetting films formed on
bitumen and its components, i.e., asphaltene and maltene, increased with temperature in
the order of asphaltene, bitumen and maltene.
3. The disjoining pressure () in the wetting films of water deceases with increasing
temperature in the order of asphaltene, bitumen and maltene.
4. Bitumen is aerated when the disjoining pressure is negative, i.e., < 0.
122 |
Virginia Tech | 5. The disjoining pressure of the water films formed on asphaltene is positive, i.e., > 0,
which in turn stabilizes bitumen-in-water emulsions and prevents bitumen from aeration.
6. The bitumen droplets acquire hydrophobicity by diffusion of maltene into the layer of
asphaltene exposed on the surface.
7. Air bubble on bitumen is coated by asphaltene, or the bubble is engulfed into bitumen, in
which asphaltene helps the free energy of spreading to be negative, i.e., G < 0. Thus,
s
asphaltene acts effectively as a spreading agent for maltene on bubble surface.
5.4 Future Work
5.4.1 Studies of the Film Thinning Kinetics Using a New Cell
In the present work, the modified thin film pressure balance (TFPB) technique has been
used to study the temperature effect on the thinning of wetting films of water on bitumen and its
components. There are two limitations associated with this technique due to the design of the glass
cell used to form a wetting film: 1) only a small amount of water (less than 0.1 mL) is in contact
with the surface of interest; 2) the surface under investigation has to be flat.
In industry, bitumen is in contact with large reservoirs of aqueous solutions and are
suspended in the solutions as droplets. Earlier photomicrographic studies show that rising bitumen
droplets in a continuous pilot plant were approximately spherical under good processing
conditions.1, 2 Hence, a new experimental set-up is suggested to study the thinning of wetting films
of water formed between bitumen and air bubble. It will be useful to modify the bubble-against-
plate (BAP) apparatus described in Chapter 4 by replacing the flat bitumen substrate with a
bitumen droplet. The new set-up is shown schematically in Figure 5.1. It will be useful to study
the effects of temperature, bubble size, bitumen doplet size, approach speed, and solution
chemistry (e.g., pH, Ca2+, Mg2+, clays) on the thinning of the wetting film of water on bitumen.
123 |
Virginia Tech | Figure 5.1 New experimental set-up designed for the study of the thinning of wetting film of
water on bitumen.
In the TFPB technique, bitumen substrate was exposed to air before contacting the water
in the cell and this is inevitable. In this way, it is possible that submicroscopic air bubbles might
form on bitumen surface since the water contact angle on bitumen surface is high (~90°). With the
new cell, one can generate a bitumen droplet in situ, i.e., in the aqueous solution without contacting
air before measurements, and ex situ, i.e., in the air and then bring the droplet into the solution,
and then compare the two sets of data to see whether the possibly adsorbed air on the bitumen
surface plays a role in the bitumen-bubble attachment. The comparison would be useful to analyze
the hydrophobic force in the bitumen aeration.
It is also suggested to do the measurements in KCl or NaCl solutions for convenient
theoretical analysis afterwards. Under a given concentration of KCl or NaCl, it is easy to calculate
the Debye length needed to determine the disjoining pressure due to the electrostatic force. Real
124 |
Virginia Tech | commercial processing water used for bitumen extraction in plants contains more than 20 mM
NaCl.3, 4
5.4.2 Measurement of the Length of the Asphaltene Tails in an Aqueous Solution
In Chapter 3, the theoretical analysis about the disjoining pressure isotherm based on the
extended DLVO theory was conducted by fitting the experimental data, where the length of the
adlayer (L) of polymers present in bitumen was estimated for the sake of best fitting. It is useful to
experimentally measure the adlayer length of polymers on bitumen, asphaltene, and maltene
surfaces, respectively, in an aqueous solution under different conditions (i.e., different temperature,
pH, electrolyte concentration, etc.), in order to 1) confirm experimentally the exposure of
asphaltene on bitumen surface by the comparison of the results obtained on above there surfaces;
2) use these values in the fitting process to better understand the hydrophobic force between
bitumen and air bubble.
The polymer adlayer length can be conveniently measured using dynamic light scattering
technique, also known as photo correlation spectroscopy (PCS), where the hydrodynamic diameter
of polymer-covered particles is compared with that of bare particles.5 Viscometric methods can
also be used to determine the polymer tail length.6 Small angle neutron scattering (SANS) method
can be used to determine the polymer segment density distribution.7
5.4.3 Measurement of Surface Tension of Maltene and Maltene/water Interfacial Tension at
Different Temperatures
As mentioned earlier in Chapter 2, there’s lack of data about the surface tension of maltene
and interfacial tension between maltene and water at different temperatures. It is therefore
recommended to measure those in the future work. The results would be useful for the calculation
125 |
Virginia Tech | of spreading coefficient for the spreading of maltene onto air bubbles and hence provide insights
on the mechanisms of the engulfment of bitumen over air bubbles.
There are a number of methods to measure surface or interfacial tension, such as Wilhelmy
plate method, pendant drop method, Du Noüy ring method, etc.
5.4.4 Thermodynamic Studies on Surface Tension Components of Bitumen and Its Components
There is also lack of data about the surface tension components, i.e., acidic, basic,
dispersion, for bitumen, maltene, and asphaltene. Determination of these components would
provide useful information to the understanding of the bitumen-air attachment and the engulfment.
5.4.5 Aeration Studies in Micron Size
In the present work, the radius of the air bubble used in the TFPB or BAP technique is
around 2 mm, which is much larger than their actual sizes in industry. It was reported that in
bitumen recovery process the mean diameter of air bubble in a pilot plant was 216 ± 140 µm, and
that of bitumen droplet was about 280 ± 140 µm.1, 2 It is preferable to do the experiments using
bubbles in the size range of that in the pilot plant. It should be noted that although the mean bubble
size fell in the mentioned range, bubbles as large as 1000 µm and as small as < 100 µm were also
observed.2 Moran et al.4 used a micropipette technique to study the interaction between bitumen
droplets and air bubbles at the 10 µm scale, roughly 10 to 40 µm in diameter.
References
1. Houlihan, R.N., Unpublished results, Syncrude Canada Ltd, Edmonton, 1976.
2. Stevenson, P., Froth Flotation of Oil Sand Bitumen, in Foam engineering: fundamentals
and applications, John Wiley & Sons, West Sussex, United Kingdom, 2012.
126 |
Virginia Tech | FINE COAL DEWATERING USING
HYPERBARIC FILTER CENTRIFUGATION
by
Serhat Keles
ABSTRACT
The solid-solid separation processes employed by modern coal preparation plants require
large amounts of process water that must be removed from the surfaces of particles using
mechanical dewatering equipment. Unfortunately, the existing processes that are used to dewater
fine particles are inefficient in terms of moisture reduction and/or solids recovery. Many coal
preparation plants are forced to discard fine coal particles because of the inability of existing
technologies to reduce the moisture content of this product to an acceptable level. In light of this
problem, a new ultrafine dewatering process called hyperbaric filter centrifugation (HFC) has
been developed. This novel method combines centrifugation and pressure filtration within a
single process to substantially reduce moistures over what can be achieved using conventional
dewatering systems.
In the current study, empirical and theoretical dewatering models were developed in
order to be able to simulate the behavior of the HFC technology. The empirical model was based
on grain-size properties. On the other hand, the theoretical model was based on fundamental
theories of filtration and centrifugation. Although the theoretical model provided a better
understanding of the working principles of the process, the empirical model produced more
accurate equilibrium moisture predictions. Therefore, the empirical model was used to further
investigate the effects of several parameters on cake moistures. As such, the empirical model was
useful for scale up and design purposes. |
Virginia Tech | The empirical dewatering model was also used to perform an economical analysis of
potential applications of the HFC technology. The model was used to investigate a variety of
new circuit designs that have the potential to be commercially applied in the coal industry. The
results clearly showed that this new technology would allow coal companies to process difficult-
to-dewater ultrafines using the HFC process, while coarser solids would be more appropriately
dewatered using conventional technologies such as vacuum filters or screenbowl centrifuges.
This “split dewatering” concept would provide substantially higher profitability due to lower
moistures and higher recoveries of ultrafine solids than could be achieved using a single
dewatering process.
Laboratory- and pilot-scale versions of this technology has been constructed and tested at
the facilities of Mining & Minerals Engineering Department of Virginia Tech. Results of this
testing program showed that 30-50% lower moisture values than the ones obtained using
conventional mechanical dewatering processes could be achieved with the HFC technology.
Based on these promising results, a pilot-scale prototype unit, which was tested successfully at
several commercial U.S. coal plants, was also constructed by Decanter Machine, Inc. Finally, the
process of developing of this novel technology was successfully completed with the sale of the
first full-scale commercial unit by Decanter Machine, Inc. to a major U.S. coal producer.
iii |
Virginia Tech | ACKNOWLEDGEMENTS
I would like to express my deepest gratitude and thanks to my advisor and my mentor Dr.
Jerry Luttrell for his guidance and valuable insight throughout this research. He supervised this
dissertation with great patience and keen interest. He contributed to it through his invaluable
suggestions and criticism. The valuable comments and suggestions of the committee members
Dr. Roe-Hoan Yoon, Dr. Greg Adel and Dr. Rick Honaker are also gratefully acknowledged.
The financial support from the U.S. Department of Energy and Mining and Minerals
Engineering Department is greatly appreciated.
I would like to extend my sincere appreciation to Mr. Jim Waddell for helping me build
all the machines that were needed for testing and for fixing things that were constantly broken.
Also, many thanks to Mr. Bob Bratton for his help with my experiments and for staying late in
the office many times to help me figure out mistakes made in Excel spreadsheets.
I wish to express my appreciation to Dr. Kerem Eraydin for his help with the
experiments, for his valuable input and suggestions about chemicals, and for keeping me in
shape.
Alpha Natural Resources, Arch Coal, Inc. and their employees are also acknowledged for
providing much needed coal samples for my experiments.
Finally, I would like to thank my parents, my brother and my fiancé for their love,
continuous moral support and encouragements throughout this study. It would not have been
possible to complete this dissertation without them.
iv |
Virginia Tech | Table 1.1 Recoverable coal reserves at the end of 2008 (BP, 2009)
Recoverable Reserve
Country
(Million short tons)
United States 262,689
Russia 173,073
China 126,214
Australia 83,996
India 64,595
Ukraine 37,339
Kazakhstan 34,502
South Africa 33,519
Poland 8,269
Brazil 7,781
Table 1.2 Leading coal producers of the world and their production numbers in 2008
(Thousand short tons) (EIA, 2010)
Anthracite Bituminous Lignite
China 539,454 2,197,069 111,461
United States 1,701 1,094,098 75,684
India 0 532,782 35,541
Australia 218 358,480 79,807
Indonesia 0 304,997 8,234
Africa 1,434 264,601 0
Russia 10,828 261,533 83,824
South Africa 1,414 258,182 0
Kazakhstan 0 115,100 4,708
Poland 0 92,217 65,665
Even though the global coal consumption increase slowed in 2008, coal still was the
fastest growing fossil fuel with 3.1% growth among the other fuels. While the domestic coal
consumption decreased as a consequence of the slowing domestic economy and poor weather in
2008, production and exports increased to levels not seen in 30 years as a result of the increased
international demand (Figure 1.2). Furthermore, coal prices also increased because of rising
international demand and the increased cost of transportation due to elevated oil prices (Freme,
2 |
Virginia Tech | 2009). Therefore, coal is likely to remain as one of the most important energy sources in the U.S.
and in the world.
1400
1200
1000
800
600
1970 1980 1990 2000 2010
Figure 1.2 Coal production and consumption in the United States (EIA, 2010)
Much of the coal consumed in the U.S. is pretreated using coal preparation operations.
These facilities are used to remove unwanted mineral matter that otherwise would decrease coal
heating value, increase transportation costs, and adversely impact environmental emissions.
Unfortunately, most of the cleaning processes used in modern coal preparation facilities are
performed in an aqueous medium. As such, a variety of dewatering processes are used in coal
preparation plants to remove excess surface moisture from clean coal products. Coarser particles
can be readily dewatered using simple screening systems, while finer particles require costly unit
operations such as centrifuges and filters (Osborne, 1988). Moisture that is not removed by these
processes reduces the heating value and increases the cost of transporting the clean coal. Excess
moisture can also create unacceptable handling problems for both the coal producer and
downstream consumers by plugging chutes, bins, and rail cars. In colder regions, coal handling
3
snoT
noilliM
Production
Consumption
Year |
Virginia Tech | can be particularly severe during winter months because of freezing (Kaytaz, Acarkan, &
Halvorsen, 1994; L. R. Lyons, 1950).
The dewatering of fine (minus 0.150 mm) coal is particularly difficult due to the large
surface area of particles in this size fraction (Sandy, Matoney, & Dahlstrom, 1991). At present,
there are no mechanical dewatering processes that can inexpensively reduce the surface moisture
of fine coal to a level that is acceptable for most industrial consumers. This shortcoming often
forces U.S. coal producers to deslime their fine coal streams and to discard the ultrafine (minus
0.044 mm) fraction into refuse impoundments. Recent estimates for the U.S. coal industry
suggest that approximately 2 billion tons of fine coal has been discarded in abandoned ponds and
500-800 million tons are in active ponds (Orr, 2002). In addition, U.S. preparation plants discard
30-40 million tons of additional fresh fine coal to ponds each year. This activity represents a loss
of valuable natural resources, loss of profit for coal producers, and the creation of a potential
environmental liability.
1.1 Literature Review
1.1.1 Coal Dewatering
The solid-solid separation processes employed by modern coal preparation plants require
large amounts of process water. After cleaning, the unwanted water must be removed from the
surfaces of the particles. Only about less than 10% of the existing U.S. coal plants still utilizes
thermal dryers for moisture control (Bratton, 2010) because obtaining new permits for new
thermal dryers is very difficult. Therefore, dewatering operations are mostly conducted by
mechanical dewatering devices in modern coal preparation plants. Depending on the size of the
coal, several types of mechanical devices are employed for coal dewatering. Coal particles larger
than 5 mm are dewatered using screens. Shaking and vibrating screens are commonly used for
4 |
Virginia Tech | this purpose. Moreover, sieve bends are generally used for preliminary dewatering of coal prior
to vibrating screens and centrifuges. 5 mm x 0.5 mm size range is sent to basket type centrifuges
for dewatering. Screenbowl centrifuges are widely used in coal industry to dewater 1 mm x 0
size range of clean coal coming from froth flotation and spirals. However, the biggest
disadvantage of screenbowl centrifuges is that they discard approximately 50% of the minus
0.044 mm material. If high coal recovery is desirable, then the fine coal (0.5 mm x 0) can be
dewatered using vacuum filters. Two types of vacuum filters traditionally used are drum filters
and disc filters. Vacuum disc filters are the most commonly employed type in coal preparation
due to their large filter areas and small footprint (Dahlstrom & Klepper, 1982; Kaytaz, et al.,
1994; Sandy, et al., 1991).
While clean coal is dewatered using the above mentioned units, coal tailings also have to
be dewatered. The first step in fines tailing dewatering is sedimentation, which is done by
thickeners. In the second step, thickened refuse sludge is sent to a dewatering device for further
dewatering. Among all the available units, belt filter presses are the most commonly used
devices to dewater fine coal refuse due to their good performance with very fine clay, which is
often the main component of fine coal refuse. While the solid refuse material obtained is
discarded to a refuse pond, clarified water recovered from thickeners and filters is recycled back
to the plant to meet the water requirements of different processes (Dahlstrom & Klepper, 1982;
Osborne, 1988).
1.1.2 Filtration theory
Filtration is used to separate liquids from solids by passing the solid-liquid mixture
through a permeable medium, which accumulates most of the solid particles while permeating
the liquid. There are two types of filters:
5 |
Virginia Tech | • Surface filters accumulate solids in the form of a cake atop a thin filter medium. Filtration
conducted using surface filters is also called cake filtration.
• Depth filters accumulate the solids inside a thicker filter medium.
Surface filters are the most commonly used type of filters in coal preparation. Although, the filter
medium is very thin, the cake formed on top of the filter medium becomes an extension to the
original medium. Consequently, the cake acts as a depth filter and continues to accumulate
particles inside and on the top as the slurry flow continues (Osborne, 1988; Svarovsky, 1990).
Cake filtration consists of several stages. First, initial bridging of particles occurs across
the filter medium (Figure 1.3(a)). Then, cake formation starts and the cake gradually grows,
while single-phase flow of water passes through the filter medium (Figure 1.3(b)). After the cake
is formed (Figure 1.3(c)), cake compression occurs if the material is compressible (Figure
1.3(d)). This stage is often not important for coal applications because fine (i.e., 0.5 mm x 0) coal
is not considered compressible (Condie, Hinkel, & Veal, 1996). Up to this point, the process is
called filtration, as there is only single-phase flow of water. After that, the dewatering part of the
process begins. Two-phase flow of air and water empties the largest pores (Figure 1.3(e)) and
finally air breakthrough occurs (Figure 1.3(f)). In this stage, small pores still contain water. If
there is a significant amount of cake cracking as result of the air breakthrough, dewatering stops
because the vacuum pressure is lost (Condie, et al., 1996; Gala, Kakwani, Chiang, Tierney, &
Klinzing, 1981).
6 |
Virginia Tech | a 180°, the work of adhesion would be minimum. Therefore, the addition of proper surface-
active dewatering aids which increase the contact angle and reduce the surface tension can help
to reduce the work of adhesion, resulting in better dewatering performance.
1.1.3 Centrifugal Dewatering Theory
Centrifugal dewatering is a solid-liquid separation technique in which solid particles are
separated from a liquid by means of a combination of sedimentation and filtration mechanisms
driven by centrifugal force. Although gravitational sedimentation and centrifugation employs the
same basic principle, which is density separation, centrifugation is a much faster process because
of the magnitude of the centrifugal force applied. Most of the centrifugal dewatering devices
used in coal industry operate at 50-3000 times of gravitational force (Osborne, 1988).
Under the effect of centrifugal force, particles in suspension move away tangentially
from the rotation axis until they are stopped by the inner surface of the centrifuge, thus forming
an annular cake. If the centrifuge has a perforated wall, the cake acts as filter medium and allows
liquid to flow out; thus sedimentation and filtration are combined in this case. However, if the
centrifuge has a solid wall, then only sedimentation occurs (Osborne, 1988; Rushton, Ward, &
Holdich, 2000a).
Generally, the centrifugal force created by a centrifuge is expressed in terms of the
number of times this force exceeds the gravitational force, or “the number of g's”, and it is
formulated as follows:
(cid:18) (cid:18) [1.16]
4! "3
01 =
4
Where F is the centrifugal force, r is the centrifuge radius, N is the number of rotations per
c
second and g is the gravitational acceleration.
12 |
Virginia Tech | The particulate bed, or the cake, formed inside a centrifuge contains three types of water:
surface water, structure water and chemical water (Lowry, 1963). Chemical water is held in the
chemical structure of the solids and cannot be removed. Surface water that lies on the particle
surfaces can be removed easily by centrifugation, while the structure water that is found inside
the capillary structure of the cake is more difficult to remove (Wakeman, 1984).
The liquid content of a cake is the most important factor among the factors that induce
the capillary pressure. The liquid can be contained in a particulate bed in three states, i.e.
pendular, funicular and capillary states as shown in Figure 1.6 States of a liquid in particulate
bed. The relationship given in Figure 1.7 shows the capillary pressure for different states. As
seen from the figure, capillary pressure is the smallest at the capillary state. As some of the liquid
is drained from the cake, the capillary pressure increases and becomes far larger in magnitude at
the funicular state than at the capillary state. If liquid drainage is continued, the capillary
pressure increases rapidly and when the saturation of the particulate bed decreases to a certain
level. Under this condition, the centrifugal force will not be large enough to overcome the
capillary pressure; therefore, drainage of the liquid will stop (Hwang, Chu, & Lu, 2001).
(a) Capillary State (b) Funicular state (c) Pendular State
Figure 1.6 States of a liquid in particulate bed
13 |
Virginia Tech | Each main group has many types of machines as follows:
• Perforate basket type
o Without a transport device
o With a transport device
(cid:1) With positive discharge system
o Vibrating basket
• Bowl type
o Solid-bowl
o Screen-bowl
1.1.4.1 Perforate Basket Centrifuges
Perforate basket centrifuges without transport devices are the simplest centrifuge designs.
They consist of a conical perforated basket with an upward pointed apex of the cone. The coal is
accelerated by the rotating bowl and the water is drained through the perforations of the basket.
Coal slides down the inclined wall and is discharged at the base of the bowl. These centrifuges
can handle 10 mm to 1 mm size range at capacities up to 50 tph, producing 6-8% moistures by
weight (Osborne, 1988; Sandy, et al., 1991).
Perforate basket centrifuges with transport devices are commonly used in coal industry.
They consist of a positive discharge mechanism, which is a solid cone with helical flights, inside
a conical perforated bowl. Wet coal enters the apex of the bowl and is transported by the rotating
flights towards the base of the bowl where it is discharged. These centrifuges can handle 10 mm
to 0.5 mm size range at 80 tph capacity, producing 6-7% moisture by weight. Another design
with a transport device, which is a horizontally rotated machine, can handle finer coal feeds of
15 |
Virginia Tech | minus 0.5 mm at 35 tph capacity and can produce a cake containing 10-12% moisture (Osborne,
1988; Sandy, et al., 1991).
Vibrating basket centrifuges, which can have either a vertical or a horizontal cone shaped
basket, are also very commonly employed in coal industry. Perforated basket of the centrifuge
vibrates in such a way that causes the coal particles move towards the discharge point. These
machines do not rotate at as high speeds as the previously mentioned centrifuges. In the vertical
axis type, the coal particles enter the centrifuge from the top through a hopper, and travel
upwards on the perforated side walls by the vibrating action. Feed size for this type of centrifuge
can be as coarse as 75 x 6.7 mm, which can be handled at 200 tph to produce 2% moisture by
weight. A finer feed of 6.7 x 0.5 mm can also be handled at up to 110 tph capacity to produce
about 7.5% moisture by weight. In the horizontal axis type, the coal particles enter the apex of
the basket and the water is drained through the perforations on the walls. The vibrating action of
the basket distributes and transports the coal through the machine. The horizontal axis type
machines have the similar capabilities to vertical types (Osborne, 1988; Sandy, et al., 1991).
1.1.4.2 Bowl Type Centrifuges
Bowl type centrifuges, which are also called as decanter centrifuges, were started being
used in the coal industry in mid 1960s with the introduction of solid-bowl centrifuge (Figure 1.8)
for treating coal refuse. These centrifuges contain two rotating elements, namely the conveyor
and the bowl. The bowl consists of a long cylindrical shaped region and a shorter cone shaped
region. The conveyor with one or more helical flights that follow the contour of the bowl rotates
at a slightly slower or faster speed than the bowl to transport the material. The transport speed is
controlled by a mechanical gear reducer that operates at a fixed gear ratio. Clearance between the
flights and the bowl is kept to a minimum. Solid particles are transported by the conveyor up the
16 |
Virginia Tech | slope of the cone above the liquid level, which is controlled by weirs at the other end of the
centrifuge. Since solid particles can be very abrasive, conveyor flights are usually lined with a
wear resistant material (Osborne, 1988; Sanders & Sutherland, 2001; Sandy, et al., 1991).
1. Feed Pipe
2. Thrust Bearings
3. Driven Sheave
4. Main Bearings
5. Vibration
Isolators
6. Feed
Compartment
7. Screw Conveyor
8. Planetary
Gearbox
9. Back drive
Figure 1.8 Components of a solid-bowl centrifuge (Decanter Machine, 2010)
A bowl centrifuge can have either a concurrent or a countercurrent feed arrangement. In
the concurrent feed design, the material enters the centrifuge at the larger cylindrical section of
the bowl and the cake moves in the same direction as the centrate towards the conical end.
Concurrent machines operate at slower speeds than countercurrent machines, therefore they are
found to be attractive in disposal of coal tailings where higher product moistures (35-45%) can
be accepted. In the countercurrent design, the material enters the centrifuge near where the
conical section starts and the cake moves in the opposite direction to centrate flow. Higher wear
and more power consumption are unavoidable with countercurrent centrifuges due to the higher
speeds at which they operate. However, lower moistures can be achieved as a result of the much
higher speeds they operate.
17 |
Virginia Tech | A screenbowl centrifuge (Figure 1.9) has an added cylindrical screen section to the solid
bowl that allows for further water drainage. The screen is made up of tungsten carbide bars that
have wedge profiles to prevent near size solid particles from getting stuck between the bars. By
design, a screenbowl centrifuge is a countercurrent machine. Feed slurry enters the acceleration
chamber, where it is brought up to speed, through a stationary pipe. Then, it is distributed inside
the bowl through the feed ports. Solids settled under the acceleration force are carried up the
slope of the cone by the helical conveyor, as in solid-bowl centrifuges. However, unlike solid-
bowl centrifuges, thickened cake of solids pass over a screen section where the remaining excess
water is filtered through the cake and discarded by the screen openings. Since the liquid drained
through the screen section has such a small volume compared to the feed slurry and it can
contain useful material, this liquid is usually recycled back to the feed (Osborne, 1988; Policow
& Orphanos, 1983).
1. Feed Pipe
2. Thrust Bearings
3. Driven Sheave
4. Main Bearings
5. Vibration Isolators
6. Feed
Compartments
7. Screen Wash
Header
8. Screw Conveyor
9. Dewatering Screen
10. Planetary
Gearbox
11. Torque Sensor
Figure 1.9 Components of a screenbowl centrifuge (Decanter Machine, 2010)
18 |
Virginia Tech | 1.1.5 Hyperbaric Centrifugation
Injection of air into a centrifuge to improve dewatering performance was first described
in the literature by Veal et al. (1998, 1999). This technology consists of the injection of a
turbulent stream of high velocity air through the coal bed formed inside a vibrating basket
centrifuge. The turbulent flow strips water from the surface of the particles. This technique can
be effective when used with a particle size range of 0.5 to 50 mm. However, it is not applicable
for fine coal.
Friedmann et al. (2004) conducted hyperbaric centrifugation experiments with a beaker
centrifuge using EPDM elastomer particles (ethylene-propylene-diene-polymethylene
elastomers), which form a highly compressible filter cake. They concluded that the hyperbaric
centrifugation can provide better results in solid-liquid separation for highly compressible
materials. They obtained the best results with respect to dewatering rate when a superimposed
static pressure was used at lower centrifugal accelerations.
According to Darcy’s law (Eq. [1.1]) the driving force for the flow is the pressure
differential. In vacuum filtration, the pressure differential is less than 1 atm, while in pressure
filtration it can be as high as 10 atm or more. In centrifugal filtration, the layer of water atop the
filter cake creates the pressure differential needed to dewater the particle bed. This pressure can
be calculated using:
[1.17]
1 (cid:18)
∆(cid:7) = 56 ℎ
2
in which ρ is the density of water, ω the angular velocity, and h is the thickness of the water
above the cake surface. One can see that ∆P becomes zero when h becomes zero (i.e., when the
water disappears from the cake surface). In fact, the pressure inside the filter cake becomes
lower than the ambient pressure when the water level drops below the surface of the cake as
19 |
Virginia Tech | shown in the example given in Figure 1.10 (Zeitsch, 1990). In the figure, the horizontal axis
represents the distance from the axis of rotation and vertical axis shows the pressure differential
across the filter cake. The cake surface is 0.6 m away from the axis of rotation. The water level
first forms close to the center of rotation and proceeds towards the cake surface with time.
Pressure differentials for three cases are provided in the figure. The top line represents the case
when the water level is 0.5 m away from the center of rotation. The middle and the bottom lines
show the pressures when the water level is 0.55 m and 0.60 m away from the center of rotation,
respectively. As can be seen, the pressure profile across the cake decreases as the water level
moves towards the cake surface. Eventually, the pressure drops below ambient, essentially
creating a negative pressure, when the water level reaches the cake surface (at 0.60 m). This
phenomenon greatly reduces the rate of dewatering during the drainage period of the centrifugal
filtration process. In order to compensate the loss of the pressure, a higher centrifugal
acceleration is typically recommended. Unfortunately, this action has practical limitations and
400
300
200
100
0
0.5 0.6 0.7 0.8 0.9
Figure 1.10 Pressure in a centrifuge filter cake under ambient pressure (Zeitsch, 1990)
20
)aPk(
erusserP
Cake Waterlayer thickness Filter
Surface above the cake: Medium
0.10 m
0.05 m
0 m
Radius Coordinate (m) |
Virginia Tech | typically leads to higher machine wear and maintenance requirements.
Another method of increasing the driving force for dewatering is to introduce compressed
gas into the centrifuge chamber. The compressed gas increases the pressure differential across
the filter cake and increases the rate of dewatering in accordance with Darcy’s law (Eq. [1.1]). A
larger pressure differential is useful for increasing the rate of dewatering during the drainage
period, which is critical in achieving lower cake moistures.
Using the concepts introduced in the preceding paragraphs, Yoon and Asmatulu (2002)
developed a novel centrifugal dewatering device for fine coal dewatering. The novel feature of
this device is the use of air pressure as an additional driving force to the centrifugal force for
removing water from a fine coal cake. With this technology, the rate of dewatering increases due
to the increased driving force. Consequently, laboratory test results showed that hyperbaric
centrifugation could provide substantially better moisture reductions than with air pressure or
centrifugation alone (Asmatulu, Luttrell, & Yoon, 2005).
1.2 Objectives
The objectives of this study are:
• To fully develop a novel fine coal dewatering technology which can be employed
commercially to recover the finest coal particles that were once discarded due to their high
moisture content and lack of economic value. This goal has been achieved by:
o conducting laboratory-scale parametric studies with batch test units so that an
improved understanding of the parameters affecting hyperbaric centrifugation could
be fully investigated,
o conducting semi-continuous bench-scale and fully-continuous pilot-scale tests to
establish engineering criteria for the design and operation of a prototype unit, and
21 |
Virginia Tech | o conducting a demonstration of the technology at an industrial site using a full-scale
continuous prototype constructed by a commercial partner.
• To successfully develop a dewatering model that can be used to accurately simulate the
hyperbaric centrifugation process by extending and improving existing centrifugal
dewatering models, and
• To demonstrate the capability of this novel technology for reducing environmental impacts
associated with coal production while simultaneously creating a potential source of new
revenue and profit for coal producers.
1.3 Organization
This dissertation consists of seven chapters. The first chapter gives an introduction to the
research and provides detailed background information obtained from various literature sources.
While the second chapter represents and discusses all the laboratory-scale test results obtained,
the third chapter gives details of the pilot-scale test program. The fourth chapter provides the
details and results of field tests conducted with a prototype unit. The fifth chapter discusses the
theoretical basis of this research project and provides two models developed in order to simulate
this novel dewatering process. The sixth chapter provides an economical evaluation of this novel
process via a case study. Finally, the seventh chapter summarizes all the findings of this research
and development project.
22 |
Virginia Tech | CHAPTER 2 A LABORATORY STUDY OF HYPERBARIC
CENTRIFUGATION
2.1 Introduction
Yoon et al. (2001; 2005) performed several series of laboratory tests to demonstrate the
effectiveness of hyperbaric centrifugation for fine coal dewatering. Based on the results obtained,
a pilot-scale test unit was constructed to study this novel technology furthermore. However, as it
will be explained in Chapter 3, the originally planned complete testing program could not be
finished because of the many technical problems faced with the pilot-scale unit. To overcome
this problem, Decanter Machine, Inc., which is the largest producer of screenbowl centrifuges in
the U.S., was chosen and licensed as a partner for developing a pilot-scale prototype hyperbaric
centrifuge. While construction of the pilot-scale prototype was being completed, it was decided
to conduct a more detailed laboratory-scale testing program to fully understand the hyperbaric
centrifugation process. For this purpose, several laboratory-scale test units were constructed and
tested using various coal samples obtained from different coal preparation plants.
2.2 Experimental
2.2.1 Samples
All slurry samples used for laboratory studies were clean coal samples (i.e. screenbowl
feed or flotation product) obtained from different commercial coal plants. These samples
typically contained 8-12% ash. A low-ash (<10%) unprocessed dry coal sample from Cucumber
Mine in West Virginia was also obtained for testing. All slurry samples were split upon receipt
by a custom-made slurry splitter (Figure 2.1) down to the amount required. This step was found
to be essential to have a representative sample of the slurry so that reproducible results could be
27 |
Virginia Tech | obtained. Past experience showed that when
dealing with a coarser sample (i.e. 1 mm
topsize), sampling from a mixed bucket was
not a trustworthy method and often resulted in
biased results. The use of a hand-held scoop to
collect feed samples was found to produce
erroneous and non-reproducible results. The
dry coal sample was also split using a riffle
sample divider. After splitting, all samples
Figure 2.1 Slurry splitter
were screened to determine size distributions.
Slurry samples were tested within 2-3 weeks upon receipt to avoid changing of the sample due to
surface oxidization.
2.2.2 Equipment Design and Setup
2.2.2.1 Hyperbaric Centrifuge Apparatus
Figure 2.2 shows a schematic of the batch test unit used in the laboratory-scale test
program. The unit was constructed with a 4-inch diameter and 3-inch height using stainless steel.
Evenly spaced holes with 3.0, 2.4 and 1.6 mm diameters were drilled on the sidewalls for water
drainage from the cake. The unit was used in two different setups with two different pressure lids
for a variety of tests.
28 |
Virginia Tech | 10 12
1 – rotating assembly
11
2 – perforated cylinder
9
8 3 – mounting shaft
4 – set screw
5 – mounting bolt
6 – filter cloth
5 7 – thickened slurry/cake
7
8 – pressure lid
9 – lid clamping bolts
10 – compressed air inlet (stationary)
4 11 – high speed bearings
6
2
12 – air control valve
3
1
Figure 2.2 Schematic of the hyperbaric centrifuge unit used in standard batch tests
The first setup was used for the standard batch tests described in Section 2.3.1. For this
setup, the centrifuge basket was placed inside in a Beckman variable-speed centrifuge machine
(Figure 2.3). This “ultracentrifuge” is capable of high rotational speeds of up to approximately
20,000 rpm. The basket was attached to the rotating shaft by means of a mounting bolt. Also, a
set screw was used to secure the basket in place to avoid any balancing issues that might arise at
high rotation speeds. Since the basket was sealed inside the centrifuge machine (Figure 2.4), very
high rotation speeds (and high-g levels) were possible without any safety concerns. Two high-
speed ball bearings were attached to the lid in order to insert and support a pressurized air line.
The air line was equipped with an on/off valve and a pressure gauge for monitoring and
controlling air pressure. Since, this test setup did not also allow for collecting the effluent, coal
recovery values could not be determined from these batch experiments.
29 |
Virginia Tech | In most of the tests, the centrifuge basket was lined with a filter cloth that had a 0.015
mm aperture size. However, a somewhat coarser filter cloth (unknown aperture size) and a 1.5
mm screen mesh were also used to study the effect of filter medium on product moisture and
solid recovery. The first test setup described above did not allow for feeding the unit while the
basket was rotating. Therefore, the basket was filled with coal slurry before the lid was securely
closed in every test. Since only a limited amount of slurry could fit inside the basket, all the
excess water contained in slurries had to be removed before testing. This was accomplished by
allowing solid particles to settle followed by removing the excess water using a syringe. On the
other hand, in the second test setup, the coal slurry could be fed to the unit while the basket was
rotating. A peristaltic pump was used for this purpose so as to minimize degradation of particle
size.
2.2.3 Hyperbaric Centrifuge Apparatus with Screen Bars
2.2.3.1 Screen Section
After experimenting with the hyperbaric batch filter centrifuge, another small scale batch
unit was built using tungsten carbide bars in the screen section. These bars were the same as
those employed in commercial screenbowl centrifuges. The aim of this study was to replace the
screen mesh or the filter cloth, which cannot be used commercially, with a wear resistant part
that is widely used in the industry. The test unit, which is shown in Figure 2.7, was constructed
from aluminum with 4.25-inch height, 5-inch outside diameter and 3.9-inch inside diameter.
Slots of approximately 1/8-inch wide were cut on the sidewalls of the basket for water removal.
T-shaped tungsten carbide bars (Figure 2.8) were glued inside the basket. These bars were
designed by the manufacturer in such a way that when two of them are put together, the gap
between them expands towards the outside (i.e., similar to the configuration used in a wedge-
32 |
Virginia Tech | The test unit was mounted on a vertical rotor shaft and fixed in place using a set screw
(Figure 2.9). A variable speed motor was used to control the rotational speed of the centrifuge. A
high-speed bearing was attached to the lid, in order to connect a pressurized air line. The air-line
was equipped with an on/off valve and a pressure gauge so that the air pressure could be
accurately monitored and controlled.
2.2.3.2 Solid-Bowl Section
One of the aims of this project was to use a feed finer than what is normally used with
screenbowl centrifuges. Some preliminary tests were run with the batch bench-scale unit with
screen bars using minus 0.15 mm flotation product, but all the particles passed through the
openings. Trials with flocculant addition also failed, because the agglomerates formed were not
large enough to be retained in the centrifuge. Commercially available screenbowl centrifuges
contain a solid-bowl section before the screen-section so that the feed can be thickened and
excessive water and ultrafine particles can be discarded for more efficient dewatering. Based on
this fact, a separate solid-bowl centrifuge was constructed to produce a thickened-feed for the
batch bench-scale centrifuge with screen bars.
The solid-bowl unit (Figure 2.10) was constructed from aluminum with 4.25-inch height,
5-inch outside diameter and 4.4-inch inside diameter. The unit was mounted on a vertical shaft
and fixed in place using a set screw. A variable-speed motor was used to control the rotational
speed of the centrifuge. A high-speed bearing was attached to the lid, in order to connect a slurry
feed line. Slurry was gravity fed into the unit. The lid of the unit (Figure 2.11) was designed in
such a way that excess water could escape the unit easily while solid particles form a cake under
the lid. Also, four small threaded-holes were drilled on the sides near the top of the centrifuge, in
34 |
Virginia Tech | All tests that were conducted with this unit consisted of two steps. In the first step,
threaded holes on the sides of the bowl were sealed using bolts, so that the water could only go
out of the holes on the top, and the centrifuge was run until there was no water leaving. Then, in
the second step, bolts on the sides were unscrewed and the centrifuge was run again to let the
remaining free water leave the unit.
2.3 Results
2.3.1 Standard Batch Tests
2.3.1.1 Effect of Coal Type
This parametric study of hyperbaric centrifugation was carried out using coal samples
from various coal preparation plants. Screenbowl feed samples taken from Tom's Creek, Moss
No. 3, Coal Clean and Weatherby coal plants, all of which process different coal seams, were
used for these tests. Tests were run as function of centrifugal acceleration (i.e. 1500, 2000 and
2700 g), rotation time (i.e. 30, 60 and 120 second) and air pressure (with and without air pressure
of 200 kPa). Air pressure was on for the whole duration of the tests with air. The particle size
distributions of all the different coal samples were determined by wet screening (see Table 2.1).
After sizing, an anionic flocculant was added to all samples for the purpose of thickening. The
Table 2.1 Size distributions of the screenbowl feed samples obtained from different plants.
Tom's Creek Coal Clean Weatherby Moss #3
Size Coal Plant Coal Plant Coal Plant Coal Plant
(Mass, %) (Mass, %) (Mass, %) (Mass, %)
+0.5 mm 39.05 45.40 53.83 1.80
0.5 x 0.25 mm 21.43 17.22 18.43 7.81
0.25 x 0.15 mm 6.21 5.44 5.76 6.04
0.15 x 0.075 mm 10.40 8.58 8.62 13.24
0.075 x 0.044 mm 5.99 5.63 4.69 13.91
-0.044 mm 16.92 17.74 8.67 57.20
36 |
Virginia Tech | excess water was decanted and discarded.
Test results obtained with the Tom's Creek coal sample are given in Table 2.2. The data
showed that single-digit moistures were easily achievable when 1500 g's or more of centrifugal
acceleration was used, even without applying air pressure. The baseline product moisture without
air was 11.5% for the lowest g level and shortest rotation time. This value was reduced down to
7.3% moisture at the highest g level and the longest rotation time. These values were further
reduced from 11.5% to 8.3% and from 7.3% down to 6.2%, respectively, when air pressure was
applied. The results also showed that the effect of air pressure on moisture reduction was more
significant at lower rotation speeds (lower g levels). For example, at 1500 g’s and 60 seconds of
rotation time, the moisture was reduced from 9.9% down to 8.1% with air, corresponding to a 1.8
percentage point reduction. However, at 2700 g’s, the moisture was reduced from 8.5% down to
7.0% with air, corresponding to a lower moisture reduction of 1.5 percentage points. Also, the
moisture reduction dropped for all g levels as the rotation time increased from 30 to 120 seconds.
Table 2.2 Test results obtained with Tom's Creek coal plant screenbowl feed (1 mm x 0).
Cake Moisture
Without Air With Air
Rotation 1500 2000 2700 1500 2000 2700
Time g's g's g's g's g's g's
(s) (%) (%) (%) (%) (%) (%)
0 47.7 47.7 47.7 47.7 47.7 47.7
30 11.5 10.3 9.6 8.3 8.1 8.1
60 9.9 9.4 8.5 8.1 7.7 7.0
120 8.9 7.8 7.3 7.0 7.1 6.2
The Coal Clean sample had a very similar size distribution to that of the Tom's Creek
sample. Even though the samples were taken from coal plants that were processing different coal
37 |
Virginia Tech | seams, the test results were very similar owning to the similar size distributions. As shown in
Table 2.3, baseline moistures obtained ranged from 11.5% for the lowest g level and shortest
rotation time down to 8.1% for highest g level with the longest rotation time. Moreover, similar
to the results obtained with Tom’s Creek sample, the moisture reduction obtained by applying air
pressure dropped with increased g levels. For example, the moisture at 1500 g’s and 60 seconds
of rotation time was reduced from 11.0% down to 8.5% with air, corresponding to a 2.5
percentage point moisture reduction. Whereas, at 2700 g’s, the moisture was reduced from 8.9%
down to 7.4%, corresponding to a lower moisture reduction of 1.5 percentage points.
Table 2.3 Test results obtained with Coal Clean coal plant screenbowl feed (1 mm x 0,
Eagle Seam).
Cake Moisture
Without Air With Air
Rotation 1500 2000 2700 1500 2000 2700
Time g's g's g's g's g's g's
(s) (%) (%) (%) (%) (%) (%)
0 50.0 50.0 50.0 50.0 50.0 50.0
30 11.5 10.8 10.0 9.3 8.8 8.0
60 11.0 9.9 8.9 8.5 7.8 7.4
120 9.9 9.1 8.1 7.4 7.1 6.8
The coal sample from the Weatherby plant was partially deslimed at 0.044 mm as a result
of the use of fine classifying cyclones utilized in the plant circuitry. Because the sample was
deslimed, all the moisture values obtained with this coal product were less than 10%. While the
baseline moisture for the lowest centrifugal field of 1500 g's and the shortest rotation time of 30
seconds was 9.4%, for the highest centrifugal field of 2700 g's and the longest rotation time of
120, it was 7.4%. When air pressure was introduced, moistures decreased down to only 8.7% and
6.5%, respectively, for the same conditions. Therefore, the lack of fines reduced the moisture
38 |
Virginia Tech | drop and severely limited the impact of air injection on moisture reduction. The smallest
moisture drop, which was 0.3 percentage points, was obtained with 2700 g's and 30 seconds
rotation time. The largest moisture drop was only about 1.0 percentage point at 2000 g's and 120
seconds rotation time.
Table 2.4 Test results obtained with Weatherby coal plant screenbowl feed (1 mm x 0,
Deslimed, Stockton Seam).
Cake Moisture
Without Air With Air
Rotation 1500 2000 2700 1500 2000 2700
Time g's g's g's g's g's g's
(s) (%) (%) (%) (%) (%) (%)
0 42.0 42.4 42.4 42.4 42.4 42.4
30 9.4 9.1 8.5 8.7 8.1 8.2
60 9.0 8.4 8.2 8.2 7.9 7.2
120 8.4 8.1 7.4 7.4 7.1 6.5
The Moss No. 3 sample was a finer sample compared to the other samples with 57.2%
minus 0.044 mm particles. As a result of the increased surface area, product moistures also
increased to values more than two times the values obtained with the other samples (see Table
2.5). For the lowest g level and the shortest rotation time, the baseline cake moisture was 26.5%.
This value was reduced down to 20.2% for the highest g level and the longest rotation time.
When air pressure was introduced, these values decreased further from 26.5% down to 21.0%
and from 20.2% down to 15.5%, respectively. Greater moisture reductions compared to the other
samples were also observed as a consequence of the increased surface area. For example, at 1500
g’s and 60 seconds of rotation time, the moisture was reduced from 25.3% to 19.0% with air,
corresponding to a 6.3 percentage point reduction. Similar to the tests with other samples,
moisture reductions with air were dropped at higher g levels. For instance, at 2700 g’s and 60
39 |
Virginia Tech | seconds of rotation time, the moisture was reduced from 21.7% to 16.7% with air, corresponding
to a lower moisture reduction of 5.0 percentage points.
Table 2.5 Test results obtained with Moss No. 3 coal plant screenbowl feed (1 mm x 0).
Cake Moisture
Without Air With Air
Rotation 1500 2000 2700 1500 2000 2700
Time g's g's g's g's g's g's
(s) (%) (%) (%) (%) (%) (%)
0 67.3 67.3 67.3 67.3 67.3 67.3
30 26.5 24.3 22.6 21.0 19.6 18.0
60 25.3 23.4 21.7 19.0 18.4 16.7
120 24.0 22.1 20.2 17.6 16.1 15.5
2.3.1.2 Effect of Dewatering Aids
Two sets of flotation concentrate samples (nominally 0.15 x 0 mm) obtained from the
East Gulf Plant of United Coal Company and the Cardinal Plant of Arch Coal Company, both
located in West Virginia, were used to test the effect of dewatering aids on cake moistures. The
main difference between the samples was the size distribution, as the East Gulf sample was much
finer than the Cardinal sample (see Table 2.6). While the East Gulf sample contained 43.24%
Table 2.6 Size distributions of the flotation concentrate samples obtained from two plants
for dewatering aid tests.
East Gulf Cardinal
Size Coal Plant Coal Plant
(Mass, %) (Mass, %)
+0.3 mm 0.34 5.04
0.3 x 0.15 mm 1.61 30.57
0.15 x 0.075 mm 17.47 30.43
0.075 x 0.044 mm 17.33 11.93
0.044 x 0.025 mm 20.01 9.74
-0.025 mm 43.24 12.29
40 |
Virginia Tech | minus 0.025 mm material, the Cardinal sample’s minus 0.025 mm material content was only
12.29%. All the centrifuge tests were run at a relatively lower centrifugal acceleration level of
500 g's to show the effect of dewatering aids, because higher centrifugal accelerations tend to
reduce the effect of other parameters. Also, rotation time was kept at 60 seconds and the air
pressure was kept at 200 kPa for the tests with air. For comparison, vacuum filtration tests were
conducted using the same samples. Several dewatering aids (i.e., reagents U, V and W) and
proprietary shear resistant flocculants (i.e., Nalco 71300, 71303, 71306 and 71321) were used in
several of the experiments.
Table 2.7 Effect of dewatering aids on moisture (East Gulf Coal Plant flotation concentrate
sample).
Air Chemical Chemical Moisture
Pressure Type Dosage* Moisture Reduction
(kPa) (lbs/t) (%) (%)
0 - 0 40.8 0.0
200 - 0 25.1 38.5
0 RU 1 40.2 1.3
0 RU 2 38.6 5.4
0 RU 4 29.7 27.3
200 RU 1 24.0 41.0
200 RU 2 20.1 50.8
200 RU 4 18.4 54.8
0 RV 1 41.7 -2.4
0 RV 2 41.7 -2.4
0 RV 4 39.8 2.4
200 RV 1 24.7 39.3
200 RV 2 24.8 39.1
200 RV 4 23.6 42.1
0 RW 1 39.6 2.9
0 RW 2 39.7 2.6
0 RW 4 37.7 7.5
200 RW 1 22.8 44.0
200 RW 2 20.0 50.9
200 RW 4 17.5 56.9
* Dosage of active chemical
41 |
Virginia Tech | Centrifugation test results obtained with the East Gulf sample are given in Table 2.7. All
the tests were conducted using 1, 2 and 4 lbs/t dosages of reagents U, V and W. Tests without
any chemical addition were also performed. Moisture reductions compared to the baseline test,
which was run without dewatering aids and without air pressure, are shown in the same table, as
well. While the baseline test produced 40.8% moisture, the lowest moisture value of 27.9%
without air injection was obtained using 4 lbs/t of reagent U corresponding to 27.3% moisture
reduction. While reagent W slightly reduced the moisture down to 37.7%, reagent V had almost
no effect and it produced 39.8% moisture at 4 lbs/t dosage. In fact, at the lower dosages of
reagent V, moisture slightly increased compared to the baseline. When air pressure was applied,
25.1% cake moisture, which corresponds to 38.5% moisture reduction, was obtained without any
dewatering aids. Reagents U and W had similar impacts on water removal and produced 18.4%
and 17.5% moistures, respectively. This level of performance corresponds to moisture reductions
of about 55-56%. Reagent V, on the other hand, caused only a slight moisture reduction
compared to the no chemical case and produced 23.6% cake moisture at the highest dosage.
Vacuum filtration test results obtained with the East Gulf Plant sample are shown in
Table 2.8. Since reagent V did not have any effect in centrifuge tests, it was not included in
filtration tests. Vacuum pressure was kept at 133 kPa for all the tests. Two cake thicknesses of 12
mm and 22-24 mm were tested. Since reagent U gave the best results in centrifuge tests and since
there was a limited amount of coal sample left, only reagent U was tested with both cake
thicknesses. Test results given in Table 2.8 show that when the cake thickness was 12 mm,
moisture was reduced to about 25% with both reagents U and W compared to the baseline
moisture of 30.3%. Also, when a 22-24 mm cake thickness was produced, the baseline moisture
of 36.5% was reduced to 31.5% with reagent U.
42 |
Virginia Tech | Table 2.8 Vacuum filtration test results (East Gulf Plant flotation concentrate sample)
Vacuum Chemical Chemical Cake Filtration
Pressure Type Dosage Thickness Time Moisture
(kPa) (lbs/t) (mm) (s) (%)
133 - 0 11 120 30.3
133 RU 2 12 120 26.3
133 RU 4 12 120 24.9
133 RW 2 12 120 25.7
133 RW 4 12 120 25.0
133 - 0 22-24 300 36.5
133 RU 2 22-24 240 32.8
133 RU 4 22-24 240 31.5
Several series of centrifugation tests were conducted using samples from the Cardinal
Plant. The first set was performed to study the effect of reagents U, V and W on cake moisture.
Table 2.9 shows the test results obtained. Since this sample was coarser than the East Gulf
sample, the need for the chemical dosage reduced due to the smaller specific surface area
(surface area per unit weight) of the sample. Therefore, 0.3, 1.0 and 1.7 lbs/t dosages were used
for these experiments. While the baseline test produced 20.0% moisture, similar moisture values
of approximately 15-16% and 11.5% were obtained when no air pressure was applied with all
dewatering aids at the lowest and the highest dosages, respectively. When air pressure was
applied, reagents U and W produced similar moisture values at all dosages. However, the lowest
moisture value of 7.8% was obtained using reagent W at the highest dosage of 1.7 lbs/t.
Moistures produced using reagent V were slightly higher than the ones with reagents U and W,
but still lower than the moisture of 13.4%, which was produced without chemicals.
43 |
Virginia Tech | Table 2.9 Effect of dewatering aids on moisture (Cardinal Coal Plant flotation concentrate
sample)
Air Chemical Chemical Moisture
Pressure Type Dosage* Moisture Reduction
(kPa) (lbs/t) (%) (%)
0 - 0 20.0 0.00
200 - 0 13.4 33.19
0 RU 0.3 14.9 25.7
0 RU 1.0 12.2 38.9
0 RU 1.7 11.7 41.5
200 RU 0.3 10.4 47.9
200 RU 1.0 8.4 58.2
200 RU 1.7 8.3 58.6
0 RV 0.3 14.9 25.8
0 RV 1.0 12.9 35.6
0 RV 1.7 11.4 43.1
200 RV 0.3 11.1 44.4
200 RV 1.0 10.4 48.0
200 RV 1.7 8.9 55.7
0 RW 0.3 16.1 19.5
0 RW 1.0 11.3 43.3
0 RW 1.7 11.6 42.0
200 RW 0.3 10.5 47.4
200 RW 1.0 8.8 55.9
200 RW 1.7 7.8 60.8
* Dosage of active chemical
Another set of tests was run with the Cardinal sample to study the effect of different
rotation times when a dewatering aid was used. Since reagents U, V and W had similar effects in
the previous set of tests, only reagent U at a mid-dosage of 1 lbs/t was used. Moreover, all the
tests were run without air. Test results given in Table 2.10 showed that when no chemical was
used, the moisture content was highly correlated to rotation time, as the moisture decreased from
35.9% at 15 seconds rotation time down to 21.7% at 60 seconds rotation time. However, when
reagent U was added to the feed slurry, even at the shortest rotation time, moisture was reduced
down to 14.8%, which corresponds to about 21 percentage points difference compared to no
chemical case. Even though the difference was less at the highest rotation time, reagent U still
44 |
Virginia Tech | produced a lower moisture value of 12.7% compared to the no chemical case. Therefore, while
the product moistures were reduced by using a dewatering aid, the amount of time required for
dewatering was also lowered.
Table 2.10 Effect of rotation time with 1 lbs/t Reagent U addition
Rotation Chemical Chemical
Time Type Dosage* Moisture
(s) (lbs/t) (% by weight)
15 - 0 35.9
30 - 0 28.1
60 - 0 21.7
15 RU 1 14.8
30 RU 1 12.6
60 RU 1 12.7
* Dosage of active chemical
The last set of centrifugation tests with the Cardinal Plant sample were conducted using
proprietary shear resistant flocculants 71300, 713003, 71306 and 71321 obtained from Nalco
Company. Test results given in Table 2.11 showed that using a proper dosage is of utmost
importance when dealing with flocculants. When no air was used, employing flocculants reduced
the moisture compared to the no chemical case at even the lowest dosage. However, while an
additional moisture reduction was obtained with all flocculants when the dosage increased from
0.00033 to 0.00055 lbs/t, increasing the dosage further to 0.0011 lbs/t resulted in elevated
moistures. For example, when 71321 was used, the baseline moisture of 22.2% was reduced to
17.1% and 16.8%, at 0.00033 lbs/t and 0.00055 lbs/t dosages, respectively. However, the
moisture value increased to 17.3% when 0.0011 lbs/t dosage employed. The same trend was also
observed when air pressure was used, only lower overall moistures were obtained due to the
elevated air pressure.
45 |
Virginia Tech | Several series of vacuum filtration tests were conducted with the Cardinal Plant sample.
The first and second sets were performed to provide a comparison to the results given in Table
2.9. While reagents U, V and W were used in the first set of tests, reagent U was not used in the
second set of tests because of the limited amount of sample available for testing. Also, only 0.3
and 1 lbs/t dosages were used in the second series of tests. The first set of tests was conducted at
120 seconds of filtration time, whereas the second set was performed at 60 seconds of filtration
time. As expected, the former produced lower moisture values due to the longer filtration time.
At 120 seconds of filtration time, reagent W produced the lowest moistures, ranging from 11.2%
at the highest chemical dosage to 12.7% at the lowest chemical dosage. Generally, moistures
decreased as the dosage increased from 0.3 to 1.7 lbs/t for all the chemicals tested. In fact, all
moistures produced with chemicals were lower than the baseline moisture of 15.0%. At 60
seconds filtration time, moistures decreased as the chemical dosage increased, resembling the
trends observed in the first set of tests. While reagent W produced the lowest moisture of 14.2%
at 1 lbs/t dosage, the moisture obtained with reagent V was 16.7% at the same dosage. The
baseline moisture of 18.7% was the highest value obtained.
Results showed that when 120 seconds filtration time was used, vacuum filtration
produced lower moistures than centrifugation alone at the lowest chemical dosage. For instance,
at the lowest dosage of reagent W, centrifugation alone produced 16.1% moisture. Whereas, a
lower moisture value of 12.7% was produced with vacuum filtration at the same conditions.
However, moisture values obtained at higher dosages with centrifugation were either similar to
or lower than the ones obtained with vacuum filtration. For example, for the same conditions
given above, centrifugation with air pressure produced 10.5% moisture. When 60 seconds
47 |
Virginia Tech | filtration time was employed, even the moistures produced with only centrifugation were lower
than vacuum filtration.
The last set of vacuum filtration tests were performed with flocculants described earlier.
Again, because of the limited amount of available samples, only flocculants 71303 and 71321,
which performed the best at centrifugation tests, were evaluated at 0.00033 and 0.00055 lbs/t
dosages. As in the case with the centrifugation tests, flocculant 71321 produced the lowest
moistures. However, this time, the lowest moistures were obtained at 0.00033 lbs/t dosage, rather
than 0.00055 lbs/t, which was the optimum dosage in the centrifuge tests. The application of
flocculants made an improvement over the baseline moisture of 18.3%. For this particular
sample, moistures of 15.3% and 17.2% were produced using flocculants 71321 and 71303,
respectively, at a dosage of 0.00033 lbs/t. While the moistures obtained with vacuum filtration
were generally lower than centrifugation only results, the application of air pressure together
with centrifugation improved moistures and resulted in 25-30% moisture reductions compared to
vacuum filtration.
Table 2.12 Vacuum filtration test results at 120 s filtration time (Cardinal Plant flotation
concentrate sample)
Vacuum Chemical Chemical Cake Filtration
Pressure Type Dosage Thickness Time Moisture
(kPa) (lbs/t) (mm) (s) (%)
133 - 0 12 120 15.0
133 RU 0.3 12 120 13.4
133 RU 1.0 12 120 12.1
133 RU 1.7 12 120 12.7
133 RV 0.3 12 120 13.8
133 RV 1.0 12 120 13.3
133 RV 1.7 12 120 13.4
133 RW 0.3 12 120 12.7
133 RW 1.0 12 120 12.1
133 RW 1.7 12 120 11.2
48 |
Virginia Tech | Table 2.13 Vacuum filtration test results at 60 s filtration time (Cardinal Plant flotation
concentrate sample)
Vacuum Chemical Chemical Cake Filtration
Pressure Type Dosage Thickness Time Moisture
(kPa) (lbs/t) (mm) (s) (%)
133 - 0 12 60 18.7
133 RV 0.3 13 60 17.2
133 RV 1.0 13 60 16.7
133 RW 0.3 13 60 14.5
133 RW 1.0 13 60 14.2
Table 2.14 Vacuum filtration test results with flocculants (Cardinal Plant flotation
concentrate sample)
Vacuum Chemical Chemical Cake Filtration
Pressure Type Dosage Thickness Time Moisture
(kPa) (lbs/t) (mm) (s) (%)
133 - 0 12 60 18.3
133 71321 0.00033 13 60 15.4
133 71321 0.00055 13 60 15.9
133 71303 0.00033 13 60 17.8
133 71303 0.00055 13 60 18.4
2.3.1.3 Effect of Air Pressure
To study the effect of air pressure on cake moisture, an additional set of dewatering tests
was conducted using the East Gulf Plant sample. The tests were performed as a function of
centrifugal acceleration and air pressure. Rotation time was held constant at 60 seconds for all
tests. The results given in Table 2.15 showed that increasing air pressure beyond 100-200 kPa at
500 g’s decreased the product moisture from 26.1% to 23.4%. Since this improvement
corresponds to only about 2.7 percentage point moisture reduction, the use of high air pressures
may not be necessary for effective dewatering. Moreover, at 3000 g's, moistures obtained at
49 |
Virginia Tech | different air pressures were practically identical. Therefore, 100-200 kPa of air pressure appears
to be adequate for hyperbaric centrifugation.
Table 2.15 Effect of air pressure on cake moisture
Centrifugal Air
Acceleration Pressure Moisture
(g's) (kPa) (% by weight)
500 0 40.8
500 100 26.1
500 200 25.8
500 400 23.4
3000 100 15.4
3000 200 14.2
3000 400 15.8
2.3.1.4 Effect of Particle Size Distribution
Another set of laboratory tests was performed to investigate the effect of feed particle size
distribution on cake moisture. To minimize the impacts of heterogeneities, a low-ash
unprocessed coal sample taken from Cucumber Mine in West Virginia was used in these
experiments. The sample was screened at 1.2 mm to eliminate large chunks of coal. After
screening, the dry sample was split down to about 400 grams. Each split was then ground in a
laboratory ball mill for different periods of time, varying from 30 seconds to 2 hours, to obtain
progressively smaller size distributions. Using these samples, hyperbaric centrifuge tests were
run at 500 and 2700 g's. These centrifugal accelerations were purposely chosen to imitate the
accelerations applied by the standard screenbowl centrifuges and the new Decanter CentribaricTM
centrifuge (described later in this document). An air pressure of about 200 kPa was applied
during the tests. The air injection and rotation time was held constant at 120 seconds for all the
tests.
50 |
Virginia Tech | The data given in Table 2.16 showed that moisture correlated very well with particle size.
For instance, when the centrifugal acceleration was 500 g’s, the coarsest product had 9.8%
moisture with air and 13.8% moisture without air, while the moisture went up to 37.7% with air
and 44.9% without air with the finest cake. The effect of different size fractions on final moisture
is more clearly illustrated in Figure 2.12 and Figure 2.13. These plots show the total moisture
content obtained as a function of the dry mass percentage below a given particle top-size. As
seen from the plots, the correlation between final product moisture and mass percent passing was
the poorest for the coarsest (i.e., minus 0.6 mm) size fraction (Figure 2.12(a) and Figure 2.13(a)).
The best correlation was obtained with the finest (minus 0.025 mm) size fraction (Figure 2.12(f)
and Figure 2.13(f)). Therefore, this fraction was chosen to further investigate the impact of size
distribution on cake moisture.
Figure 2.13(f) shows that at 2700 g's, when no air pressure was used, the moisture content
increased linearly with an increase in the amount of minus 0.025 mm particles. When air
pressure was applied, there was a constant 2 percentage-point moisture improvement until the
minus 0.025 mm solids accounted for about 60% of the total sample mass. After this point, the
moisture remained steady around 25% with air injection, while it kept increasing without air
injection. The absolute difference between adding air and not adding air was about 12 absolute
percentage points when the amount of minus 0.025 mm size fraction reached 99%.
Figure 2.12(f) shows that, at 500 g's, all the moisture values increased due to the decrease
in centrifugal acceleration. The moisture difference between with and without air cases started at
about 4 percentage points for the coarsest cake and increased up to about 17 percentage points
when the minus 0.025 mm solids accounted for about 70% of the total sample mass. After this
point, the difference started to decrease, because of the rapid change in moisture when air
51 |
Virginia Tech | pressure was applied. As a result, decreasing the centrifugal acceleration from 2700 to 500 g's
rendered air less effective when the cake was too fine, thus causing the two performance curves
to merge.
2.3.2 Modified Batch Tests
2.3.2.1 Effect of Media Type and Dewatering Kinetics
Several series of tests were run to examine the kinetics of hyperbaric centrifugation. The
coal used for these tests was the screenbowl feed sample (1 mm x 0 mm) taken from Tom's
Creek Coal Plant, which was described in section 2.2.1. The first two sets were run with either a
standard filter cloth, while a screen mesh with 1.5 mm aperture was used as the filter media for
the last set of runs. The feed charge was held relatively constant at 485-495 gm using either 29%
or 35% solids contents. Two different levels of centrifugal acceleration (i.e., 300 and 800 g’s)
were used during the tests. Flocculant and coagulant were added to all test samples, in order to
increase the solids recovery. Air pressure was also varied from 0 to 200 kPa.
After a few trial tests, it was seen that the test setup was not precise enough to record the
small amount of water weight removed from the cake by air injection. Therefore, in the first set
of tests, instead of recording the weight continuously, the drum was removed and weighed every
10 to 20 seconds after the air injection started until the end of the test run. Since this procedure
was time consuming and the curves needed were obtained using the first set of tests, only the
final weight of the drum was measured for the rest of the tests.
The test results given in Tables 2.17-2.19 compare the cake moisture and solid recovery
values under different conditions. The biggest impact of air injection was achieved when filter
cloth was used, due to higher solid recoveries. While the maximum moisture drop was 6
55 |
Virginia Tech | 2.3.2.2 Solid Bowl Tests
This set of tests was run to identify optimum conditions for producing a thickened feed in
a solid-bowl centrifuge. Ideally, this material would be passed to a filter-screen section at a high
solids content to ensure that a high recovery of solids could be maintained. The test unit
described previously in section 2.2.3.2 was used for this portion of the experimentation.
Parameters examined in the study included feed charge, feed rate, feed percent solids and
centrifugal acceleration. A flotation product sample (0.15 x 0 mm) taken from Tom’s Creek coal
plant was used for the tests.
The results given in Table 2.18 and Figures 2.17-2.24 showed that changing the amount
of feed charge had almost no effect on either the percent solids of the product or the solids loss to
Table 2.18 Solid bowl test results
Feed Feed Centrifugal Feed Feed Product Solids loss
Charge Charge Force Rate Solid Content Solid Content to Effluent
(l) (dry g) (g's) (%) (%) (%) (%)
0.50 78.03 760 10 15.0 64.3 3.0
0.70 109.25 760 10 15.0 64.1 2.8
1.00 156.07 760 10 15.0 64.3 2.8
1.25 195.09 760 10 15.0 65.9 2.8
1.50 234.10 760 10 15.0 62.9 2.9
1.00 156.07 100 10 15.0 N/A 100.0
1.00 156.07 365 10 15.0 64.7 3.6
1.00 156.07 760 10 15.0 67.0 2.7
1.00 156.07 1045 10 15.0 66.0 2.0
1.00 156.07 1080 10 15.0 67.3 1.5
1.00 156.07 760 10 15.0 65.3 2.6
1.00 156.07 760 30 15.0 67.4 4.0
1.00 156.07 760 50 15.0 64.2 4.0
1.00 156.07 760 70 15.0 64.7 2.9
1.00 156.07 760 100 15.0 74.4 5.5
3.13 156.07 760 10 5.0 66.8 3.8
1.52 156.07 760 10 10.0 65.4 2.8
1.00 156.07 760 10 15.0 64.7 2.6
0.58 156.07 760 10 25.0 62.6 0.7
0.35 156.07 760 10 40.0 66.0 0.4
60 |
Virginia Tech | 80
70
60
0 10 20 30 40 50
Figure 2.23 Effect of feed solid content on Figure 2.24 Effect of feed solid content on
product solid content solids loss to effluent
the effluent. In contrast, the loss of solids to effluent significantly decreased from about 4%
down to 1.5% when the centrifugal acceleration was increased. At the same time, the solids
content of the product slightly increased from about 64% to 67%, while the loss of solids
increased significantly from about 2.5% to 5.5%, due to the increasing feed rate. Finally,
increasing the feed solids content decreased the solids loss from about 4% down to nearly zero,
while the product solids content held constant at about 65%.
2.4 Discussion
2.4.1 Solids Recovery and Product Moisture
Standard batch tests conducted with different coal types showed that the product moisture
greatly depended on particle size distribution of the feed slurry. Employing finer feed samples
resulted in higher product moistures as a result of the increased specific surface area. Moreover,
the application of air pressure was more effective when the feed was finer, which suggests that
the higher capillary pressure created by smaller capillaries due to finer particle sizes could only
be overcome by applying pressurized air in addition to the centrifugal acceleration.
Moisture reductions obtained using slurries with different size distributions at various
levels of centrifugal accelerations are given in Figure 2.25. According to these plots, similar
62
)%(
tnetnoC
diloS
Effect of Solid %
6
4
2
0
0 10 20 30 40 50
Solid %
)%(
ssoL
diloS
Effect of Solid %
Solid % |
Virginia Tech | moisture reductions were obtained using 500 and 2700 g's when the product contained less than
40% minus 0.025 mm material (compared to the baseline moisture obtained at 500 g's without
using air pressure). Therefore, it may be unnecessary to use a costly high-g machine when the
same level of performance can be obtained using a cost effective lower-g machine for products
containing less than 60% minus 0.025 mm material. However, when the product contained more
than 60% minus 0.025 mm material, applying 2700 g's with air was more effective than the other
options. These results indicate that the hyperbaric centrifuge is best suited for the dewatering of
coals containing substantial amounts of ultrafine solids.
50
40
30
20
10
0
0 20 40 60 80 100
Figure 2.25 Effect of amount of fines on moisture reduction
Tests conducted with various filter media showed that solid recovery can drop
dramatically (down to 50-60%) when a conventional screenbowl feed is used with a coarse
screen mesh. Fortunately, solids recoveries were improved by using filter cloth without greatly
impacting moisture. However, since it is not possible to use filter cloth in industrial applications,
63
)%(
noitcudeR
erutsioM
500 g's (With Air)
2700 g's (Without Air)
2700 g's (With Air)
minus 0.025 mm (%) |
Virginia Tech | reported to effluent at all the conditions tested. Generally, solids loss to the effluent increased as
the residence time decreased. Also, as shown in Figure 2.27, the loss of plus 0.044 mm particles
to effluent was found to be correlated to the feed time and, therefore, the feed rate. While the loss
was lower than 0.2% at feed times longer than 0.25 minutes, it increased up to 1.5% as the feed
time became shorter.
2.4.2 Reproducibility of Test Results
Although a full experimental design was not performed for a complete statistical analysis,
some of the tests described in section 2.3.1.4 were repeated to determine the reproducibility of
the data. Average moisture values were calculated for each repeated test and these values were
used in the charts given in Figure 2.28. Error bars for each repeated test were then determined by
adding and subtracting three times the standard deviation value to this average value, which gave
a 99.7% confidence interval. As can be seen, the reproducibility of results without air pressure
was high. The calculated relative errors approximately 0.3 percentage points for 2700 g's and 0.7
percentage points for 500 g's when no air was used. Although the addition of air pressure
(a) (b)
50
40
30
20
10
0
0 50 100
Figure 2.28 Charts showing reproducibility of test results at (a) 2700 g's and (b) 500 g's
65
)%(
erutsioM
50
Without Air
40
With Air
30
20
10
0
0 50 100
minus 0.025 mm (%)
)%(
erutsioM
Without Air
With Air
minus 0.025 mm (%) |
Virginia Tech | increased the relative error values to about 0.7 and 1.2 percentage points for 2700 g's and 500 g's,
respectively, these variations were still far from being very significant when compared to
variations resulting from manual changes to air pressure or centrifugal acceleration.
2.4.3 Engineering Criteria for Machine Design
2.4.3.1 Rate Constants
The kinetic tests run with different filter media showed that the kinetics of centrifugal
dewatering is quite different than filtering. The following filtration equation was derived by
Dahlstrom from the commonly employed Poiseuille equation (Dahlstrom, 1985):
[2.1]
(cid:3) 9:;(cid:2) 9"
= +
(cid:2)⁄(cid:8) 2∆(cid:7)(cid:8) ∆(cid:7)
where V is the volume of filtrate, A is the area of filtration, t is the time, ∆P is the pressure drop
across the filter, µ is the liquid viscosity, a is the specific cake resistance, w is the weight of dry
cake solids per unit volume of filtrate and r is the resistance of filter cloth and drainage network.
Eq. [2.1] yields a straight line plot of t/V/A vs. V/A as given in Figure 2.29. However,
when the same plot was created using the data obtained from centrifugal kinetics tests (Figure
2.30), it was not possible to obtain a straight line relationship, because centrifugal acceleration
0.8
0.6
0.4
0.2
0.0
0 50 100 150 200 250
Figure 2.29 t/V/A vs. V/A (filtration theory) Figure 2.30 t/V/A vs. V/A (Measured)
66
V/t
0.8
0.6
0.4
0.2
0.0
0 50 100 150 200 250
V
V/t
V |
Virginia Tech | changes with time as opposed to the constant pressure used in conventional filtering. Therefore,
the Dahlstrom filtration equation cannot be used directly to describe hyperbaric centrifugation.
In the current study, a multiple-term rate equation was used to describe the centrifugal
dewatering process. The model assumes that a portion of water was removed quickly from the
cake, while another portion was removed slower. Also, another portion of water was assumed to
be non-removable by mechanical means, thus staying inside the cake as final product moisture.
The resultant dewatering equation can be expressed as:
[2.2]
(cid:26)⁄(cid:26)< = ∅> ∙@AB(cid:14)−(cid:5)> ∙(cid:3)(cid:17)+∅C ∙@ABD−(cid:5)C ∙(cid:3)E+∅F
where W is the weight of water remaining in cake, W is the initial weight of water in feed, K
0 s
and K are the rate constants for slow removed and fast removed water from the cake,
t
respectively; φ , φ and φ , are the proportions of slow removed, fast removed and not removed
s f n
water and t is the time. In general, this expression provided a good correlation between the
weight of water remaining in the cake and the dewatering time.
During all of the kinetic tests, the centrifuge was run for a certain amount of time to
remove most of the water that could be removed by only centrifugation. Then, air pressure was
applied to remove an additional amount of residual water. Therefore, curve fitting was done for
every centrifugal acceleration applied using all available data in the relevant set by simply
eliminating the values obtained with air. The results of curve fitting based on Eq. [2.2] are given
in Figures 2.30-2.32. Even though the moisture and recovery values were reasonable for Tests 3
of the first and second sets, kinetic data recorded were not in agreement with rest of the tests run
at the same speed. Also, data for Test 3 of the second set of tests contained an excess amount of
scatter. This problem can attributed to the difficulty of recording data with a sensitive digital
balance connected to centrifuge running at thousands of rpm. Therefore, the data for Test 3 from
67 |
Virginia Tech | both sets were not included in the plots, since, possibly, for these particular tests, the weights
recorded were not accurate.
The kinetic test results showed that the first and second sets of tests that were run with
filter cloth had only one dewatering rate, while the third set run with the screen mesh had two
rates. More solids and more fines were recovered during the tests with filter cloth, because filter
cloth had a much smaller opening size than screen mesh. After a certain amount of time, fines
blinded the filter cloth and dewatering rate diminished; thus, there was no slow rate. On the other
hand, the screen mesh was coarse enough for dewatering not to stop, but continue at a slower
rate. Therefore, the tests with screen mesh had two different dewatering rates.
0 50 100 150 200
1.0
0.1
Figure 2.31 Rate curve fit for the first set of kinetic tests at 300 g's
68
W/W
0
300 g's (Set 1)
Estimated
Test 1
Test 2
Test 4
(cid:26)⁄(cid:26)< = 0.810∙@AB(cid:14)−0.096∙(cid:3)(cid:17)+0.190
Time (sec) |
Virginia Tech | As explained in Section 2.3.2.1, the effect of pressurized air injection was studied by
measuring cake weight at every 10-20 seconds (Figure 2.14(d) and (f)) at 300 g's. These tests
readily demonstrated that air injection provided a further decrease in product moisture. Eq. [2.2]
was applied to these data to determine the kinetics of water removal by pressurized air injection
during centrifugation. The results given in Figure 2.34 suggest that this process was also a two-
rate process. The bulk of the water was removed quickly during the first 10 seconds of air
injection with a very high rate constant of 9.0 sec-1. After 10 seconds, air lost its effectiveness
and the rate constant declined dramatically down to about 0.2 sec-1. Therefore, these results
indicate that injecting air for a short amount of time can be just as efficient as applying air
pressure during the entire centrifugation time.
0 10 20 30 40 50 60 70
1.00
0.10
Figure 2.34 Rate curve fit for the air injection part of the kinetic tests at 300 g's
71
W/W
0
300 g's (Air injection)
Test 1
Test 4
Estimated
(cid:26)⁄(cid:26)< = 9.000∙@AB(cid:14)− 0.804∙(cid:3)(cid:17)+0.202∙@AB(cid:14)−0.062∙(cid:3)(cid:17)
+0.134
Time (sec) |
Virginia Tech | 2.5 Conclusions
A detailed parametric study of hyperbaric centrifugation was conducted using several
laboratory-scale test units. Tests performed with different coal types showed that it was possible
to achieve very low moistures using hyperbaric centrifugation. Moreover, tests performed using
different size distributions showed that as long as the product contained less than about 60%
minus 0.025 mm particles, a lower centrifugal acceleration of about 500 g's with air pressure was
as effective as employing a high centrifugal acceleration of 2700 g's. Therefore, it was concluded
that when the fines content of a product was below a certain amount, the use of a high-g
centrifuge is probably not necessary. However, when the minus 0.025 mm particle content was
above 60%, the use of a high centrifugal acceleration together with air pressure was found to be
essential to achieve good moisture values.
Tests conducted using several surface active dewatering aids and flocculants showed that
moisture reductions as much as about 60% (compared to no-chemical and no-air cases) were
possible when dewatering aids were used along with air pressure. Although it was not possible to
measure solid recoveries because of the limitations of the test setups, it was visually observed
that the use of chemical additives (especially flocculants) vastly improved solids recoveries.
Test results obtained with the modified batch test setup showed that centrifugal
dewatering process could be described as a kinetic process. While the bulk of the water was
removed quickly from the cake during centrifugation, a smaller amount was removed at a slower
rate and a certain amount was not removed at all, which represented the final moisture content of
the product. Therefore, it was necessary to employ a rate equation with multiple rate constants to
represent the different stages of cake formation and moisture removal.
72 |
Virginia Tech | CHAPTER 3 A PILOT-SCALE STUDY OF HYPERBARIC
CENTRIFUGATION
3.1 Introduction
In the light of previous studies done on hyperbaric centrifugation at Virginia Tech
(Asmatulu et. al., 2005), a two-stage pilot-scale unit was designed and built. The design of the
pilot-scale unit incorporated two integrated processes: (i) filtration in a pressurized rotating filter
basket (pressure chamber) with scraper discharge and (ii) slurry pre-thickening and retreatment
of recycled fines in an intermittent feed tank. The rotating pressure chamber is mounted along
the central axis of a hollow shaft. This shaft is driven by a variable speed motor. A second
hollow shaft is located within the drive shaft to allow feed and air to enter the pressure chamber.
The second shaft is also connected to two sets of end plates that serve to seal the pressure
chamber on one end and to discharge the cake on the other. A conceptual drawing of the rotating
filter basket is shown in Figure 3.1. For comparison, a photograph of the assembled pilot-scale
unit is provided in Figure 3.2.
3.2 Experimental
During a typical production cycle, the basket (see Figure 3.4) is rotated at speeds up to
319 g’s using a variable speed motor and drive belt (see Figure 3.5). Slurry from the feed tank is
pumped into a distributor pipe located within the center of the shaft of the rotating basket. The
basket is constructed using wedge-shaped tungsten carbide bars spaced at 0.1 mm intervals. All
particles larger than the bar spacing are retained along with most of the fines (via entrapment in
the cake). The feeding cycle is continued until a desired cake thickness is achieved (Figure
3.3(a)). The feed flow is then diverted back to the feed tank via an automated three-way valve
74 |
Virginia Tech | (see Figure 3.6). Compressed air is then introduced into the filter chamber to create a differential
pressure across the cake. The combined force of the centrifugal field and air pressure drives out
nearly all of the surface moisture and some ultrafine solids through the spacing between the
basket bars (Figure 3.3(b)). This effluent stream is passed back to the feed tank where the solids
are concentrated by flocculation and recycled back to centrifuge to maintain a high recovery of
solids. After this pressure drying cycle, the air pressure is released and the cake is discharged by
a pneumatically driven scraper (Figure 3.3(c)). The edges of the scraper are sealed by rubber O-
Rings to provide an excellent seal and to serve as a replaceable wear element. After discharge is
completed, the scraper returns to its original position and the entire cycle of dewatering is
repeated. The entire operation requires 30-120 seconds, depending on the range of operating
conditions selected, feed particle size distribution, and target moisture level. The entire
sequencing of the feeding, drying and discharge cycles is fully automated and controlled using a
programmable logic controller (see Figure 3.7).
75 |
Virginia Tech | After conducting the tests that are given in sections 3.3.1 through 3.3.3 with the setup
mentioned above, the feeding mechanism was modified to allow recycling of the effluent slurry
that drains through the screen section back to the feed inlet of the pilot-scale unit. The recycling
of the effluent for a certain amount of time was found to be necessary in order to increase the
capture of fine solids by the cake. With this system, a known amount of coal slurry was pumped
from a 6-inch diameter by 17-inch tall feed tank to the centrifuge unit. Effluent was collected in
the same tank and was pumped back to the centrifuge. After the effluent became clear, the valve
at the bottom of the tank was closed and pumping stopped.
Coal samples from two different coal plants were used to test the pilot-scale unit. The
first sample, which was used with tests described in sections 3.3.1 through 3.3.4, was a non-
deslimed screenbowl centrifuge feed (nominally 1 mm x 0) obtained from Tom's Creek Coal
Preparation Plant. The second sample, which was used with the tests mentioned in section 3.3.5,
was a deslimed screenbowl centrifuge feed (nominally 1 mm x 0.15 mm) from Pinnacle Pond
Reclamation Plant. The size distributions of the samples are given in Table 3.1.
Table 3.1 Size distributions of samples tested with the pilot-scale unit
Size Tom's Creek Pinnacle
(%) (%)
+0.6 mm 26.54 16.08
0.6 x 0.15 mm 33.53 72.38
0.15 x 0.044 mm 17.76 5.48
-0.044 mm 22.17 6.06
79 |
Virginia Tech | 3.3 Results and Discussion
3.3.1 Effect of Rotation Speed and Rotation Time
Filling time, which controls the amount of feed slurry entering the test unit, was kept
constant at 15 seconds for all test runs. Unfortunately, the baseline tests (without air injection)
could not be run due to plugging of the feed line that occurred when no air was introduced.
Therefore, air was on for the whole duration of each test. Tests were performed as a function of
centrifugal acceleration (i.e. 142 and 319 g’s) and rotation time (i.e. 10, 60 and 120 s).
The results from the pilot-scale tests are summarized in Table 3.2. The data show that the
moisture dropped from just over 20% at the lowest air injection time and rotation speed to below
11% at the largest air injection time and rotation speed. The data are plotted in Figure 3.8 for
comparison. This plot shows that an increase in rotation time from 10 to 120 s incrementally
improved the moisture removal by about 5 percentage points for the lowest centrifugal
acceleration of 142 g’s. The incremental improvement was somewhat smaller (i.e., about 3
percentage points) for the higher centrifugal acceleration of 319 g’s. Moreover, increasing
centrifugal acceleration from 142 g’s to 319 g’s caused the moisture to decrease about 7
percentage points for the lowest rotation time and 4 percentage points for the highest rotation
time.
Table 3.2 Effect of Rotation Speed and Rotation Time
Rotation Rotation Centrifugal Moisture
Time Speed Acceleration Content
(s) (rpm) (g's) (%)
10 1000 142 20.2
60 1000 142 18.3
120 1000 142 15.0
10 1500 319 13.9
60 1500 319 13.2
120 1500 319 10.9
80 |
Virginia Tech | 22
20
18
16
14
12
10
0 50 100 150
Figure 3.8 Effect of Rotation Speed and Rotation Time
3.3.2 Effect of Rotation Time and Air Injection
Several test runs were also conducted to investigate the effects of rotation time (i.e. 5, 10,
30, and 120 s) and air pressure (i.e. with and without air) on the performance of the pilot-scale
test unit. The centrifugal acceleration was kept at 319 g’s during these tests. The results are
summarized in Table 3.3. A comparison of the data is also provided in Figure 3.9. The test data
indicate that the injection of air provided approximately 2 percentage points reduction in
moisture with rotation times of 5, 10 and 30 seconds. However, all of the moisture values
obtained in these tests were already very low (around 12% or less). In fact, the values obtained
with air injection at centrifugation drying times of 30 s or greater were all single-digit values.
This can be explained by low solid recovery values obtained. Because the coal slurry was not
thickened before it was fed to the unit, most of the fine solids were lost during centrifugation.
Therefore, only 55-60% of solids were recovered during all the tests. The absence of fines drove
moisture values down to below 12%. Moreover, since moisture values obtained at 120 s rotation
81
)%(
erutsioM
1000 rpm
1500 rpm
Rotation time (s) |
Virginia Tech | centrifuge. Although, a more extensive test program with several cake thicknesses and with and
without air injection was planned for this section, only a fraction of these tests could actually be
conducted because of several mechanical failures occurred during testing. Tests were conducted
as a function of centrifugal acceleration (i.e. 142 g’s and 319 g’s), rotation time (i.e. 15, 30, 60
and 120 s) and air pressure (i.e. with and without air).
The test results given in Table 3.4 and Figure 3.10 show that the moisture of 18.7%
obtained at the shortest rotation time of 15 seconds with 142 g’s reduced to 16.1% when the
centrifugal acceleration increased to 319 g’s. A further moisture reduction down to 13.2% was
achieved when air was injected. While the same trend continued at the other rotation speeds, all
moisture values improved as the rotation time increased. At the longest rotation time of 120
seconds, 14.1% moisture value obtained with 142 g’s reduced down to 10.0% at 319 g’s with air
injection.
As a result of thickening the feed slurry, higher solid recoveries of more than 90% were
obtained. Since no effluent sample was collected during testing, solid recovery values were
Table 3.4 Equilibrium moistures and solids recoveries obtained when samples were
thickened
Rotation Rotation Centrifugal Air Moisture Solid
Time Speed Acceleration Injection Content Recovery
(s) (rpm) (g's) (%) (%)
15 1000 142 No 18.7 92.9
30 1000 142 No 16.3 97.5
60 1000 142 No 15.1 96.0
120 1000 142 No 14.1 96.2
15 1500 319 No 16.1 68.6
30 1500 319 No 14.6 89.4
60 1500 319 No 13.0 88.4
120 1500 319 No 12.3 73.7
15 1500 319 Yes 13.2 98.6
30 1500 319 Yes 11.9 72.5
60 1500 319 Yes 11.2 91.4
120 1500 319 Yes 10.0 96.8
83 |
Virginia Tech | calculated by dividing the dry cake weight by dry feed weight. However, this approach created
potentially inaccurate recovery values, because feed samples were not taken for all the tests.
Therefore, although the solid recoveries seem to be generally improved by thickening, the
accuracy of these values is arguable.
20
18
16
14
12
10
8
0 50 100 150
Figure 3.10 Effect of thickening on product moisture
3.3.4 Effect of Feed Size Distribution
Screenbowl centrifuges are widely used in coal industry to dewater fine coal feeds
containing nominally 1 mm x 0 size particles. However, a screenbowl centrifuge can only
recover about 50% of the minus 0.044 mm size fraction. Therefore, one of the expectations from
the hyperbaric centrifuge technology was the ability of dewatering the minus 0.044 mm size
fraction effectively while keeping solid recoveries high. To examine this possibility, the effect of
feed size distribution was one of the major parameters examined in the pilot-scale tests.
Unfortunately, all the test attempts with finer size feeds failed due to several factors. First of all,
when a feed finer than 0.5 mm was used, all the particles passed through the screen openings and
a cake could not be formed. Secondly, although flocculant addition seemed to be effective with
84
)%(
erutsioM
1000 rpm (Without Air)
1500 rpm (Without Air)
1500 rpm (With Air)
Rotation Time (s) |
Virginia Tech | the coarser feed as explained in the previous section, the sizes of the particle aggregates formed
with finer feeds were not sufficiently big to facilitate a cake formation. Even though a shear-
resistant flocculant was used, the high shear forces introduced by the feed pump caused these
particle aggregates to break, thus they were not able reach the size needed for cake formation.
Therefore, the recirculation of the screen effluent appeared to be the only viable option for
improving solids recoveries.
3.3.5 Effect of Recirculation
After running the tests that were given in previous sections, the test setup was changed to
allow recirculation of screen effluent back to the feed port. This was done using the feeding
arrangement explained in section 3.2. Recirculation was considered to be necessary to maximize
the solid recovery. First, the non-deslimed Tom's Creek sample (Table 3.1) was tested. However,
none of the tests could be completed, because recirculation caused fine particles to form a slime
coating atop the cake. This coating prevented the passage of pressurized air through the cake,
thus causing the airline to plug and burst several times. Moreover, using a flocculant to form
larger agglomerates did not work either, since these weak agglomerates broke after passing
through the feed pump several times. Consequently, it was decided to use a deslimed sample to
see the effect of recirculation. The results are described in the next section.
3.3.6 Effect of Desliming
These tests were conducted using the deslimed screenbowl feed obtained from the
Pinnacle Pond Reclamation plant. The size distribution of the sample is given in Table 3.1. The
centrifugal acceleration was maintained at 319 g’s for all tests. After 30 seconds of recirculation,
the cake was centrifuged for an additional 30 seconds before ending the test run. The test results
85 |
Virginia Tech | given in Table 3.5 indicate that the use of the deslimed sample, which contained very few
particles finer than 0.044 mm, resulted in very low values of cake moisture (i.e., 5-7%). Also, by
recycling the effluent, very high solid recoveries were achieved (i.e., 97-99%). Moreover,
injecting air at different pressures lowered the moistures by 1-2 percentage points compared to
the baseline values. Therefore, single-digit moistures could be achieved by desliming ahead of
the filter centrifuge.
Table 3.5 Effect of desliming and recirculation
Air Moisture Solid
Pressure Content Recovery
(kPa) (%) (%)
0 6.9 97.7
93 5.9 97.2
107 5.6 99.1
133 5.1 97.4
3.4 Discussion
3.4.1 Solids Recovery and Product Moisture
The tests conducted with the pilot-scale unit showed that thickening the feed slurry is a
necessity for achieving high solid recoveries. Also, recirculation of the effluent stream back to
the feed stream gave a chance to fine particles to be captured by the cake, thus increasing the
recovery of solids. However, increased solid recovery caused product moisture to escalate about
2.7 to 5.4 percentage points depending on centrifugation period (see Table 3.3 and Table 3.4).
Considering the increased moisture values were still below 20% and higher recoveries will help
with the capacity requirements, the somewhat higher moisture values were found to be
acceptable.
86 |
Virginia Tech | 3.4.2 Engineering Criteria for Machine Design
Ability of running a continuous operation is essential for maximizing the capacity in
industrial applications. However, the design of the pilot-scale unit did not allow for a truly
continuous operation. First of all, it was not possible to feed the unit continuously because of the
unit's cyclic operation procedure that was explained in section 3.2. Secondly, coal slurry had to
be thickened outside and pumped into the unit, which caused problems with flocculation since
the agglomerates broke due to the high shear forces inside the feed pump.
The pilot-scale unit consisted of two sections, a screen section, which was also used as a
pressure chamber, and a discharge section. The need for thickening the slurry outside the unit
could be eliminated by adding a third "solid" section to the rotating unit, which is similar to the
solid-bowl part of a screenbowl centrifuge. Therefore, coal slurry would be fed directly into the
unit and be thickened inside. Integrating the thickening operation to the centrifuge would have
many advantages over using a conventional thickener. First of all, the space required for a
thickening tank would be eliminated. Also, thickening can be done much faster and more
efficiently with centrifugation. Furthermore, feeding slurry directly to the unit can effectively
reduce the amount of chemicals needed for thickening. Also, since the cake would already be
formed in the solid section, the problems encountered with cake formation using fine particles
can be overcome. Addition of a third section to the centrifuge unit would not eliminate the need
for a feed tank because of the semi-continuous nature of the unit. However, a smaller buffer tank
can be used instead of a wide thickening tank. The only downside to using a solid bowl section
as part of the hyperbaric centrifuge is that it would need to operate at a very high speed to
provide the centrifugal forces necessary to capture ultrafine solids. Calculations suggest that the
87 |
Virginia Tech | required centrifugal field is likely to be in the order to 2,000 to 3,000 g’s for the effective capture
of minus 0.044 mm solids now lost in standard screenbowl machines operating at 500 g.
3.5 Conclusions
A pilot-scale centrifuge that utilizes pressurized air was built and tested under several
conditions. The results showed that higher rotational speeds and longer centrifugation times were
needed for producing lower moistures. Also, injecting pressurized air through the cake further
improved the moisture.
It was also concluded that thickening was an essential step for centrifugal dewatering, as
it increased the recovery of solids. Also, it was found that solids recovery could be further
increased when effluent stream was re-circulated back to the feed stream. However, this exercise
caused a slime coating to form atop the cake when a non-deslimed feed was used due to the
breakage of flocculated agglomerates inside the feed pump. Integration of the thickening process
via the addition of a solid bowl section during centrifugation can help with this problem by
eliminating the need for this step.
Originally, a more detailed and a complete parametric study of the pilot-scale unit was
planned. However, all the attempts for running complete series of tests were unsuccessful due to
the fundamental operational problems. For example, the hollow shaft that was used for material
feeding and other feed lines and valves could not be designed properly for such a small machine
to allow an easy passage of the typical feed material (1 mm x 0). As a result, the lines were
plugged several times during testing and led to some failures. Due to the mechanical and design
problems faced, it was decided pursue the development of a prototype machine that would
include both solid- and screen-bowl sections.
88 |
Virginia Tech | similar to other centrifugal dewatering processes in terms of applicability, operation and
maintenance. (Figure 4.4).
Feed
Product
Main Effluent Screen Drain
Figure 4.4 Diagram of the prototype hyperbaric centrifuge
4.2.2 Test Procedure
In total, seven different feed mixture combinations were evaluated at this particular test
site. The coal used as the feed to the test unit was Alma A coal from the Mountaineer II Mine.
The 6-inch cyclone overflow stream (minus 0.044 mm) of the plant was floated with a 20-
inchx10-ft pilot scale flotation column, and it was mixed in various proportions, which are given
in Table 4.1, with the plant’s screenbowl centrifuge screen drain to prepare the feed for every
batch. Note that a progressively finer feed stream was used in each subsequent test run. This
feeding arrangement made it possible to investigate the effects of a wide range of particle size
distributions on dewatering performance and solid/slurry throughput capacity. The mixture was
kept agitated in a holding tank during the tests. For the last test, instead of screenbowl screen
drain, plant’s column flotation concentrate (nominal 0.25 x 0.044 mm) was mixed with leftover
slurry from the previous test.
92 |
Virginia Tech | Table 4.1 Feed Mixtures
Test Mixing Ratio
1 50 % Column concentrate / 50% Screenbowl screen drain
2 60 % Column concentrate / 40% Screenbowl screen drain
3 70 % Column concentrate / 30% Screenbowl screen drain
4 80 % Column concentrate / 20% Screenbowl screen drain
5 90 % Column concentrate / 10% Screenbowl screen drain
6 100 % Column concentrate / 0% Screenbowl screen drain
7 NA% Column concentrate / NA% Plant column concentrate
Timed samples from feed, product, screen effluent and main effluent streams were taken
during test runs. While the entire product and screen effluent streams were collected for each
test, only 2-4 cuts were taken from feed and main effluent streams to fill up two 5 gallon buckets
for each stream. After cutting the sample from the main effluent, a full stream sample of the feed
was taken by diverting the feed stream from the centrifuge to a bucket. Feed stream was
redirected back to the centrifuge and sufficient time was given to reach steady-state before
cutting the next set of samples. The holding tank was filled up before every test at the desired
mixing ratio. All the samples were sealed after each test to prevent any evaporation. Analyses of
the samples were done by a commercial laboratory. Size distributions and ash data for all the
feeds are given in Table 4.2 and Table 4.3, respectively.
Table 4.2 Feed size distributions for all tests
Weight (%)
+0.5 0.5 x 0.15 0.15 x 0.044 0.044 x 0.025 -0.025
Test mm mm mm mm mm
1 16.63 26.08 20.47 0.58 36.24
2 10.51 17.56 23.28 7.35 41.30
3 8.07 14.63 22.53 8.78 45.99
4 5.52 9.97 21.45 9.20 53.86
5 2.77 7.37 16.58 11.13 62.15
6 0.18 1.40 20.42 3.70 74.30
7 0.01 0.44 3.20 1.89 94.46
93 |
Virginia Tech | Table 4.3 Feed ash values for all tests
Ash (%)
+0.5 0.5 x 0.15 0.15 x 0.044 0.044 x 0.025 -0.025
Test mm mm mm mm mm
1 7.65 5.43 6.68 20.50 10.21
2 10.59 5.88 5.41 7.68 7.83
3 8.60 6.27 6.47 8.97 8.12
4 10.83 6.80 5.29 6.55 7.94
5 12.90 7.30 4.28 5.03 6.86
6 29.09 7.54 2.33 3.25 6.67
7 28.65 4.61 3.60 3.85 7.19
4.2.3 Performance Analysis
To create a reliable set of data out of the raw data obtained from the laboratory, a steady-
state mass balance routine that was set up in an Excel spreadsheet was used. This routine makes
the least possible change to the measured values, so that the mass of components (i.e. solid,
water, ash, etc.) entering the unit is equal to the mass of components leaving the unit. Different
amounts of confidence were given to measured values, using weighted sum-of-squares, to be
able to let the routine selectively adjust these values.
It was observed that the feed flow rates calculated from the timed feed samples were
higher than the combination of product and effluent stream flow rates. Even though the feed
pump was a positive displacement type, it was assumed that there was enough wear on the rotor
of the pump that allowed surging while it was operating under pressure. Whereas, when the full
feed stream was diverted from the centrifuge to a bucket during sampling, resulting pressure
drop on the pump caused the flow rate to increase. Therefore, very little dependence was put on
the measured feed flow rate numbers when mass balancing. Also, according to the laboratory
analysis, 0.044 x 0.025 mm size fractions of every sample had unrealistically small mass values
compared to the upper and lower size fractions. When mass balancing was tried, the routine
94 |
Virginia Tech | 4.3 Results
Summary of test results is given in Table 4.4. As can be seen, a low moisture value of
13.7% was achieved at the first test run. This particular run employed a relatively coarse
feedstock prepared using a 50/50 mixture of screenbowl screen drain (minus 0.5 mm) and
column flotation concentrate (minus 0.044 mm). From this baseline test run, the moisture
climbed up to 19.4% as the mass percent of ultra-fines (minus 0.025 mm) increased to its
maximum value of 94.5% by weight. The fact that the moisture in the final run was still below
20% was considered to be very impressive given the extreme fineness of the feed material used
in the final test run. Conventional dewatering processes (e.g., screenbowl centrifuges) typically
generate moistures >50% for this ultrafine size class.
The prototype test data also demonstrated that very high recoveries of ultrafine solids
could be achieved while maintaining low moisture contents. The highest recovery of dry solids
(i.e., 99.9%) was obtained for the coarsest feed mixture containing only 36.2% minus 0.025 mm,
while a slightly lower recovery (i.e., 97.9%) was obtained for the finest feed mixture containing
94.5% minus 0.025 mm.
Table 4.4 Cardinal Plant test results
No Recirculation With Recirculation
(Measured) (Predicted)
Solid Solid
Test Condition Moisture Moisture
Recovery Recovery
(%) (%) (%) (%)
1 50 / 50 : Col Conc / Scrn Drain 13.3 96.1 13.7 99.9
2 60 / 40 : Col Conc / Scrn Drain 15.0 95.5 15.1 98.9
3 70 / 30 : Col Conc / Scrn Drain 16.2 93.4 17.0 99.9
4 80 / 20 : Col Conc / Scrn Drain 18.6 93.8 18.6 99.2
5 90 / 10 : Col Conc / Scrn Drain 19.4 94.5 19.7 99.4
6 100 / 0 : Col Conc / Scrn Drain 18.7 83.2 18.3 97.3
7 NA/NA : Col Conc / Plant Col 19.7 89.4 19.4 97.9
Conc
96 |
Virginia Tech | 4.4 Discussion
It was originally anticipated that a linear correlation would be observed between the
moisture values and the amount of ultrafine solids in the product. This correlation has been
observed for most of conventional dewatering processes used in the coal industry (Arnold,
1999). However, as shown in Figure 4.5, the moisture obtained using the CentribaricTM
technology increased linearly until the mass percent of minus 0.025 mm solids reached 60%. The
moisture then reached a plateau at about 19.4% and did not increase further as the amount of
minus 0.025 mm solids increased. This surprising trend was also found to be reproducible in
laboratory tests conducted with air injection. No such plateau was observed in the laboratory
tests where no air injection was employed (i.e., the moisture continued to increase in a linear
fashion as the amount of ultrafines increased). As such, the CentribaricTM technology appears to
be ideally suited for dewatering of ultrafine feeds to low moisture values.
4.5 Conclusions
The prototype hyperbaric centrifuge unit that was built by Decanter Machine Inc. was
tested with various feed conditions at Arch Coal’s Cardinal Plant. Test results showed that
moistures of 13.3-19.7%, which are very low numbers compared to conventional filtration, with
high recoveries of 83.2-96.1% were easily achievable when screen drain of the test unit was
discarded. Whereas, slightly higher but still low moistures from 13.7% to 19.4% with 97.3-
99.9% recoveries were predicted to be achievable if the screen drain was re-circulated. Also,
mixing fine (minus 0.044 mm) column concentrate with a coarser stream was found to be
effective on reducing moisture and increasing solid recovery. To avoid a substantially higher
capacity requirement, a smaller ratio of coarse/fine can be used as the test results show that even
97 |
Virginia Tech | CHAPTER 5 MODELING OF A HYPERBARIC FILTER
CENTRIFUGE
5.1 Introduction
Coal cleaning is a water based process after which a product with certain amount of
moisture is produced. Excess process water is removed by dewatering equipment at the end of
the cleaning process. Dewatering is an important step of coal cleaning, because coal producers
can only ship a limited amount of moisture due to contractual requirements. Therefore,
successfully simulating dewatering equipment using a reliable and accurate model would greatly
benefit coal producers, as the value of coal can be estimated under different conditions even
before actually running the dewatering equipment.
The final moisture content of a cake depends on many variables such as the physical
properties of particles (i.e. size, shape, surface properties, etc.) and operating parameters of the
dewatering equipment. Because of the large number of particles involved with randomly
distributed sizes and shapes, it is impossible to completely describe this complex system. Instead,
the best estimates of process variables can be used to make an accurate prediction.
Dewatering models commonly involve a balance of de-saturating forces that push the
water out of the cake and capillary forces that retain water inside the cake. The applied de-
saturating forces can be either mechanical as in conventional centrifuge machines or can be
hydrodynamic as in pressurized air injection. These forces can also be used in conjunction with
each other. Even though there are several empirical and theoretical models developed by
researchers to predict moisture content of centrifuge and filter products, there is no model
currently available in the technical literature for a hyperbaric centrifuge. Therefore, the purpose
99 |
Virginia Tech | of this chapter was to develop a model for hyperbaric centrifugation. For this purpose, two types
of models were developed. The first type of model, which is explained in section 5.2, was
developed to explain the pressure distribution inside an HFC cake. This model made it possible
to understand why injection of pressurized air helped with dewatering. The second type, which is
explained in section 5.3, was intended for prediction of moisture content of HFC products and it
involved development of two different models, one of which was a theoretical model and the
other one was an empirical model.
5.2 Pressure Distribution Model
The centrifugal dewatering process can be characterized by a filter cake submerged in an
annulus of clear liquid (Figure 5.1). Zeitsch (1990) developed a model to predict centrifugal
filtration rate based on the first principle considerations. According to Zeitsch, the fundamental
law of filtration as formulated by Darcy (1856) is not applicable to centrifugal filtration as it fails
to take mass forces into account. Darcy's law suggests that there has to be a pressure difference
in order to move a liquid through a porous medium, as follows:
[5.1]
(cid:5)∆(cid:7)(cid:8)
N =
(cid:9)O
where Q is the flow rate of liquid passing through the cake, K is the permeability of the bed, ∆P
is the pressure drop across the cake, A is the filter area, η is the viscosity of the fluid and H is the
cake thickness. With reference to Figure 5.1, during centrifugation when the liquid layer reaches
the cake surface (r = r ), pressures at r and r will be equal to the ambient pressure. Therefore,
0 S S B
according to Eq. [5.1], the liquid should not be flowing at this point because the pressure
difference across the cake is zero. However, the liquid will keep flowing because of the acting
centrifugal force.
100 |
Virginia Tech | 5.3 Cake Moisture Models
5.3.1 Empirical Dewatering Model
The moisture content of a centrifuge product is calculated in two steps. First, the
minimum water content, which is achievable when a de-saturating force acts for a very long
time, is calculated and then this value is adjusted upwards to allow for the fact that there is only a
finite amount of time available for dewatering. Moreover, when carrying out moisture content
calculations, saturation (S) values are used instead of moisture values for ease of computation.
The saturation is estimated by using a dimensionless number called capillary number
(N ). The capillary number represents a ratio of de-saturating forces to water retaining forces.
c
Several researchers such as Brownell and Katz (1947), Haruni and Storrow (1953), Dombrowski
and Brownell (1954) defined capillary number using the permeability factor K, but Wakeman et
al. (1976) showed that the capillary number and the previously used permeability expressions to
correlate residual saturation data were inadequate because of the inability of Brown (1950)
method to calculate K for centrifugally spun beds. However, Swindells (1979) replaced
permeability term, K, with the squared mean particle size term, d 2, which resulted in a better
p
correlation:
(cid:18) [5.4]
5Y3Z4(cid:1)[
3(cid:22) =
\cos+
where ρ is the density of water, N is the number of g's, d is the mean particle diameter, σ is
w g p
surface tension of water and θ is the contact angle. In the current study, this equation was
modified to allow for pressurized air injection as follows:
(cid:18)
[5.5]
D5Y3Z4+(cid:7)-WX⁄OE(cid:1)[
3(cid:22) =
\cos+
103 |
Virginia Tech | where P is the air pressure and H is the cake thickness. d can be calculated using the following
air p
equation:
[5.6]
1
(cid:1)[ =
F (cid:24) ∗
∑W`a(cid:28)D∑^`a0W^E/(cid:8)W(cid:29)
Above equation gives the diameter of a particle with the same surface area per unit volume as the
total surface are per unit volume of collection of all particles.
It is important to determine the mean particle size accurately, since it greatly affects the
simulation results. Size distribution of a particulate system is commonly determined by wet
screening. However, because of the limitations of this method the size distribution below 0.025
mm cannot be measured. Moreover, a practical method for determining the size distribution of
this size fraction does not exist. Therefore, when conducting simulations, the well known Gates-
Gaudin-Schumann (GGS) particle size distribution equation was fitted to size distributions
obtained for different feed samples to determine the distribution below 0.025 mm. This allowed
for calculating a more realistic mean particle size, especially for the finer size distributions.
While fits obtained using GGS equation were not the perfect, it was deemed to be sufficiently
accurate for the current study.
Using the data he obtained during his tests and the data available in the literature, Ledger
(1999) suggested an empirical relationship that yielded a good correlation between the capillary
number and the residual saturation, which is the minimum saturation achievable:
[5.7]
(cid:19)X =
0.051@cde⁄fa.(cid:12)
+0.0045
However, due to the limited amount of time available for dewatering, a higher saturation, which
is called effective saturation, S , is normally seen in commercial centrifuges. Brown (1950)
e
derived a relationship that defines saturation, S , in terms of S and S:
e r
104 |
Virginia Tech | Simulation results given in Table 5.6 and Table 5.7 showed that, as expected, for both
coarse and fine coals, the highest moisture value would be obtained at the lowest g level without
air injection and the lowest moisture value would be obtained at the highest g level with air
injection. Moreover, it would be possible to avoid using higher g levels to achieve lower
moisture values by application of more air pressure. For example, using 200 kPa of air pressure
at 500 g's would produce the same moisture content as 3000 g's would produce without air
pressure for both coarse and fine cakes (values highlighted in Table 5.6 and Table 5.7).
Experimental results discussed in section 2.3.1.4 that were run at 500 g's and 2700 g's also
confirmed this conclusion. In those tests, using 200 kPa air pressure at 500 g's had produced
similar moistures to the ones obtained at 2700 g's without air pressure.
Table 5.6 Moisture values obtained with coal A using empirical dewatering simulations at
different air pressures and centrifugal accelerations
P (kPa)\g's 100 300 500 1000 2000 3000
air
0 32.4 27.4 25.1 22.2 19.6 18.2
67 22.5 21.8 21.1 19.9 18.4 17.4
133 20.1 19.7 19.3 18.6 17.6 16.8
200 18.7 18.4 18.2 17.7 16.9 16.3
267 17.8 17.6 17.4 17.0 16.4 15.9
333 17.1 16.9 16.8 16.5 16.0 15.5
400 16.5 16.4 16.3 16.1 15.6 15.2
Table 5.7 Moisture values obtained with coal B using empirical dewatering simulations at
different air pressures and centrifugal accelerations
P (kPa)\g's 100 300 500 1000 2000 3000
air
0 40.5 38.7 37.4 35.2 32.7 31.1
67 35.5 34.8 34.2 33.0 31.3 30.1
133 33.2 32.7 32.4 31.6 30.3 29.4
200 31.7 31.4 31.1 30.5 29.5 28.7
267 30.6 30.4 30.2 29.7 28.9 28.2
333 29.7 29.6 29.4 29.0 28.3 27.7
400 16.5 16.4 16.3 16.1 15.6 15.2
117 |
Virginia Tech | The next set of simulation runs were conducted to do a sensitivity analysis of different
parameters that were used in the empirical model. The first step of the sensitivity analysis was to
establish a baseline moisture value by running the empirical simulation routine using average
values of all the input parameters. New moisture values were then predicted by changing each
input parameter value by ±10%. Finally, the percent change of the baseline moisture was
calculated for each parameter using the new moisture values. Therefore, this study was
conducted to determine the magnitude of the influence of different input parameters on moisture
prediction of the empirical model.
The results of the sensitivity analysis given in Table 5.8 showed that the most influential
parameter was the cake porosity, as a 10% increase in the porosity caused a 16.9% increase in
the moisture and a 10% decrease caused a 15.4% decrease in moisture. Moreover, increasing
cake thickness by 10% increased the moisture content by 2.7%, while the moisture decreased by
3.0% with a 10% decrease in cake thickness. While the rest of the parameters had similar
magnitudes of effects, changing surface tension and contact angle had very small effect on
moisture. However, as it was explained in section 1.1.2.2, it is a known fact that lower moisture
values can be achieved by adjusting surface tension and contact angle values using dewatering
Table 5.8 Magnitude of change in moisture caused by each parameter
Baseline values % ∆Moisture with % ∆Moisture with
Inputs of parameters +10% change -10% change
ε 0.5 16.9 -15.4
σ 0.072 0.0 0.0
µ 0.001 1.6 -1.7
L 0.0127 2.7 -3.0
t 60 -1.5 1.7
N 500 -0.4 0.4
g
Pair 30 -1.2 1.3
θ 60 0.0 0.0
K 10-14 -1.5 1.7
118 |
Virginia Tech | chemicals in accordance with filtration laws. Even though surface tension and contact angle
values are included in Eq. [5.4], which is used to calculate the capillary number, these values are
rendered ineffective when the capillary number is divided by the value of 91.3 in Eq. [5.7], since
the value of the capillary number was usually much lower than 91.3. Therefore, this was found to
be one of the shortcomings of the empirical model and further investigation of this shortcoming
is suggested for future work.
The final set of simulation runs were conducted to provide a more detailed look at the
effects of different parameters on moisture. The results of these runs are shown in Figures 5.5 -
5.11. Since the sensitivity analysis showed that contact angle and surface tension values had very
small effects on moisture when the empirical model was used, these parameters were not
included in the figures.
It was observed from these plots that the porosity and the moisture content had a straight
line relationship, while the correlation of the moisture with other parameters was more in a
logarithmic fashion. It was also observed that an increase in porosity, cake thickness and
viscosity values caused the moisture content to increase, while an increase in centrifugal
acceleration, air pressure, permeability and time values caused the moisture content to decrease.
Moreover, as it was observed with the sensitivity analysis, Figure 5.10 and Figure 5.11 showed
that increasing air pressure caused a bigger drop of moisture than it was caused by increasing
centrifugal acceleration.
119 |
Virginia Tech | 5.4.2 Theoretical Dewatering Model Simulations
A simulation routine for the theoretical dewatering model was setup in Microsoft Excel
using equations described in section 5.4 and was run using the same set of coals and parameters
that were used to run the empirical dewatering model. Simulations were conducted with and
without 200 kPa air injection at 500 and 2700 g's. Results obtained with simulation routine were
then compared to experimental results. All the input parameters used in these simulations are
given in Table 5.9.
Table 5.9 Parameters used in dynamic dewatering simulations
Parameter Unit Value
ρ kg/m3 1400
s
ρ kg/m3 1000
w
ε -- 0.45-0.55
σ N/m 0.072
µ Pa.s 0.001
H m 0.0085
t s 120
N -- 500-2700
g
P Pascal 0-206,843
air
(psi) (0-30)
g m/s2 9.81
θ ° 60
Conducting simulations using this theoretical model revealed serious issues related to the
model. As can be seen from the charts given in Figure 5.12, predicted values were mostly off and
did not match the experimental results. The model was able to predict relatively accurate
moistures at 2700 g's without air injection with a low R2 value of 0.794. However, all the other
simulations conducted at different conditions (i.e. 500 g's with and without air and 2700 g's with
air), produced negative R2 values, which showed that there was no correlation between predicted
and measured values. The model had difficulty with producing decent results, especially with
122 |
Virginia Tech | as 0%. Furthermore, according to the model, the equilibrium moisture content first decreased
while the cake thickness increased from 0.0064 m to 0.0127 m. However, after this point the
moisture increased with increased cake thickness.
As a result, it was concluded that there was a fundamental problem with this model,
which could not be solved. While the effects of different parameters on cake moistures were
generally reasonable, the equilibrium moisture predictions were unreasonable. An investigation
directed at improving the theoretical model is suggested as a topic for future research.
5.5 Mechanical Degradation
Mechanical degradation of particles is unavoidable in centrifugation due to the forces
involved in this process. This effect was not taken into consideration when the empirical and
theoretical model simulations given in previous sections were conducted, because those size
distributions were obtained from products of real HFC tests, which were already degraded.
However, it is essential to use a degradation model when dealing with feed size distributions, i.e.
size reduction of the feed inside a centrifuge has to be taken into account before any attempt to
model centrifugal dewatering process.
Lyons (1951) did a systematic study of mechanical degradation and he found out that
larger particles were subject to size reduction more than finer particles. He also found out that
the size distribution of newly created fine particles was similar to the ones already existed in the
feed.
Particles existing in the feed stream of a centrifuge can be divided into n size ranges,
where size 1 is the coarsest size fraction and size n is the finest. It is assumed that the fraction of
size class i that can be degraded is linearly related to the average particle size by the following
relation:
126 |
Virginia Tech | [5.53]
¨W = (cid:13).(cid:8)W; 0 < ¨W < 0.9
where D is the fraction of size class i degraded, A is the average particle size of the size class i
i i
and k is a parameter unique to the equipment. Tierney and Gottfried (1983) reported that an
d
average k value of 98.4 m-1 can be used for most centrifuges. However, when simulations given
d
in the following sections were conducted, it was observed that a k value of 400 m-1 was required
d
to match data from commercially available centrifuges. Therefore, a value of 400 m-1 was used
for all the empirical simulations conducted in the present work.
The degraded fraction of size class i, given by D, is distributed among all size classes
i
that are finer than the size class i, in proportion to the fractions of these size classes in the feed.
Moreover, the distribution will be different for each gravity class j. Therefore, the contribution to
size class p (p > i) due to the degradation of size class i in gravity class j is given by:
F
[5.54]
«[,W,^ = 0W,^¨W0[,^ ¬ D0o,^E
o`W{a
where F is the weight fraction of feed in size class i and gravity class j before degradation. The
i,j
total contribution to size p from each size class coarser than p can be obtained by summing all
G values from i = 1 to p - 1. The corrected fraction of feed in size class p and gravity class j
p,i,j
after degradation can be determined by adding the summation of G values to amount that was
p,i,j
not degraded in size class p.
[ca
[5.55]
∗
0[,^ = ¬D«[,W,^E+D1−¨[E0[,^
W`a
5.6 Conclusions
An empirical dewatering model and a theoretical dewatering model were developed to
describe the process of dewatering during hyperbaric centrifugation. While very accurate results
127 |
Virginia Tech | NOMENCLATURE
A: Surface area of filter cake, m2 S : Specific surface area defined by Eq.
0
[5.15]
A: Average size, mean of upper and S : Skewness of the particle size
i K
lower mesh sizes, m distribution
b: Radius of a capillary, m S : Residual Saturation
r
c , c : Fitting parameters used in Eq. [5.14] S : Effective Saturation
1 2 e
C : Coefficient of variation s: Diameter of a capillary, m
DP
C: Defined by Eq. [5.10] s*: Critical pore diameter, m
t
D: Fraction of particles of size i which ŝ: Predominant pore diameter, m
i
is degraded
D : Particle diameter, m t: Time, s
p
F F* : Weight fraction of feed in size range V : Volume of water that cannot be
i,j i,j R
i and gravity range j before and after removed from the filter cake, m3
degradation
g(s): Pore size distribution function V : Volume of water that is not removed
E
from the filter cake due to time
limitation, m3
g: Gravitational acceleration, m/s2 y: Dimensionless parameter defined by
Eq. [5.17]
G : Fraction of feed size i and gravity j y(t,s): Momentary height of liquid in a
p,i,j
which appears in size p after capillary with diameter s at time t, m
degradation
h : Equilibrium height of a liquid in a η: Viscosity of the liquid, Pa.s
0
pore, m
H: Cake thickness, m ρ , ρ : Density o water and solids, kg/m3
w s
k : Degradation constant, dimensionless σ: Surface tension of liquid, N/m
d
K: Permeability of filter cake, m2 θ: Contact angle
N : Capillary number, dimensionless ε: Porosity or the void fraction
c
N : Number of g's τ: Tortuosity of particles
g
P : Air pressure, kPa ζ: Parameter defined by Eq. [5.27]
air
P: Pressure, Pa β: Parameter defined by Eq. [5.28]
r: Radius coordinate starting at the X: Parameter defined by Eq. [5.29]
pore axis, m
S: Total Saturation
129 |
Virginia Tech | CHAPTER 6 ECONOMICS OF FINE COAL DEWATERING
6.1 Introduction
Modern coal preparation plants typically incorporate a series of sequential unit operations
for sizing, cleaning, and dewatering. This sequence of steps is repeated for each size fraction.
This subdivision is necessary since the cleaning processes used in modern plants have a limited
range of applicability in terms of particle size. As a result, modern plants may include as many as
four separate processing circuits for treating the coarse (plus 50 mm), small (50 x 1 mm), fine (1
x 0.15 mm), and ultrafine (minus 0.15 mm) material.
In the U.S., coarse and small size coal is typically cleaned using dense medium processes
such as vessels and cyclones. The coarser products are dewatered using screens, while smaller
coal is dewatered using centrifugal dryers. Fine coal is usually upgraded using water-based
density separators such as spirals and water-only cyclones. Ultrafine coal is either cleaned using
froth flotation or discarded as waste. The fine and ultrafine coal products are dewatered using
various combinations of centrifugal dryers, screenbowl centrifuges and/or filters. After
dewatering, the clean coal products from the parallel circuits are usually blended together prior
to shipment to market.
An important task for plant managers is to select operating points for each plant process
to ensure that the coal products from the plant meet quality criteria dictated by sales contracts.
These specifications may include limits on ash, sulfur, moisture, heat content, or other quality
indicators. Unfortunately, plant operators often choose set points that provide an identical clean
coal quality in every plant circuit. As discussed later, operation of all circuits at a constant
132 |
Virginia Tech | product quality typically does not optimize plant yield. Moreover, managers are often forced to
select set points that are less than ideal for a given coal due to contractual limitations that may or
may not be justified from economic considerations. A mismatch between the inherent
cleanability of a given coal (including its tendency to retain moisture) and the specifications in a
sales contract can substantially lower overall plant yield and overall profitability.
6.2 Plant Optimization
6.2.1 Optimization by Constant Incremental Ash
Many articles in the technical literature advocate the use of simulation routines to identify
optimum set points for processes used in coal preparation plants (Peng and Luckie, 1991; King,
1999). Such data are used to determine the total plant yield and clean coal quality that may be
obtained for any combination of operating points. The optimum points are those that provide the
highest total plant yield at the desired target quality. However, the simulations require detailed
data describing the washability of the feed coals and empirical and/or theoretical partition data
describing the separating capabilities of the cleaning equipment. Unfortunately, the lack of such
detailed data makes this approach impractical to implement in modern facilities. This trial-and-
error approach also provides little insight regarding the underlying principles that control the
optimization process.
A more attractive method for plant optimization is to employ constant incremental ash.
This concept, which has long been recognized in coal preparation (Mayer, 1950; Dell, 1956;
Abbott, 1981; Rayner, 1987), states that the clean coal yield for parallel operations is highest
when all circuits are operated at the same incremental ash. This statement is true for any number
of parallel circuits and is independent of the size and washability characteristics of the feed coal.
Mathematical proof of the incremental ash concept has been done by Luttrell et al. (2003).
133 |
Virginia Tech | However, the fundamental basis for this concept can be easily demonstrated using the simple
illustrations provided in Figure 6.1(a) and Figure 6.1(b). In each diagram, the same two feed
streams (“A” and “B”) are to be treated. The two feeds may represent coal fed to two different
cleaning operations in the same plant, the streams fed to two different plants that both service the
same customer, or the same cleaning unit at two different points in time. The feed coals are
comprised of various amounts of carbonaceous matter (black) and ash-bearing matter (gray). In
this example, the washability of Feed “A” makes it relatively easy to upgrade since most of the
particles are well liberated. In contrast, Feed “B” has a difficult washability due to the presence
of a large percentage of middlings (composite) particles.
Figure 6.1(a) shows the result that is obtained when the separation is conducted to
Feed Feed Feed Feed
“A” “B” “A” “B”
Cleaning Cleaning Cleaning Cleaning
Y = 9/12 Y = 7/13 Y = 7/12 Y = 11/13
= 75.0% = 53.8% = 58.3% = 84.6%
A = 2.25/9 A = 1.75/7 A = 0.75/7 A = 3.75/11
= 25.0% = 25.0% = 10.7% = 34.1%
Y = 16/25 Y = 18/25
Combined = 64.0% Combined = 72.0%
“A+B” “A+B”
A = 4/16 A = 4.5/18
= 25.0% = 25.0%
(a) (b)
Figure 6.1 Clean coal products obtained by (a) treating parallel streams at constant
cumulative ash and (b) treating parallel streams at constant incremental ash
134 |
Virginia Tech | provide a target ash content of 25% for both streams. For Feed “A”, the operating point is raised
from left to right until a total of 9 units (blocks) are recovered out of a total of 12 units. This
operating point produces a yield of 75% (i.e., 9/12=75%) at the desired target ash of 25% (i.e.,
2.25/9 = 25%). For Feed “B”, a yield of only 53.8% (i.e., 7/13=53.8%) can be realized before the
target ash of 25% (i.e., 1.75/7=25%) is exceeded. The lower yield is due to the poorer liberation
of Feed “B”. The clean products from these two streams are blended together to produce a
combined yield of 64% at the desired target ash of 25%. At first inspection, this appears to be an
acceptable result since the target grade is met. However, a closer examination shows that
particles containing 75% ash reported to clean coal when treating Feed “A”, while at the same
time many particles containing just 50% ash were discarded when treating Feed “B”. This
combination is obviously not optimum since higher quality particles were discarded so that
poorer quality particles could be taken.
The incremental ash concept dictates that maximum yield can only be obtained when the
ash content of the last unit recovered from each feed stream is identical. To test this principle, the
feed streams shown in Figure 6.1(b) are each separated so that the last unit recovered in both
cases contained 50% ash. This combination of operating points decreased the ash content of Feed
“A” to 10.7% and increased the ash content of Feed “B” to 34.1%. However, when the two
streams are blended together, the combined clean coal product still meets the required target ash
of 25%. More importantly, this new set of operating points provides a higher combined yield
(i.e., 72% versus 64%). In fact, separation at a constant incremental ash will always provide the
maximum yield of coal at a given ash content.
135 |
Virginia Tech | 6.2.2 Incremental Ash and Specific Gravity
Incremental ash is a mathematical term that cannot be directly monitored in operating
industrial plants. However, this value can be estimated for ideal separations. This approximation
assumes that the particles in run-of-mine coals contain only two components, i.e., a low-density
ash-free carbonaceous component and a high-density pure-ash mineral component. Under this
assumption, the ash content of an individual particle must increase linearly with the reciprocal of
the composite particle density (ρ) according to the expression:
[6.1]
100 5(cid:22) +5-
(cid:8)xℎ(cid:14)%(cid:17) = € +5-(cid:129)
5(cid:22) −5- 5
where ρ and ρ are the densities of the light (carbonaceous) and dense (ash bearing) components,
c a
respectively (Abbot and Miles, 1990). The utility of this expression can be demonstrated by the
data shown in Figure 6.2. This illustration shows the ash contents of narrowly partitioned density
fractions obtained from experimental float-sink tests conducted on six different size fractions of
a run-of-mine coal. For particles coarser than
100
0.6 mm, the data show that the same
80
individual ash is obtained for any specific
60
gravity regardless of the size fraction treated.
The deviation noted for the fractions finer
40
than 0.6 mm can normally be attributed to
20
inefficiencies in the experimental float-sink
0
procedures at finer particle sizes. In any case,
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9
1 / Mean Specific Gravity
plant optimization requires that all parallel
Figure 6.2 Relationship between specific
circuits be operated at the same incremental gravity and individual ash
content of different size classes
136
)%(
tnetnoC
hsA
laudividnI
Plus 1-1/4
1-1/4x3/4
3/4x1/4
1/4x28M
28x48M
48x100M |
Virginia Tech | ash. Since Eq. [6.1] shows that incremental ash is fixed by specific gravity, this implies that total
plant yield is maximized when all parallel circuits are operated at the same specific gravity
cutpoint (Clarkson, 1992). This statement is true regardless of the size distribution or washability
characteristics of the feed coal, provided that very efficient separations are maintained in each
circuit. If the separation is less efficient (which is typical of fine coal separators), a correction
must be made to account for the impacts of misplaced material on the incremental quality. This
correction generally requires that less efficient units be operated at a slightly higher specific
gravity cutpoint (Luttrell et al., 2003).
6.2.3 Effect of Moisture on Optimization
The incremental ash approach to plant optimization works well for processing
metallurgical coals where ash is the primary quality specification. Unfortunately, most coal is
sold in the steam (utility) market where other quality specifications are also important (e.g.
moisture, heat content, etc.). In particular, the combined ash and moisture content of a coal is
very important in this market since this sum dictates the heat content of the coal. Therefore, in
the case of steam market, optimization requires that the “incremental value” of the coal be held
constant. In other words, the market value of the next particle of coal recovered should be the
identical for all circuits. As such, the incremental value concept includes not only incremental
ash, but also incremental moisture and, if appropriate, other incremental values that impact the
monetary value of the coal product (e.g., sulfur). The inclusion of moisture in the incremental
value approach typically dictates that density cutpoints for operations treating coarser particles
be set higher than the density cutpoints used to treat finer particles. This difference is due to the
fact that moisture content increases (and coal value decreases) as the particle becomes finer. As
137 |
Virginia Tech | can be seen from the following example, the increased moisture can have a dramatic impact on
the operating points needed for plant yield optimization.
6.2.4 Incremental Quality and Coal Value
Any discussion of plant optimization must address the issue of coal value. Although coal
payments in the steam (utility) market are typically reported on a per ton basis, price adjustments
are nearly always applied to compensate for variations in the heat content of the as-received
product. For example, consider the tabular analysis illustrated in Table 6.1(a) and Table 6.1(b).
In this simplified example, two scenarios for the same cleaning operation are shown. In the first
scenario, it is assumed that the product is dry; therefore, moisture has no effect on the coal value.
In the second case, the product contains 16.67% moisture, which drives down both the heat
content and the market value of the coal. In this example, pure particles of carbonaceous matter
(SG=1.3) were assigned a heat value of 15,000 BTU/lb (dry ash-free basis), while pure particles
of rock (SG=2.5) were assigned no heat value. The sales agreement for this example dictates a
market price of $30.00 per ton of as-received clean coal with quality specifications of 12,500
BTU/lb and 10% ash (as-received). The coal payment schedule is prorated at $0.20 per 100 BTU
above/below 12,500 BTU/lb and $0.50 per 1% ash above/below 10% ash. A $10.00 per ton sales
related cost has also been applied to account for miscellaneous fees, taxes and other sales
expenses associated with the coal production.
The calculations for the first case indicate that pure coal particles carry a $5.00 per ton
BTU premium (i.e., [15,000–12,500] x $0.20/100 = $5.00) and $5.00 per ton ash premium (i.e.,
[10-0] x $0.50/1 = $5.00). After subtracting a sales related cost of $10.00 per ton, the net value
of these high grade particles is $30.00 per ton (i.e., $30.00 base + $10.00 premium - $10.00 sales
= $30.00). In contrast, the pure rock particles carry a $25.00 per ton BTU penalty (i.e., [0–
138 |
Virginia Tech | 12,500]/100 x $0.20 = $25.00) and $45.00 per ton ash penalty (i.e., [10-100]/1 x $0.50 = -
$45.00). Thus, after subtracting a sales related cost of $10.00 per ton, these low-grade particles
have a net negative value of minus $50.00 per ton (i.e., $30.00 base – $70.00 penalty – $10.00
sales = -$50.00). In other words, the producer gains $30.00 for every ton of pure carbonaceous
material shipped to the customer and loses $50.00 per ton of pure rock shipped. Composite
particles containing mixed amounts of coal and rock have market values between these two
limits. In order to maximize profitability, the coal producer for this example should obviously
ship only those particles that have a positive net value and should never ship particles that have a
negative net value. In this example, the “zero value” particle, which contains just enough heat
content to justify its shipment to market, occurs at a density of 1.75 SG (i.e., an incremental ash
of 37.5%). Therefore, for this simplistic example, which does not include the effect of moisture
on coal value, the plant operator should set and maintain an effective specific gravity cutpoint of
1.75 SG to ensure that profitability is maximized.
The second scenario is identical to the first except the impact of moisture has been
included in the value calculation. In this scenario, the heat value of particles drops with the
presence of moisture, thus driving down the net value of each coal particle. The pure particles of
carbonaceous matter (SG=1.3) now have a heat value of 12,500 BTU/lb (i.e., 15,000 x [100-
16.67]/100 = 12,500 BTU/lb). Therefore, these particles no longer carry a BTU premium. They
still carry the same ash premium of $5.00 per ton, since the ash content did not change. (In
reality, the presence of moisture would reduce the as-received ash content slightly). Thus, after
subtracting the sales cost of $10 per ton, these particles now have a lower net value of $25 per
ton (i.e., $30 base + $5 premium – $10 sales = $25.00). The pure rock particles (SG=2.5) still
have the same net value of -$50 since moisture does not further reduce the heat value of 0
139 |
Virginia Tech | Luttrell, 2005).
This above example is important because it shows that the specific gravity for optimum
plant operation is controlled by the coal sales agreement and is largely independent of the coal
washability. Moreover, penalties attributable to other factors that influence the monetary value of
coal such as moisture, sulfur, etc., can shift the optimum density for separations to values that are
lower than those predicted by the well-known incremental ash concept. Although this example
has been oversimplified, the underlying optimization principle illustrated by this analysis can
conceptually be applied to any generalized case if the contractual details are available. Such an
example is examined in the following case study.
6.3 Case Study
The optimization of plant yield via proper management of incremental value and the
effect of moisture on incremental value can be best illustrated by means of a case study. The
study was conducted to evaluate three different circuit designs that treated a minus 1 mm coal
stream (see Figure 6.3). The first circuit design was a standard fine coal dewatering circuit used
at U.S. coal plants with desliming. In this circuit design, underflow stream of a 15-inch
classifying hydrocyclone (nominal 1 x 0.150 mm) was sent to a spiral for cleaning, while the
overflow stream (nominal 0.150 x 0 mm) was sent to a 6-inch desliming hydrocyclone. The
underflow stream (nominal 0.150 x 0.044 mm) of this hydrocyclone was cleaned with flotation
and the clean product was then dewatered using a screenbowl centrifuge along with the spiral
product. Moreover, the overflow stream (nominal 0.044 x 0 mm) of the desliming hydrocyclone
and the main effluent stream of the screenbowl centrifuge were discarded. In the second circuit
configuration, the desliming hydrocyclone was removed and the whole overflow stream of the
15-inch hydrocyclone was cleaned using flotation. Again, the underflow stream of this
141 |
Virginia Tech | CHAPTER 7 SUMMARY AND CONCLUSIONS
Several series of laboratory-scale, pilot-scale and field tests were conducted using a novel
fine-particle-dewatering technology called hyperbaric filter centrifugation. Based on the results
obtained from this work, Decanter Machine Inc. constructed and sold a full-scale prototype that
has now been successfully demonstrated at an industrial plant site. The technology is now
marketed under the trade name CentribaricTM centrifuge.
Several laboratory-scale test units were constructed for the purpose of conducting a
detailed parametric study of hyperbaric filter centrifugation. Test results obtained with samples
containing different size distributions showed that as long as the dewatered product contained
less than about 60% minus 0.025 mm particles, centrifugation could be improved through the
injection of pressurized air without having to resort to very high centrifugal fields. For example,
a centrifugal acceleration of 500 g's with air pressure was found to be as effective for dewatering
as a high centrifugal acceleration of 2700 g's without air pressure. Therefore, it was concluded
that the use of a costly high-g centrifuge could be avoided when the fines content of a product
was below a certain amount. However, when the minus 0.025 mm particle content was above
60%, the use of a high centrifugal acceleration with air pressure was found to be essential to
achieve low moisture values. Moreover, tests performed using different coal samples obtained
from commercial coal plants showed that it was possible to achieve single-digit moisture values
when desliming was performed ahead of hyperbaric centrifugation. Furthermore, tests conducted
to study the effects of several surface active dewatering aids and flocculants showed that
moisture reductions as large as 60% (compared to no-chemical, and no-air cases) were possible
by using chemicals in conjunction with air injection.
149 |
Virginia Tech | Dewatering kinetics tests performed at the laboratory-scale showed that centrifugal
dewatering process could be described as a rate process. While the bulk of the water was
removed quickly from the cake during centrifugation, a smaller amount was removed at a slower
rate. Moreover, a certain amount of water was not removed at all, which represented the final
moisture content of the product. The application of air pressure assisted in the removal of water
at both fast and slow rates from the cake and lowered the final equilibrium cake moisture.
Pilot-scale test results showed that higher rotational speeds and longer centrifugation
times were beneficial for producing lower moistures. Also, injecting pressurized air through the
cake further improved the moisture. It was also concluded that thickening was an essential step
for centrifugal dewatering, as it increased the recovery of solids. The experiments also showed
that solids recovery could be further increased when the effluent stream was re-circulated back to
the feed stream. Although a more detailed and a complete parametric study of the pilot-scale unit
was originally intended, all attempts to run a complete series of tests were unsuccessful due to
the frequent operational issues associated with the use of a small pilot-scale machine.
In light of the problems associated with the pilot-scale tests, a continuous prototype
hyperbaric centrifuge was built by Decanter Machine Inc. for use in field testing. The prototype
unit successfully demonstrated that low moisture values could be obtained using hyperbaric
centrifugation. Low moistures values in the range of 13 to 20% were obtainable with high solid
recoveries of 83-96%. Slightly higher, but still low, moistures ranging from 14% to 19% with
97-99% recoveries were predicted to be realistically achievable by circulating the screen drain
back to the feed stream. Also, mixing fine (minus 0.044 mm) column concentrate with a coarser
stream was found to be effective in reducing moisture and increasing solids recovery. It was
concluded that a smaller ratio of coarse-to-fine particles could be used to avoid a substantial
150 |
Virginia Tech | SCALE-UP OF USING NOVEL DEWATERING AIDS
By
Mert Kerem Eraydin
ABSTRACT
Coal preparation plants use large quantities of water for cleaning processes. Upon
cleaning, the spent water must be removed such that the final product moisture level meets
market constraints. However, removal of free water from the surface of fine particles is difficult
and costly, and often the results are less than desirable. Fine particles inherently have very large
surface areas, and hence retain large amounts of water. Increased amounts of fines also cause
denser particle packing, which creates relatively small capillaries in filter cakes and, thus, cause
slower dewatering kinetics. As a result, dewatering costs for fine particles are much higher than
for dewatering coarse particles.
Considering the technical and economic issues associated with dewatering coal and
mineral fines, an extensive matrix of laboratory- and pilot-scale dewatering tests have been
conducted to evaluate the use of novel dewatering aids. The reagents are designed to lower the
surface tension of water, increase the hydrophobicity of the particles to be dewatered, and
increase the capillary radius by hydrophobic coagulation. All of these are designed to lower the
moisture of the filter cakes produced in mechanical dewatering processes. Laboratory-scale
dewatering tests confirmed that using the novel dewatering aids can lower the final cake
moisture of coal by 20-50%, while increasing the dewatering kinetics. Several on-site, pilot-
scale tests were conducted to demonstrate that the process of using the novel dewatering aids can
be scaled. |
Virginia Tech | Based on the laboratory- and pilot-scale tests conducted, a scale-up model for the process
of using the novel dewatering aids has been developed. It can predict the final cake moistures as
a function of vacuum pressure, filtration time and specific cake weight. The model can be useful
for the scale-up of vacuum disc filters (VDF) and horizontal belt filters (HBF). Simulation
results indicate that dewatering aids can be very effective, especially when used in conjunction
with HBF due to its ability to control cake thickness and drying cycle time independently.
In light of the promising laboratory- and pilot-scale test results, an industrial
demonstration of the novel dewatering aids has been conducted at the Smith Branch
impoundment site, which contains 2.9 million tons of recoverable coal. When the reagent was
used for dewatering flotation products using a VDF, the moisture content was reduced from 26 to
20% at 0.5 lb/ton of reagent addition and to 17.5% at 1 lb/ton. The use of the dewatering aid
also improved the kinetics of dewatering, increased the throughput, and reduced the power
consumption of vacuum pumps by 30%.
The novel dewatering aids were also tested successfully for dewatering of kaolin clays.
In this case, the mineral was treated with a cationic surfactant before adding the dewatering aids.
This two-step hydrophobization process was able to reduce the cake moisture and also increase
the throughput. |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.