Solid-Liquid Separation and CCD washing
MTRL 358
Solid-Liquid Separation
2012
Introduction
This is a critically important operation in both mineral
processing and at various stages in hydrometallurgical extraction.
It is often quite costly, requiring large equipment and facilities.
Poor liquid-solid separation characteristics can undo what would
otherwise be a successful process. This requires careful
attention.
Water may need to be reclaimed for a couple of reasons. First to
adjust process stream characteristics so that they are suitable for
subsequent treatment (applies to both tailings and other process
streams). Second to recover or recycle water, which reduces
consumption (very important in dry areas) and to prevent
contamination of fresh water sources. On the other hand solids may
need to be removed from a solution in order to obtain a pure metal
product, or to further process the solids themselves.
Slurries, suspensions and tailings may contain from 10% to
almost 100% water by weight. Dewatering is accomplished by four
basic methods:
(a) Screening (larger particles retained on screens as water
passes through)
(b) Sedimentation (decantation or thickening - relies on the
same principles as
sedimentation classification)
(c) Filtration
(d) Thermal drying (may be costly since the heat capacity of
water is high)
The effectiveness of water removal as a function of particle
size (on a log scale) is shown in the figure below. The vertical
axis lists volume % water retained in the dewatered product. There
are numerous types of screens. Screening a slurry can very simply
remove excess water from a desired size fraction. The smaller the
particle size, the greater the total surface area per unit weight
and the more water that is retained by the solids. A sieve bend is
a type of screen that uses a curved surface. Slurry flows over the
concave surface and in doing so experiences some centrifugal force
due to the curvature. They are therefore somewhat more efficient
than flat screens.
There are numerous types of classifiers. Classifiers have use in
separating small particles from large particles or particles on the
basis of density differences. They may also be used for dewatering.
In this case the slurry may be fed into a long trough. Solids
settle and are dragged up an incline by spirals, rakes etc. A
disadvantage is that the aqueous phase may be significantly
diluted. Very fine particles may also be deliberately removed from
the pulp, with the aqueous phase (desliming). A thickener is a type
of classifier. According to the figure a thickener can achieve
about 75-80% water content by volume in the discharged underflow
slurry. This seems rather large, but it should be remembered that
minerals are much more dense than water. In many instances
thickeners can achieve 50% water retention by weight. Relatively
new paste thickeners use much deeper, narrower vessels and have
solids (mud) zones that are several meters thick; much more than in
conventional thickeners. This produces a much denser slurry (paste)
that contains less water and exhibits non-Newtonian flow properties
(rheology). An important type of classifier not indicated in the
diagram is a hydroclone. These use centrifugal force to accentuate
density differences. They are used for size separations, and for
dewatering as well. Centrifuges are accelerated sedimentation
devices that spin rapidly, also accentuating density differences.
In a hydroclone, the machine stays fixed and the slurry is
introduced tangentially into the cylindrical housing.
Filtration is commonly practiced in hydrometallurgy. Common
types of filters are drum filters (explained later), belt filters
and filter presses. Drum and belt filters employ reduced pressure
on the underside of a porous membrane to draw off liquid. Filter
presses use positive pressure on the side where the slurry contacts
the filter cloth. Vacuum filters, for instance, are good at
removing fine material and lower water content to as low as 10% by
volume.
It is apparent from the figure that no matter what the particle
size range, no mechanical dewatering system can removal all water.
Further, as particle size decreases mechanical methods become less
effective, due to the increased surface area per unit mass.
Reduction in water content to low levels ultimately requires
thermal drying. This is increasingly expensive as the cost of
energy increases. Feeds to pyrometallurgical smelters may required
thermal drying. In some instances this is also required for
hydrometallurgical processes, but not commonly.
Figure 1. Overview of dewatering methods [1].
In sedimentation the settling rates of very small particles (a
few microns in diameter) may be very slow under gravity alone.
Flocculants may be added which act to agglomerate fine particles
into more massive aggregates, which settle more rapidly. A wide
variety of flocculants are available. Many are polymers with
charged sites that attract oppositely surface-charged particles.
This is illustrated in the figure below. This is an important
technique in sedimentation. The objective now is not to separate
one type of solid from another, but to separate all the solids from
the water or solution. Thickeners are the most important type of
sedimentation unit. They are similar in design and size to
sedimentation classifiers. Diameters on the order of ~100 m exist.
They are cylindrical and have conical bottoms. A variety of designs
and types exist. They operate continuously. The overflow is the
substantially clarified water/solution. The underflow slurry is
higher in solids concentration than the feed slurry is, but is
still quite high in water or solution content. A radially aligned
rake turns, moving along the bottom to direct the solids to the
discharge.
Figure 2. Schematic illustration of how a flocculant gathers
small particles into a floc and makes a denser, more easily settled
agglomerate.
CCD Washing
When a solid-liquid separation must be performed it may be
preferable to use a thickener. Often solids settle readily under
gravity, but filtration and washing turn out to be uneconomically
slow. Additionally, as the tonnage of slurry to be separated
increases, the cost of the thickener per tonne of material
decreases. But, to filter and wash a large tonnage of slurry will
require several filters in parallel. (Filters have relatively low
capacities.) The capital cost of filters is proportional to the
throughput. A single series of thickeners will often suffice to
handle a large tonnage throughput. Thickeners can be made very
large.
The solution in a slurry of leached ore often needs to be
separated from the solids. Passing the slurry through a single
thickener would still leave a substantial fraction of the solution
(an hence valuable solute) going out the bottom of the thickener
with the solids. This residual solution can be recovered by washing
the solids. This is achieved by counter-current decantation. In
this technique a series of thickeners are used. This is illustrated
in the diagram below. Overflow from a thickener becomes the wash
solution to the preceding thickener. It is preferable to have the
underflow contain as high percent solids as possible. This
minimizes the fraction of valuable metal in the underflow solution
exiting with the solids. The final underflow slurry still contains
some solute in solution. This is minimized when the underflow
slurry has the highest possible solids content (less solution, and
hence less solute), and with increasing number of wash
thickeners.
In the schematic diagram below the total number of stages here
is just three. In practice many more stages may be used, depending
on the application. The number of wash stages in this example above
is two, after the first , or lead (lead as in front, not as in Pb)
thickener, to which feed slurry is added and from which the final
clarified solution overflows. (The number of wash stages is always
1 less than the total number of thickeners in the circuit.)
Wash
Figure 3. Schematic illustration of a simple CCD circuit.
water (or spent solution, low in the solute of interest) enters
the last thickener and moves up to preceding thickeners. Underflow
solids move from the first thickener to be combined with the wash
solution as shown. The solids move in one direction and the
solution overflow in the other. Hence the term counter-current
decantation, or CCD. )The term decantation refers to the idea of
pouring off a solution from a settled mass of solids.) It is
important that the concentration of solution be uniform in the
thickeners. This is achieved by ensuring good mixing of the
underflow with the wash stream, as indicated by the confluence of
streams in the diagram. If the flow of solutions and slurry is
turbulent, a good rule of thumb is that the combined streams should
flow through a distance equal to the radius of the thickener. Then
mixing of the streams is sufficient to provide a uniform
composition solution in the slurry. It is this mixing that results
in the overflow solution and underflow solution form a thickener
having the same composition! If mixing is inadequate there will be
some loss of efficiency. It is often necessary to add flocculating
agents to aggregate small particles in order to facilitate
settling. Too strongly flocculated particles may trap some solution
phase in the floc and this can result in excessive loss of solute
to the final underflow.
Test work needs to be done for new processes in order to
determine the settling rates of solids, for example after leaching
of an ore. Another critical variable is the efficiency of the
circuit as indicated by the recovery. Recovery is the ratio of the
mass of solute in the final overflow (the pregnant solution) to the
total mass of solute in the feed solution (the feed solution,
again, being associated with a slurry). The remainder is lost to
the final underflow solution exiting with the solids slurry leaving
the final thickener. A key parameter for evaluating efficiency is
the concentration of solids in the underflow. And the higher the
solids content, the less the amount of solution, and therefore
solute, that exits with the underflow slurry from each thickener.
Then a lower number of stages may be required to effect the same
degree of recovery. Washing results in a substantial dilution of
the final recovered solution. The final PLS solute concentration is
typically still >50% of the starting value. With enough
thickeners virtually complete recovery can be obtained. The
theoretical limit is 100%, albeit with a very large number of
thickeners. In practice, since thickeners are capital intensive,
there is an economic limit. A tradeoff between technical
feasibility and economic feasibility must be made. Each additional
stage recovers a smaller percentage of the solute; the law of
diminishing returns.
To calculate recovery and efficiency a mass balance approach may
be used. Consider a thickener into which a leach slurry enters. The
overflow is the pregnant solution. In this example there are four
stages, three of which are wash stages. In practice there may be
several more stages. This is a variable. A diagram illustrating the
solution flows is shown below.
Figure 4. Schematic illustration of solution flows and
concentrations in CCD streams.
The feed enters the lead thickener (number 0) with a solution
flow rate of F in convenient units. The concentration of a solute
is C*. The pregnant solution overflow exits at a flow rate of V
(> F) and a lower concentration C0 (because of dilution by wash
water). Wash water enters the final thickener (number 3) at a
concentration of solute = C4. If there is none of the solute in the
water then C4 = 0. The wash water flow rate is designated O. This
is also the overflow flow rate from each thickener, except number
0. (If this were not so, then solution would accumulate or deplete
in the circuit.) It is assumed that the solute concentration in
each thickener is uniform. The concentration in solution associated
with thickener 2 is designated C2 and so on. Underflow from each
thickener exits at a flow rate U of solution phase. The solids flow
rate out the bottom adds to the total mass flow, but the mass
balance is being carried out on the basis of solution. The
concentration of solute in solution exiting thickener 2 is C2 for
both the overflow and the underflow.
Now mass balance equations can be written for the solute. The
mass balance properly requires knowledge of the mass flow rates and
concentrations expressed in terms of mass of solute per unit mass
of solution. However, mass flow rates may not always be readily
available; pumps deliver known volumes per unit time. If the
solution and wash densities are close to 1 g/mL, which in many
instances is valid, the concentrations in mass/volume (e.g. Kg/m3)
and flow rates in volume/time may be used. If the density of the
leach solution, for example, is significantly higher than one,
which may occur with concentrated solutions, then volumetric
concentrations and flow rates will introduce a degree of error.
Incorporating solution densities into the calculations then will
overcome this error, but make the equations more complicated. For
the sake of simplicity, in this discussion we will assume solution
densities are 1 g/mL. Then 1 Kg of water or solution = 1 L
volume.
A mass balance equation may be set up for each stage. At steady
state, what enters must equal what leaves:
For 0FC* + OC1 = VC0 + UC0
(1)
For 1UC0 + OC2 = OC1 + U C1
(2)
For 2UC1 + OC3 = OC2 + UC2
(3)
For 3 UC2 + OC4 = OC3 + UC3
(4)In general for n wash stages,
UCn-1 + OCn+1 = OCn + UCn
(5)
Note the pattern of concentrations on the left and right. This
symmetry leads to convenient simplification of the math later. The
equations may be rearranged as follows:
FC*/U + (O/U)C1 = (V/U + 1)C0
(6)
C0 + (O/U)C2 = (O/U + 1)C1
(7)
C1 + (O/U)C3 = (O/U + 1)C2
(8)
C2 + (O/U)C4 = (O/U + 1)C3
(9)
The ratio O/U is called the overflow-underflow ratio, also
called the wash ratio. This very important variable is the
volumetric ratio of the overflow volume from each wash thickener to
the underflow solution volume from each thickener. Let this be x.
If C4 is zero, for wash water,
C3 = C2
(10)
x + 1
Now we have C3 in terms of C2. Next we express C1 in terms of
C2, and so on. In other words, start with the nth thickener and
work back. Substitute (10) into (8):
C1 + xC2 = (x + 1)C2
(11)
x+1
Rearranging gives:
C2 = C1
(12)
x + 1 - x
x + 1Similarly, substituting (12) into (7) yields:
(13)
Finally, substituting (13) into (6) leads to:
(14)
Note the repeating pattern in x terms! There is one set of these
per wash thickener. Recovery of solute from the 0th thickener is
expressed as:
Recovery = VC0/FC*
(15)
Multiply this by 100 and we have the recovery in %. Note also
that the pregnant solution overflow rate must obey the
equation:
F + O - U = V
(16)
assuming there are no changes in volume upon mixing. This is the
mass balance for solution flows. Recovery then is:
(17)
The concentration C* disappears, which makes sense; fractional
recovery cannot depend on concentration. It is just a ratio of
masses. F disappears in (17) too, but reappears later when V is
converted into F + O - U as per (16). Substituting (16) into (17)
and multiplying by (1/U)/(1/U) to convert O terms into O/U = x
terms leads to:
(18)It turns out that the function,
(19)
is precisely equivalent to:
xn+1 - x
(20)
xn+1 - 1
where n = the number of wash stages (the total number of
thickeners less one). Recovery then is:
R = F/U + x - 1
(21)
F/U + x - xn+1 - x
xn+1 - 1
When terms are collected for equation (19) hairy polynomials
result, one in the numerator and one in the denominator.
Multiplying these by (x-1)/(x-1) results in equation (20). Hence
the n+1 powers. It turns out that these numerators and denominators
are always factors of xn+1-x and xn+1-1, respectively; x-1 is the
other factor.
Equation (20) is called the wash efficiency. It indicates the
effectiveness of the total of all the washing thickeners. Note that
as the number of stages increases the wash efficiency gets closer
to 1, i.e. 100% recovery:
(xn+1 - x) ( 1
(22)
(xn+1 - 1)as n gets large, and if x > 1. Note too that as O/U
increases the wash efficiency improves. The higher O/U is, the less
the amount of solution present with the solids in the underflow. In
other words, the higher the solids content of the underflow, the
better the wash efficiency and recovery. This is reasonable; the
less of the solution that is going out the bottom with the solids,
the more of it is being recovered with the overflow from each
thickener. Hence the O/U ratio is absolutely critical to
performance of thickeners.
Equation (14) may be rearranged with the same considerations as
above to read,
FC*
C0 = U
(23)
F + x - xn+1 - x
U xn+1 - 1
A CCD wash circuit is designed around experimental data obtained
during testwork to engineer a process. If it turns out that the %
solids in the underflow is significantly less than the testwork
suggested, then wash efficiency and total recovery will be
correspondingly poor. This results in loss of metal values and it
can be economically fatal. In conclusion, recovery and wash
efficiency are improved by increasing the number of wash stages and
increasing the % solids in the underflow. Each additional thickener
improves recovery, but to an ever decreasing extent, as noted
previously. With the equations above the effect can be quantified.
Sometimes a thickener malfunctions (the walls fall off). While it
is being repaired it must be bypassed. This lowers the number of
stages and lowers recovery. If this happens too often, overall
performance will be decreased. When solute is also present in the
wash solution
The equations developed above assumed that the wash water
contained none of the solute of interest. Then C4 in equation (9)
was zero, or more generally, Cn+1 = 0. If this is not the case then
the algerbra is a bit more complicated, but the equations can still
be readily solved. The same method is used. The value of the solute
concentration in the wash solution must be known. All the same
assumptions as previously apply. The solute concentration in the
PLS from the zeroth thickener is then found to be,
(24)Note that there are two terms:
FC* xn+1(x - 1) Cn+1
(25)
U + xn+1 - 1
F + x - xn+1 - x F + x - xn+1 - x
U xn+1 - 1 U xn+1 - 1
The first is identical to equation (23). The second is entirely
due to the solute in the wash solution. The two contributions are
completely separable.
Recovery becomes a matter of how you want to define it. Ignoring
the contribution from the wash solution, the recovery of solute
that is fed into the lead thickener is the same as it was before
and is expressed by equation (21). The wash efficiency is also
still the same as equation (20). If this were not so, then we could
increase recovery of solute from the feed just by adding some of it
to the wash water. That would violate the principle that you cannot
get something for nothing, thermodynamically speaking. We can write
and equation for recovery of solute that incorporates the solute
added both from the feed and the wash solution. Then,
R = VC0
(26) FC* + OCn+1Substituting equation (24) for C0 and collecting
terms finally yields the equation below,
(27)For Cn+1 = 0, i.e. wash water, the ratio on the right goes
to unity and the familiar recovery from equation (21) results. The
ratio on the right is the factor due to solute being present in the
wash solution. In the limit of large n the ratio on the right tends
toward,
FC* + Cn+1 x - Cn+1
(28)
U
FC + Cn+1 x
U
Thus the ratio on the right in equation (27) is always 1 is
used, but W >4 would require too much time. The void volume
within the filter cake is very small compared to the filtrate
volume. Hence even with W = 4 the dilution of the filtrate is very
small. With CCD wash circuits the extent of dilution is
substantially greater.Economic Aspects of Solid-Liquid
Separation
Some aspects of this have already been mentioned. These are
typically rather slow processes. Capital costs are therefore high.
Rates of removal of water tend to decrease rapidly as the water
content decreases (law of diminishing returns again). Going from 1%
to 0.1% water may take as long as going from 50% to 1% did.
Likewise the same idea applies to removing suspended solids from a
solution. The more clear the liquid needs to be, the greater the
time and effort required. Sedimentation processes have moderate
electricity requirements. Flocculating reagents may be quite
costly, but they tend to be added at low to moderate
concentrations. Drying costs may be quite high since high amounts
of energy are required to evaporate water. This is not commonly
employed in hydrometallurgy.
The highest solid-liquid separation costs are encountered just
after leaching of fine solids in a reactor (tank or autoclave
etc.). After initial separation to obtain a leach solution, the
solids must be washed and then dewatered again to recover valuable
metals in solution. The leach solution itself must be clarified to
a high degree (very low suspended solids content) for it to be
suitable for metal recovery. This may involve considerable capital
cost. Good lab testing is always required. Generally the cost of
solid-liquid separation increases with decreasing particle size. It
is preferable to avoid having to conduct solid-liquid separations
on very fine particles. References[1] Hayes, P.C., Process
Principles in Minerals and Materials Production," Hayes Publ. Co.,
1983, p. 117.
[2] Retrieved May 17/12 at:(a)
http://www.komline.com/docs/rotary_drum_vacuum_filter.html(b)
http://www.solidliquid-separation.com/VacuumFilters/Drum/drum.htm
PAGE 18