![Swenson walker crystallizer pdf download Swenson walker crystallizer pdf download](/uploads/1/2/1/4/121419355/973984525.jpg)
Bạn đang xem bản rút gọn của tài liệu. Xem và tải ngay bản đầy đủ của tài liệu tại đây (28.3 MB, 329 trang )
Swenson Walker Crystallizer Pdf Filler
116
CRYSTALLIZER SELECTION AND DESIGN
such cases, the total heat released on crystallization may be a
substantial portion of the total heat removed in a cooling-type
crystallization. In evaporative-type crystallizers, the heat of crystallization is usually negligible when compared with the heat
required for vaporizing the solvent.
The heat effects in a crystallization process can be computed
by two methods:
offers significant advantages and permits modifications and recalculation quite easily. The following equation is based on a material
balance (Perry 1973) and may be used for direct yield calculations.
1. A heat balance can be made wherein the individual heat effects,
such as the sensible heat required for heating or cooling the
solution, the heat of crystallization of the product formed, and
the latent heat of vaporization for liquid vaporized can be
combined into an equation for the total heat effects.
2. An enthalpy balance can be made in which the total enthalpy of
all leaving streams minus the total enthalpy of all entering
streams is equal to the heat absorbed from external sources by
the process.
where
The advantage of the enthalpy-concentration diagram method is
that both heat and mass effects are taken into account simultaneously. This method has only limited use, however, because of the
difficulty in obtaining enthalpy-concentration data that have been
published for only a few systems. In using either method, however,
it is necessary to make a corresponding mass balance since the heat
effects in a crystallization process are related to the quantities of
solids produced through the heat of crystallization.
5.1.3.
YIELD OF A CRYSTALLIZATION PROCESS
In most cases, the process of crystallization is slow and the final
mother hquor is in contact with a sufficiently large crystal surface so
that the concentration of the mother Hquor is substantially that of a
saturated solution at the final temperature in the process. In such
cases, it is normal to calculate the yield from the initial solution
composition and the solubility of the material at the final temperature. Some solutions, such as sucrose and water, supersaturate easily
and do not attain equilibrium with respect to the final mother liquor
temperature. In such cases, an assumption must be made as to the
mother hquor composition at the end temperature.
If evaporative crystallization is involved, the solvent removed
must be taken into account in determining the final crystal yield. If
the crystals removed from solution are hydrated, then the water of
crystallization in the crystals must be taken into account in making
the final calculation since this water is not available for retaining
solute in the solution. The crystal yield is also influenced in many
plants by the removal of some mother liquor with the crystals
being separated in the centrifuge or on the filter. Typically the
product separated on the centrifuge or filter carries with it adhering
mother liquor that is in the range of 2-10% of the weight of the
crystals.
The actual yield is computed from a knowledge of the solubility of the system at various temperatures under consideration and
may be calculated by algebraic equations or trial and error calculations. When the calculations are made by hand, it is generally
preferable to use the trial and error system since it permits easy
adjustments for relatively small deviations found in practice, such
as the addition of wash water, instrument and purge water, or
liquid purges from the system. While the calculation of yield for
an anhydrous salt formed by cooling a saturated solution can be
calculated quite simply, the calculation of yield for systems crystallizing hydrated salts by evaporation is considerably more complex
and often requires use of a solubility curve for the hydrated salt
plotted in terms of solubihty per pound of 'free' water. If calculations are to be made with a digital computer, the formula method
can be used; however, calculation via the spread sheet method
C= R
lOOWo-S{Ho-E)
100 - S{R - 1)
(5.1)
C- • weight of crystals in final magma, kg
mole weight of hydrated crystal
Rmole weight of anhydrous crystal
S = solubility at the final mother liquor temperature
(anhydrous basis) in kg/100 kg solvent
Wo = weight of anhydrous solute in the original batch
Ho = total weight of solvent at the beginning of the batch, kg
J? = evaporation, kg
5.1.4.
FRACTIONAL CRYSTALLIZATION
When two or more solutes are dissolved in the solvent, it is sometimes
possible to separate these into pure components or separate one and
leave the other in solution. Whether or not this can be done depends
upon the solubihty and phase relations of the system under consideration. How to plot and use this data is explained by Fitch (1970),
and Campbell and Smith (1951). It is helpful if one of the components has a much more rapid change in solubility with temperature
than does the other. A typical example, which is practiced on a large
scale, is the separation of KCl and NaCl from water solution. A
simplified phase diagram for this system is shown in Figure 5.1. In
this case, the solubihty of NaCl is plotted on the Y-axis as parts per
100 parts of solvent, and the solubihty of the KCl is plotted on the
X-axis in the same units. The isotherms show a marked decrease in
solubihty for each component as the other component is increased.
This example is typical for many inorganic salts.
The hne marked CE in Figure 5.1 is the mixed salt hne and
corresponds to compositions of the solution where both sohd
phases are in equihbrium. To make a separation of the two solutes,
it is necessary to operate on one side of this line or the other. When
crystallizing solid-phase NaCl, the solution compositions lie above
the hne CE. When crystallizing KCl only, the solution composition
lie below the line. Practically, it is possible to approach the line
very closely (within a few percent) and still precipitate the solid
phase indicated without coprecipitation of the other material. On
this curve, evaporation from a solution of any specific composition
is represented by movement away from the origin, and dilution of
the system is represented by movement from the composition in
question towards the origin.
A typical separation could be represented as follows. Starting
at E with a saturated brine at 100 °C, a small amount of water is
added to dissolve any traces of solid-phase NaCl that might be
present so as to be sure the solids precipitated initially are KCl.
This dilution is represented by the hne E-H. Evaporative coohng
along line H-G represents the precipitation of KCl. During this
evaporative cooling, part of the water evaporated must be added
back to the crystallizer to prevent coprecipitation of NaCl. The
final composition at G can be calculated by the NaCl/KCl/water
ratios and the known amount of NaCl in the incoming solution at
point E. The solution at point G may be concentrated by evaporation at 100 °C. During this process the solution will increase in
concentration with respect to both components until point I is
reached. At this point, NaCl will precipitate and the solution will
become more concentrated with KCl, as indicated by the line I-E,
until the original point E is reached. If concentration is carried on
beyond point E, a mixture of NaCl and KCl will precipitate.
5.1. FUNDAMENTALS
35
NaCl
^^''^-^30^C.^
'^'*'««*l,„^
30
6
E
' ' ' V^'^I
y%
25
§
BO'O.
X
15
2
5
NJOO'C.
20
5
O
117
/
W
/
40*C. ^
*c.
V
^
/
10
15
KCl,
20
25
30
35
40
PARTS/100 PARTS H20
Figure 5.1 Mutual solubility of KCl and NaCl in water.
5.1.5.
NUCLEATION
In all crystallization phenomena, the crystals must first form and
then grow. The formation of a new solid phase may occur on either
an inert particle surface or in the solution itself, and is called
nucleation. The increase in size of the nucleus by layer-upon-layer
addition of solute is called growth. Both nucleation and crystal
growth have supersaturation as a common driving force. If the
solution remains (only) saturated, the crystals will not grow. If it
becomes unsaturated, the crystal phase present will gradually disappear.
Most early work on crystallization and nucleation was done
with extremely pure solutions. There are numerous literature references in the high degree of supersaturation created and the energies
associated with the formation of nuclei under such conditions; this
work is summarized in Chapter 2 of this volume. Within the last
few years there has been a substantial rethinking of this aspect of
crystallization. Most modern students of the subject agree that
there are at least four identifiable sources of nucleation that must
be considered in industrial crystaUization:
1.
2.
3.
4.
Homogeneous nucleation.
Heterogeneous nucleation.
Attrition.
Contact nucleation (secondary nucleation).
In addition to these, one must consider intentional and accidental
seeding as a surrogate form of nucleation, which is so often
encountered industrially that it is convenient to consider it with
other types of nucleation.
Homogeneous and Heterogeneous Nucleation. Homogeneous
and heterogeneous nucleation occur at very high levels of supersaturation, either in the solution, or in the case of heterogeneous
nucleation, on other inert particles that are present in the form of
crystals or as amorphous solid material. Most industrial crystallizers of the types in commercial use operate at levels of supersaturation far below those at which these types of seeding are
expected, except under startup conditions.
Attrition. Attrition causes small particles of crystalUne material that have already been formed to be broken from identifiable
crystals, and thereby, to be added to the body of soHds that are
present as individual or discrete particles. Their presence increases
as the level of mechanical energy input in the system increases.
They can be formed by contact of the crystals with a pump
impeller or due to impact of the slurry on tube sheets, piping, or
vessel walls. In some crystallizer designs, the attrition is sufficiently
large so that if the crystallization rate is reduced to a very low
value, the product size progressively decreases as particles separate
from the existing soHds and become new particles. Attrition can
occur whether or not supersaturation is present.
Contact or Secondary Nucleation. Contact or secondary
nucleation is another significant source of crystallizer seed. It
occurs when crystals touch each other, some metal object, such
as a pump impeller, vessel walls, or piping while the crystals are
growing. The amount of nucleation caused is proportional to the
energy expended when striking the crystal and the degree of supersaturation. It is a significant source of nucleation in most industrial
crystallization equipment and was first quantified by Clotz and
McCabe (1971). The effects have been described by Ottens and
118
CRYSTALLIZER SELECTION AND DESIGN
deJong (1972) and Bennett et al, (1973). In recent years, this
phenomenon has been investigated by a number of authors,
including Rush et al. (1980), Grootscholten et al. (1984), Shaw et
al. (1973), and Ness and White (1976).
This phenomenon differs from attrition in that it is proportional to and dependent upon super saturation.
5.1.6.
POPULATION DENSITY BALANCE
Modern industrial crystallization theory dates essentially from the
work of Randolph and Larson (1962), who developed the concept
of population density insofar as it applies to mixed suspension,
mixed product removal crystallization equipment. A detailed treatment of the development of these concepts is given in Chapter 4
of this volume. The population density concept is useful because it
allows the user to take the data developed from a crystal screen
analysis along with knowledge of the operating parameters of the
crystallizer, such as the retention (drawdown) time, slurry density,
and active volume, to calculate the growth rate and nucleation
rate. From multiple sets of data it is possible to determine the
kinetics of the system and thereby have sufficient data to predictably model the effect of other changes and modifications in the
system. Although these equations are limited to fully mixed suspensions such as the forced circulation and draft tube baffle (DTB)
crystallizers described in Section 5.3.3 and 5.3.4 (and do not apply
to agitated trough crystallizers, static tanks, etc.), they are important because this type of equipment accounts for the vast bulk of
the important commercial applications. At the present time, mathematical analogs for other types of crystallizers are either lacking
or not as well developed.
5.1.7. CRYSTAL SIZE DISTRIBUTION
As developed in Chapter 4 of this volume, if a crystallizer (such as
shown in Figure 5.2) is operating at steady-state, is thoroughly
mixed, is discharging a representative sample of the suspended
BAROMETRIC CONDENSER
NON-CONDENSABLE
GAS OUTLET
WATER OUTLET
PRODUCT SLURRY
DISCHARGE
CIRCULATION
PUMP
-DRAIN
S E S N FORCED CIRCULATION CRYSTALLIZER
W NQ
Figure 5.2 Swenson forced-circulation crystallizer.
5.1. FUNDAMENTALS
magma, and if McCabe's AL law is applicable, a straight line plot
of In n versus L will result. The intercept of this straight line with
the size corresponding to L = 0 is the nuclei population density
and the slope of this straight line is equal to (—l/Gt). The growth
rate (G) so calculated is the diametral growth rate, which is twice
the facial growth rate. The retention time (t) used in these calculations is sometimes referred to as the drawdown time and is the
slurry containing (active) volume of the crystallizer divided by the
slurry discharge rate. The population density in the crystallizer
under these conditions is shown in Eq. (5.2).
(5.2)
n = «^exp(-^/^^>
By definition, therefore, the nucleation rate, which is the birth of
crystals at size L = 0 is shown in Eq. (5.3). As will be shown later,
B^ is also a function of other properties of the system and the
crystallizer.
£f^ = n^G
(5.3)
no./h/volume
The total number of crystals, the total area of the crystals, and the
mass of the crystals in the sample are shown in equations below.
/•OO
Total no. of crystals: N=
Jo
n^Qxp'^'^'dL = nGt) (5.4)
POO
Total area of crystals: A=
KL^n^Qxp~^^^^dL
Jo
= 2KnGtf
(5.5)
/•OO
Total mass of crystals: M =
k^L^n^Qxp'^'^^dL
Jo
= 6pKnGty
(5.6)
Since nucleation is also a function of supersaturation, it is frequently expressed as a power law model similar to that shown in
Eq. (5.7) when the crystallizer design effects on the nucleation rate
can be neglected.
EP = knM-'G'
(5.7)
This power law model contains an expression for slurry density
and, as will be shown later, there are other considerations that
could have been added that have a demonstrated influence on
nucleation rate. In order to satisfy any given condition, the mass of
crystals per unit volume of the crystallizer, as shown in Eq. (5.7), must
be consistent insofar as the nuclei population density and growth
rate are concerned with that shown in Eq. (5.2).
Growth rate, therefore, is a constrained variable and the
correct solution to any new set of conditions involves a simultaneous knowledge of both the nucleation rate and the growth rate.
5.1.8. CRYSTAL WEIGHT DISTRIBUTION
Once the growth rate at any given retention time is known and if
the plot of In n versus L is a straight line, then the crystal weight
distribution may be computed by the following equations.
X = LIGt
(Dimensionless size)
Mx
Weight fraction: Wx =
M
(5.8)
Mass to size X
X
Total sample mass or mass at L = C D
Wx = - exp
119
This equation can be easily solved for many different L values to
give the crystal size distribution. The solution to this equation is
also given in table form in Appendix A of the book Theory of
Particulate Processes (Randolph and Larson 1971). The generalized solution to Eq. (5.9) for certain specific properties, such as the
dominant particle, is given in Eq. (5.10).
Dominant particle (area wt avg): Lo = 3Gt
(5.10)
The average particle by weight may be computed from Eq.
(5.11). This is a very useful equation for quick estimates of size
versus growth rate and vice versa.
Average particle by weight: Lav = 3.61Gt
(5.11)
The techniques for evaluating growth and nucleation rates
have been widely used in the analysis of forced-circulation crystallizers and other systems that are being operated continuously, are
thoroughly mixed, and where growth follows the AL law. There is
good agreement between field measurements and results predicted
by the theory. The nucleation sensitivity parameter (/) shown in
Eq. (5.7) is believed to be characteristic of many systems and is
often the same in large-scale and in laboratory tests.
NonUnearity of the plot of In n versus L is frequently encountered for a variety of reasons. In a thoroughly mixed suspension, a
straight line plot of In n versus L will generally not occur until 6-10
retention periods have passed following start-up or serious perturbations in operating conditions. Since the suspension is thoroughly
mixed, larger particles present must have time to grow, and frequently the coarsest fractions have a retention period many times
the average. Until these larger crystals are present at equilibrium,
the straight line will show a downward curve at the larger-crystal
fractions. An illustration of the curved plots and the ideal plot is
shown in Figure 5.3.
Also, during start-up conditions with systems that nucleate
heavily or with systems that exhibit cycling behavior with regard to
crystal size, it will be found that lower-than-equilibrium values of the
finer crystal sizes will be present for a time during the growth cycle.
Many crystallizers employ classification of the product discharge either intentionally through the use of elutriation legs or
hydrocyclones on the discharge stream, or unintentionally because
of inappropriate locations for the slurry discharge connection. The
use of an elutriation leg or cyclone to remove the coarsest solids in
the system will cause a concave curve in the larger crystal sizes,
representing a decrease in the population density of the coarsest
fractions. Randolph et al. (1968) generaUzed on the degree of
crystal size distribution narrowing that could be obtained using a
combination of classified product removal and vessel staging.
If growth of the crystals does not follow McCabe's AL law,
then an underUning assumption of the derivation of the population
distribution shown in Eq. (5.2) is not fulfilled. A generalized
treatment of the curvature that can exist in the plot of In n versus
L under such conditions is given by Abegg et al. (1968).
Multiple staging of crystallizers where nucleation per stage is
small will result in a significant narrowing of the crystal size
distribution. At least one large-scale continuous sugar crystallization system works in this manner and a theoretical treatment of the
subject is given by Abegg and Balakrishnan (1971).
Concave curvature upwards in the small crystal sizes has been
noted by many experimenters (Sikdar and Randolph 1976), and the
size-dependent growth rate theory proposed by Bujak (1976)
appears to have a solid basis. Recent work by Berglund and Larson
(1992), Human et al. (1982), Randolph and White (1977), and Zumstem and Rousseau (1986) make useful comments on this theory.
120
CRYSTALLIZER SELECTION AND DESIGN
© IDEAL DISTRIBUTION MSMPR CRYST.
(DSELECTIVE LARGE CRYSTAL DISCHARGE
(D 'BUJAKIAN' BEHAVIOR
0 G R O W T H RATE INCREASES WITH SIZE
©CRYSTALLIZER WITH FINES REMOVAL AND
CLASSIFIED DISCHARGE
Figure 5.3 Typical plots of the population density (n) versus length (L).
5.1.9. CONTACT NUCLEATION
As mentioned earlier, the nucleation rate is actually not a simple
power law model of growth rate and slurry density, but should
contain terms dependent upon contact nucleation effects caused by
the pump or propeller circulator and the motion of crystals striking each other within the crystallizer suspension. This is shown
algebraically in Eq. (5.12).
^ = K„M^G' + BP
(5.12)
where
^ = nucleation rate (total)
^ = contact nucleation
G = growth rate mm/h
M = slurry density g/1
/, j , Kn = experimentally determined constants
Work has been done by Bennett et al. (1973) to indicate the
correlation between mechanical nucleation and crystallizer physical parameters as shown in Eq. (5.13).
Efi = Ket
TIPS'
TO
f
LdL
(5.13)
Jo
where
TIPS = tip speed of pump or impeller
active volume
= time/turnover
T0 =
circulation rate
The nucleation term Pf^ used in Eq. (5.13) includes terms containing specific design perimeters from the crystallizer. If a straight-line
plot of In n versus L is obtained, Eq. (5.13) can be solved as shown
below.
Pll = KepvPG
TIPS^
TO
(5.14)
As shown in Eq. (5.14), small changes in the dominant particle size
have a fifth-order influence on contact nucleation, other factors
being equal, thereby explaining the historical character of most
forced-circulation type systems to operate stably with a particle
size distribution (for a given pump speed) that is changed very little
by changes in slurry density or retention time.
Another mechanism that must be mentioned in connection with
nucleation in real crystallization equipment is attrition. In a classic
paper by Randolph (1969), this subject is reviewed. Randolph
demonstrated that crystal breakage, classified product removal,
and decrease in linear growth rate with crystal size all skew the crystal
size distribution, producing a narrower size range of crystals than
would be expected in a mixed suspension, mixed product removal
system. This skewing results in a curve of the plot of In n versus L.
There probably exists no good way to separate the influence of
attrition or breakage from contact nucleation at the present time
other than by observation of the product crystals and measurements of the crystal size distribution or coefficient of variation.
With ammonium sulfate, there is relatively little observable distortion of the crystal habit by attrition. In the earher work by Bennett
et al. (1973), there was significant rounding of the salt particles in
the largest size range and at the highest pump speeds.
Whether the increased nucleation mechanism is due to the
McCabe-type contact nucleation, attrition, or the action of high
liquid turbulence of the Estrin type, the final result will be a smaller
particle size and probably a narrowing of the size distribution as
the pump speed increases.
5.1.10. CRYSTALLIZERS WITH FINES REMOVAL
Crystallizers of the draft tube baffle type that employ fines removal
and destruction devices can also be analyzed by the techniques
described previously, as shown in Perry (1973) and by Larson
(1978).
Jo
[rP exp^' LF/GIF) ](exp(-^/^^))L3^L
(5.15)
5.1. FUNDAMENTALS
121
-SLOPE = — ^ (FINES)
GTF
Ln n°eff
SLOPE = —^ (PRODUCT)
Figure 5.4 Plot of In n versus L for a crystallizer with fines destruction.
As shown in the above Figure 5.4, the size removed by the
baffle (crystals up to Lp) determines the effective nucleation rate
and growth rate, whereas the size of the product crystals is represented by a straight line of lower slope, but at a longer retention. An
extension of this line defines the effective nucleation rate. In rp eff.
As long as the particle size removed by the baffle is relatively small
compared with the product size crystals, the overall distribution
can be represented by Eq. (5.15).
The weight size fraction from a crystallizer of the DTB type
for particles up to size X is
Wx =
6Kpn^e^-^^l''^Gtf [l - exp-^ (^i + x + ^ + ^ ) ]
M = slurry density, g/1
X = L/Gt (dimensionlesssize)
Figure 5.5 shows the calculated and experimental data for an
ammonium sulfate crystallizer operating at a retention time of
5.4 h. The calculated size distribution for various Lf sizes at constant
CALCULATED SIZE DISTRIBUTION
AT G=0.554 mm/HR, Ln n° = 18.663
MT=527 C|/L,TF=0.094
• EXPERIMENTAL DATA AT 0.28 = LF
LF=0.29
LF=0.270-v
1.60
(5.16)
where
LF=0.35
1.80
M
10
*v
5
LF=0.250-^
1.40
LF = 0 . 2 2 5 - ^
^X
^
V
1.00
N
N >
4
^
14 I
^ v
x ' ^x '
^
X
0 80
^
^
0.60
U.4U
X
20
28
v , •N,
^ - ^ ^
35
48
^ ^
65
0.20
100
150
0
0.01
10
30
50
70
90
99.99
CUMULATIVE % RETAINED
Figure 5.5 Calculated size distribution of (NH4)2S04 in a DTB crystaUizer as a function of Lf.
122
CRYSTALLIZER SELECTION AND DESIGN
CALCULATED SIZE DISTRIBUTION
AT 6=0.554 mm/HR, Ln n*=]8.663
Mj-527 g/l, 1^=0.27 mm
T=5.f
'^Nf^
TF=O.IO-
^
30
50
70
CUMULATIVE % RETAINED
Figure 5.6 Calculated size distribution of (NH4)2S04 in a DTB crystallizer as a function of Tf.
mother liquor retention times of tp = 0.094h (5.64min) is shown.
The influence of this change is very great—much greater than that
of either tf or /, which are shown in Figure 5.6.
5.1.11.
CYCLIC BEHAVIOR
With control techniques such as fines destruction, the particle size
distribution within a crystallizer body may be varied through
relatively wide ranges by changing the velocity behind the baffle
and hence, the diameter of the particle withdrawn. The quantity of
liquor removed with the particle separated and its residence time
within the crystallizer body is another important variable. Experience has shown that such systems, when pushed beyond their
capacity, can produce cyclic crystal size behavior due to homogeneous nucleation that occurs as the supersaturation rises beyond
the metastable Hmit.
Numerous attempts have been made by operators and
designers to stabilize such systems; the normal technique for doing
this is by an appropriate selection of the maximum particle size
separated by the baffle (Lf) and the flow through the fines destruction system. A pioneering paper by Randolph et al. (1973) developed the mathematical basis for instability in systems of this type,
and predicted instability in such systems as a function of the fines
dissolving parameter and the nucleation sensitivity parameter.
Experimental verification of these concepts was demonstrated by
Randolph and Beckman (1977) for a KCl crystalHzer.
To control a crystallizer of this type in such a way that cycling
can be avoided is extremely difficult and for the most part has only
been conducted with Umited success. The retention dme required
to grow large KCl or (NH4)2S04 crystals in industrial equipment is
in the order of 4-6 h. The time required for overnucleation when
the system exceeds the metastable zone is a matter of minutes. The
kind of operator attention required to make adjustments in the
fines destruction system that would be effective in this kind of a
crystalHzer requires greater operator attention than can be normally expected. In addition, the presence of excess fines in the fines
destruction system is difficult to detect visually until the particles
are 200 mesh or larger in size.
In the last few years, laser particle size measuring equipment
capable of being used for on-line measurements in the fines
destruction system of crystallizers has become available. Such
systems can respond to particle diameters down to a few microns
in size even when the solution is not clear. These systems have been
demonstrated as effective tools in on-line control in crystalHzation
equipment by Rovang and Randolph (1980), Randolph et al.
(1977), Randolph and Low (1982), and Bennett and Randolph
(1983).
Figure 5.7 shows a cooling KCl crystallizer of the DTB type,
including a fines destruction system with variable velocity and flow
control. The baffle is divided by vertical baffles into several areas
and the recycle flow can be changed, thus giving independent
control over tf (fines destruction retention) and the separation
velocity behind the baffle.
Figure 5.8 shows an on-line particle size counter installed in
the fines destruction system of such a crystallizer, and arranged for
continuous monitoring of the nuclei population density and a
feedback control system for regulating the fines destruction system.
The use of an in-line laser particle counting system is surprisingly
insensitive to the presence of small amounts of foreign material;
the characteristics of the feed required for such a system are
compatible with the liquid delivered by the fines separation baffle
of a DTB-type crystallizer. With sampling dmes of around
3-6 min, this equipment is capable of generating the basic data
required for continuous evaluation of the nuclei population density. Coupled with a modern highspeed distributed computer control system, calculations may be readily made and the control
5.2. SELECTION OF A CRYSTALLIZER
123
TO AIR
EJECTORS
VAPOR HEADLIQUID LEVEL
MOTHER
LIQUOR
DILUTION
WATER
BAFFLE
SETTLING AREA
FEED FROM
PRECEDING
STAGE
RECYCLE
PROPELLER FLOW
DRIVE
•d
NEXT
CRYSTALLIZER
STAGE
MOTHER
PRODUCT
LIQUOR PUMP
PUMP
Figure 5.7 Flows and instrumentation on a typical DTB crystallizer.
SAMPLE PUMP l o n a i
LASER
y
SAMPLE CELL
i-^-r-EI
1
'
-a-*.
^
WASHOUT ^ M . L TRACED
SAMPLE LINES
Figure 5.8 Particle size measurement and control of a DTB
crystallizer.
algorithm may be updated as required to achieve the best control
characteristics. Use of this system is an outstanding example of the
application of complex mathematical modeling to practical industrial equipment to achieve optimum results.
5.2.
SELECTION OF A CRYSTALLIZER
A crystallizer is an apparatus in which an environment can be
created suitable for the formation and growth of crystalline materials. Paramount in the design of such equipment is the means that are
chosen to create supersaturation at a temperature level that produces
a desired or proper hydrate or composition of the product crystals.
5.2.1.
INFORMATION REQUIRED FOR EVALUATION
Before a potential crystallizer application can be properly evaluated, it is necessary to have certain basic information regarding the
material to be crystaUized and its mother liquor. Typical solubility
curves are shown in Figure 5.9. Is the material a hydrated or
anhydrous material? What is the solubility of the compound in
water or any other solvents under consideration and how does this
change with temperature and pH? Are there other compounds in
the solution that coprecipitate or remain in solution, if so, how
does their presence affect the solubility of the main component?
What will be the influence of the impurities on the crystal habit
and the growth and nucleation rate? What are the physical properties of the solution and the crystal? What vapor pressure of water
exists over the saturated solution and crystals as a function of
temperature? What is the heat of crystallization? What is the
production rate and on what basis is this production rate computed? What materials of construction can be used in contact with
the solutions at various temperatures? What utilities will be used at
the crystallizer location and what are the costs associated with
these utiUties? Is the final product to be blended or mixed with
other crystaUine materials or solids? What size of product crystal is
required and how can this material be separated from the mother
Hquor and dried? How can these solids or mixtures be handled and
stored without undue breakage or caking?
5.2.2.
SOLUBILITY
In working with materials of very steep solubility, such as Glauber's salt or other hydrated salts, it is common to use cooHng-type
crystallizers since they permit very high overall yields and reduce
the overall energy requirements of the separation process. With
materials whose solubility curves have a normal but moderate
slope, many choices are available including evaporative cooling,
surface cooling, or constant temperature evaporation. With
materials having flat or inverted solubility, it is necessary to use
evaporative crystallization. This can be done by causing water to
be evaporated from the solution at a wide variety of temperatures,
depending on the economics, the crystal phase desired, and the
materials of construction available. Evaporative crystallization is
generally done at constant temperature but can also be performed
at relatively low temperatures in solar ponds.
In general, cost considerations dictate that cooUng crystallization be used in cases where the precipitated salts are highly
124
CRYSTALLIZER SELECTION AND DESIGN
4) BENZaC ACID
DMT
TEMPERATURE °C
Figure 5.9 Solubility versus temperature for various compounds.
hydrated or where the solubility curve is relatively steep. The
reason for this is that to produce soUds by cooling only requires
removal of the heat of crystallization and the sensible heat of
cooling the mother Hquor. In general, these effects are quite small,
typically 0.7cal/gm°C for cooling the solution and from 18-55
kcal/kg of product crystallized. To produce a pound of crystals by
evaporation requires vaporization of water that is a function of the
solubiUty, but typically might represent 1100 kcal/kg of product
crystallized for a material whose solubility is 33%. Obviously there
are other factors that are highly specific to each case under consideration, but typically, surface cooling requires somewhat higher
investment than does evaporative cooling. In addition, the operating
cycles of surface-cooled equipment tend to be shorter than those
of evaporative crystallizers.
5.2.3. SCALE OF OPERATION
The scale of operation often has an overriding importance on the
selection of the equipment because of the means used for heat
transfer. For very small-scale crystallization work it is common to
use radiation. The capacity of such equipment varies from a few liters
up to several hundreds of Uters per day (of solution cooled). For
operation on scales up to several thousand liters per day, it is possible
to use tanks with water-cooled coils and an agitator. For large-scale
applications where the quantity of solution is thousands of liters
per day, it is almost universal practice to use vacuum evaporation
to remove the solvent; this is true whether the solution is cooled by
adiabatic evaporation or in equipment where crystallization occurs
because of isothermal evaporation.
5.2.4. BATCH OR CONTINUOUS OPERATION
Another consideration is whether the crystallization should be
carried out on a batch or continuous basis. The present tendency in
most processing plants is to use continuous equipment wherever
possible. Continuous equipment permits adjusting the operating
conditions to a relatively fine degree to get the best results in terms
of overall energy usage and product characteristics. It permits the
use of a smaller operating labor force and results in a continuous
utility demand that minimizes the size of boilers, cooling towers,
and power generation facilities. It also minimizes the capital
investment not only in the crystallizer but in the feed storage
and product hquor storage facilities. Continuous separating
devices for the product crystals have been developed to the point
where they can operate reliably for long periods of time and the
drying of crystalline products is also generally done on a continuous basis.
Batch handhng of wet or semidry crystalline materials presents considerable difficulties as compared with the storing and
handling of dried crystalline materials. On the other hand, a train
of continuous processing equipment, such as just described, has
economic apphcation only on relatively large production rates
from approximately 50 tons of product per day and upward.
Batch crystallization still has some range of application where
production rates are very small, where very expensive materials are
being handled and losses must be kept to an absolute minimum, or
where a careful inventory of materials is required. Wall deposits
in batch crystallizers are generally removed by each new charge of
unsaturated feed. Batch operation also has useful application
where the cooling range is very wide, such as in handling material
whose initial feed concentration corresponds to relatively high
pressure and whose final mother liquor temperature corresponds
to room temperatures or values significantly below it. In such
systems the use of batch crystallization avoids the shock introduced to the system by mixing high-temperature feed solutions
with relatively low-temperature mother liquor in continuous
equipment.
Another area of application is when the final liquor temperature is so low that it requires the use of extensive vapor compression equipment. In such systems, the use of batch equipment is
frequently more economical than anything else, other than a relatively large number of multiple-stage series crystallizers.
5.2.5. MULTISTAGE CRYSTALLIZERS
The use of multistage crystallization equipment, while not analogous to multistage evaporation equipment, nevertheless permits
certain economics in operation. It is normally employed when a
5.2. SELECTION OF A CRYSTALLIZER
RETURN TO
^ COOLING TOWER
COOLING WATER
REHEATED
. MOTHER LIQUOR
BAROMETRIC
CONDENSER
FEED
LIQUOR
Wml
i^'^r^ f^'^ {g'^fe {^'^fe ]Bim MB,
SWENSON 5-STAGE DTN POTASH CRYSTALLIZER
FILTRATE
TANK
Figure 5.10 Swenson five-stage DTB potash crystallizer.
MAKE-UP
STEAM
CONDENSATE
TO
CENTRIFUGE
FEED
SOLUTION
SINGLE STAGE
CENTRIFUGAL
VAPOR COMPRESSOR 4
Figure 5.11 Swenson single-effect recompression crystallizer.
125
126
CRYSTALLIZER SELECTION AND DESIGN
large flow at a high temperature and concentration is cooled to
produce crystals and the mother liquors are returned to a dissolving
or leaching station for reconcentration. In such systems, it is possible to use multistage cooling so that the mother liquor from the
final crystallizer can be heated by crystallizers in the first one or
several stages of the process. This heating can be done in either
barometric heaters or in tube and shell-type exchangers. The use of
this type of system not only reduces the steam required to raise the
leach solution to its operating temperature, but also reduces the
cooling water required on the last one or two stages of crystallization. Shown in Figure 5.10 is afive-stagepotash cooling crystallizer with three stages of mother liquor reheating. Systems
employing more than 12 stages of cooling have been built and
successfully operated.
5.2.6.
MECHANICAL VAPOR RECOMPRESSION
As the cost of energy has increased in recent years, attention is
again being directed to mechanical vapor recompression, which by
its nature, permits substitution of electrical energy for evaporation
and crystaUization rather than requiring heat energy (steam). A
typical recompression crystallizer flow sheet is shown in Figure
5.11. In a single-stage evaporative crystallizer operating at approximately atmospheric pressure, the amount of heat energy necessary
to remove a kilogram of water to produce the equivalent in crystal
product is approximately 555 kcal. If the water evaporated is compressed by a mechanical compressor of high efficiency to a pressure
where it can be condensed in the heat exchanger so as to supply the
energy needed to sustain the process, then the equivalent power for
this compression is about 6.3 kcal (Bennett 1978).
Although this technique is limited to those materials having
relatively low boiling point elevation and to those cases wherein a
significant amount of heat input is required to produce the evaporation required for a given crystaUization step, it nevertheless
offers an attractive technique for reducing the use of heat energy
and substituting mechanical energy (or electrical energy) in those
cases where there is a cost advantage in doing so. This technique
finds many applications in the crystallization of sodium sulfate,
NaCl, and sodium carbonate monohydrate.
As shown in Figure 5.12, the amount of vapor compressed per
horse-power decreases rapidly with increasing AT. Normal design
considerations, therefore, dictate that recompression evaporators
have a relatively large amount of heat transfer surface so as to minimize the power cost. In order to maintain adequate tube velocity
for heat transfer and suspension of crystals, this increased surface
requires a larger amount of internal recirculation within the crystallizer
body, which results in a lower supersaturation of the fluid pumped
through the tubes. One beneficial consequence is that with low
temperature differences between the liquid temperature and the
steam temperature, as indicated in Figure 5.13, the supersaturation
created at the tube wall with materials of inverse solubility tends
to be small. This effect is enhanced by operation where the solubility
curve is as flat as possible. For instance, increasing the Hquid
temperature from 71 to 100 °C would significantly reduce the
supersaturation of the solution in Figure 5.13. Such design parameters
tend to increase the operating cycle of the equipment when handling
solutions of inverted solubility that can frequently be troublesome
if higher ATs are used, as is often the case in normal evaporative
crystallizers.
5.2.7.
REACTIVE CRYSTALLIZERS
When a solid-phase crystalline material results from the reaction of
two components, it is often advantageous to perform this reacdon
in a reactive crystallizer rather than in a separate reactor. This is
commonly done, for example, in the case of ammonium sulfate
where liquid or gaseous ammonia and concentrated sulfuric acid are
reacted in the crystallizer to produce ammonium sulfate crystals,
as shown in Figure 5.14. The reaction is carried on in a Hquid
suspension of growing crystals and the heat of reaction is removed
by vaporizing water that is condensed and recycled to maintain the
Hquid balance. Other examples are the production of (NH4)2HP04
from NH3 and H3PO4 the oxidation of calcium sulfite by oxygen
(air), and the neutralization of waste acid by lime.
The reactants can be mixed in the circulation piping of a
forced-circulation-type crystallizer, or 'Oslo' crystallizer (Figure
8
X
§
30.5
s
5
AT (°F) = STEAM TEMP.-VAPOR TEMP.
VAPOR TEMP > 21 OF.
30.0
7r
^v^.^
83*
il
• SUPERSATURAT ION
r TUBE WAl.L
80
85
90
95
100
105
SOLUTION TEMPERATURE, 'C,
Figure 5.12 Pounds per hour of vapor compressed per BHP versus
temperature difference.
Figure 5.13 Maximum supersaturation at 12 °C. GTD.
110
Main Office Swenson Technology, Inc. 26000 S Whiting Way Monee, IL USA [email protected] P: +1 708 587 2300 F: +1 708 587 2225. Swenson-Walker Crystallizer. It consists of an open trough with a semicircular bottom having a cooling water jacket welded outside. It is about 2 ft wide and 10 ft long. The hot concentrated solution to be crystallized is fed at one end of the trough and cooling water usually flows through the jacket in counter current to the solution. SWENSON-WALKER CRYSTALLIZER. Crystallization for PU 3rd semester pharmaceutical. (PPT, KEY, PDF) logging in. Swenson- Walker crystallizer 60 This is common type of continuous. Master of Crystallography and Crystallization October 2012. Master of Crystallography and Crystallization October 2012-Sevilla. The Swenson-Walker crystallizer. Swenson-walker crystallizer Description: It consists of an open trough (A) 2 ft wide, with a semicylindrical bottom. A water jacket (B) is welded to the outside surface of the trough. “DINESH” Swenson Walker Crystallizer is designed to demonstrate and stimulate the phenomena of crystallization. Swenson Walker crystallizer is a continuous.