. Cl7rm. G i g .

, 15(3) 424-428
C / 7 i ~./. (2007)

An Innovated Tower-fluidized Bed Prilling Process

WU Yuan(f%E)*,BAO Chuanping(@,f$T) and ZHOU Yuxin(M 35%)

College of Chemical Engineering & Pharmacy, Wuhan Institute of Technology, Wuhan 430073, China

Abstract Cooling-solidification of sprayed droplets is one of major methods for prilling of melt. Traditionally,
this is carried out in an empty tower, and the equipment requirement for producing larger particles is very high, re-
sulting in not only significant cost increasing but also difficulties in transporting melt etc. Based on analysis and
simulation, a new prilling process is developed for the melt prilling, which combines a tower with a fluidized bed so
that the height of equipment is greatly decreased, and it exhibits satisfactory performance in industrial application.
Mathematical model for tower prilling, its simulated results, the structure of the equipment for the innovated prill-
ing process and its application are addressed.
Keywords prilling, melt, spray tower, fluidized bed

1 INTRODUCTION wall for spray, as widely used in urea and ammonium

It becomes clear that many solid products from
chemical and other process industries, especially those
produced in huge quantity, tend to be made into gran-
ules, which are more convenient for use, storage, and
transportation because of their much smaller specific
surface area and larger bulk density.
Up to date, cooling-solidification of sprayed
droplets is one of the major processes for granulation
from melt, in which melt is sprayed into droplets with
requested sizes and then cooled and solidified to yield
granular solid product. Traditionally, the process is
carried out in an empty tower, i.e., melt is sprayed at
the top of the tower; the droplets formed are cooled
and solidified as they fall downwards, and solid parti-
cles are collected at the bottom. This process has been
used successfully in industries for many years, e.g.,
for urea and ammonium nitrate, and production capac-
ity of 11 single unit can be as large as 7 W 8 0 t per hour
at present. The major disadvantage of the process is
the requirement of a very high tower, resulting in cost
increase and difficulties in transporting melt, opera-
tion and maintenance. Also, it is almost impossible to
obtain the products with size greater than 2mm in this
way because of unacceptable height of tower.
On the basis of analysis and simulation, a new
process combining tower prilling with fluidized bed
for the same purpose is developed and patented, which
has been applied in industry and exhibited very good
perforinances.
nitrate production, while the second raises the re-
quirement for a high tower. As particle size increases,
the settling velocity of particles and the resistance of
surface solid shell to heat transfer increase exponen-
tially and, at the same time, the specific interface area
decreases significantly. So the height of the tower re-
quested is a strong function of particle size.
2.2 Mathematical model for tower-prilling
2.2.1 Shrinking unsolidified core model
The behaviors of melt or solution droplets during
evaporation-solidification in a gas stream may vary
from substance to substance. While in the experiments
carried out by Wu et al.[11, a solid film non-permeable
to gas is found to form first on the surface of solution
droplets of most substances tested, which contributes
great resistance to heat and mass transfer, yielding the
four-period drying mechanism. For generalization, it is
considered that, after entering a cool gas stream, the
surface of melt droplet solidifies first, and then the solid
layer grows towards the center as the droplet falls down
until completely solidified. The cooling-solidification
process of droplet can be described with the shnnking
unsolidified core model shown in Fig. 1. If the differ-
ence between melt and solid densities is small, the so-
lidified particle would be dense, while when the differ-
ence is large it may have a hollow central core.

2 ANALYSIS FOR TOWER PRILLING

2.1 General consideration
The key problems having to be solved in prilling
of melt are: (1) The melt must be sprayed into droplets
with uniform size as requested; and (2) The sprayed
droplets must be fully solidified and cooled in order to
avoid bonding of particles with each other and caking shrinking intcrlace unsolidilicd core
on the wall, and to ensure that the product collected at Figure 1 Shrinking unsolidified core model
the boltom can be packed directly.
Industrially, the first problem has been solved The volumetric ratio of droplets/particles to gas
successfully by employing rotary cup with perforated in prilling towers is normally very small (around 0. I %

Received 2006-09-05, accepted 2007-0 1-29.

* To whom correspondence should be addressed. E-mail: wuyuan@public.wh.hb.cn
 An Innovated Tower-fluidizedBed Prilling Process
only) so that the effects of droplets/particles on each
other in both heat transfer and movement can be ne-
glected, and the behaviors of droplets/particles can be
examined by analyzing a single droplet/particle with
the average size, dp.
For understanding the variations of the states of
droplet/particle and gas stream along tower height,
one-dimensional model is accurate enough. In addi-
tion, the following assumptions are made in the model
establishment: (a) The droplet/particle is spherical. (b)
The droplet/particle moves downward at its terminal
velocity, ut, throughout whole the process. In fact, in
the case of using rotary cup with perforated wall for
spray, the initial velocity of droplet in the vertical di-
rection is near its terminal one and thus the time of
unstable movement with varying velocity is negligible.
(c) Since no mass transfer happens during the process
of melt prilling and the shape of droplet is quickly fixed
due to solid shell formed on the surface, the drop-
let/particle is assumed to have a constant density, pp.
2.2.2 Movement equation of droplet/particle
The relationships predicting terminal velocity, ut,
in various flow regimes are well known[2,3]. For the
actual velocity of droplet/particle, up, moving in
counter-current gas flow, the velocity of the latter, uo,
should be subtracted from u,:
u P = u t -uo (1)
Or, in differential form,
dH
~ = ut - uo
dr
2.2.3 Thermal behavior of droplet/particle
In the tower, the droplet/particle undergoes three
periods of different natures: (a) Droplet cooling before
solid at the surface appears; (b) After solidification
starts at surface, the interface of melt core shrinks to-
wards the center until completely solidified, during
which the resistance of solid shell to heat transfer
cannot be neglected; and (c) Further cooling after
complete solidification.
In period (a), the variation of droplet temperature,
T,,, is determined by the heat balance:
n
;dzppcpLdTp =-h&i(TP -
That is
(3)
In all the calculations below, the film heat trans-
fer coefficient h is predicted by Ranz-Marshall's equa-
tion[41:
Nu = 2.0 + 0.6Re"'Pr"3 (4)
When T,, drops down to the melting point, T,, solidi-
fication at surface begins and period (a) is finished.
This can be taken as a criterion to determine the time
the first period lasted.
In period (b), only solidification occurs, and so
the essential problem to be considered is the heat
transfer between the unsolidified core and the gas flow.
425
Since movement of melt molecules at higher tem-
perature is relatively violent, the core can be assumed
to be uniform in temperature, which is kept at T,". The
equation for heat transfer can be written as
q 2 =U&i(Tn - T g ) (5)
where the overall heat transfer coefficient, U , is de-
termined by
The variation of the solid layer thickness, 6, with
time is determined according to the heat removed
from the particle in this period, Q?, which meets the
relationship below:
-n[ d ; - ( d p -26,']ps =- Q2 (7)
6 -AH,
and Q2 can be found out by integrating Eq.(5), as
It will be more convenient for simulation to ex-
press 6 i n a time-derivative. To do this, Eq.(7) is dif-
ferentiated with respect to time to yield
The time of period (b) ending, t?, can be deter-
mined from the following equation:
In the case where the difference between the
densities of solid and melt is negligible, the following
can be simply used for the final thickness, &:
In period (c) only heat transfer between solid
particle and gas stream happens. Certainly, there exists
temperature gradient inside the particle. Also, the
process is unstable because of the movement yielding
variation of gas temperature around the particle.
However, this period is of less importance in com-
parison with the others, especially period (b). So, as
the first order approximation, a pseudo steady state is
assumed and the heat transfer is considered to occur
between the gas stream and the particle core, and the
overall transfer coefficient is assumed to have the
same value as that at the end of period (b), U12.Thus,
the heat transfer rate in this period, ql, is written as
For systems where Eq.( 1 1) holds, Ut2can be expressed
by
And the variation in the particle temperature is deter-
mined by the differential heat balance below:
Chin. J. Ch. E. 15(3) 424 (2007)
426 Chin. J. Ch. E. (Vol. 15, No.3)

Tgf =

2.2.4 Temperuture variation of gas stream

The variation of gas temperature is resulted from
heat transfer between all the droplets/particles and gas
stream[5), and so it should be determined from heat
balance in the whole system, as shown in Fig.2. The
heat loss through wall can be neglected without sig-
nificant error, if the following facts are taken into ac-
count: (a) The prilling tower is of large diameter gen-
erally in order to avoid caking on the wall and bond-
ing of particles with each other, (b) The gas tempera-
ture is relatively low.
mBCPg
(20)
Thus, the conditions expressed by Eq.( 18) become the
initial conditions:
Tp(0) = Tpo 9 T, (0) = T,f (21)
2.3 Major results of simulation with the model
With the model developed above, simulation is
made for urea prilling as a typical case. The operating
conditions and properties data used are listed in Table
1, where the heat conductivity of chalk, As =
',
2.651 X 1OSkW.m-'.K- is used to substitute approxi-
mately that of urea because the value for urea was not
found. All the other data listed in the table are accurate,
and most of them are taken from industrial operations.

Table 1 Operating conditions and property data involved

in urea prilling
Initial temperature of melt Tpo= 140°C
Temperature of product Tpf<70"C
Temperature of cooling air Tgo=25"C
Air flow rate 9000m3.[t (NH2)2COI-'
Air velocity uo= 1 .Om.s-'
Average size of product dp= I .2mm
Figure 2 Heat balance for granulation in a tower
Melting heat of urea AHm=224kJ.kg I

Specific heat of melt c,,,,= 1.334kJ.kg-l.K I

The equations for gas temperature variations in
various periods are derived as follows:
In period (a):
Specific heat of solid urea cps=1.334kJ.kg l.K-'
Density of urea particles pp= 1 3 4 0 k g K '
Density of air pg= 1 . 0 5 6 k g 6 '
Heat conductivity of air Lg=2.651 X IO-'kW.m-'.K I

Heat conductivity of solid L,=2.651 X 10 'kW.m '.K '

In period (b): Specific heat of air cpg= 1.082W.kg-l.K '
Viscosity of air ~ , = 2 . 0x I O - ~ P ~ S

The computed results are given in Figs3 and 4.

In period (c): Curves 1 and 2 in Fig.3 reflect the variations in tem-
peratures of unsolidified core of the droplet/particle
and air along the tower height, while the curve in

The conditions determining the solutions of the I60 I 1

above model equations are

Tp(0) = Tp" 9 T, 0 3 ) = T,,
To convert the problem into an initial value one, the
(18)
L- xo
60
solidification
period i- \
................................
'

overall heat balance below can be used:

40 ~

mglcl,f (Tgj - T,U) = 20 : 1 1 I I '_(I

0 I0 20 30 40 50 60
[
mp c p L P('o -Tm j +(-urn
+ c p s (G - ~ p f j] (19) H, in
Figure 3 ,Resultscomputed for tower prilling of urea
where the subscripts 0 and f denote initial and final (for urea with dp= 1.2mm)
states, respectively. Solving Eq. (19) for Tgfyields I-temperature of particle core; 2-temperature of air

June, 2007
 An Innovated Tower-fluidized Bed Prilling Process
0 It is important that strength of the solid shell
0 10 20 30 40 50 60
H, m
Figure 4 Relative thickness of solid shell versus particle
falling distance
(for urea with dp= 1.2mm)
Fig.4 exhibits the behavior of the relative thickness of
solid shell, &, which is defined as
6, = 26/d, (22)
and hr= I .O suggests the melt has completely solidified.
The following are important from these figures:
(a) The height is mainly (over 90%) consumed in so-
lidification of melt; and (b) The effective height
needed for product average-sized 1 . 2 m should be no
less than 60m. In addition, totally over 10m auxiliary
height is also needed for installing the rotary spray
cup, product discharging etc. As well known the prill-
ing tower for urea product consisting of particles sized
1.0 to 1.5mm is usually about 75m high. Therefore,
the simulated results are correct and reasonable.
It is of significance to understand the influence of
particle size of the product on tower height needed.
Fig3 gives the results of comparative calculation, for
which all the other data are the same'as those yielding
Figs.3 and 4. As can be seen, the tower height in-
creases exponentially as particle size increases. It can
thus be concluded that to produce product with larger
size, say, >,2mm, by the traditional prilling process is
almost impossible in practice because of unacceptable
height of the tower.
particle size of product, m m
Figure 5 Relationship of effective height of tower needed
versus urea particle size
3 COMBINED TOWER-FLUIDIZED BED PRI-
I,LING
3.1 Basic consideration of the innovation
As concluded in the last section, the major height
of tower is used for solidification of melt in the tradi-
tional prilling process. On the other hand, solidifica-
tion starts generally at the surface and then extends
towards center of droplet. It seems possible to divert
the further solidification and cooling of particles from
427
empty tower into a fluidized bed when a solid shell
strong enough has formed. If so, a combined
tower-fluidized bed can be used for prilling of melt, and
even solution. Because of very long residence time,
great hold-up of particles per unit volume, and very
intensive heat transfer in the fluidized bed in compari-
son with in an empty tower, the total height of the pro-
posed equipment can be shortened significantly.
must be high enough to avoid breakup of the drop-
letdparticles when they fall into the fluidized bed. The
strength calls for a certain thickness, and the necessary
value for it must be related to particle size. However,
to determine the value experimentally for the relative
thickness, 6,, needed is difficult, at least, at the present.
Empirically the value of 6,=0.2-0.3 may be feasible.
If so, it can be seen from Fig.4 that the effective
height of tower can be shortened by 50% to 60%.
3.2 Structure of the innovated equipment
The combined tower-fluidized bed granulator
(TFBG) newly designed (Wu et a1.[6])is sketched in
Fig.6. Essentially it is the combination of an empty
tower with a co-axial tapered fluidized bed.
air o u t
r s
melt
-
Figure 6 A brief view of TFBG
I-tower; 2-taper Huidized bed;
3 i o t a r y cup with perforated wall
The granulator employs a rotary cup far spraying
melt, which is of the reversed, truncated cone-shape
with perforated side-wall and closed bottom. On ro-
tating, the melt continuously fed in the cup is pushed
by the centrifugal force to flow through the small ori-
fices on the side-wall to fomi many pillars with
requested uniform diameters, and then the pillars are
broken into droplets due to hydrodynamic instability.
The height of the tower is decided according to
Chin. J. Ch. E. 15(3)424 (2007j
428 Chin. J. Ch. E. (Vol. 15, No.3)

Table 2 Essential operating condition for prilling of sebacic acid in combined tower-fluidized bed
Melt tempera- Sizes of particles Temperature of Temperature of Air flow rate, Temperature of Amount of en-
ture, OC in product, mm product, “C cooling air, “C m’.t- I exhaust air, -C trained powder, %
dexigncd - I50 0.34.5 40 -30 3.55 62 2.0
measured - I50 0.3-0.5 -30 - 30 -3 70-75 (0. I

the principles described above as the effective one, which diverts the further solidification and cooling of
and the principles for determining the auxiliary upper droplets/particles after a solid shell strong enough hav-
height ;ire essentially the same as for traditional tower ing formed from the empty tower into a co-axial fluid-
granulator. The co-axial tapered fluidized bed has a ized bed, resulting in greatly shortened height of the
perforated plate bottom as the gas distributor, for equipment and many advantages, such as lower air flow
which the design method is conventional and can be rate, energy saving in melt transportation, convenience
found in monographs on fluidization and textbooks of operation and product discharging etc. The process is
chemical reaction engineering[3,7]. It is clear that, by applied industrially for sebacic acid with a production
the combination of tower with a co-axial taper fluidized of 2000t per year, and exhibits very good performance.
bed, the total height of the new equipment is much
shortened, resulting in much lower cost, energy saving NOMENCLATURE
in transporting melt, and convenience in operation. Be- CP specific heat, kJ.kg-’.K I
side the short height, it has the following advantages: d diameter, m
g gravity acceleration (9.8 I m.s ’1
( I ) Much lower flow rate of cooling air. This is
H height droplet/particle dropped down, m
because temperature of exiting air can be higher, and -AH, melting heat, kJ.kg- I
because the diameter of fluidized bed is much smaller h film heat transfer coefficient, kW.m ’.K ’
than the: tower, so lower rate of cooling air can main- m mass flow rate, k g Y I
tain the particles in fluidization. Nu Nusselt number
(2) Fewer entrained fine powder in exhaust air re- Pr Pranlt number
sulted from lower superficial velocity of air in the tower. Q heat transferred from droplet/particlc, kJ
( 3 ) Fully cooling of product, resulting in more 4 heat transfer rate, kJ.s-’
convenience in package and little bonding during stor- Re Reynolds number
age. T temperature, K or “C
t time, s
(4) More convenient discharge of product be- U overall heat transfer coefficient, kW.m ’.K I
cause of good fluidity of fluidized materials. U velocity, m.s
s thickness of solid shell, mm
3.3 Industrial application and performance 1 heat conductivity, kW.m ‘.K I
The innovated prilling process described has P viscosity, P a s
been applied industrially for sebacic acid with a pro- P density,
- kg.m
- ’
duction of 2000t per year. The effective height of the Subscripts
f final
combined tower-fluidized bed granulator determined g gas or air
according to the method above is 9m, and the total L liquid phase
height 13m. In comparison, a tower granulator in op- m melt, melting
eration in another company for the same product with P droplet/particle
s solid
the same size range is of a total height over 40m. In t terminal
practice, the equipment is operated very stably. The 0 superficial, initial
operation data designed and measured are listed in I ,2,3 period number
Table 2 The measured data of lower product tem-
perature, higher temperature of exhaust air and much REFERENCES
smaller amount of entrained powder show that the Wu, Y., Zheng. C., Fu, J.. “Evaporation of aqueous solu-
performance of the equipment newly designed is even tion droplets Gf some inorganic salts”, J . Chrin. I d . Eng.
a little better than predicted, suggesting well reliability (China),34(4), 342-352( 1983). (in Chinese)
Perryry, R.H., Green, D.. Chemical Engineer’s Handbook,
of the design method used. 6th ed., McGraw-Hill, New York (1984).
Kunii, D., Levenspiel, O., Fluidization Engineering, John
4 CONCLUSIONS Wiley & Sons, Inc., New York (1969).
The results simulated for the traditional tower Ranz, W.E., Marshall, W.R., “Evaporation from droplets”,
Chem. Eng. Prog., 48(3), 141, 173 (lY52).
prilling of melt show that significantly high tower is Wu, Y., Fu, J., “A mathematical model for countercurrent
necessary to ensure sprayed droplets being completely spray drying of highly concentrated and superheated
solidified and fully cooled, and that to produce prod- MAP solution”, In: DRYING ’86, Hemisphere, New
uct of larger size, say 3 2 m m , is almost impossible York, 305-3 13(1986).
because of unacceptable height of tower. On the basis Wu, Y., Zhou, Y., Xu, J., Bao, C., Shu, A , , “Spray
tower-fluidized bed granulator” China Pat., ZL
of an analysis for the traditional prilling process with 03241473.0 (2003). (in Chinese)
the shrinking unsolidified core model, an innovated Wu, Y., Chemical Reaction Engineering, Press of Dalian
tower-fluidized bed prilling process is developed, Institute of Ocean Shipping, Dalian (1992). (in Chinese)

.June,2007