BahmanTechnology

ZareNezhad
Journal of the University of Chemical
and Metallurgy, 42, 1, 2007, 45-50

EFFECTS OF PROCESS VARIABLES ON THE CAKING TENDENCY OF PRILLED UREA

IN THE WAREHOUSE OF UREA PRODUCTION PLANTS
Bahman ZareNezhad

Department of Chemical Engineering,

Shiraz University of Technology,
Modarres Street 71555-313, Shiraz, Iran
E-mail: zarenezhad@yahoo.com

Received 18 July 2006

Accepted 10 January2007

ABSTRACT
The caking of urea fertilizer is investigated in a warehouse of a large scale petrochemical plant. It is found that the
caking tendency below the recommended value of 5 %mass can be achieved if the prill coefficient of variation and the
temperature difference between the product and warehouse ambient are kept at 0.07 and 4 oC, respectively. A first order
kinetic model is also suggested for description of the urea caking process. The caking tendency is well predicted by the
proposed correlation with the overall deviation of 3 %mass.
Keywords: caking, urea, prilling, warehouse, petrochemical plant.

INTRODUCTION
Urea is in many ways the most convenient form
for fixed nitrogen. It has the highest nitrogen content
available in a solid fertilizer (46 %). It is easy to produce as prills and easily transported in bulk or bags
with no explosive hazard. The change from bagged to
bulk handling and the increase in production and handling capacities require that the physical quality of fertilizers should meet higher demands in order to assure
customer acceptance. To be assured of good physical
quality of prilled urea, the product should have a low
caking tendency during storage [1].
In the usual operation of a petrochemical plant,
crystalline urea prills with the size range of 1-4 mm are
dried to a low moisture content (less than about
0.2 %mass), piled up in the warehouse and then packaged
in multiply bags. It is always found that the mass of

prills tends to set up into a single solid cake when stored

in the ordinary manner. In some instances, a hard solid
cake is formed in a matter of hours when the bagged
urea is subjected to very slight pressure, on the order of
0.5 bar [2]. Thus the caking tendency makes the urea
fertilizer quite unsuitable for technical applications and
reduces the profitability of the process drastically.
The main cause of caking is the growth of crystal
bonds on the contact points of the prills [3]. This process can only take place in the presence of liquid phase,
in this case saturated aqueous urea solution. The caking
tendency of urea prills is adversely influenced by:
The amount of liquid phase
- water (moisture). The caking tendency increases
with the water content. Difference in the prill water
content causes water to migrate through the pile. This
too increases the caking tendency.

45

Journal of the University of Chemical Technology and Metallurgy, 42, 1, 2007

- free ammonia. Excessive amounts of free ammonia increase the amount of the liquid phase in the
prills. Ammonia loss in turn leads to the crystallization
and essentially to the agglomeration or lumping of prills.
- prill temperature. The caking tendency decreases
with lower storage temperatures (less liquid phase).
What should be avoided are high prill temperatures at
the bottom of the prilling tower (higher than 60oC).
Migration of water
- migration of water between prills. This causes
crystallization and the formation of crystal bonds resulting in a higher caking tendency. The migration of
the water from one prill to another is brought about by
differences in the partial vapour pressure of water in
the prills resulting from the differences in temperature
and the water content. A more uniform prill size distribution is desirable.
- water absorption from the ambient air and loss
back. It is worth noting that 1 m3 of urea may contain 1
to 3 kg of water whereas 1 m3 of air at 40oC contains
only 50 g of water, and that water loss is at least harmful with respect to the caking as is the absorption. Thus
the temperature difference between the warehouse ambient and the product could be an important issue.
High proportion of unround and small prills,
dust and fines
The caking tendency increases with the number
and size of contact points between the prills. Similar to
the case of the migration of water between prills, a more
uniform prill size distribution may lead to a less caking
tendency.
Poor mechanical strength
This is particularly important when the product
needs to be stored later, after being subjected to a variety of handling operations. Poor mechanical strength
leads to disintegration and dust formation and, thus, as
pointed earlier, to increased caking. The problem can
be alleviated by proper prilling and seeding so as to
prevent subcooling of the prills [4].
The caking tendency of urea prills can be combated by adding specific amount of anticaking agents
before (e.g. formaldehyde compounds) or after (e.g.
polypropylene glycol or mineral oils) the prilling pro-

46

cess [3,5,6]. Unfortunately the use of anticaking agents

may render urea unsuitable for technical applications.
Also the recycling part of the treated urea (recovered
dust, fines and non-standard prills) lead to the formation of foam in the evaporators and deactivation of the
catalyst in the urea synthesis unit [7]. Therefore it is
preferred to reduce the rate of urea caking by controlling the process variables rather than using the anticaking
agents. In this work we are trying to find the ranges of
necessary operating conditions to keep the caking tendency in the urea warehouse below the recommended
standard value.
EXPERIMENTAL
The experiments are carried out in an urea warehouse of a 1500 MTPD production plant. The ammonia
content of urea melt is kept at the lowest possible value
of 50 ppm while the maximum dedusting and fine removal is applied. The average prill moisture content was
kept at 0.2 %mass and the crushing strength and impact
resistance of urea prills have been kept at the maximum
possible values of 25 bar and 82 %, respectively [1]. Considering all these and previously discussed influences, it
can be concluded that the rate of caking in the warehouse is mainly affected by prill size distribution expressed as coefficient of variation (CV) and temperature
difference between urea bulk and warehouse ambient (T).
Effects of these two parameters on the urea caking tendency are rarely discussed in the literature.
The warehouse stack test is used to determine
the caking tendency of urea prills because this stack
test substantially approaches actual field storage conditions. In these tests, the standard 50 kg bags are employed as in routine commercial production. In a typical experiment, 50 kg of prilled urea entering the warehouse with definite CV and T is quickly transferred to
a test bag. These test bags are placed horizontally on a
special pallet and are arranged on the pallet in layers of
four bags each, one bag along each side of the pallet.
Six layers of bags are placed on the pallet. The pallet of
test material is then placed in warehouse storage, and
weight is added by stacking two full pallets of commercially bagged product (6 x 4 = 24 bags/pallet) on the
top of the pallet of the tested material (i.e. test pallet).
The tested pallet is stored in this condition for a given

Bahman ZareNezhad

period of time (from 1 to 10 weeks). At the end of the

storage period, the bags are carefully removed from the
pallet and each is dropped once from a height of about
1 meter to separate any loosely held agglomeration of
particles that may have formed during storage. Each bag
is opened at one end, and its content is poured onto a 2
mesh U.S. Standard screen which retains any lumps of
caked material that may have formed. The lumps are
retrieved and weighed. The weight, in kg, of lumps retained on the 2 mesh US Standard screen constitutes a
quantitative measure of the ureas caking tendency and,
hence of ureas free-flowing characteristics. According
to the EFMA Standard [1], urea is substantially freeflowing if, on being subjected to the warehouse stack
test, about 5 kg or less lumps (particles which fail to
pass a 2 mesh U.S. Standard screen) are produced per
100 kg of urea tested after 6 weeks storage period.
A series of experiments are carried out at different operating covering the ranges of the process variables shown in Table 1.
Modelling of the caking process
In this work a first order kinetic model is used
for description of the caking process. It is assumed that
the rate of change of non-caked prills into lumps is first
order with respect to the mass of non-caked prills. This
can be written as:

dm
= km ,
dt

(1)

where
m = mass of non-caked prills at time t, kg
t = storage time, weeks
k = caking rate constant, weeks-1.

1 m
ln
=k,
t mo

(2)

where mo is the initial mass of non-caked urea prills.

Using the experimental data, the values of k can
be determined by correlating the left side of Eq. (2) in
terms of CV and T by the following equation:

k = (a + bT x ) + (c + dT y )CV z

(3)

where:
a, b, c, d, x, y, z are parameter in Eq. (3)
CV = /Lm - coefficient of variation [-]
- variance, mm.
The coefficients and exponents in Eq. (3) are estimated by multiple non-linear regression analysis of
all experimental data and the results are tabulated in
Table 2. The values of the coefficient of determination,
R2, show that a good correlation is obtained. The standard error in parameter estimates is determined at
95 % confidence limits.
Using Eqs. (2, 3) and the parameters, the caking
tendency, described as the weight percent of conversion
of initial mass of prilled urea into caked lumps, can be
determined as follows:
caking tendency = [1- exp ( kt)] x100

(4)

Predicted caking tendencies by using Eq. (4) are

compared with experimental data and discussed in the
following section.
RESULTS AND DISCUSSION

Eq. (1) can be integrated and written in the following form:

The measured caking tendencies at a fixed T of

8oC and different prill coefficients of variation are displayed in Fig 1. As shown in this
figure, the caking tendency inTable 1. Ranges of run conditions for investigating the caking phenomena.
creases from 9.5 %mass to 47.9 %mass
after 6 weeks as the CV is increased
Run number
T (oC)
CV
R1 - R4
0
0.07, 0.14, 0.21, 0.28 from 0.07 to 0.28. It is practically
R5 - R8
4
0.07, 0.14, 0.21, 0.28 impossible to keep the caking tenR9 - R12
8
0.07, 0.14, 0.21, 0.28 dency below standard recomR13 - R16
12
0.07, 0.14, 0.21, 0.28 mended value of 5 %mass for the

47

Journal of the University of Chemical Technology and Metallurgy, 42, 1, 2007

%
$

70%

8C
?TT == 8?C

C V = 0 .0 7

C V = 0 .1 4

60%

+3 %

#

C V = 0 .2 8

Predicted cak ing tenden cy, wt%

C ak in g t en den cy

C V = 0 .2 1

M odel

"
!



50%

-3%
40%

30%

20%

10%




"

&

t , wee k s



0%
0%

10%

20%

30 %

40%

50%

60 %

70%

Mea sured ca king ten dency, wt%

Fig. 1. Effect of prill coefficient of variation on the urea caking

tendency in the plant warehouse at T of 8oC.

Fig. 3. Comparison of predicted and measured urea caking

tendency.

60%

8C
? T=T 4= ?C

CV=0.07
50%

CV=0.14

Caking tendency

CV=0.21
40%

CV=0.28
Model

30%

20%

10%

0%
0

t, wee ks

10

12

Fig. 2. Effect of prill coefficient of variation on the urea caking

tendency in the plant warehouse at T of 4oC.

ranges of process variables shown in this figure. According to Fig 1, at higher prill CV, the caking tendency is more severe. An increase in prill CV from 0.07
to 0.14 leads to a 11 %mass increase in the caking tendency after 10 weeks, while for a further CV increase
from 0.14 to 0.21 %, a 19 %mass increase in caking tendency is observed at the end of the same time period,
which is quite significant. This occurs because increasing values for the coefficient of variation result in an
increasing abundance of both small and large diameters
(relative to mean diameter) in the population of urea
prills. This amplifies the rate of moisture migration from
large to small prills and speeds up the rate of caking.
Also the small prills posses more external surface area
and this further increases the rate of agglomeration and
caking. Table 3 shows how the < 0.6 and > 3.0 mm size
fractions and the < 1.0 mm and > 2.6 mm size fractions decrease as the spread of prills with 1.8 mm mean
diameter decreases from 24 % to 10 %. As the coefficient of variation is decreased, the coarse and fine frac-

Table 2. Calculated values of the parameters for Eq. (3).

a
6.98810-3
0.0002

48

Constants and exponents for k in Eq. (3)

b
c
d
x
5.66410-4
1.055
0.08764
1.1597
0.00007
0.12
0.003
0.17
2
R -value = 0.989

y
1.3856
0.11

z
2.6020
0.34

Bahman ZareNezhad

Table 3. Reduction of coarse and fine fractions with decreasing the coefficient of
variation.

Coefficient of
variation

Standard deviation
(mm )

Fraction (%mass )
< 0.6 mm and
> 3.0 mm

Fraction (%mass )
< 1.0 mm and
> 2.6 mm

0.24
0.22
0.20
0.18
0.16
0.13
0.10

0.432
0.396
0.360
0.329
0.288
0.234
0.180

0.27
0.13
0.05
none
none
none
none

3.22
2.17
1.32
0.75
0.27
0.03
none

tions are reduced such that a normal Gaussian distribution is approached and thus caking tendency diminishes as explained before.
In Fig. 2 the caking tendencies at a fixed T of 4oC
and different prill coefficient of variations are also compared together. As shown the caking tendency is decreased
from 11 to 4.8 %mass as the prill coefficient of variation is
decreased from 0.14 to 0.07 after 6 weeks storage. According to this figure, the caking tendency after 6 weeks can be
kept below the recommended value of 5 %mass, if the T
and CV are controlled at 4oC and 0.07, respectively.
The measured and predicted caking tendencies
at T of 4 and 8oC and different prill coefficient of
variation are compared in Figs. 1 and 2, respectively.
Eq. (4) is used to predict the caking tendency of urea
prills in the warehouse. As shown in these figures the
predicted and measured values are in good agreement
at different operating conditions. Predicted results well
reproduce the experimental measurements.
The predicted caking tendencies according to Eq.
(4) are also compared with experimental data for all
runs in Fig. 3 at a relative deviation of 3 %. This
figure clearly indicates that the caking process is well
described by the proposed model.
CONCLUSIONS
Controlling the caking tendency of urea prills
during the storage in the warehouse is of prime importance to keep the product quality at the highest possible
level. It is shown that even for an ammonia concentra-

tion of 50 ppm, average prill moisture content of 2 %mass,

crushing strength of 25 bar and impact resistance of
82 % and maximum dedusting and fine removal, the
urea caking tendency could be severe in the warehouse
after 6 weeks. It is found that in addition to above process variables it is necessary to keep the T and CV at
4oC and 0.07, respectively, in order to maintain the caking tendency below the maximum permitted value of
5 %mass. Thus controlling the temperature difference between prilled product and the warehouse ambient and
also the granulometery of the urea prills are essential to
minimize the caking tendency of the urea product. The
suggested temperature difference can be achieved by
warehouse insulation and space heating, while the proposed prill coefficient of variation can be obtained by
installation of a vibrating melt spray nozzle at the top
of prilling tower. A first order kinetic model is also
suggested for description of urea caking process. The
effects of temperature difference ( T) and prill coefficient of variation (CV) on the product caking tendency
are well predicted by the proposed correlation.
REFERENCES
1. European Fertilizer Manufacturers Association, Production of Urea and Urea Ammonium Nitrate,
Booklet No 5, 1995.
2. D.Gerald, North American Fertilizer Capacity Data,
Tennessee Valley Authority, 2003.
3. M.E. Pozin, Fertilizer Manufacture, First Edition,
Mir Publishers, 1986.

49

Journal of the University of Chemical Technology and Metallurgy, 42, 1, 2007

4. B.ZareNezhad, Static and Dynamic Strength of Prilled

Urea in Petrochemical Plants, Research Report No
230-23-7, NIOC, Iran, 2004.
5. D.W. Leyshon, Granulation Plant Screening, J.of
Phosphorous and Potassium, 221, 1999.

50

6. D.W. Leyshon, Crushing the Oversize, J.of Phosphorous and potassium, 227, 2000.
7. B.ZareNezhad, Controlling the Rate of Caking in
the Urea Warehouse, Research Report, National
Petrochemical Company, Iran, 2003.