Capítulo 7.

INTRODUCCIÓN AL BALANCE DE POBLACIÓN

7.1. Introducción. Analogía con reactores químicos

En la Figura 7.1 se presenta un esquema de un reactor químico donde ingresa y

egresa una mezcla multicomponente. Si se realizarán experimentos sería posible
medir, entre otras cosas, las composiciones de los reactivos y productos a la entrada y
salida del reactor.

REACTOR
QUÍMICO
F0, mol/h Fs, mol/h
C0i, mol/m3 Csi, mol/m3
v0, m3/h vs, m3/h
Figura 7.1. Reactor químico continuo con alimentación multicomponente.

Supongamos que se lleva a cabo la siguiente reacción:

A +B →C+D (7.1)
y que se conoce las composiciones de los reactivos y productos tal como se muestra
en la Tabla 7.1. La misma información tabular la podemos representar gráficamente tal
como se muestra en la Figura 7.2.

Tabla 7.1. Composiciones a la entrada y salida del reactor.

Entrada Salida
C0i %molar Csi % molar
A 80 A 60
B 20 B 0
C 0 C 20
D 0 D 20

6.1
 Rector Químico Continuo
100

Concentraciones de entrada, %
90 Concentraciones de entrada
80 Concentraciones de salida
70
60

molar
50
40
30
20
10
0
0 1
A 2
B 3
C 4
D 5
Componentes de la mezcla

Figura 7.2. Representación gráfica de la Tabla 7.1

Para poder predecir las concentraciones de salida, conociendo las de entrada,

es necesario establecer:
ƒ Tipo de reacción química, estequiometría y cinética.
ƒ Tipo de reactor/ tipo de flujo (TAC; RT; BATCH, etc.)
ƒ Datos geométricos de la unidad de reacción, datos variables operativas
de entrada y propiedades de la mezcla.
Con esta información es posible plantear los balances molares por componente, el
balance de energía y cantidad de movimiento, y establecer así todas las variables a la
salida del reactor.

ni* Unidad de ni*

Aumento de
Tamaño
dpi (GRANULADORES) dpi
PSD de entrada PSD de salida

ni* Unidad de ni*

Disminución
de Tamaño
(MOLINOS)
dpi dpi
PSD de entrada PSD de salida
Figura 7.3. Granuladores y Molinos.

6.2
 La Figura 7.3 muestra la operación de granuladores y molinos. Básicamente los
granuladores son equipos en los cuales se aumenta el tamaño de partículas de la
distribución entrante a la unidad (ver Tabla 7.2). Hay diferentes tipos de granuladores
los cuales será presentados en el próximo capítulo. Un ejemplo de granulador sería el
equipo descripto para recubrir los rocklets. Por su parte, como es conocido por todos,
los molinos (en el capítulo 9 serán presentados algunos ejemplos) reducen el tamaño
de las partículas que ingresan al equipo (ver Tabla 7.3). Un ejemplo que se puede citar
es la molienda de trigo para la obtención de harina. La pregunta que surge es ¿qué
ecuación constitutiva nos permite establecer la calidad del producto de estos
equipos, en función de las variables operativas y de diseño?. Así como los
balances molares por componente, conociendo propiedades del sistema, permiten
establecer la distribución de productos a la salida del reactor, el BALANCE DE
POBLACIÓN es la herramienta que nos permite calcular distribuciones de tamaño de
partículas (en inglés, Particle Size Distributions- PSDs) a la salida de un equipo dado.

Tabla 7.2. PSDs entrada/ salida granuladores.

Entrada Salida Granulador

70
dpi ni* x100/NT dpi ni* x100/NT 60
Entrada
Salida

0 0 0 0 50

40
ni* %

1 20 1 20 30

20
2 60 2 60
10

3 20 3 20 0
0 1 2 3 4 5
dpi
4 0 4 0

Tabla 7.3. PSDs entrada/ salida molinos.

Entrada Salida Molino

dpi ni* x100/NT dpi ni* x100/NT 70

Entrada
60
Salida
0 0 0 0 50

40
1 10 1 25
ni* %

30

2 20 2 50 20

10
3 60 3 25 0
0 1 2 3 4 5
4 10 4 0 dpi

6.3
 Aire, YS , TaS

Semillas, X0,
BM dp0 , Ts0
+ Granulador Producto, XS,
BE dpS , TsS
+
BCM Aire, Y0 , Ta0
m& sol
Información para resolver el PBE

Masa Masa
PBE
dp dp
PSD de entrada PSD de salida
Figura 7.4. Balances necesarios para definir completamente un granulador.

En síntesis si quisiéramos modelar un granulador, por ejemplo uno de lecho

fluidizado como el que se muestra en la Figura 7.4, se requeriría plantear los balances
de masa, energía, cantidad de movimiento y por último el balance de población. Así
como los balances molares requieren de información cinética, patrón de flujo, datos
geométricos, etc. para describir las composiciones a la salida de una unidad de
reacción, el balance de población (en inglés, Population Balance Equation – PBE)
requiere de información cinética para describir la performance de unidades de
aumento o reducción de tamaño.

7.2. Deducción del Balance de Población

Como texto para la deducción del balance de población utilizaremos la siguiente

publicación que se adjunta al apunte: “Teaching Population Balances for ChE
Students: Application to granulation processes”, Chemical Engineering Education, Vol.
41, No. 3, 209-217, Summer 2007.

6.4
 ChE curriculum

TEACHING POPULATION BALANCES

FOR Che STUDENTS:
Application to Granulation Processes

Verónica Bucalá and Juliana Piña

I
PLAPIQUI (UNS-CONICET) • (8000) Bahía Blanca, ARGENTINA
n many chemical engineering degree programs world- During instruction of the PBE formulation, we found that
wide, particle and powder technology is not afforded the analogies with chemical reaction principles (well known
same attention in the curricula as processes and technolo- by the alumni) helped the students to understand this “new”
gies incorporating liquids and gases.[1] Consequently, it is not constitutive equation.
surprising that plants handling solids perform less optimally In this work we are particularly focused on the approach
than those processing only liquids and gases.[2] There is there- to teaching PBE formulation in the context of granulation
fore need for new courses in particle science and technology processes.
in the established chemical engineering curricula.
Chemical engineers are used to handle mass, energy, and Verónica Bucalá is a professor of chemical engineering at Universidad
momentum balances in modeling and designing equipment
Nacional del Sur (Bahía Blanca, Argentina). She received her B.S. and
Ph.D. degrees in chemical engineering from the same university. She
for the chemical industry. Often, however, they are not as held a postdoctoral research fellow position at Massachusetts Institute of
familiar with the population balance equation (PBE) to de- Technology, Cambridge, Mass. Her research interests are in the area of
chemical reaction engineering and simulation of solids processes.
scribe important attributes of particulate streams (e.g., particle
size distributions). While the PBE is generally agreed to be Juliana Piña is an assistant researcher in the Chemical Engineering De-
partment at the Universidad Nacional del Sur (Bahía Blanca, Argentina).
difficult to solve, many students find even formulating it to She received her B.S. and Ph.D. degrees in chemical engineering from
be very complicated. the same university. She held a postdoctoral research fellow position at
University of Western Ontario, London, Canada. Her research interests
An optional course for education in particle technology has include modeling and simulation of catalytic chemical reactors and
been introduced in the last year of the chemical engineering granulation processses.

program at the Universidad Nacional del Sur, Argentina. © Copyright ChE Division of ASEE 2007

Vol. 41, No. 3, Summer 2007 209

The need for PBE modeling of the one particle can produce more than two fragments due to colli-
granulation process sions with other particles and/or with the granulator walls.[4, 5]
Figure 1 shows a schematic diagram of a granulation unit. The PBE has to capture all the granulation rate processes
The seeds—i.e., particles whose diameters are smaller than to predict the granulometry of the final product. Before dis-
those of the granular product—may be fed to the system
either initially (batch processes) or continuously (continuous
granulators). Inside the granulation “box,” these particles
undergo an effective growth. A liquid solution (e.g., a concen- Seeds
trated solution or melt) is sprayed into the granulation unit.
Granulation
Depending on the type of granulator, air can be fed into the Unit
unit to fluidize the bed.[3]
Inside the granulator the particles are involved in many Granular
mechanisms of size enlargement or reduction. Depending Product
Liquid
on the process, they can occur alone or simultaneously. This m sol
is the reason why the seeds’ particle size distribution (PSD)
evolves to a different exit or final PSD (see Figure 1). The PBE
is the tool that allows, for example, predicting the granules’ Seeds PSD Product PSD
size distribution. Particle
Masa Particle
Masa
Property Property
PBE
Granulation rate processes
The PBE formulation requires understanding of the phe- dp dp

nomena the particles are subjected to during the granulation Figure 1. A schematic diagram of a granulation unit and
evolution. Figure 2 schematically shows the nucleation, its relationship with the PBE.
layering, coalescence, attrition, and breakage processes. All
these processes are well explained by, among others, Litster,
et al.,[3] and Rhodes.[4] Nucleation
Nucleation is the formation of new seeds from liquid or fine
powder feed. New granules can be formed when the liquid Layering +
drops (produced in the spray) solidify before they reach the
surface of the seeds. This mechanism is discrete. This adjec-
tive means that the new nuclei just appear, i.e., they are not Coalescence +
produced gradually. Layering increases the granule size by
coating the particle surface with drops produced in the spray
zone. The growth is differential, i.e., the particle size augments Attrition +
progressively. Particles may also undergo coalescence, i.e.,
two particles agglomerate to give a bigger one, this being a Breakage +
discrete phenomenon. By means of the attrition mechanism
the particles suffer surface wearing, a differential granule size
reduction. The granule breakage is also a discrete process, where Figure 2. Granulation rate processes.

TABLE 1
Sieve Analysis of a Single Sample, Using Different Sets of Sieves
Fine Grid Coarse Grid 1 Coarse Grid 2
Size Range, mm Count, # n(dp), #/mm dpav, mm Size Range, mm Count, # n(dp), #/mm dpav, mm Size Range, mm Count, # n(dp), #/mm dpav, mm
0.00-0.21 0 0.00 0.11 0.00-0.297 0 0.00 0.15 0.00-0.21 0 0.00 0.11
0.21-0.297 0 0.00 0.25 0.297-0.59 30 102.39 0.44 0.21-0.42 10 47.62 0.32
0.297-0.42 10 81.30 0.36 0.59-1.19 90 150.00 0.89 0.42-0.84 50 119.05 0.63
0.42-0.59 20 117.65 0.51 1.19-2.00 160 197.53 1.60 0.84-1.68 160 190.48 1.26
0.59-0.84 30 120.00 0.72 2.00-3.36 40 29.41 2.68 1.68-2.38 90 128.57 2.03
0.84-1.19 60 171.43 1.02 3.36-4.76 0 0.00 4.06 2.38-4.76 10 4.20 3.57
1.19-1.68 100 204.08 1.44 4.76-5.66 0 0.00 5.21 4.76-5.66 0 0.00 5.21
1.68-2.00 60 187.50 1.84 NT=320 NT=320
2.00-2.38 30 78.95 2.19
2.38-3.36 10 10.20 2.87
3.36-4.76 0 0.00 4.06
4.76-5.66 0 0.00 5.21
NT=320

210 Chemical Engineering Education

secting the population balance equation, it is necessary to employing 11 sieves (fine grid). Secondly, some sieves were
understand how to represent particle populations properly. In extracted from the original set of sieves, and the same sample
the next section different particle size distributions (PSDs) was studied by using just 6 sieves (coarse grid 1). Finally, the
will be discussed. sample was analyzed by using 6 different sieves from those
of coarse grid 1 (coarse grid 2). From this information, the
Particle size distributions histogram of frequency (count) vs. particle size can be built
The particle size distributions are presented in many books, for the three grids presented in Table 1. Figure 3 shows the
such as Litster, et al.,[3], Rhodes,[4] and Randolph and Larson.[6] histograms that were thus obtained. Even though the same
The proper particle size distribution to compare populations population has been analyzed, the histograms are very differ-
is the density function as suggested, for example, by Litster, ent. Therefore, the graphs of count vs. particle size cannot be
et al.[3] The students, however,—and even chemical engineers used to evaluate size population similarities. The number of
working in industry—frequently use the plot of number of particles is often plotted as a function of the average diameter
particles (or mass fraction) as a function of mean diameters of the size interval. As can be seen in Figure 4, an inspection of
to represent the size distribution of a given particle popula- the curves does not indicate that the distributions are similar,
tion. In this section, the need for using the density function in agreement with the histograms of frequency (Figure 3).
is presented through an example. This didactic explanation The density function (n) is a particle size distribution that
was found understandable by all the students that participate allows comparison of populations,[3] and it is defined in a
in the classes. The working example was sufficiently clear to discrete form as:
prove the need for using the density function as a description
of the particle size distribution. n*i
n i (dp ) = (1)
Table 1 shows different data size analyses performed to ∆dp i
the same particle sample. First, the population was analyzed
where n*i is the number of particles between two contigu-
ous sizes (dpi and dpi+1) and ∆dpi is the interval width (dpi+1
175 − dpi). The continuous density function can be expressed as
150 Fine Grid follows:
Count, #

125
dN
100 n(dp ) = (2 )
75 d(dp )
50
25 where N is the number cumulative distribution and has the
0 units of number of particles (#). The density function n (#/L;
where L indicates a generic length unit) represents the number
175 of particles per unit of particle size. The continuous density
150 Coarse Grid 1 function verifies the following equation:
Count, #

125 ∞

100 ∫ n(dp)d(dp) = N T ( 3)
75 0

50
25 180
0 Fine Grid
160
Coarse Grid 1
175 140
Coarse Grid 2
150 Coarse Grid 2 120
Count, #

Count, #

125 100
100 80
75 60
50 40
25
20
0
0
0 1 2 3 4 5 6
0 1 2 3 4 5 6
dp, mm dpav, mm

Figure 3. Histogram of frequency (number) Figure 4. Particle number as a function of the

vs. particle size. average diameter.
Vol. 41, No. 3, Summer 2007 211
 Often the frequency or density distribution is expressed as Since Eq. (3) has to be satisfied, from the discrete representa-
a normalized distribution: tion (histogram) it is possible to obtain a continuous density
function that preserves the area under the curve n(dp) vs. dp
n(dp )
∞ ∞

∫ d(dp ) = ∫ f (dp )d(dp ) = 1 (4 ) (this is the continuous curve that appears in Figure 5). As
0
NT 0 can be noticed, the original particle number attributed to a
size interval in the histogram representation can be assigned
where NT and f(dp) are the total number of particles (#) and
approximately to the average diameter of the size interval for
the normalized density function (L−1), respectively.
the continuous curve. That is the reason why the continuous
Eq. (3) indicates that the total area under the curve n(dp) density function is often plotted as a function of the arithmetic
vs. dp has to be equal to the total number of particles. Figure mean of the size range.
5 shows the histogram of the density function, calculated
Figure 6 shows the number density function for the fine and
according to Eq. (1), for the fine grid presented in Table 1.
coarse grids of Table 1. It is clear that the data of coarse grids
250 1 and 2 track well the density function calculated from the
fine grid. This fact indicates that independent of the number
200 Fine Grid of sieves employed in an experimental analysis or the grid
points selected in a numerical procedure, the density function
of a unique sample has to be equal. The density distribution is
n(dp), #/mm

150
independent of the interval widths used for experimentation
100 or numerical analysis. This point is very important, since the
density function is commonly used to formulate the popula-
50 tion balance equation.
When agglomeration takes place, particles of different
0 volume coalesce to give a bigger one. As seen in Figure 7,
0 1 2 3 4 5 6 the new particle has a volume equivalent to the sum of the
dp, mm volumes of the individual ones. For this reason, the PBE
commonly uses the density function in terms of volume
Figure 5. Density function as a function of the
rather than particle diameter.[4, 7, 8] When the volume is se-
particle diameter.
lected as the representative size of the particles, the density
250 function becomes:
Fine Grid
n*i
200 Coarse Grid 1 n i (Vp ) = (5 )
Coarse Grid 2 ∆Vp i
n(dp), #/mm

where ∆Vpi is the volume interval width. For Eq. (5), the
150

density function n has the units of #/L3.

100
At this point, the students should have in mind that the
50
density function is the proper distribution for characterizing
particulate systems and that the density function based on
particle volume is appropriate to represent coalescence pro-
0
0 1 2 3 4 5 6
cesses. Certainly, these items have been already treated in
the literature[3, 5, 6]; nevertheless students often have difficulty
dpav, mm
understanding these two key points. The order of explanation
Figure 6. Density function as a function of the average of the topics was found adequate in the performed teaching
diameter for the sample analyzed with experiences.
different sets of sieves.
Population balance equation
The ideal granulation units are classified according to their
+ flow pattern into perfectly mixed and plug flow granulators.
These are exactly analogous to the Continuous Stirred Tank
Reactor (CSTR) and the Plug Flow Reactor (PFR) in chemi-
V1 V2 V=V1+V2 cal reactors, respectively.[3] The PBE for size enlargement
processes has been introduced in several books and publica-
Figure 7. Preservation of volumes in the tions, among others, Litster, et al.,[3] Rhodes,[4] Randolph and
coalescence process.
212 Chemical Engineering Education
Larson,[6] and Heinrich, et al.[9, 10] Even though the PBE derivation can be small size. Even though it is not shown in Figure
found in the literature, the internal coordinates (particle properties) are 9, the particle population generated by nucleation
hard to visualize for many students. In the following sections, the flow of has a density function distribution. Therefore,
particles in the internal coordinates is specifically discussed, focusing on the nuclei can enter several size compartments.
the clear definition of the control volume to derive the PBE. Moreover, Layering and attrition cause the differential
the granulation rate processes and flow of particles are compared with particle enlargement and diminution. Therefore,
the chemical reactions and flow patterns occurring in chemical reactors, the particles gradually leave the size clusters to
respectively. This didactic strategy, employed to some extent by Litster, et the contiguous ones.[11]
al.,[3] was found very useful in teaching the subject to advanced chemical A similar conceptualization of the flow of
engineering students. particles shown in Figure 9 was introduced by
Heinrich, et al.[9] In this paper, however, a com-
Perfectly mixed granulators plete description of all the feasible flows into and
Figure 8 shows what can be seen through a window in a perfectly mixed out of the particles “baskets” is given, facilitating
granulator. Inside the unit, particles of different diameters are located the student’s comprehension.
everywhere. The perfectly mixed condition implies that the particle size Once the influence of the granulation rate
distribution is identical throughout the granulator volume. processes on the movement of particles from one
If the population of Figure 8, which is identical in any position of the size “basket” to others is understood, the PBE
granulator, is classified in baskets of different sizes, the particles contained formulation can be discussed.
in the granulator of Figure 8 can be schematized as shown in Figure 9. For the PBE derivation, the frequency distri-
For a granulator like the one presented in Figure 1, a seed particle dis- bution per unit of granulator volume (n+) is also
tribution is fed continuously to the system. In the “baskets” representation
of the particle population inside the granulator (Figure 9), the addition
contributes particles of different sizes that have to be classified in the
size compartments. Similarly, a granular product extraction involves the
removal of particles from different size boxes. In a common granulation
process, the additions will specifically increase the number of particles
in the smaller-size boxes.
If two particles coalesce to give a bigger one, in a discrete manner,
both particles will abandon their respective size ‘baskets” to increase
the number of particles of a compartment that stores bigger particles.
The effect of the binary breakage is opposite to the one caused by the
agglomeration or coalescence phenomenon. Coalescence and breakage
produce the birth of particles for some sizes and simultaneously the death
of particles for other sizes. Nucleation leads to the birth of particles of

Coalescence m Figure 8. Conceptual-

ization of a perfectly
Additions
Layering mixed granulator.

Nucleation
b Figure 9. Classifica-
tion of the particles
according to the volume
Vp1 Vp2 Vpi-1 Vpi Vpi+1 size.

Extractions
Breakage Attrition

Vp
Vol. 41, No. 3, Summer 2007 213
commonly used. This distribution is related to the density PBE for a perfectly mixed granulator is obtained:
function n according to the following expression:
∂  n+ (Vp )V ∂ (G − A)n+ (Vp )V
n(Vp ) −  −
n+ (Vp ) = (6 ) ∂t ∂Vp
V
+n +birth (Vp )V − n +death (Vp )V
where V is the volume of the granulation unit. Therefore, +Q in n+in (Vp ) − Q out n+out (Vp ) = 0 (9 )
n+(Vp) has #/L6 units.
The symbol n + (Vp ) is reserved for the frequency dis- As mentioned above, n birth and n +death are terms analogous to
+

tribution per unit of volume and time (#/L6 t), and is used the chemical reaction terms in chemical reactors.[1] The birth
to represent birth and death rates; n + (Vp ) is completely and death coalescence and the breakage rates all require the
equivalent to the chemical reaction rate of a chemical reactor. use of theoretical or empirical models that depend on n+, as
In order to understand these new properties better, Table 2 the chemical reactions are functions of the concentration of
shows the analogy of the particle size distributions presented different species.
in this work with the variables commonly used in the design Since the granulation is supposed to be perfectly mixed,
of chemical reactors. the outlet density function of the population can be as-
Bearing in mind the concepts introduced in Figure 9, the sumed to be equal to the distribution inside the granulator,
particle number balance for a generic “basket” of volume n+out (Vp ) = n+ (Vp ). Considering this relationship and Eq.
∆Vp can be expressed as: (6), Eq. (9) becomes:
Number of particles Number of particles ∂n ∂ [ (G − A)n ]
− −
 −  ∂t ∂Vp
in time t  in time t+∆t 
+n birth − n death
Number of particles"in" Number of particles "out"
+  −  Q Q
by layering/attrrition  by layering/attrition  + in n in − out n = 0 (10 )
V V
Number of particles "in"  Number of particles 
+  − 
by coales./break./nuccleation "out" by coales./break. A similar derivation of the PBE can be found elsewhere.[3,4]
Number of particles Number of particles The sequential presentation of the process conceptualization,
+  −  shown in Figure 9 and the PBE derivation, were found helpful
"in" by additions  "out" by extraction
ns 
while teaching, however. From our experience, this compre-
=0 hensive view of the internal coordinates makes understanding
(7) Eq. (10) easier.
Eq. (10) has the following two first derivative terms: the
n (Vp )  ∆Vp   V t − n (Vp ) ∆Vp V
+ + accumulation term for unsteady state behavior, and the dif-
t+∆t
ferential term of layering and attrition. The second first deriva-
+(G − A) n+ (Vp ) V  ∆t Vp tive represents a convective term in the Vp direction because
−(G − A) n (Vp ) V ∆t
+ there is a plug flow of particles in the volume particle axis.
Vp+∆Vp
Therefore, a granulator that is perfectly mixed with respect
+n +
birth (Vp ) ∆Vp V ∆t − n +
death (Vp ) ∆Vp V  ∆t
+Q in   n (Vp ) ∆Vp   ∆t
+
in
Seeds

−Q out n+out (Vp ) ∆Vp   ∆t = 0

(8 )

where G is the growth or layering rate (L3/t); A, the attrition

rate (L3/t); n birth(Vp), the birth of particles of class Vp by
+

coalescence, breakage, and nucleation; n +death(Vp), the death

of particles of class Vp by coalescence and breakage; Qin and
Qout are the inlet and outlet particle volumetric flow rate (L3/t);
and n+in (Vp ) and n+out (Vp ) are the density distributions of the Granular
product
seeds and granular product streams (#/L6).
Dividing the entire resulting Eq. (6) by ∆Vp and ∆t and
taking the limit as ∆Vp and ∆t tend to zero, the differential Figure 10. Drum granulator scheme.

214 Chemical Engineering Education

to the real flow pattern exhibits a convective term in a new all the rate processes that can occur during granulation.
coordinate. The particle size is recognized as an internal co- For a volume element ∆Vp A ∆z, where A is the granulator
ordinate; while the spatial coordinates of the equipment are cross section (L2), a number balance turns into:
commonly called external coordinates.[6, 12]
Number of particles Number of particles
In short, a perfectly mixed granulator behaves as a plug  − 
in time t  in time t+∆t 
flow for the chosen particle property the population balance
is focused on. This analogy was found effective in teach- Number of particles"in" Number of particles "out"
ing the process to advanced chemical engineering students, +  − 
by layering/attrrition  by layering/attrition 
who are familiar with chemical reaction engineering. In
the example described, the number variation was evaluated Number of particles "in"  Number of particles 
+  − 
for changes in the particle volume. Other particles proper- by coales./break./nuccleation "out" by coales./break.
ties, however,—such as porosity and density—can also be Number of particles  Number of particles 
considered. The size classification given in Figure 9 can be +  − 
"in" by convective flow "out" by conv
vective flow
adapted to other particle properties for processes with several
internal coordinates. =0
(11)
Plug flow granulators
The drum granulators are basically inclined cylinders that n+ (Vp )  ∆Vp   A   ∆z t − n+ (Vp ) ∆Vp   A ∆z t+∆t
are rotated to facilitate the movement of particles toward one
end of the unit (see Figure 10). As a rough approximation, it +(G − A) n (Vp ) A  ∆z ∆t
+
Vp −
can be assumed that the granules flow through the drum in (G − A) n+ (Vp ) A ∆z ∆t Vp+∆Vp  
plug flow.[3] In this section, the PBE is derived for this type
of ideal granulator. +n +
birth (Vp ) ∆Vp A ∆z ∆t
As is the case for chemical reactors, the flow of particles in −n +
death (Vp )∆VpA   ∆z∆t
plug flow granulators can be associated with a series of perfectly +vz A   n+ (Vp ) ∆Vp   ∆t z
mixed units, as shown in Figure 11. For a ∆z length element,
−vz   A   n (Vp ) ∆Vp   ∆t
+
=0
the particles can also be classified according to their sizes, as z+∆z

schematized in Figure 12 (see page 216). The real flow of (12 )

particles only occurs in the axial direction, though a flow of
particles in the Vp direction is also needed in order to capture where vz is the velocity of the particles (L/t).
Dividing the entire resulting Eq. (12) by A, ∆Vp, ∆z and
∆t and taking the limit as ∆Vp,
Granular ∆z and ∆t tend to zero, the dif-
product Seeds ferential PBE for a plug flow
granulator becomes:

∂n+ ∂ (G − A)n 

+

− −
∂t ∂Vp

z ∂  vz n 
 +

−  + n +birth
∂z
Figure 11. Conceptualization of a plug flow of particles. −n +death = 0
(13)
TABLE 2
Analogy Between the Properties of Granulators and Chemical Reactors
Eq. (13) shows the fol-
Granulators Reactors lowing three first derivative
terms: the accumulation
Symbol Description Units Symbol Description Units
term, and two convective
n Frequency distribution #/L3 NA Moles of the A specie mol terms in the Vp (internal) and
z (external) coordinates.
Frequency distribution per 6
n+ #/L CA Concentration mol/L3
unit of volume It is important to stress
n  Frequency distribution per
#/L6 t rA Reaction rate mol/L3 t that the control volume to
unit of volume and time
Vol. 41, No. 3, Summer 2007 215
 Attrition
Vp
Layering

Vpi+1
Vpi

Internal coordinate
Breakage

Vpi-1
Coalescence

Vp2
Vp1
Convective
Convective flow out flow in
Nucleation
Granular
product Seeds

z External coordinate

Figure 12. Change of particle size in internal and external coordinates.

derive the PBE for this system is not clearly presented in the where vx, vy, and vz are the velocities of the particles in the
literature. The conceptualization of the control volume given external coordinates x, y and z of the granulation unit; vi cor-
in Figure 12 lets the students understand in a rapid and easy responds to the rates of the selected particle properties that
way the PBE for plug flow granulators. change in a differential manner; xi are the internal coordinates,
while m symbolizes the number of internal coordinates chosen
Generalized PBE for ideal granulators to represent particle properties. The units of vx, vy, and vz are
The PBE given by Eq. (13) can be generalized to a system L/t, while vi has the units of the internal coordinates xi per
where convective flow may occur in all the granulator real unit of time.
coordinates, considering that many particle properties may
change during the granulation. Following the same line of about the teaching experiences
reasoning as the one for plug flow granulator, the generalized As mentioned, the described material was used to teach
PBE equation can be obtained[6]: the formulation of the population balance equation applied to
granulation processes in the framework of an optional course
∂n+ ∂  vx   n  ∂  vy n  ∂  vz n 
+ + +

− − − − given for advanced students of chemical engineering. The

∂t ∂x ∂y ∂z course is an elective in the last year of the chemical engineer-
m ∂  v n+ 
 i     + ing career at the Universidad Nacional del Sur, Bahía Blanca,
−∑  + n birth − n +death = 0 Argentina. The course entitled “Solids Processing” covers
∂x i
1
the following main topics: particle size analysis, particles in
(14 ) fluids, fluidization, solids conveying, gas/solid and solid sepa-
ration, solids storage, solids caking, solids mixing, particle
216 Chemical Engineering Education
size comminution and enlargement. The population balance Acknowledgments
equation is introduced when the last two chapters are covered. This work has been supported by the Consejo de Investiga-
The approach described in this contribution is particularly ciones Científicas y Técnicas (CONICET), Agencia Nacional
useful for the last chapter (size enlargement). The emphasis de Promoción Científica y Tecnológica (ANPCyT, Grant:
on granulation processes is motivated by the presence of an PICT 25541) and Universidad Nacional del Sur, Argentina.
industry that produces about 1 million tons of granulated
urea/year in our city. It is important to note that the material References
presented in this work was also successfully employed to teach
1. Fitzpatrick, J.J., N. Zumaeta, and E.P. Byrne, “Teaching Particle and
the “meaning and potential uses” of the PBE for granulation Powder Technology in Chemical Engineering at University College
processes to engineers working at fertilizer plants. Cork,” Fifth World Congress on Particle Technology, Orlando, FL,
April 2006
The textbooks used in the course are, among others, 2. Bell, T.A., “Challenges in the Scale-Up of Particulate Processes—An
the ones written by Rodhes,[4] Litster, et al.,[3] Seville, et Industrial Perspective,” Powder Technology, 150, 60-71 (2005)
al.,[13] and Kunii and Levenspiel.[14] The homework and 3. Litster, J., B. Ennis, and L. Liu, The Science and Engineering of Granu-
exams assigned to the students include the development lation Processes, Particle Technology Series, 15, Kluwer Academic
Publishers, London (2004)
of the population balance equation for various granulation 4. Rhodes, M., Introduction to Particle Technology, John Wiley & Sons,
equipments and different particle properties. Regarding West Sussex, England (2003)
the solution of PBEs, only the simple cases for which 5. Iveson, S.M., J.J. Litster, K. Hapgood, and B.J. Ennis, “Nucleation,
analytical solutions can be found are covered in the un- Growth, and Breakage Phenomena in Agitated Wet Granulation Pro-
cesses: A Review,” Powder Technology, 117, 3-39 (2001)
dergraduate course. 6. Randolph, A.D., and M.A. Larson, Theory of Particulate Processes,
Academic Press, New York (1971)
Conclusions 7. Ramkrishna, D., Population Balances. Theory and Applications to Par-
ticulate Systems in Engineering, Academic Press, San Diego (2000)
The method presented in this paper for introducing the 8. Scarlett, B., “Particle Populations—To Balance or Not To Balance,
derivation of population balance equations was used to That is the Question,” Powder Technology, 125, 1-4 (2002)
teach the subject both to chemical engineers that work in 9. Heinrich, S., M. Peglow, and L. Mörl, “Unsteady and Steady State
a granulation plant and to chemical engineering students. Particle Size Distributions in Batch and Continuous Fluidized Bed
Granulation Systems,” Chem. Eng. J., 86, 223-231 (2002)
The conceptualization of the internal coordinates was clear 10. Heinrich, S., M. Peglow, M. Ihlow, and L. Mörl, “Particle Population
and comprehensible for all of them. Our approach can be Modeling in Fluidized Bed-Spray Granulation—Analysis of the Steady
considered valuable as a teaching strategy to explain the State and Unsteady Behavior,” Powder Technology, 130, 154-161
PBE in the context of granulation processes. In particular, (2003)
11. Cameron, I.T., F.Y. Wang, C.D. Immanuel, and F. Stepenek, “Process
the analogy with chemical reactors is useful because this is Systems Modeling and Applications in Granulation: A Review,” Chem.
a subject well known by the advanced chemical engineer- Eng. Sci., 60, 3723-3750 (2005)
ing students. Probably the PBE is not as extensively used 12. Ramkrishna, D., and A.W. Mahoney, “Population Balance Modeling:
as it could be, due to early difficulties understanding its Promise for the Future,” Chem. Eng. Sci., 57, 595-606 (2002)
13. Seville, J.P.K, U. Tüzün, and R. Clift, “Processing of Particulate Sol-
derivation and then solving the final equation. The pres- ids,” Particle Technology Series, Blackie Academic and Professional,
ent contribution may help students/teachers that want to London (1997)
start learning/teaching population balances for model size 14. Kunii, D., and O. Levenspiel, “Fluidization Engineering,” Butterworth-
enlargement processes. Heinemann: Newton, MA (1991) p

Vol. 41, No. 3, Summer 2007 217