You are on page 1of 3

We are concerned with systems consisting of particles dispersed in an environmental phase, which

we shall refer to as the continuous phase. The particles may interact between themselves as well
as with the continuous phase. Such behavior may vary from particle to particle depending upon a
number of "properties" that may be associated with the particle. The variables representing such
properties may be either discrete or continuous.^ The discreteness or continuity of the property
pertains to its variation from particle to particle.

There are several examples of discrete variables. First, a population consisting of particles of
different materials may be distinguished by assigning a variable taking on discrete values each
representing a particular material. Second, one may, merely for the sake of convenience, take a
discrete view as is often done by engineers in characterizing a continuous spectrum of particle
sizes by discrete mesh sizes (obtained by sieve analysis). Third, the particle may itself be
distinguished by a discrete number of entities in it. An example is that of a liquid phase emulsion
droplet in a precipitation process containing a limited number of precipitated particles.

Continuous variables may be encountered more frequently in population balance analysis. They
often arise as a natural solution to dealing with indefinite or variable discreteness. For example, a
particle-splitting process where the products of sphtting could conceivably have any size smaller
than the parent particle is most naturally handled by assigning particle size as a continuous
variable. The external coordinates denoting the position vector of (the centroid of) a particle
describing continuous motion through space represent continuous variables. The temperature of a
particle in a fluidized bed is another example of a continuous variable.

In following the temporal evolution of the particulate system, we shall regard time as varying
continuously and inquire into the rate of change of the particle state variables. It is more
convenient to deal with continuous variables in this regard. A fundamental assumption here is that
the rate of change of state of any particle is a function only of the state of the particle in question
and the local continuous phase variables. Thus we exclude the possibility of direct interactions
between particles, although indirect interaction between particles via the continuous phase is
indeed accounted for because of the dependence of particle behavior on the "local" continuous
phase variables. In order to enable such a local characterization of the continuous phase variables,
it is necessary to assume that the particles are considerably smaller than the length scale in which
the continuous phase quantities vary. The continuous phase variables may be assumed to satisfy
the usual transport equations with due regard to interaction with the particulate phase. Thus, such
transport equations will be coupled with the population balance equation.

2.

We shall be primarily concerned with particle phase variables that are continuous. The choice of
the particle state variables depends on the application. For example, chemical engineers
concerned with the modeling of crystallizers will be interested in predicting and controlling the
size distribution of crystals in the product. The particles in this application are of course crystals,
and the size of the crystal is the main particle state. A growing crystal changes its size at a rate
often determined only by the size of the crystal (besides the prevailing supersaturation in the
continuous phase with respect to the crystallizing solute). Thus, we need only specify the size of
the crystal to predict its growth rate. The particle state is therefore characterized in this example
by a single quantity, viz., crystal size. Notice that the local supersaturation, although important to
determining the growth rate of a crystal, is a continuous phase variable and hence does not enter
the characterization of the particle state.

Consider another example. Suppose we are interested in following the total number of cells in a
population of bacteria that are multiplying by binary division. Assume that the cells do not divide
until after a certain age has been reached. In this case, it becomes essential to define cell age as
the particle state although it is not of explicit interest originally. Thus, the identification of age as
the particle state in this case was dictated by its influence on the birth rate.

In general we may conclude that the choice of the particle state is determined by the variables
needed to specify (i) the rate of change of those of direct interest to the application, and (ii) the
birth and death processes.^ The particle state may generally be characterized by a finite
dimensional vector, although in some cases it may not be sufficient. For example, in a diffusive
mass transfer process of a solute from a population of liquid droplets to a surrounding continuous
phase (e.g., hquid-liquid extraction) one would require a concentration profile in the droplet to
calculate the transport rate. In this case, the concentration profile would be an infinite
dimensional vector. Although mathematical machinery is conceivable for dealing with infinite
dimensional state vectors, it is often possible to use finite dimensional approximations such as a
truncated Fourier series expansion. Thus it is adequate for most practical apphcations to assume
that the particle state can be described by a finite dimensional vector.

The dependence of particle processes (i) and (ii) on the current particle alone of course
implies that we are neglecting memory effects. In other words, the choice of the particle
state must be suitably made to support this assumption.

The finite dimensional state vector can accommodate the description of particles with
considerable internal structure. For example, consider a cell with m compartments. Each
compartment may be considered as well mixed containing a total of n quantities. Suppose
now the cell changes its state by interaction between its compartments and with the
environment. The particle state can be described by a partitioned vector [Xi,X2,...,x j where
x^ represents the vector of n components in the ith compartment. It is also interesting to
observe that a finite dimensional vector is adequate to describe particles with spatial,
internal morphology where several discrete components may be located anywhere within
the particle relative to, say, the centroid of the particle. In this case, the elements in the
partitioned vector above may be interpreted as position vectors of such components.
It is convenient to distinguish between external coordinates r = (r^, r2, r^), which may be used to
denote the position vector of the particle (as determined by that of its centroid), and internal
coordinates x = (x^, X25 • • •' ^d) representing d different quantities associated with the particle.
The particle state vector (x, r) accounts for both internal and external coordinates. We shall
further let Q^ represent the domain of internal coordinates, and Q^ be the domain of external
coordinates, which is the set of points in physical space in which the particles are present. These
domains may be bounded or may have infinite boundaries

The particle population may be regarded as being randomly distributed in the particle state space,
which includes both physical space and the space of internal coordinates. Our immediate concern,
however, will be about large populations, which will display relatively deterministic behavior
because the random behavior of individual particles will be averaged out. We conclude this
section with the observation that many problems in population balance may feature particles
distributed only according to their size or some other scalar variable. We shall exploit the
simplicity of such problems to demonstrate concepts applicable to the more general problems of
population balance.

The continuous phase variables, which affect the behavior of each particle, may be collated into a
finite c-dimensional vector field. We thus define a continuous phase vector Y(r, t) = lY^{r, t), ¥2(1,
t),..., Y^{r, t)], which is clearly a function only of the external coordinates r and time t. The
evolution of this field in space and time is governed by the laws of transport and interaction with
the particles. The actual governing equations must involve the number density of particles in the
particulate phase, which must first be identified.

In some applications, a continuous phase balance may not be necessary because interaction
between the population and the continuous phase may not bring about any (or a substantial
enough) change in the continuous phase. In such cases, analysis of the population involves only
the population balance equation.

You might also like