48 8

Multiphase Reactors

FALLING FILM REACTOR

In the membrane reactor a wall of area
separates the phases, and this area is generally
fixed by the geometry of the reactor using planar or cylindrical membranes. However,
most multiphase reactors do not have fixed boundaries separating phases, but rather allow
the boundary between phases to be the interfacial area between insoluble phases. This is
commonly a variable-area boundary whose area will depend on flow conditions of the
phases, as shown in Figure 12-7.
Before we deal with these situations, it is instructive to consider a fixed-area version
of a membraneless reactor, the falling film reactor. We cannot think of many applications of
this reactor type because one usually benefits considerably by using configurations where
the surface area is as large as possible, but the falling film reactor leads naturally to the
description of many variable-area multiphase reactors.
We consider a vertical cylindrical tube of length
and diameter
(radius
with
liquid admitted at the top such that it forms a
that coats the walls of the tube.
We also add a gas into the top of the tube
or into the bottom (countercurrent).
This is a standard unit in extraction processes called a wetted-wall column.
In analysis of the wetted-wall column one finds the velocity profile in the film u(R)
and then places a solute A in one of the phases and calculates its extraction to the other phase
by first integrating across the liquid film to find
and the average concentration, and
then integrating along the column to find
and then
When used as a chemical
reactor, we assume that species A is a reactant that reacts with another solute in the liquid
or with a catalyst in the liquid or on the walls of the column.
Since we do not want to consider the fluid mechanics of this reactor in detail, we will
assume that the liquid falls with a constant average velocity
and forms a liquid film of
thickness
with
the radius in the tube at the surface of the liquid film. If these
assumptions hold, then
and
are independent of the position in the tube of
length
(actually the height, since the tube must be vertical). Similarly, if the liquid film
thickness is constant, then the cross section occupied by the gas is constant, and the velocity
of the gas
is also independent of if the density of the gas is constant. These distances
are sketched in Figure 12-8.
The interface between the gas and liquid is at
and its area is

Figure 12-7 Sketches of four types of gas-liquid

multiphase reactors.

bubble
column

catalytic wall
falling film

trickle
be d

Falling Film Reactor

4 89

Figure 12-8 Sketch of a falling film reactor with positions and indicated.
At the right are sketched possible concentration profiles of a reactant A, which is
supplied in the gas but reacts in the liquid.

Since the phases are gas and liquid in this section, we use g and rather than and to
designate phases. The cross section of the tube is
the gas will occupy a cross-sectional
area
and the liquid will occupy a cross-sectional area
The volumes
occupied by the phases are
=
=
so that the residence times are
L

Thus we have defined all the parameters for the reactor mass balances in gas and liquid
phases in terms of the inputs to the tube. In all this analysis we make the assumption that
all parameters are independent of height
in the column.
Next we add solute A to the gas phase and solute B to the liquid phase and allow the
reaction
+

+ De

to occur to form products C and D. We assume that the reactions are irreversible, and we
usually assume that the products can easily escape from this phase so the specific products
are not important. We further assume that reaction occurs only in the liquid phase with a rate

so that

= 0.

Multiphase Reactors

The reactant A must be transferred from the gas phase into the liquid phase in order
to react with B. The rate of mass transfer per unit of area is

where
is the equilibrium distribution of A between gas and liquid phases, as discussed
previously. The area of this interface is
We can now write the mass-balance equations for the phases as

We included the term

= 0 to indicate that there is no reaction in the gas phase. The mass
transfer rates obviously have opposite signs, and we have to multiply the mass transfer
flux by [area/volume], where the volume is that occupied by that phase. Note that the mass
transfer term after dividing out becomes proportional to
Since the reactor volume is
proportional to
while the surface area for mass transfer is proportional to R, the falling
film column obviously becomes less efficient for larger reactor sizes. This is a fundamental
problem with the falling film reactor in that small tubes give high mass transfer rates but
low total production of product.
The mass transfer coefficient is usually much lower in the liquid phase, and therefore
is a function of R, the distance from the wall to the interface. One would have to
solve for the steady-state profile
and find its average CA (Z ) to insert into the
mass-balance equations simultaneously to find
in each phase.
Example 12-1 An aqueous solution containing IO ppm by weight of an organic
of molecular weight 120 is to be removed by air oxidation in a l-cm-diameter falling
film reactor at 25C. The liquid flows at an average velocity of 10
and forms a film
1 mm thick on the wall, while the air at 1 atm flows at an average velocity of 2
The reaction in the liquid phase has the stoichiometry A +
products with a rate
=

(a) What are the compositions when the reaction has gone to completion?
We assume that both phases are in nearly plug flow, and we need first to calculate
the flow rates of each reactant.
=

x
x

= 2.83

x 2 = 1.005

The feed concentrations are

8.33 x
assuming the density of water is

moles/cm3

1 g/cm3 and
= 8.58 x

moles/cm

Falling Film Reactor

491

Therefore, the molar flow rates are

=

= 2.83 x 8.33 x

2.36 x

= 1.005 x 8.58 x
Thus we see that there is an excess of
conversion. This gives

8.62 x
so all A will be consumed at complete

(there are 2 moles of

consumed per mole of A reacted), so that if the reaction
goes to completion, we have
= 0
= 8.14 x

moles/cm3

which we will assume to be a negligible change in

(b) Write the equation(s) that must be solved to find the reactor length required for a given
organic removal.
At these flow rates the
in the air is not depleted, and we do not need to worry
about the mass balance of the air in the gas phase. We only need to solve the
equation for the organic in the liquid phase, which is
e

(c) How many tubes in parallel are required to process 100

of water? What will
be the diameter of the tube bundle if the tubes are packed in a square array and the tube
wall thickness is neglected?
One tube processes 2.83

so to process 100 liters/mm we will need

100 x
590 tubes
2.83 x 60
If these are packed in a square configuration with a negligible tube wall thickness,
each tube occupies an area 1
in the tube bundle. The cross-sectional area of
the tube bundle is 590
so the diameter of the column is
= 27.4 cm

1 ft

(d) Estimate the reactor length needed to reduce the organic concentration to 1 ppm if the
reaction rate in the liquid is infinite and the reaction is limited by the mass transfer of
in
the gas. Assume
0.1
sec.
This doesnt seem to be a good approximation because mass transfer in gases is
almost always much faster than in liquids. If it were true, the Sherwood number

492

Multiphase

Reactors

would be
we have

for laminar flow of the air flowing in the 0.8-cm-diameter tube. Thus

=
We have to solve for A in the liquid phase,
=

The interfacial area per volume of reactor is

A
V

and the equation becomes

Co,
dz =
The right-hand side is constant because Co, does not change significantly, so this
equation can be simply integrated to yield
0.031 cm
Co,
This answer is clearly nonsense because we would need a reactor length of less
than 1 millimeter!
(e) Estimate the reactor length needed to reduce the organic concentration to 1 ppm if
mass transfer is sufficiently fast that the liquid solution remains saturated with
and the
rate coefficient
is 1 x
liter/mole sec. The Henrys Law constant for
in water is
1x
atm.
Now mass transfer of
is assumed to be fast so that the liquid remains saturated
with
and the process is assumed to be limited by the reaction in the liquid
phase. Therefore, we have to solve the equation
e

We need the

concentration in the liquid phase. We assume Henrys Law

=

so that
0.2
which gives
Recall that Co,

x 32

= 5.56 x
moles/liter.
nearly constant so this equation can be rearranged to yield
10
x 5.56 x

10 = 415 cm =

= 16ft

Bubble Column Reactors

493

a reasonable column height.

Estimate the reactor length needed to reduce the organic concentration to 1 ppm if the
reaction actually occurs on a catalyst on the wall of the reactor tube. Assume that the reaction
rate on the wall is infinite and the process is limited by the diffusion of
through the liquid,
with
=
and that the concentration profile is linear.
Now we assume that reaction occurs on the exterior surface of the tube. The
equation to be solved is
Al
= -v 2

= -3.47 x

moles/liter

The area for mass transfer per unit volume of reactor in the liquid is given by
A
11.1
V
We now solve the preceding differential equation, noting that all terms on the
right-hand side are constants to obtain the reactor length needed,
L =

3.47 x
This is a very tall column.

= 2160 cm = 22 meters = 70 ft

(g) How would the answers to (c), (d), and (e) change if the pressure of air were increased
from 1 to 2 atm?
Doubling Co, will double the rate of the reaction and the rate of
mass transfer
to the wall of the reactor. Therefore, the reactor length needed will be half of those
calculated in each of the previous situations.

BUBBLE COLUMN REACTORS

We just noted that the falling film reactor has the problem that the surface area increases as
the perimeter between gas and liquid while the volume increases as the cross section
area
volume

2
gas volume

1
D

In order to design an efficient reactor using a falling film reactor, we would need to have
many small tubes in parallel so that the interfacial area can be large. This is difficult
to accomplish with flow down tubes, but it is easy to accomplish with rising bubbles
or falling drops. The interfacial area is not now the area of
between gas

494

Multiphase

Reactors
and liquid but rather the area of the bubbles or drops, which can be large in a

diameter tube.
Consider a liquid flowing downward in a tube with gas bubbles introduced at the
bottom. The bubbles rise because they are less dense than the liquid. At the top the gas is
separated from the liquid in the head space, while pure liquid is withdrawn at the bottom,
as shown in Figure 12-9.
An important example of bubble column reactors is in partial oxidation of
such as p-xylene to make terephthalic acid, a monomer for polyester (PET),
p-xylene

terephthalic acid

where p-xylene and terephthalic acid are liquids and

is a gas. In partial oxidation such
as these the reaction occurs homogeneously (either with free-radical intermediates or with
a homogeneous catalyst) in the liquid phase, and the desired product remains in the liquid
phase, where it is separated from the other organic. It is usually important to have no
in
the product because it can be explosive: so the reactor must be designed so that all
in
the bubbles has reacted and none reaches the head space, where it could react explosively
with the organic products.
Another example of bubble columns is hydrogenation of olefins and other
such as reduction of dinitrotoluene
dinitrotoluene

diaminotoluene

which is an intermediate in production of polyurethanes and in the hydrogenation of fats

and oils,
+

In this case the reaction occurs only in the liquid phase (the oils have negligible vapor
pressure), but reaction requires a solid catalyst such as finely divided Ni, which is suspended
in the liquid. Thus we see that this is in fact a three-phase reactor containing
gas, organic
liquid reactants and products, and solid catalyst.

liquid

gas

liquid

gas
liquid
gas
bubble
column

tower

Figure 12-9 Bubble column and spray tower reactors. Large drop or bubble areas increase reactant mass
transfer,

Bubble Column Reactors

495

Figure
Sketches of reactant concentration
around a spherical bubble or drop that reacts
after migrating from the gas phase into the liquid
phase in bubble column and spray tower reactors.

The concentration profile for a reactant A which must migrate from a drop or
bubble into the continuous phase to react might be as shown in Figure 12-10. There is
a concentration drop around the spherical drop or bubble because it is migrating outward,
but, as with a planar gas-liquid interface in the falling film reactor, there should be a
discontinuity in
at the interface due to the solubility of species A and a consequent
equilibrium distribution between phases.
Bubble columns rely on nozzles, mixing plates, and impellers within the reactor to
control the bubble size, which determines the interfacial area between gas and liquid phases.
Clearly, the interfacial area can be varied over a wide range by suitable design of the mixer
and flow pattern.
Consider a bubble phase occupying a fraction
in a reactor with volume
=
Suppose all bubbles have radius
so that each bubble has a volume
If there are
bubbles per unit volume of reactor, then the volume occupied by bubbles is

If the bubbles supply reactant into the liquid phase, then the bubbles are decreasing in
size because of reaction; so we need to find
or
of the bubbles as they rise in
the reactor. Thus we have the problem of a sphere that varies in diameter as the reaction
proceeds. We considered this in Chapter 9 where we were concerned primarily with reaction
of solid spheres instead of liquid spheres. The bubbles usually have a distribution of sizes
because larger bubbles usually rise faster than small ones, and they can coalesce and be
redispersed by mixers. However, to keep the problem simple, we will assume that all
particles have the same size. For the reactant A supplied from the bubble, we have to solve
the equation
=
just as we did for a reacting solid concentration profiles of reactant A and product B reacting
around a spherical particle are shown in Figure 12-l 1.

4 96

Multiphase Reactors

Sketches of concentration profiles of a reactant A around a drop where the reaction is in the surrounding continuous
phase. Arrows indicate directions of reactant mass flow.

As with reactions of solids, we first assume a steady-state concentration profile of

to
and then calculate the variation of
versus time as the reaction
proceeds.

Example 12-2 An aqueous solution contains 10 ppm by weight of an organic contaminant

molecular weight 120, which must be removed by air oxidation in a lo-cm-diameter
bubble column reactor at 25C. The liquid flows downward in the tube at an average velocity
1
The air at 1 atm is admitted at 0.1
and is injected as bubbles 1 mm
diameter, which rise at 2
Assume no coalescence or breakup and that
gas and
liquid are in plug flow. The reaction in the liquid phase has the stoichiometry A +
products with a rate
(a) What are the compositions when the reaction has gone to completion?
The volumetric flow rates are
=

= 1x

78.5

The feed concentrations are

8.33 x

moles/cm3
= 8.58 x

The molar feed rates are therefore

=

= 6.5 x

2 =

= 100 x 8 x

This gives
0
=

= 7.8 x

=8x

moles/cm3

Bubble Column Reactors

4 97

This means that

decreases by only -3% in the reactor if the reaction goes to
is constant throughout the reactor.
completion, so we can assume
(b) Write the equation(s) that must be solved to find the length for a given organic removal.
Since the
concentration remains constant, we do not need to solve a
balance equation in the gas phase, and
and countercurrent flows should
give the same answer if velocities remain constant. The equation to be solved is
therefore

(c) Assuming that mass transfer of

in the liquid around
bubbles limits the rate,
calculate the reactor length required to reduce the contaminant to 1 ppm.
Since mass transfer in the liquid around the bubbles limits the rate, the
balance equation on A in the liquid phase becomes
A
,

dz
for flow over a sphere we assume

= 2.0,

A
SO

that

2 x
=
=
D
The right-hand side of the differential equation is constant; so it can be integrated
to yield
k

We next need A/ V, the mass transfer area per unit volume of the reactor. The
residence time of a bubble is
The volume of each bubble is
and the
surface area of a bubble is
The volumetric flow rate of gas is

Therefore, the surface area of bubbles per unit volume of liquid in the reactor is
A
V
Substituting, we obtain
L =

8.33 x
x 38.2 x 2 x

=
x 0.9 x 1
x 1.11 x

38.2

177 cm = 1.77 m

(d) Estimate the reactor length required to reduce the organic concentration to 1 ppm if the
concentration of
remains at equilibrium in the solution.
If the

remains in equilibrium, the equation to be solved is

Multiphase

Reactors
e

Since

is constant, the differential equation can be integrated to yield

10
x 1.11 x

cm = 2.07 m

Mixing in drop or bubble phases

If an isolated drop or bubble rises or falls in the reactor, then the flow pattern in this phase
is clearly unmixed, and this phase should be described as a PFTR. However, drops and
bubbles may not have simple trajectories because of stirring in the reactor, and also drops
and bubbles can coalesce and breakup as they move through the reactor.
These possible flow patterns of a drop or bubble phase are shown in Figure 12-12. At
the left is shown an isolated bubble which rises (turn the drawing over for a falling drop),
a clearly unmixed situation. If the bubble is in a continuous phase which is being stirred,
then the bubble will swirl around the reactor, and its residence time will not be fixed but
will be a distributed function. In the limit of very rapid stirring, the residence time of drops
or bubbles will have a residence time distribution

p(t)
which is the residence time distribution in a CSTR. We derived this expression in Chapter 8
where we considered the distribution of residence times of molecules in the reactor, but
now we consider the residence time of isolated bubbles or drops. If bubbles flow generally
upward, but continuously break up and coalesce, the residence time distribution of the
species in this pase is narrow or roughly that of a

However with stirring and coalescence and breakup, both effects tend to mix the
contents of the bubbles or drops, and this situation should be handled using the CSTR mass
balance equation. As you might expect, for a real drop or bubble reactor the residence time
distribution might not be given accurately by either of these limits, and it might be necessary
to measure the RDT to correctly describe the flow pattern in the discontinuous phase.
Figure 12-12 Sketches of possible flow patterns of bubbles rising through a liquid phase
in a bubble column. Stirring of the continuous
phase will cause the residence time distribution
to be broadened, and coalescence and breakup
of drops will cause mixing between bubbles.
Both of these effects cause the residence time
distribution in the bubble phase to approach that
of a CSTR. For falling drops in a spray tower,
the situation is similar but now the drops fall
instead of rising in the reactor.