Batch To Continuous Processing
Batch To Continuous Processing
Batch To Continuous Processing
From Batch to
Continuous Processing
1,000
Reactor size
10 mL
100 mL
100
Flow, L/h
Continuous flow
reactors can provide
many benefits over
batch processes.
This article answers
why and how
1L
10 L
100 L
10
1,000 L
10,000 L
0
0
Robert Ashe
AM Technology
10
100
1,000
10,000
100,000
1,000,000
Reaction time, s
FIGURE 1. Reactor sizes for a range of throughputs and reaction times are shown
Flow reactors
A small chemical reactor has a higher
ratio of heat transfer area to working volume than a large reactor. It
can also deliver more mixing energy
per unit volume without bending the
agitator shaft. Since the outcome of
most chemical reactions is linked in
some way to mixing or heat transfer
(or both), small reactors have inherent
advantages over large ones. Depending on the reaction type, these advantages can contribute toward substantially lower capital and operating
costs. Small batch vessels, however,
are impractical at the industrial scale
since hundreds or even thousands of
process cycles would be required for
34
FIGURE 2. In plug flow, the fluid velocity across the face of a channel is uniform
substitute for testing under flow conditions. The equipment used for scaleup development should be as similar
as possible to the full sized reactor.
This generally favors the use of large
flow reactors for scaleup work rather
than microreactors (other than where
microreactors can be used at the production scale).
Reactor capacity
The volumetric capacity of a flow reactor should be as small as possible.
Apart from the obvious benefits of
small footprint and higher performance, small flow reactors have
proportionately lower startup and
shutdown losses. The physical size is
calculated from the relationship:
Reactor size (L) = Throughput (L/s)
Reaction time (s)
Reactor sizes for a range of throughputs and reaction times are shown in
Figure 1.
A flow reactor can be scaled up by
a process of numbering up (or scaling
out). For this, the channel size is kept
constant and capacity is increased by
using multiple parallel channels of
similar size. The size of the scaled up
system can be predicted accurately
when numbering up, since the functional capabilities of the channels remain unchanged.
The reactor can also be scaled up
by increasing the channel length, although scope for this is limited by
pressure drop. In some cases, longer
channels will give better performance
(due to higher velocities) and therefore require smaller working capacities than predicted.
Increasing the tube diameter is often
the only practical way of scaling up,
especially where high reactor volumes
are required. Increasing the diameter,
however, often reduces performance,
and therefore the scaled system may
have to be larger than predicted to
compensate for the difference.
When scaling up, it is preferable to
use the maximum tube diameter possible, subject to satisfactory performance. Short, large-diameter tubes
have lower fabrication costs, a reduced
tendency to block and a lower pressure
drop. They are also easier to clean.
35
Feature Report
Plug flow
Plug flow describes a flow pattern
where the fluid velocity across the
face of a channel is uniform (Figure 2).
It also implies no back mixing and all
fluid elements having the same residence time. While plug flow is required
for most applications, ideal plug flow
cannot be achieved in practice due to
the effects of molecular diffusion, surface drag, bends and obstructions in
the channel.
Microreactors typically have channel diameters of less than 0.5 mm. In
these systems, the plug flow characteristics benefit from low fluid velocities
[2]. This is a useful characteristic as
the reaction time can be varied from
less than a second to many minutes in
relatively short channels.
In larger diameter channels, good
plug-flow characteristics benefit from
high fluid velocities [2], and the channel length has to be varied to suit the
reaction time. Good radial mixing is
also desirable to prevent the fluid
from separating into zones of fast
and slow moving fluid. Radial
mixing can be achieved by means of
baffles, static mixing elements, dynamic mixing and turbulent flow. A
good approximation to plug flow can
also be achieved by using multiple
stirred tanks in series, and the higher
the number of tanks used the better.
Above ten stages, however, the cost of
additional stages starts to outweigh
the incremental benefits.
Plug flow is determined by experimental methods such as dye injection
tests. These data need to be treated
with care, because the quality of plug
flow required will change with the reaction rate. The observed quality of
plug flow can also change as the physical properties of the fluid changes.
Plug flow provides the means for
separating reacted and unreacted material (Figure 3a). Without plug flow,
36
Mixing
Mixing can be characterized in many
ways, such as applied mixing energy,
uniformity of mixing, time to achieve
a uniform blend, Reynolds number
or mass transfer characteristics for
Heat transfer
coefficient
W/m2/C
1,000
10
100
100
10
1,000
0.1
10
100
Tube diameter, mm
FIGURE 4. Cooling capacities for different tube sizes and heat-transfer coefficients
are shown here
1. No cooling
Temperature , C
100
2. Single cooling
temperature,
two tube sizes
80
60
40
20
0
0
10
Heat transfer
In a tubular flow reactor, heat is added
or removed by a cooling jacket. The
amount of heat that can be transferred
per unit volume of product in a tube
(Qv) is determined by the relationship:
Qv = (4 U T)/D
The heat transfer coefficient (U) is
typically in the range of 501,500 W/
m2/C. There is little scope for varying
this since the parameters that determine it are fixed by other needs.
The heat transfer rate can be controlled by varying the temperature
difference between the process fluid
37
Feature Report
Figure 5 illustrates different cooling
strategies for an exothermic reaction
in a 10-L reactor:
In the first case, no cooling is applied, and the tube diameter is 38
mm. The overall reactor length is
9 m, and the product temperature
rises to over 100C
In the second case, a single coolant
temperature of 0C is used. The first
cooling stage (2 L) uses a 5-mm-dia.
tube, and the second stage (8 L) uses
a 38-mm-dia. tube. The overall reactor length is 109 m and the process
temperature is held at 4060C
In the third case, a single tube of
5-mm dia. is used. The first cooling
stage (3 L) uses a cooling jacket at
10C, and the second stage (7 L)
uses a cooling jacket at 40C. The
overall reactor length is 509 m and
the temperature is held at 4060C
In the fourth case, a single tube of
5-mm dia. is used, and the cooling
jacket is at 10C throughout. The
overall reactor length is 509 m, and
the process material is progressively
cooled to below 20C
The first case is the simplest and
cheapest solution since the tube length
is short and no cooling system is required. In the second and third cases,
the process temperature is controlled
within a range of 20C but the second
case has the advantage of shorter tube
length. In the fourth case, the product
is subject to severe overcooling, which
will affect the reaction rate (and hence
loss of yield).
FIGURE 6. This
dynamically mixed
plug-flow reactor
uses transverse
mixing in tubes
Conclusion
The batch to continuous debate encompasses a broad spectrum of views.
At one extreme are those who remain
fully committed to batch methods. For
some applications batch reactors are
clearly a better solution or the only option. Processes with sticky materials
for example are inherently unsuited
to flow systems as are some slurries
with large solids. In the absence of
heat transfer or mixing constraints,
batch reactors are also cheaper than
flow reactors (although significantly
less energy efficient). The resistance
to change from batch to continuous
is often justified on the grounds that
existing batch processes are well optimized and offer little scope for improvement. This is true for some processes and particularly so when the
cost of switching from batch to continuous is factored in.
Where there are mixing or heat
transfer constraints however, the commercial benefits of flow reactors are
often clear cut and compelling. It has
been suggested that as much as 60%
of commercially important non-polymer reactions could be run more efficiently in flow reactors [6] (and higher
if peripheral benefits are included). To
quantify these benefits, however, account needs to be taken of many factors that contribute to lower cost of
goods. It also has to be recognized that
most processes in use today were developed for batch reactors and would
39
Feature Report
not necessarily be the first choice of
method given the option of more-capable flow reactors.
At the other extreme of the batch
to continuous debate are advocates
of fully continuous processes where
multiple steps are carried out within
a single process train. The difficulty
here is that good residence-time control is critical to many flow processes.
Problems in one stage of a continuous
process will impact the entire process
train, and the system will only be as
good as the least reliable step. As process equipment and knowhow develop
References
1. Nichols, J., Innovations in Process Technology for Manufacture of APIs and BCPs,
Pharm. Eng., Vol. 28, No 5, Sept./Oct. 2008.
2. Levenspiel, O., Bischoff, K., Backmixing in
the Design of Chemical Reactors Vol. 51,
Dec. 1959.
3. Eslinger, J., De Dietrich Process Systems,
Bulletin BP 8, F-61170, Sept. 28, 2004.
4. Capretto L. and others, Micromixing Within
Microfluidic Devices, Topics in Current
Chemistry, Vol. 304, pp. 2768, 2011.
5. Levenspiel, O., Chemical Reaction Engineering, 2nd Ed., John Wiley and Sons, 1972.
6. Calabrese, G.S. and Pissavini, S., From batch
to continuous flow processing in chemicals
manufacturing, AIChE J., Vol. 57, No. 4,
pp. 828834, 2011.
7. Roberge, D.M. and others, Microreactor Technology and Continuous processes in the fine
chemical and pharmaceutical Industry: is
the Revolution Underway?, OPRD, Vol. 12,
pp. 905910, 2008.
Author
Robert Ashe (AM Technology,
The Heath Business & Technical Park, Runcorn,Cheshire,
WA7 4QX, U.K.; Phone: +44
(0)
1928.515454;
Email:
[email protected])
is a chemical engineer with
over 35 years of experience in
fine chemicals, pharmaceuticals and food manufacturing.
Much of his career has been
devoted to chemical plant design in high-value chemical manufacturing. In
2000, he co-founded AM Technology with David
Morris. This company specializes in the manufacture of flow reactors and holds a broad portfolio of patents for dynamic mixing in tubes.
Circle 1 on p. 60 or go to adlinks.che.com/40272-01
40