Professional Documents
Culture Documents
Bubble Fermentor VG
Bubble Fermentor VG
Contents
1. Introduction
2. First characterization of bubble columns
3. Mass transfer
4. Oxygen Transfer Efficiency
S
5. Examples of bubble column reactors
R
6. Airlift Loop Reactors
AP S
7. Examples of airlift loop reactors
TE
H S
8. Heat transfer in bubble columns and loop reactors.
9. Conclusions
C OL
Glossary
Bibliography
Biography Sketch
E -E
Summary
PL O
Bubble columns and airlift loop bioreactors are attractive alternatives to stirred tank
bioreactors for large-scale manufacturing of commodity products such as yeast, citric
M SC
acid etc. Their most important advantage as compared to stirred tank bioreactors is that
they allow oxygen mass transfer at considerably lower power input levels and, as they
most often contain no moving parts, that their requirements concerning investment and
SA NE
maintenance costs are lower. Bubble columns and airlift loop reactors can be built with
much larger working volumes than stirred tank reactors. Their advantages become
effective predominantly at scales larger than 50-100 m³. Here we focus the attention on
U
the dynamics of bubble columns and explain the fluid mixing in the turbulent flows in
these multiphase reactors as well as reasons for their mass and heat transfer
performance. The text is illustrated with a number of real examples from large-scale
industrial applications.
1. Introduction
Bioreactors are containments used to provide microbial, animal, or plant cells the
environment for optimally producing the desired products. Experiences gained over
many decades showed the cells‟ environment can best be kept under control when they
are suspended in a liquid culture medium from which they receive all their nutrients.
Particularly, temperature, pH, and the nutrient concentrations can easily be adjusted in
a continuous liquid phase. Hence, in most bioreactors the cells are suspended in liquid
media.
The only important component that cannot be fed to such a culture at an arbitrarily
high rate is the oxygen that the cells need in most production processes. In nearly all
practical cases oxygen is supplied by bubbling air through the liquid. In order to keep
the cross section for the transfer of oxygen from the gas bubbles into the liquid phase
S
high enough, a high gas throughput is necessary, preferentially using very small air
R
bubbles.
AP S
TE
H S
In the classic catalytic chemistry, production processes are characterized by high
space-time yields, i.e. in a reactor with a given culture volume, the formation of much
C OL
product mass per production time. Chemists, however, often produce under high
pressures and temperatures, frequently they need organic solvents, and many times
they form toxic byproducts. Biotech production processes on the other hand use
E -E
extremely specific catalysts, in most cases the enzymes active in living cells, and they
work under comparably very mild conditions. However, the space-time-yields, are
often considerably smaller. Hence, large reaction volumes and reactor sizes are
PL O
Since optimal conditions must be provided to all cells, homogenization of the culture
medium is essential. In practice one tries to achieve this by intensively mixing the
culture. This is the reason why most bioreactors from the smallest up to the production
SA NE
scale of 100 m³ and even larger are designed as stirred tank bioreactors. There an
agitation system, consisting of one or more impellers mounted on a single central shaft
driven by an electromotor, is used to induce turbulent motions in the culture in order to
U
support both mixing and oxygen mass transfer. A clear advantage of aerated stirred
tank reactors is that the power needed to maintain the turbulent motion in the culture
can be adjusted independently (i) by the agitator and (ii) by means of the gas phase that
is pressed into the culture by an air compressor.
Unfortunately, stirring at high power levels not only leads to high running costs but, at
large scale equipment, also to high investment and maintenance costs. At culture
volumes above 100 m³, where predominantly cheap bulk products are being produced,
the energy cost factors have an important influence on the process economics. In some
practical cases, the stirrer is thus shut down during cultivation phases where the broth
viscosity is low and not too high oxygen transfer rates are required, in order to reduce
the operation cost. Consequently, wherever possible in bulk product manufacturing,
impellers are not installed at all and the tanks are operated as so-called bubble
columns. Bubble columns find wide-spread applications, for instance in baker‟s yeast
production and even in systems that cultivate Aspergillus niger, a mold used, e.g., for
the production of citric acid. At reactor dimensions larger than about 200-400 [m³] one
will seldom find standard stirred tank reactors. Conversely, at smaller production
scales lower than 50 [m³] one practically will not find bubble columns. In this article
we will thus discuss bubble columns as alternatives to large aerated stirred tank
bioreactors. We will not only mention the standard bubble columns but also airlift loop
reactors that are variants which are obtained by modifications of the vessel geometry.
Many excellent professional surveys appeared in recent years on this topic (e.g.
Heijnen & van‟t Riet 1984, Deckwer 1985, Chisti 1989, Schügerl 1997, Merchuk &
Garcia Camacho 2010). Here we do not aim in copying them. Our focus is placed on
insights into the dynamics of these reactors and not on the flood of engineering
correlations derived for them.
S
R
Bubble columns are usually constructed as cylindrical vessels initially filled with
AP S
liquid medium to which gas bubbles are continuously supplied at their bottoms. Apart
TE
H S
from their length to diameter ratios, the only important construction detail is the gas-
sparger forming the bubbles. The important point for understanding bubble columns is
C OL
that the flow in these vessels is density-driven. Sparging bubbles in the liquid lowers
the density of the culture locally which leads to convective flows within the entire
column.
E -E
PL O
M SC
SA NE
U
As depicted in Figure 1, the bubbles rise in the culture while exchanging volatile
components with the liquid phase. The sizes of the bubbles are usually in the order of
some millimeters and hence much larger than suspended microorganisms which are of
diameters in the micron-range. The cells are even smaller than the smallest eddies in
the turbulent liquid flow, which are in the order of magnitude of 0.1 mm. From the
fluid mechanical point of view, the bubble column bioreactor can thus be considered a
two-phase gas-liquid flow system.
It must be stressed that the turbulent fluid motions needed in bioreactors for mixing the
reaction system are energetically only supplied by the gas phase. From that point of
view the gas flow rate or gas throughput vg is the important quantity. Chemical and
biochemical reaction engineers, however, prefer a slightly different key operational
variable for bubble columns, namely the superficial gas velocity us defined as
vg
us [m/s] (1)
A
where A [m²] is the cross sectional area of the column. The reason for this choice is
that (i) nearly all important characteristics of the bubble column reactor are simple
S
functions of the superficial gas velocity and (ii) it is directly proportional to the (scale-
R
independent) specific power Paer V input into the column:
AP S
TE
H S
1 Paer
us [m/s] (2)
g V
C OL
[kg/m³] is the liquid density, g [m/s²] the gravitational acceleration, Paer W the
E -E
power input into the dispersion by the air compressor and V [m³] the culture volume.
Not only g is constant, for a given system, is considered practically constant as
PL O
well.
M SC
At low superficial gas velocities or air throughputs, the bubble column depicts a quite
homogeneous picture, which was termed the bubbly flow regime. The bubble sizes do
not vary very much. With increasing us the flow becomes more and more turbulent
SA NE
and the bubble size distributions become broader. At high superficial gas velocities one
gets a quite heterogeneous flow picture and then one speaks of the heterogeneous flow
regime, often also referred to as the churn turbulent phase. The transition between both
phases appears at about us 6 [cm s-1]. Figure 2 gives an impression of what is meant
U
Noteworthy, turbulence may also appear in the bubbly flow regime. Whether or not
turbulence appears depends on the Reynolds number, Re , in fluid flow systems and in
the liquids of lower viscosity present in most big production reactors, Re usually well
exceeds its critical value for the appearance of turbulence even at quite low flow
velocities which appear at superficial gas velocities which may be an order of
magnitude smaller than 6 cm/s. Hence, the very big bubble columns need not be
operated at us 6 cm in order to obtain good turbulent mixing.
S
R
AP S
TE
H S
C OL
E -E
PL O
large bubble columns: Left “bubbly flow”, Right “churn turbulent flow” (Lübbert
2008)
SA NE
0.6us0.7
(3)
Where us us uus , with the unit velocity uus 1 [m s-1], a quantity which is needed
for dimensional correctness. This correlation gives a quite a coarse estimate but at least
a hint to the fact that rises sub-proportionally with us .
The main construction parameter of large bubble columns is the aspect ratio, i.e. the
height H of the column in units of its tank diameter T . In practice both types of
columns can be found, large aspect ratio columns and small aspect ratio columns. The
difference in their behavior, when they are employed as bioreactors, will be discussed
later.
Figure 3 shows a typical example of slender bubble columns. These columns are
operated at Citro-Misr Co. in Ramadan City, Egypt for citric acid production. In recent
years, the citric acid industry replaced nearly all aerated stirred tank reactors that were
formerly used by big bubble columns. Figure 3 shows the 600 [m³] reactors during the
construction phase. The columns are about 30 [m] in height and 5 [m] in diameter. The
plant was designed to produce 36.5 [t] citric acid monohydrate (CAM) per day.
S
R
AP S
TE
H S
C OL
E -E
PL O
M SC
SA NE
U
Figure 3. Citric acid production reactors during their erection (Photo Courtesy of
Vogelbusch Vienna)
3. Mass Transfer
One of the most important issues in bioprocess reaction engineering is, as already
mentioned in the introduction, the gas-liquid mass transfer. The oxygen mass flux
passes several steps from the interior of the air bubble via the physical gas-liquid
interfacial area towards the bulk of the liquid phase. One of these steps, the transport
across the liquid side boundary layer at the gas-liquid interface, makes up by far the
biggest influence on the transfer rates, hence it proved to be sufficient considering only
this one.
The physical mechanism that dictates the flux across this resistance is molecular
diffusion. Fick‟s first law thus applies. The simple graph in Figure 4 helps explaining
the various quantities involved in this gas-liquid mass transfer law. The diffusive
oxygen flux is driven by the concentration difference O O* O across the
boundary layer, where O* [mol/m³] is the solubility of oxygen in water, which is
determined by Henry‟s law, and O the oxygen concentration dissolved in the bulk
fluid. The volume-related oxygen transfer rate, OTR , in the case that no biochemical
reactions influences the transfer, is expressed by the simple transport law
S
The transport resistance is R 1 k1a D / a , where k1 is the mass transfer
R
AP S
coefficient, the strength of the boundary layer, and D the diffusion coefficient of
TE
H S
oxygen in water. As k1 and a most often cannot be determined separately, one focuses
C OL
the attention to the product k1a and takes it as the key transport coefficient.
E -E
PL O
M SC
SA NE
U
k1a mainly depends on the superficial gas velocity us and to a smaller extend on the
broth viscosity .
kl a Kus 1 h (5)
The general behavior of the oxygen mass transfer as described by these equations is
not doubted; however, the parameters K , , and are neither independent of the
scale nor of the position in the reactor. In big reactors, the assumption of a constant O*
along the axis of the column cannot be justified as the bubbles lose oxygen when they
rise through a tall reactor and the hydrostatic pressure becomes smaller on their way.
This leads to an expansion of the bubble volume and thus to an increase of the
transport cross section for the gas-liquid mass transfer. Simple models assume a
logarithmic decay of the oxygen volume fraction in the bubbles and take the
logarithmic mean for the corresponding O* values determined with Henry‟s law. This
however needs additional information about the oxygen volume fraction in the bubbles
when they disengage from the culture. Another point is the dependency of k1 from the
interfacial properties. Much disputation about that question can be found in the open
literature. van‟t Riet and Tramper (1991) state that most data support Higbie‟s
approach (1935), where k1 is dependent on the square root of the diffusion constant of
the solute in the solvent, here Dow , the diffusion coefficient of oxygen in water:
4Dow uB
kl (6)
S
pd B
R
AP S
Further key influence parameters are the representative bubble diameter d B and bubble
TE
H S
rise velocity u B as well as the total pressure p .
C OL
With respect to the gas-liquid mass transfer, the second important factor of the product
k1a is the specific interfacial area a [m² m-3], i.e. the transport cross section presented
E -E
by the bubbles per culture volume. Bubble diameter d B and thus, a, may change due to
coalescence and redispersion processes. Most often not sufficiently recognized is the
coalescence behavior of the bubbles in the two-phase flow system. Usually, culture
PL O
broths are coalescence repressing (!) two phase systems. We thus must often add
antifoam agents to promote coalescence in order to avoid foam formation! The
M SC
practical consequence is that in bubble columns the bubble size is less influenced by
secondary coalescence/redispersion processes, e.g., by eddies in the turbulent flow.
Hence, as opposed to aerated stirred tank bioreactors, the primary gas sparger is of
SA NE
Spargers are often constructed as big wheels made from stainless steel tubes. The air
U
supply goes to the hub and is dispersed into the surrounding liquid through small bores
distributed across the entire tube system. Generally, the aim is generating small
bubbles at the bottom of the reactors. The bubble sizes depend on the pressure
difference p between gas and liquid and the bore diameter, d h , the equivalent bubble
diameter d B is known to become smaller with smaller bores and higher p . Small
hole-diameters at the sparger should thus be of advantage. However, in biotechnology
practice, the hole must not be too small, since then practical problems with clogging
may lead to severe disadvantages. Hence, in practice one often finds simple sparger
constructions based on ring tubes or tree-like tube arrangements with bores in the
millimeter range.
In order to enhance the specific interfacial area, a, several construction details were
considered. The first was to reduce the bore diameters and increase the air pressure
within the sparger in order to form small gas jets from each bore. This is shown in
Figure 5, where the gas is pressed in form of jets through small orifices at the end of
the spokes. It depicts the jet sparger for the big bubble column reactors such as the one
shown in Figure 3. These jets quickly decay to a series of small bubbles.
S
R
AP S
TE
H S
C OL
Figure 5. Examples of gas spargers for large citric acid production bioreactors
(Courtesy of Vogelbusch, Vienna).
E -E
nozzles that supersonic conditions appeared at the nozzle exits. In this way very small
bubbles were generated. Such an arrangement is shown in Figure 6. Then, however, it
M SC
must be guaranteed that air is continuously pressed at high pressure levels through
these nozzles in order to avoid clogging.
SA NE
U
Figure 6. Dispersion by a supersonic air jet expanding into the liquid phase. Right the
geometry of the nozzle arrangement showing that there are seven narrow bores in the
stainless steel nozzle screwed onto an air pressure tube (Courtesy of DSM).
Particularly in biological wastewater plants where the aeration cost is dominant, many
alternative air dispersion systems where investigated and used in real practice. All of
them were led by the idea that the total power needed to finely disperse the air must be
reduced. So two-phase flow arrangements were constructed where the liquid phase and
the gas phase where brought together within narrow momentum exchange spaces from
where the dispersions are then expanded into the culture broth. An important example
is the development of Bayer AG Leverkusen in the 1980s. This is shown in Figure 7.
S
R
AP S
TE
H S
C OL
Figure 7. Momentum exchange two-phase gas liquid aeration nozzles developed by
Bayer in the 1970s (Courtesy of Bayer AG)
E -E
Here, as the Figure shows, energy supplied with the liquid flow is used to cut down the
gas fluid elements into many small bubbles. On the right of the Figure it is shown that
PL O
4 such two-phase nozzles are arranged in one aeration block. Many such blocks were
installed on the flat bottom of a waste water treatment tower. Similar systems were
M SC
developed at all big chemical companies at that time, e.g. Hoechst AG, BASF AG, ICI
Inc. etc. Their common idea was using an auxiliary local (!) energy source to support
air dispersion.
SA NE
Further alternatives are widely distributed in the yeast manufacturing industries. These
are mechanically supported air dispersion systems. An example is shown in Figure 8.
The aerator looks like an impeller with 13 long narrow blades. They rotate and are
U
driven from below. What looks like an impeller shaft is a tube that allows sucking air
from the reactor‟s head space to the low pressure regions behind the rotating hollow
blades of this sparger. The horizontal metal bars above this dynamic sparger are flow
baffles that prevent rotation of the entire liquid in the column. Again, this is a
dynamically aerated bubble column, not a stirred tank reactor. The arrangement shown
serves for both, aeration (sucking air from the head space) and gas dispersion.
-
-
-
Bibliography
Bowker, R.P.G., (1982), Technology Assessment of the Deep Shaft Biological Reactor, Wastewater
Research Division, Municipal Environmental Research Laboratory, US Environmental Protection
Agency Cincinnati, OH [Deep shaft reactors applied to biological wastewater treatment]
S
Chisti, Y. (1989) Airlift bioreactors, Elsevier Applied Science, London [Basic book on airlift reactors]
R
AP S
Deckwer, W.D. (1985), Reaktionstechnik in Blasensäulen, Salle+Sauerländer, Frankfurt/Aarau
[Standard book on bubble column reactors from the reaction engineering point-of-view]
TE
H S
Devanathan, N., Dudukovic, M., Lapin, A., Lübbert, A. (1995), Chaotic Flow in Bubble Column
Reactors, Chem.Eng.Sci., 50, 2661-2667 [Experimental approach to turbulent motions in bubble column
C OL
reactors]
Freedman, J.F., Davidson, J.F. (1969), Hold-up and liquid circulation in bubble columns. Trans.
IChemE 47, 251-262 [Basic reference for the mean flow pattern in bubble columns]
E -E
George, S., Larsson, G., Olsson, K., Enfors, S.O. (1998), Comparison of the Baker‟s yeast process
performance in laboratory, and production scale, Bioprocess Engineering, 18, 135-142 [Biotechnology
of baker‟s yeast production]
PL O
Giovannettone, J.P., Gulliver, J.S. (2008), Gas Transfer and Liquid Dispersion Inside a Deep Airlift
M SC
Reactor, AIChE Journal, 54, 850-861 [Good reference showing the mixing behavior of long airlift
reactors with an external loop geometry]
Godo, S., Junghans, K., Lapin, A., Lübbert, A. (2004), Dynamics of the flow in bubble columns, pp.53-
66 in: Bubbly Flows, Analysis, Modelling and Calculation, Sommerfeld, M., ed., Springer, Berlin, ISBN
SA NE
ICI (1965), The ICI Deep Shaft Effluent Treatment Process, ICI Limited, Cleveland [Description of the
deep shaft reactor by the developers]
Karaffa, L., Kubicek, C.P. (2003), Aspergillus niger citric acid accumulation: do we understand this well
working black box?, Appl Microbiol Biotechnol, 61, 189-196
Kubicek CP, Röhr M. (1989), Citric acid fermentation. Crit.Rev.Biotechnol., 4, 331–373 [Basic review
of modern citric acid production]
Kubicek, C.P., Karaffa, L. (2006), Organic acids, Chap.15 in Basic Biotechnology, 3rd ed., Rutledge, C.,
Kristiansen, B., eds., Cambridge University Press, Cambridge
Lapin, A., Lübbert, A. (1994), Numerical simulation of the dynamics of two-phase gas-liquid flows in
bubble columns, Chem.Eng.Sci., 49, 3661-3674 [Computational fluid dynamics for bubble columns
based on an Euler-Lagrange methodology]
Lübbert, A., Paaschen, T., Lapin, A. (1996), Fluid dynamics in bubble column bioreactors: Experiments
and numerical simulations, Biotechnol.&Bioeng.,52, 248-258
Lapin, A., Paaschen, T., Junghans, K., Lübbert, A. (2002), Bubble column fluid dynamics: Flow
S
structures in slender columns with large-diameter ring-spargers, Chem.Eng.Sci., 57, 1419-1424
Mandenius, C.F., Eklöv, T., Lundström, I. (1997), Sensor fusion with on-line gas emission multisensor
R
AP S
arrays and standard process measuring devices in baker‟s yeast manufacturing process,
Biotechnol.Bioeng., 55, 427-438
TE
H S
Merchuk, J.C., Garcia Camacho, F. (2010), Bioreactors, airlift reactors, pp.851-912 in: Encyclopedia of
industrial biotechnology, Wiley [Review on airlift loop reactors]
C OL
Rice, R. G., Barbe, D. T., & Geary, N. W. (1990). Correlation of nonvertically and entrance effects in
bubble columns. A.I.Ch.E. Journal, 36, 1421–1424 [This paper addresses the influence of the vertical
alignment of a slender bubble column]
E -E
Roels, J.A. (1983), Energetics and kinetics in biotechnology, Elsevier Biomedical, Amsterdam
Schügerl, K. (1997), Three-phase-biofluidization: Application of three-phase fluidization in the
PL O
van‟t Riet, K., Tramper, J. (1991), Basic Bioreactor Design, Marcel Dekker, New York [Basic textbook
for bio-engineers]
Wiebe, M. (2002), Myco-protein from Fusarium venenatum: a well-established product for human
U
Biography Sketch
Andreas Lübbert studied Physics at the Technical University Hannover, Germany. His doctoral theses
he wrote at the same University under the guidance of Professor Karl Schügerl about elementary
chemical reactions. He received his venia legendi for the subject „Technical Chemistry‟ at the University
S
R
AP S
TE
H S
C OL
E -E
PL O
M SC
SA NE
U