You are on page 1of 5

Bioprocess Engineering 8 (1992) 49-53

Bi0pr0cessEngineering
9 Springer-Verlag 1992

Improved scale-up strategies of bioreactors


L.-K. Ju and G.G. Chase, Akron, U S A

Abstract. Effective scale-up is essential for successful bioprocessing.


While it is desirable to keep as many operating parameters constant
as possible during the scale-up, the number of constant parameters
realizable is limited by the degrees of freedom in designing the
large-scale operation. Scale-up of aerobic fermentations is often
carried out on the basis of a constant oxygen transfer coefficient,
kL a, to ensure the same oxygen supply rate to support normal
growth and metabolism of the desired high cell populations. In this
paper, it is proposed to replace the scale-up criterion of constant k L
by a more direct and meaningful criterion of equal oxygen transfer
rate at a predetermined value of dissolved oxygen concentration.
This can be achieved by using different oxygen partial pressures in
the influent gas streams for different scales of operation. One more
degree of freedom, i.e., gas-phase oxygen partial pressure, is thus
added to the process of scale-up. Accordingly, one more operating
factor can be maintained constant during scale-up. It can be used to
regulate the power consumption in large-scale fermentors for economical considerations or to describe the fluid mixing more precisely. Examples are given to show that the results of optimization
achieved in the bench-scale study can be translated to the production-scale fermentor more successfully with only a small change in
the gas-phase oxygen partial pressure employed in the bench-scale
operation.

kg/m" s
kg/m 3

Liquid viscosity
Liquid density

1 Introduction
Bioprocesses are usually developed in three stages or scales:
(1) Bench or laboratory scale, where basic screening procedures are carried out; (2) pilot plant, where the optimal
operating conditions are ascertained; and (3) plant scale,
where the process is brought to economic fruition. Scale-up
means reproducing in plant-scale equipment the results from
a successful fermentation made in laboratory- or pilot-scale
equipment [1]. The scale-up process thus directly influences
the production capacity and efficiency of a bioprocess.

2 Criteria of scale-up
List of symbols
a

m2/m 3

Cz

mole/m 3

C*

mole/m 3

Di

DT

HL

kL
N

mole/m 3 - atm
m/s
1/s

N,

OTR

mole/s - m 3

Pg
Po
pg
Q

kW
kW
atm
m3/s

Re i

TQ
V
vs

Joule
m3
m/s

Specific gas/liquid interfacial area


Dissolved oxygen concentration in bulk
liquid phase
Equilibrium oxygen concentration at gas/
liquid interface
Impeller diameter
Bioreactor diameter
Henry's-law constant
Liquid-phase mass transfer coefficient
Impeller agitation speed
Number of impellers
Oxygen transfer rate per unit volume of
the medium
Power input in aerated fermentation
Power input in non-gassed fermentation
Gas-phase oxygen partial pressure
Volumetric gas flow rate
Impeller Reynolds number
Torque applied to the mixer shaft
Liquid volume
Superficial gas velocity

Fermentations in general can be evaluated by the following


physical characteristics: Mixing time, shear, pH, heat transfer and mass transfer. Although the physical characteristics
are inter-related, in scale-up and reactor design the mixing
time, shear and mass transfer aspects are coupled while pH
and heat transfer [2] are considered separately.
Unlike mass transfer which takes place throughout the
broth, heat transfer occurs only at the boundary surfaces of
heat exchange. It is possible to provide the required heat
transfer capacity of a large-scale fermentor by methods
independent of process scale-up, such as by using a refrigerant rather than cooling water or by using an external heat
exchanger. Consequently, equal heat transfer capacity traditionally has not been used as a basis for translating experimental results between different scales of operation. Similarly, proper p H control can be achieved independently by
automatic addition of concentrate acids and bases in fermentation systems with an adequate dispersing mechanism.
Therefore, pH does not constrain scale-up.

50

Bioprocess Engineering 8 (1992)

The process characteristics which have been suggested to


be maintained constant during scale-up include [3-13]:
1.
2.
3.
4.
5.
6.
7.
8.
9.

Reactor geometry;
volumetric oxygen transfer coefficient, k L a;
maximum shear;
power input per unit volume of liquid, Po/ld
volumetric gas flow rate per unit volume of liquid,
O / v or V V M ;
superficial gas velocity, v~;
mixing time;
impeller Reynolds number, Rei= ~ ND~/#, and
momentum factor.

To combat the problems of poor mixing in large fermentors, especially with viscous non-Newtonian systems, Fox
and Gex [4] proposed to employ equal mixing times in fermentors of different" scales. Although important for fast
chemical reactions, the requirement of constant mixing time
is unnecessary for normal fermentations involving slow biological reactions. Furthermore, it has been shown that the
volumetric power input required to maintain equal mixing
time increases as the 2/3-th power of the system volume [5]
and can become prohibitive on scale-up. Experience proves
that the power input thus determined is much more than
required for practical scale-up. A reduction of the volumetric
power input with the increasing system scale in actual practice has been confirmed in a survey of European fermentation plants [6]. Criteria 8 and 9, suggested by Rushton [7]
and Blakebrough and Sambamurthy [8] respectively, have
not generally worked for fermentations. One main reason is
that no consideration is given to the effect of aeration on the
process results. Consequently, only criteria 1 - 6 are commonly considered for scale-up of bioreactors.
Criterion 1 is based on the fact that almost all of the
existent empirical or semi-empirical correlations for scale-up
are developed experimentally with geometrically similar reactors of different scales. When translating experimental results between reactors with different geometries, one needs
to consider the validity of correlations chosen from the literature and the necessity of developing or modifying the correlations for the special situation. For nongeometric scale-up,
a range of DI/DT, the ratio of impeller diameter to fermentor
diameter, capable of delivering acceptably good gas dispersion has been suggested by Oldshue [14].
The ultimate performance of most aerobic fermentations
is often oxygen limited. To ensure equal oxygen transfer rates
at various scales of operation, scale-up of aerobic fermentations is most commonly conducted with the criterion of
constant k L a. Criterion 3 is of critical importance to fermentations of shear-sensitive organisms. For example, mycelial
fermentations have been found to be affected by shear or the
absolute value of the impeller tip speed [9].
Equal power per unit liquid volume (Po/V) has been used
in many antibiotic fermentations as the primary scale-up
parameter. Typically, a Po/V value of 1.0 to 2.0 kW/m 3 is
used [10, 11]. It has been found to be the most successful

scale-up rule for mixing in shear-sensitive operations, such


as gas dispersion in liquids which is one of the most important characteristics of aerobic fermentations [5, 12].
Criteria 5 and 6 address the importance of aeration rate
to scale-up of aerobic fermentations. Volumetric gas flow
rate per unit liquid volume (VVM), due to its influence on
mass transfer in bioreactors subjected to bubble aeration
without mechanical agitation [10], has been accepted as a
traditional scale-up criterion even for mechanically agitated
vessels [3]. On the other hand, the superficial gas velocity is
also an important factor in bioreactor design. It strongly
affects the mixing energy required to disperse the gas stream
and, in all cases, has an upper limit before impeller overloading or liquid blow-out action begins [13]. The upper limit of
vs is strongly system-dependent. According to Cooper et al.
[15], overloading occurs at vs>2.5 x 10 -2 m/s for vaneddisk impellers. When overloaded, the impeller cannot disperse all the gas supplied, short-circuiting occurs, the gas
rises as big bubbles to the surface and the pumping action of
the impeller diminishes.
Unfortunately, criteria 5 and 6 are contradictory to each
other when applied to scale-up of geometrically similar
bioreactors. If it is desirable to maintain equal V V M , then
the superficial gas velocity through the vessel will increase
directly with the scale ratio and, consequently, may lead to
impeller overloading or liquid blow-out. Therefore, the
choice and/or compromise between criteria 5 and 6 have to
be carefully weighed for the individual process involved.

3 Conventional scale-up strategies


While all these criteria are important, it has been pointed out
that when a particular scale-up strategy is carried out by
maintaining a specific set of parameters constant, o t h e r
parameters cannot be controlled and may change substantially in unexpected ways [14, 16]. This may have undesired
effects on the yield, because so many factors affect microbial
growth and/or product formation.
Obviously, one should first analyze the bioprocess and
list the physical characteristics desired to be kept constant
during scale-up in the order of criticality to the performance
of the bioprocess. The number of realizable factors, however,
is limited by the degrees of freedom available in process
scale-up. Three degrees of freedom have generally been recognized. They are: N, the impeller speed; Q, the volumetric
gas flow rate; and Di/D T, the ratio of impeller diameter to
fermentor diameter (or, in a broader sense, the reactor geometry). Accordingly, the maximum number of criteria that can
be maintained constant in a conventional scale-up strategy
is three. For example, the following combinations of scale-up
criteria have been suggested in the literature:
Combination I [~]:
- Geometric similarity (or c o n s t a n t DI/DT) , constant k L a,
and constant Q/V (or VVM);
- N determined by the k L a correlation.

L.-K. Ju and G.G. Chase: Improved scale-up strategies of bioreactors


Combination 2 [1]:
- Geometric similarity, constant ICEa, and constant maximum shear (or constant impeller tip speed NDi);

- Q calculated from the kz a correlation.


Combination 3 [3]:
- Constant kL a, constant impeller tip speed N D i , and constant Q/V;
- D t / D T adjusted within the limits suggested by Oldshue
[141.
While other combinations can be generated from the previous list of scale-up criteria, it is apparent that constant k L a
has been included in all the above approaches reported in
the literature. However, sacrifices have been made in other
important design factors such as geometric similarity, shear
sensitivity of organisms, and superficial gas velocity in view
of impeller overloading or liquid blow-out. Furthermore,
power input, which strongly influences the operating cost
and mixing of the fermentor, has not been seriously considered.

4 Improved scale-up strategies for aerobic bioreactors

The reason that strong emphasis has been put on maintaining constant volumetric oxygen transfer coefficient, k z a, in
process scale-up is to ensure the same oxygen supply rate to
satisfy the oxygen demand of the desired cell population.
However, the oxygen transfer rate in a typical submerged
aerobic fermentor is actually the product of k z a and
(C*-CL); the latter represents the difference between the
equilibrium concentration at gas/liquid interface and the
dissolved oxygen concentration in bulk liquid phase. The
maximum oxygen transfer rate per unit volume (OTRmax) is
achieved when C L-~ 0: all oxygen entering the bulk solution
is rapidly consumed, i.e.,
OTRma x = k L a C*

(1)

The effects of dissolved oxygen concentration, CL, on cell


growth, metabolism and product formation have been investigated extensively in the literature for various organisms
[17]. In general, above a certain oxygen concentration, called
the "critical oxygen concentration", the cell metabolic machinery is saturated with oxygen. In this case, sufficient oxygen is available to accept immediately all electron pairs
which pass through the respiratory chain, so that some other
biochemical process within the cell is rate-limiting. Otherwise, even temporary depletion of dissolved oxygen in aerobic bioprocesses could mean irreversible cell damage. In this
respect, it is worth noting that the same microbial species
may show large variations in its oxygen requirements, depending on the oxygen concentration to which it has been
adapted [18]. In some cases, such as tryptophan [19] and
L-glutamic acid [20] productions, optimal results are
achieved when a constant oxygen concentration can be
maintained throughout the course of fermentation.

51
Instead of the constant oxygen transfer coefficient (k z a),
a more meaningful scale-up criterion for aerobic fermentations is the constant oxygen transfer rate per unit volume at
a predetermined value of CL. For many microbial fermentations, this value of CL can be reasonably set as zero because
of the very low critical oxygen concentrations of the microorganisms such as yeast, Escherichia coli, Pseudomonas
denitrificans, etc. [17]. Equal maximum oxygen transfer rate
is the scale-up criterion suggested for these fermentations. In
other cases, the optimal values of CL have to be determined
in the bench-scale study.
In this work, improved scale-up strategies for aerobic
fermentations are proposed to be developed by keeping constant oxygen transfer rate, at the predetermined value of CL,
in different scales of operation through the use of different
oxygen partial pressures in influent gas streams. The gasphase oxygen partial pressure, pg, regulates C* via Henry's
law, i.e.,
(2)

C* = H L pg ,

where H L is Henry's-law constant of the liquid medium.


Besides being more meaningful, this approach adds one
more degree of freedom to the design of fermentations. Accordingly, one more operation factor can be maintained
constant in scale-up. It can be one of the sacrificed factors in
the above-mentioned approaches, or one of the following
criteria suggested in the literature for better control of the
power input and mixing in large fermentors:
1. The power input given by Einsele [6], based on data
obtained in a survey of European fermentation plants having volumes from 0.5 to 300 m 3, i.e.,
PJV~

V -~

(3)

2. Constant power input per unit liquid volume, for fluid


mixing in shear-sensitive operations such as gas dispersion
in normal aerobic fermentations [5].
3. Constant torque intensity, i.e., Te/V,, for fluid mixing in
flow-sensitive operations such as fermentations with high
viscosity and/or dense solid suspensions [21].
Two examples are given in the Appendix to demonstrate
the determination of gas-phase oxygen partial pressures to
be used in the bench-scale fermentations for the scale-up
from a 10-1 vessel to a 10,000-1 fermentor. Four criteria, i.e.,
geometric similarity, constant impeller tip speed, power input suggested by Einsele [6], and constant maximum oxygen
transfer rate, are employed in the first example. In the second
example, the scale-up is conducted with the criterion of constant Q / V replacing the Einsele correlation for power input.
Compared to the strategies suggested in the literature, i.e.,
Combinations 1-3 given in the previous section, the first
example offers better control of power input in the largescale processes; the second example represents a direct improvement over Combination 3 by carrying out the scale-up
in geometrically similar fermentors and avoid the uncertain-

52

Bioprocess Engineering 8 (1992)

ties in applying literature correlations to reactors with different geometries.


It is shown that lower gas-phase oxygen partial pressures
(i.e., 0.19 a n d 0.15 atm for Examples I and 2, respectively)
should be used in the l a b o r a t o r y experiments. In batch
fermentations, cells are exposed to varying oxygen tensions
in m e d i a anyway. The slight reduction of gas-phase oxygen
partial pressure in the bench-scale fermentation is not expected to cause any significant microbiological effects such
as changes in cell metabolism a n d morphology. Therefore,
the i m p r o v e d scale-up strategies reported in the w o r k will
lead to m o r e precise translation of the optimal results
achieved in bench-scale experiments to production-scale
processes.

Q
]Q
~s=(N/~
l\Ns,12'5 (~z_~)Dt,23.21(gssV/)-2"5 = 103"21 "
The ratio of superficial gas velocity is thus:
v.,., =

v~,, \ Q , ] \ D , , J
(Note: vs.~should also be evaluated to determine whether it is so high
as to cause impeller overloading or liquid blow-out in the large-scale
fermentor.) Accordingly, the ratio of gas-phase oxygen partial pressure, given in Eq. (5), can be calculated as:

P~
Po,s

( t 0 - i " 1 1 ) -0"77 (t01"21) -0'67 = 1.10.

While the use of oxygen-enriched gas in production plants may not


be economically favorable, the laboratory-scale fermentations can
be easily made with nitrogen-diluted air having oxygen partial pressure of 0.19 (=0.21/1.10) atm.

Appendix
Determination of the ratio of gas-phase oxygen partial pressures
employed in different scales of fermentation for successful scale-up.
Example 1
It is assumed desirable to scale up a bioprocess from results obtained in a 10-1 vessel to a 10,000-1 fermentor based on the following
combination of scale-up criteria:
1. Geometric similarity, i.e.,

D~,t DTI
=

Rearranging Eqs. (4), (6) and (7) leads to:

The previous scale-up from a 10-1 fermentor to a t0,000-1 fermentor


is to be carried out with the criterion of constant Q/V replacing the
Einsele correlation for power input.
With constant Q/V,

Q, = __v, =
Qs v~

1o

and

(V/~ 1/3
=

Example 2

(Q, (o,42=1o

=1o,

v,,~ \ Q J \O,,,]
where subscripts l and s represents large- and small-scale, respectively.

Combining Eqs. (6) and (7) leads to

2. Constant impeller tip speed, i.e.,

Di's

0.1 .

Ns D~,~
3. Empirical power input per unit volume suggested by Einsele [6],
i.e.,

(Pg/V)I (V/~- ~
(e./v)~, = \ v d

= 10-1"11 .

(4)
= (101"944 1 0 - 3 ) -0.77 (10) -0.67 = 1 . 3 9 .

4. Constant maximum oxygen transfer rate, i.e.,


--0.77

(kL a)s
p,,,

(kL a),

-0.67

L(v./v)J

'

(5)

where the kL a correlation proposed by Fukuda et al. [22, 23] for


low-viscosity fermentations under turbulent mixing, i.e.,
k L a = (2.0+2.8 Ni) (Po/V)~

(vs)~

/p,o2 ND~\ ~
where Po is the nongassed power input and can be estimated from
the following relationship for geometrically similar vessels [7]:
when

Rei > 104 9

Consequently, to successfully translate ez~perimental results to the


production process made with air, the gas-phase oxygen partial
pressure of 0.15 (= 0.21/1.39) atm should be used in the laboratoryscale fermentation.

References

has been used. v~ is the superficial gas velocity.


Michel and Miller [24] showed that for Newtonian fluids a good
estimate of Po can be obtained from the relationship:

Po ~ N3 D~

where NJN~=0.1 on the basis of constant impeller tip speed. The


ratio of gas-phase oxygen partial pressure between the two scales,
which is required to maintain a constant maximum oxygen transfer
rate, can then be calculated by using Eq. (5):

(7)

i. Hubbard, D. W.: Scale-up strategies for bioreactors. In: Ho,


C. S.; Oldshue, J. Y. (Eds.): Biotechnology processes. Scale-up
and mixing, pp. 168-184. New York: AIChE 1987
2. Cooney, C. L.: Bioreactors: Design and operation. Science 219
(1983) 728-733
3. Wang, D. I. C.; Cooney, C. L.; Demain, A. L.; Dunnill, P.;
Humphrey, A. E.; Lilly, M. D.: Fermentation and enzyme technology, pp. 194-211. New York: John Wiley & Sons, Inc. 1979
4. Fox, F. A.; Gex, V. E.: Single-phase blending of liquids. AIChE
Journal 2 (1956) 539-544

L.-K. Ju and G.G. Chase: Improved scale-up strategies of bioreactors


5. Uhl, V. W.; Von Essen, J. A.: Scale-up of fluid mixing equipment.
In: Uhl, V. W.; Gray, J. B. (Eds.): Mixing: Theory and practice,
Vol. III, pp. 155-167. New York: Academic Press 1986
6. Einsele, A.: Scaling of bioreactors, theory and reality. Paper 4.13
presented at the 5th International Fermentation Symposium,
Berlin 1976
7. Rushton, J. H.: Mixing - present theory and practice. Chem.
Eng. Prog. 49 (1953) 161-168 and 267 275
8. Blakebrough, N.; Sambamurthy, K.: Mass transfer and mixing
rates in fermentation vessels. Biotechnol. Bioeng. 8 (1966) 25- 32
9. Steel, R.; Maxon, W. D.: Dissolved oxygen measurements in
pilot and production-scale novobiocin fermentations. Biotechnol. Bioeng. 4 (1962) 231-240
10. Aiba, S.; Humphrey, A. E.; Milfis, N. E: Biochemical Engineering, 2nd edition, pp. 195-217. New York: Academic Press 1973
11. Gaden, E. L : Aeration and agitation in fermentation. Sci. Rep.
Ist. Super. Sanita 1 (1961) 161
12. Penney, W. R.: Recent trends in mixing equipment. Chem. Eng.
78(7) (1971) 86-98
13. Moo-Young, M.; Blanch, H. W.: Design of biochemical reactors.
Mass transfer criteria for simple and complex systems. In:
Ghose, T. K.; Fiechter, A.; Blakebrough, N. (Eds.): Advances
in biochemical engineering, Vol. 19, pp. 1-69. Berlin: SpringerVerlag 1981
14. Oldshue, J. Y.: Fermentation mixing scale-up technique. Bioteehnol. Bioeng. 8 (1966) 3-24
15. Cooper, C. M.; Fernstrom, G. A.; Miller, S. A.: Performance of
agitated gas-liquid contactors. Ind. Eng. Chem. 36 (1944) 504509
16. Bailey, J. E.; Ollis, D. E: Biochemical engineering fundamentals,
2nd edition, pp. 508-512. New York: McGraw-Hill 1986
17. Finn, R. K.: Agitation and aeration. In: Blakebrough, N. (Ed.):
Biochemical and biological engineering science, Vol. 1, pp. 6999. New York: Academic Press 1967

53
18. Taguchi, H.; Miyamoto, S.: Power requirement in non-Newtonian fermentation broth. Biotechnol. Bioeng. 8 (1966) 43-54
19. Niitsu, H.; Fujita, M.; Terui, G.: Tryptophan fermentation with
an improved strain of Hansenula anomala under the control
of dissolved oxygen concentration. J. Ferm. Teehnol. 47 (1969)
194-202
20. Sumino, Y.; Kanzaki, T.; Fukuda, H.: Oxygen transfer in L-glutamic acid fermentation by an oleic acid-requiring organism. II.
Effects of dissolved oxygen. J. Ferm. Technol. 46 (1968) 10401047
21. Connolly, J. R.; Winter, R. L.: Approaches to mixing operation
scale-up. Chem. Eng. Prog. 65(8)(1969) 70-78
22. Fukuda, H.; Sumino, Y.; Kansaki, T.: Scale-up of fermentors.
I. Modified equations for volumetric oxygen transfer coefficient.
J. Ferment. Tech. (Japan) 46 (1968) 829-837
23. Fukuda, H.; Sumino, Y.; Kansaki, T.: Scale-up of fermentors.
II. Modified equations for power requirement. J. Ferment. Teeh.
(Japan) 46 (1968) 838-845
24. Michel, B. J.; Miller, S. A.: Power requirements of gas-liqnid
systems. AIChE Journal 8 (1962) 262-266
Received October 16, 1991

Dr. Lu-Kwang Ju (corresponding author)


Dr. G.G. Chase
Department of Chemical Engineering
The University of Akron
Akron, Ohio 44325-3906
U.S.A.

You might also like