Feature Report

Mass Transfer in
Fermentation Scaleup
Representative volume element inside fermenter

As fermenters are
scaled up to huge sizes,
mass transfer is a key
consideration

Fgas

1 cm

er
et

Jim Gregory and Bob Green

Fluor
Nicolle Courtemanche
and Richard Kehn
SPX Flow Technology, Lightnin

ow big can a fermenter get?

And what would the biggest
fermenter look like? The answers to these questions depend upon how the requirements
of heat transfer, mass transfer (gasto-liquid), and momentum transfer
(mixing) are met. In an earlier article (Heat Transfer for Huge-Scale
Fermentation, Chem. Eng., November 2013, pp. 4446) the authors described how heat-transfer requirements can cause jackets to become
ineffective at large scale, which
drives the need for external heat
exchangers. This article examines
the issues that arise with mass and
momentum transfer at huge scales.
The concerns associated with mass
transfer at huge scales also influence the type and size of pilot- and
demonstration-plant facilities that
are used in scaleup.
Many useful chemicals can be produced by microbes that require oxygen to grow. An aerobic fermenter
is used to grow these microbes and
create the right conditions for them
to produce these chemicals. This
type of fermenter is essentially a
mass transfer device that promotes
the transfer of oxygen from gas bub44

Vtotal = 10L

Vgas = 0.6/Fbubble
Vliquid = Vtotal Vgas

1 meter

Figure 1. A representative volume element of the fermenter is depicted in this

sketch, where V is volume and F is the gas flowrate

Table 1. Number of fermenters required for a given diameter

10

15

20

Volume, 1,000 gal w/v 28

Fermenter diameter, ft

63

113 176 254 345 451 571 705 853 1,015

25

30

35

40

45

50

55
9

60

Number of fermenters
required

284 126 71

45

32

23

18

14

11

Harvest interval, h

0.4 1.0 1.8

2.7

3.9

5.4

7.0

8.9

11.0 13.3 15.8

18

13

Required seed fermenters

Number of seed trains

110 49

28

S-1 Seed,1,000 gal w/v 2.8 6.3 11.3 17.6 25.4 34.5 45.1 57.1 70.5 85.3 101.5
S-2 Seed,1,000 gal w/v 0.3 0.6 1.1

1.8

2.5

3.5

4.5

5.7

7.0

8.5

10.2

S-3 Seed, gal w/v

28

63

113 176 254 345 451 571 705 853 1,015

S-4 Seed, gal w/v

11.3 17.6 25.4 34.5 45.1 57.1 70.5 85.3 101.5

S-5 Seed, gal w/v

bles into the liquid medium where

the microbes live. Often the rate of
oxygen transfer is the limiting factor in the whole manufacturing process. That is why maximum oxygentransfer rate is a key to a successful
fermenter design.

A hypothetical process

Outlining a hypothetical fermentation process, such as the following

one, gives a sense of the need for
huge fermenters:
Objective: Make 100,000 ton/yr of
product

Chemical Engineering www.che.com March 2014

8.5

10.2

Assumptions:
The final fermenter broth is 5%
w/w product after 100 hours of
incubation time
Use 10% inoculum
Seed stages incubate for 36 h with
a 12-h turnaround time
The fermenter specific gravity is
equal to 1.02
The maximum fill of the fermenter
is 80%
The maximum fermenter straightside height is 60 ft
The oxygen uptake rate is 100
mmole/L/h

SPX

Seed trains: A fermentation process typically

involves inoculating a
batch of sterile growth
media with a seed,
which consists of viable
microbes of the desired
type. A 1-mL vial could
inoculate a 100-mL flask,
which would grow enough
to inoculate a 10-L vessel, which would grow to
inoculate a 1,000-L tank
and so on. In this way, a
Figure 2. This configuration of a combined radialproduction fermenter reand axial-impeller system is typical to provide mixing
in an aerobic fermenter
quires a series of smaller
fermenters to produce a
The fermenter turnaround time sufficient volume of inoculum. Since
is 25 h (to harvest, clean, sanitize, the seed fermenters operate in sefill and inoculate)
ries, they are often referred to as a
Planned down time is 30 days for seed train.
an annual overhaul, plus 15 days
Oxygen transfer
of contingency
A fermenters oxygen transfer rate
The downstream yield is 95%
(OTR) is a function of the oxygen
Calculations:
1. Fermenter production require- transfer driving force, the surment = (100,000 ton/yr)/(95% face area across which the oxygen
flows, and the resistance to oxyyield) = 105,000 ton/yr
2. Fermenter broth required = gen transfer:
[(105,000 ton/yr)(2,000 lb/ton)]/
(0.05 ton product/ton broth) = OTR = kL a(CbubbleCliquid) (1)
4,200,000,000 lb broth/yr
3. Fermenter volumetric produc- where OTR is the oxygen transfer
tion = (4,200 million lb broth/yr)/ rate in mmol/h; kL = conductance
[(8.34 lb/gal)(1.02)] = 494,000,000 (reciprocal of resistance) to oxygen
gal/yr = 1,540,000 gal/d = 64,300 transfer; a is the surface area of oxygen transfer in square feet; and C
gal/h = 1,070 gal/min
4. Total fermenter capacity re- is the oxygen concentration.
This means that the oxygen transquirement (working volume) =
(64,300 gal/h)(125 h/fermenter fer rate can be increased by increasing kL, a, or the change in C.
cycle) = 8,000,000 gal
How many fermenters would be
needed to offer 8,000,000 gallons The effect of tank height
of net tank capacity? Table 1 offers One of the primary constraints assome options for the number of fer- sociated with mass transfer in fermenters required versus fermenter menters is that bubbles rise only so
size, using the assumption that the fast. No matter how much air is infermenter height is limited to 60 ft. troduced at the bottom, the bubbles
Ten-foot-diameter fermenters are will rise at a rate dependent on the
known to be capable of production bubble size and the liquid density
rates of 100 mmole/L/h and are and viscosity, not on the rate of air
economical, but that size would re- being blown into the tank. The efquire 284 fermenters and 110 seed fect is that increasing airflow intrains (see next section). It is hard creases the availability of air in
to believe this would be an economi- the fermenter. The inventory of air
cal plant design. If the fermenters at any time, the void fraction, discould be 60 ft in diameter, then places product.
For a very large fermenter with
there would be eight of them and
water-like fermentation broth, the
four seed trains.

average-sized air bubbles could

rise at a rate of about 0.6 meters
per second (m/s). That means that
a superficial air velocity of 0.3 m/s
results in a fermenter that is 50%
liquid and 50% air bubbles. That is
not a very productive fermenter.
As the gas bubbles rise, oxygen
is transferred from the air to the
liquid. The average oxygen concentration in the gas phase goes down
with increasing height.
Consider a representative volume
element of the fermenter that is one
meter per side and one centimeter
tall as in Figure 1. Assume that the
oxygen uptake rate is 100 mmol O2/
L/h throughout the fermenter; the
superficial gas rate is 0.1 m/s (0.1
m3/s per square meter of horizontal
surface); and the bubble rise velocity for this system is 0.6 m/s. The
maximum fermenter height can be
calculated as follows:
1. O
 xygen supplied to the bottom
square meter column element =
[(0.1 m3/s)(1,000 L/m3)(0.209 mol
O2/mol air)]/(24.5 L/mol air at
25C) = 0.83 mol O2/s
2. The void fraction in a representative volume element = (0.1 m/s)/
(0.6 m/s) = 0.17
3. The liquid volume in a representative volume element = (1m)(1m)
(0.01m)(1,000 L/m3)(1 0.17) =
8.3 L
4. Oxygen consumed by each volume
element =[(100 mmol O2/L/h)(8.3
L)]/[(1,000 mmol/mol)(3,600 s/h)
= 0.00023 mol O2/s per volume
element
5. The number of volume elements
in column of liquid = 0.83 mol/s)/
(0.00023 mol O2/s/volume element) = 3,600 elements = 3,600
cm = 36 m = 118 ft
It makes no sense to scale this process up to a height of above 36 m
because the oxygen is completely
depleted from the sparge air at
that height. Actually, the oxygen
concentration would never drop to
zero, because the oxygen transfer
driving force falls along with the
oxygen concentration, so the top of
the fermenter suffers from diminishing returns.
In the above calculation it has
been assumed that there is neg-

Chemical Engineering www.che.com March 2014

45

Guidelines for pilot plant testing

Feature Report
ligible axial mixing of the liquid.
This type of mixing in an actual
fermenter (Figure 2) would serve to
move dissolved oxygen that is near
the bottom to upper levels where
it is needed, and to move oxygendeleted liquid that is near the top
to flow downward. This movement
increases the oxygen-transfer driving force near the bottom of the
fermenter. However, as shown in
Figure 3B, the improvement in the
oxygen transfer rate near the bottom of the fermenter causes the oxygen in the gas to run out sooner.
The fermenter should not be designed as tall as 36 m, because of
the very poor oxygen transfer in
the upper part of the fermenter at
such heights. Thus, huge fermenters need to grow fat, not tall.
High gas flowrates to the fermenter increase the number of
bubbles, which increases the bubble
surface area and thereby increases
kLa. In addition, with more airflow
the oxygen concentration depletes
more slowly, thereby increasing
the overall oxygen-transfer driving
force. However, since the bubbles
rise only so fast, the increasing airflow will decrease the liquid volume
in the tank. An increase in gas flowrate will also increase the agitator
size. The more air there is, the more
the impellers will have to disperse,
and the higher the mixer motor
power will be. This presents an interesting optimization problem.
What is the optimum air flowrate?

Demonstration scale

Scaleup is about business risk. In

order to evaluate the risk involved,
it is important to determine what elements of the design involve performance uncertainty. An intermediate-scale demonstration plant might
be required to prove that scaleup
considerations are well understood.
Thanks to the use of external heat
exchangers, the heat transfer coefficients (U), the effective heat-transfer area (A), and the temperature
driving forces (T) are all known, so
that heat (Q) can be calculated:
Q = (U)(A)(Tlog mean)

(2)

The above analysis shows that,

46

1. The minimum volume should be 250 gal (950 L) for scalable mass-transfer testing.
A 20-gal tank can be used to evaluate blending and impeller placement
2. Liquid-level-to-tank-height ratio, and tank geometry should be similar to full scale
3. Baffles and heating coils on pilot scale should be similar to full-scale tank
4. Test the fluid with the organism, if possible. If not, use water, knowing the oxygen
transfer rate results will be different
5. The gas and sparging system should be similar (the same would be better) as the
one to be used on full scale
6. Make sure the sparge location is under the main gas-dispersing impeller
7. Use a rotameter with capabilities to fluctuate the gas flowrate over a range (use at
least four different flowrates)
8. One flowrate should be the same vessel volumes per minute as the full scale
achieving the same superficial gas velocity will be difficult
9. Different styles and diameter of impellers should be tested. Include the ability to
adjust location of the impellers
10. Variable-speed drive should be used to alter speed to test four different power levels
11. Use a tachometer to measure the operating speed of the shaft and impeller
12. Use a torque sensor to record mixer horsepower while the test is running
13. Dissolved oxygen probe locations should be at the top and the bottom of the tank.
Keep them away from baffles and any other dead spots
14. Take note of how important the location of the lower impeller is in relation to the
sparger
15. Make sure the tank will be tall enough to account for the gas hold-up. The hold-up
will increase the liquid level, sometimes significantly, if the mixer has produced a
well-dispersed system
16. Acid/base indicator or conductivity probes can be used for qualitative blend-time
evaluation

for heat transfer, the design factors are already well understood
and predictable, and thus present a
low risk to the project. In the case
of mass transfer, however, the mass
transfer conductance used in Equation (1) is not well known for fermenters above about 100,000 gallons. Pilot testing is required.

Pilot-scale testing

Pilot work is critical for any new

process. For fermentation applications, pilot work is required to understand how the organism will
behave under specific process conditions. The information studied on
the pilot scale for a fermenter must
include the following: mass transfer, gas dispersion and blending. All
three are of equal importance.
If the mass transfer requirements
are not met, the organisms in the
fermenter will die because there is
not enough power available to force
the liquid/gas boundary layer transfer to take place.
If the gas dispersion requirements
are not met, the air is not properly
distributed throughout the vessel
and again, the organisms will die. If
the tank is not well blended, the nutrients that are added to the vessel,
the heat transfer and the pH will
not be uniform. The organism will
not survive in this environment.

Chemical Engineering www.che.com March 2014

All of these are undesirable results.

The information gleaned from the
pilot work is used to successfully
model the full-scale operation. Pilot
plant work will determine what
impeller style(s), diameter(s) and
power levels are required for the
agitator to successfully perform.
Proper experiment set-up and
execution will make sure repeatable results are achieved on the full
scale. The specific parameters that
must be examined are: tank geometry, baffle and coil arrangement
and gas-sparging system. The tank
geometry ratios, and baffle and coil
arrangements should be similar
between full scale and pilot scale.
Pilot testing should be done with
the exact process fluid to be used on
the full scale, or a fluid with very
similar properties. The liquid-levelto-tank-diameter ratio should be
constant in scaleup, as should the
type of gas and sparge system. The
lower impeller should be located at
a specific distance above the sparger
and that ratio should remain unchanged between scales.
Traditional laboratory-scale testing is performed at a minimum volume range between 20 to 250 gal.
When considering pilot scale work,
a tank with a minimum volume of
750 gal, or a 4-ft-dia. 8-ft tank
should be considered.

Vent

Ga

MXR-51
Compressed
air

Figure 3. The oxygen transfer

rates and the amount of dissolved
oxygen in the liquid are strongly
affected by mixing in a fermenter

For the 1-million-gal scale, a

larger test volume would be recommended. Here, the minimum would
be 10- to 12-ft-dia. vessels. A torque
sensor affixed to the shaft that records data while the test is running
is a necessity. Reading power using
an ampere or watt meter is not recommended, especially during pilot
testing. A tachometer that can accurately measure the lower shaft
speed is required.
A rotameter with capabilities of
adjusting the gas flowrate over a
specific range is also required. At a
minimum, four different gas flowrates should be examined. One gas
flowrate should be the same vessel
volumes per minute (VVM) as the
full-scale. (VVM is a unit of gas
flowrate widely used in the fermentation industry.) It will be difficult
to achieve the same superficial gas
velocities at full and pilot scales.
While at the pilot scale, the style
and diameter of impeller(s) should
be reviewed for optimum performance. A few different styles and
different diameters should be available to test. The impellers should be
adjustable, so that their positions
on the shaft can be changed while
running different experiments.
Dissolved oxygen (DO) probes
should be located at the top and
bottom of the fermentation tank.
The probes must be kept away
from baffles or other potential
low velocity or dead areas within
the vessel. It is also necessary to
study the gas hold-up volume on

Elevation

sp
ha
s

Elevation

Li

FCV
1

qu

id

as

ph

ph

as

as

SP
FIT
1

Liquid phase

C* C
oxygen transfer
driving force

C* C
20.9

Oxygen concentration

0.0

A. When mass-transfer effects are greater

than axial mixing effects, the oxygen transfer
rate is uniform and the dissolved oxygen
varies with fermenter height

the smaller scale to make sure the

full-scale vessel will be tall enough
to account for the increase in gas
liquid volume.
Many of these points are summarized in the box on Guidelines For
Pilot Plant Testing on p.46.

Final thoughts

When designing very large fermenters, care must be taken to avoid designs that are so tall that the upper
portion of the fermenter is ineffective.
Care must also be taken to provide
adequate mixing and mass transfer

20.9

Oxygen concentration

0.0

B. When axial mixing is sufficient to make

mass-transfer limiting, then the dissolved
oxygen is uniform throughout the fermenter,
and mass-transfer rates are higher at the
bottom where oxygen is introduced

in very wide fermenter designs. Since

unusual tank geometry is required
for very large fermenters, scaleup ratios are smaller, so large demonstration-scale testing is beneficial.
Agitation is a big expense. Using
a microbe that can tolerate low or
zero dissolved oxygen is highly advantageous, because of the higher
mass-transfer driving force that results. Also, a microbe that does not
require oxygen to produce product
has a clear economic advantage by
reducing agitation costs.

Edited by Dorothy Lozowski

Authors
Jim Gregory is a process
engineer at Fluor Corp. (100
Fluor Daniel Dr., Greenville,
SC 29607-2762; Email: jim.
[email protected]). He holds
a B.A. in biophysics and a
B.S.Ch.E. from the University of Connecticut, and an
M.Sc. in biochemical engineering from Rutgers University. He has experience
in the design and operation
of industrial microbiological processes ranging
from human-cell-line monoclonal antibodies to
diesel fuel.

C.R. Green (Bob) is director of design development at

Fluor Corp. (same address as
left; Email: bob.green@fluor.
com). He holds a B.S. in mechanical engineering from
North Carolina State and
an M.E. in mechanical engineering from the University
of South Carolina. He is a
registered professional engineer in six states. Green has
experience in the design and startup of microbiological processes, including human-cell-line
monoclonal antibodies, amino acids, bacteria,
biofuels and biochemicals.

Nicolle Courtemanche is a
senior application engineer at
SPX Flow Technology (Lightnin brand; 135 Mt. Read Blvd.,
Rochester, NY 14611; Email:
nicolle.courtemanche@spx.
com), a segment of SPX that
designs, manufactures and
installs engineered solutions
used to process, blend, meter
and transport fluids, in addition to air and gas filtration
and dehydration. Nicolle holds a B.S.Ch.E. from
the University of New Hampshire. Her areas of
mixing expertise include, pulp and paper, biotech, pharmaceuticals and other chemical process industries.

Richard Kehn is manager of

Research and Development at
SPX Flow Technology (Lightnin brand; same address as
left; Email: richard.kehn@spx.
com). Kehn holds a B.S.Ch.E.
from Rensselaer Polytechnic
Institute and is pursuing an
M.E. degree in mechanical engineering with a concentration
in computational fluid dynamics (CFD) from Rochester Institute of Technology. Kehn has been the author or coauthor of ten technical papers regarding mixing,
covering low-viscosity blending, solids suspension,
copper solvent extraction, slurry-tank-agitator
design and CFD. His areas of mixing expertise
include mineral processing, water and wastewater treatment, pulp and paper, and experimental
methods including scaleup and scale-down.

Chemical Engineering www.che.com March 2014

47