Doran (2013) - Mixing

C H A P T E R
8
Mixing
The physical operation of mixing can determine the success of bioprocesses. In fermentations, single- and
multiple-phase mixing occurs in fluids with a range of rheologies. Mixing controls the access of cells to dis-
solved nutrients and oxygen, and plays a critical role in controlling the culture temperature. The equip-
ment used for mixing has a significant effect on agitation efficiency, power requirements, and operating
costs. A consequence of mixing operations is the development of hydrodynamic forces in the fluid. These
forces are responsible for important processes in fermenters such as bubble break-up and dispersion; how-
ever, cell damage can also occur and must be avoided. Problems with mixing are a major cause of produc-
tivity loss after commercial scale-up of bioprocesses.
This chapter draws on material introduced in Chapter 7 about fluid properties and flow
behaviour. In turn, as mixing underpins effective heat and mass transfer in bioprocesses,
this chapter provides the foundations for detailed treatment of these subjects in Chapters 9
and 10.
8.1 FUNCTIONS OF MIXING

Mixing is a physical operation that reduces nonuniformities in fluid by eliminating gra-
dients of concentration, temperature, and other properties. Mixing is accomplished by
interchanging material between different locations to produce a mingling of components.
If a system is perfectly mixed, there is a random, homogeneous distribution of system
properties. Mixing is used in bioprocesses to:
• Blend soluble components of liquid media such as sugars
• Disperse gases such as air through liquids in the form of small bubbles
• Maintain suspension of solid particles such as cells and cell aggregates
• Where necessary, disperse immiscible liquids to form an emulsion or suspension of fine
droplets
• Promote heat transfer to or from liquids
Mixing is one of the most important operations in bioprocessing. To create an optimal
environment for cell culture, bioreactors must provide the cells with access to all
Bioprocess Engineering Principles, Second Edition 255 © 2013 Elsevier Ltd. All rights reserved.
256 8. MIXING
substrates, including oxygen in aerobic cultures. It is not enough to just fill the fermenter
with nutrient-rich medium; unless the culture is mixed, zones of nutrient depletion will
develop as the cells rapidly consume the materials they need within their local environ-
ment. This problem is heightened if mixing does not maintain a uniform suspension of
biomass; substrate concentrations can quickly drop to zero within layers of settled cells.
We rely on good mixing to distribute any material added during the fermentation, such as
fresh medium to feed the cells or concentrated acid or alkali to control the culture pH. If
these materials are not mixed rapidly throughout the reactor, their concentration can build
up to toxic levels near the feed point with deleterious consequences for the cells in that
region. Another important function of mixing is heat transfer. Bioreactors must be capable
of transferring heat to or from the broth rapidly enough so that the desired temperature is
maintained. Cooling water is used to take up excess heat from fermentations; the rate of
heat transfer from the broth to the cooling water depends on mixing conditions.
Mixing can be achieved in many different ways. In this chapter we will concentrate on
the most common mixing technique in bioprocessing: mechanical agitation using an
impeller.
8.2 MIXING EQUIPMENT
Mixing is carried out most often in cylindrical stirred tanks, such as that shown in
Figure 8.1. Baffles, which are vertical strips of metal mounted against the wall of the tank,
are installed to reduce gross vortexing and swirling of the liquid. Mixing is achieved using
an impeller mounted on a centrally located stirrer shaft. The stirrer shaft is driven rapidly
by the stirrer motor; the effect of the rotating impeller is to pump the liquid and create a
regular flow pattern. Liquid is forced away from the impeller, circulates through the ves-
sel, and periodically returns to the impeller region. In gassed stirred tanks such as bioreac-
tors used for aerobic culture, gas is introduced into the vessel by means of a sparger
located beneath the impeller.
The equipment chosen for mixing operations exerts a significant influence on the out-
come of the process. Aspects of this equipment are outlined in the following sections. The
operating characteristics of different impellers are described in detail in Section 8.4.
8.2.1 Vessel Geometry and Liquid Height

The shape of the base of stirred tanks affects the efficiency of mixing. Several base
shapes are shown in Figure 8.2. If possible, the base should be rounded at the edges rather
than flat; this eliminates sharp corners and pockets into which fluid currents may not pen-
etrate, and discourages the formation of stagnant zones. The energy required to suspend
solids in stirred tanks is sensitive to the shape of the vessel base: depending on the type of
impeller and the flow pattern generated, the modified geometries shown in Figure 8.2(b)
through (e) can be used to enhance particle suspension compared with the flat-bottom
tank of Figure 8.2(a). In contrast, sloping sides or a conical base such as that shown in
3. PHYSICAL PROCESSES
8.2 MIXING EQUIPMENT 257
Stirrer motor FIGURE 8.1 Typical configuration of a stirred
and gearbox tank.
drive unit
Stirrer
shaft Tank
Baffle
Impeller
Sparger
Air
Figure 8.2(f) promotes settling of solids and should be avoided if solids suspension is
required.
Other geometric specifications for stirred tanks are shown in Figure 8.3. For efficient
mixing with a single impeller of diameter Di equal to a 1/4 to 1/2 the tank diameter DT,
the height of liquid in the tank HL should be no more than 1.0 to 1.25 DT. Because the
intensity of mixing decreases quickly as fluid moves away from the impeller zone, large
volumes of liquid in the upper parts of the vessel distant from the impeller are difficult to
mix and should be avoided.
Another aspect of vessel geometry influencing mixing efficiency is the clearance Ci
between the impeller and the lowest point of the tank floor (Figure 8.3). This clearance
affects solids suspension, gas bubble dispersion, and hydrodynamic stability. In most prac-
tical stirring operations, Ci is within the range 1/6 to 1/2 the tank diameter.
8.2.2 Baffles
Baffles are standard equipment in stirred tanks. They assist mixing and create turbu-
lence in the fluid by breaking up the circular flow generated by rotation of the impeller.
258 8. MIXING
FIGURE 8.2 Different profiles for the base of stirred vessels:

(a) flat; (b) dished; (c) round; (d) contoured; (e) cone-and-fillet;
(f) conical.
(a) (b)
(c) (d)
Cone
Fillet
(e) (f)
Baffles are attached to the inside vertical walls of the tank by means of welded brackets.
Four equally spaced baffles are usually sufficient to prevent liquid swirling and vortex for-
mation. The optimum baffle width WBF depends on the impeller design and fluid viscos-
ity, but is of the order 1/10 to 1/12 the tank diameter. For clean, low-viscosity liquids,
baffles are attached perpendicular to the wall as illustrated in Figure 8.4(a). Alternatively,
as shown in Figures 8.4(b) and (c), baffles may be mounted away from the wall with clear-
ance CBF ! 1/50 DT, or set at an angle. These arrangements prevent the development of
stagnant zones and sedimentation along the inner edge of the baffle during mixing of vis-
cous fluids or fluids containing suspended cells or particles.
8.2.3 Sparger
There exists a large variety of sparger designs. These include simple open pipes, perfo-
rated tubes, porous diffusers, and complex two-phase injector devices. Point spargers, such
as open pipe spargers, release bubbles at only one location in the vessel. Other sparger
designs such as ring spargers have multiple gas outlets so that bubbles are released simul-
taneously from various locations. Bubbles leaving the sparger usually fall within a rela-
tively narrow size range depending on the sparger type. However, as the bubbles rise
8.2 MIXING EQUIPMENT 259
FIGURE 8.3 Some geometric specifications for a
stirred tank.
CBF
DT
HL
LB
WBF
WB
Di SS
Ci
DS
CBF = baffle clearance HL = liquid height

Ci = impeller clearance LB = blade length
Di = impeller diameter SS = sparger separation
DS = sparger diameter WB = blade width
DT = tank diameter WBF = baffle width
from the sparger into the impeller zone, they are subjected to very high shear forces from
operation of the stirrer that cause bubble break-up. The resulting small bubbles are flung
out by the impeller into the bulk liquid for dispersion throughout the vessel. Although the
type of sparger used has a relatively minor influence on the mixing process in most stirred
tanks, the diameter DS of large ring spargers and the separation SS between the sparger and
impeller (Figure 8.3) can have an important influence on the efficiency of gas dispersion.
8.2.4 Stirrer Shaft

The primary function of the stirrer shaft is to transmit torque from the stirrer motor to
the impeller. Torque is the tendency of a force to cause an object to rotate. The magnitude
260 8. MIXING
Tank wall Baffle FIGURE 8.4 Baffle arrangements: (a) baffles attached
to the wall for low-viscosity liquids; (b) baffles set away
from the wall for moderate-viscosity liquids; (c) baffles
set away from the wall and at an angle for high-viscosity
liquids.
From F.A. Holland and F.S. Chapman, 1966, Liquid
Mixing and Processing in Stirred Tanks, Reinhold,
New York.
Flo
w
(a)
Flo
w
(b)
Flo
w
(c)
of the torque around the shaft axis is related to the power required for operation of the
impeller. The stirrer shaft also performs other mechanical functions: it resists the bending
forces created by the impeller, it limits any lateral deflections, and it supports the impeller
weight. These functions must all be achieved without excessive vibration.
In typical mixing operations, the impeller is attached to a vertical stirrer shaft that
passes from the motor through the top of the vessel. However, when headplate access is at
a premium because of other devices and instruments located on top of the tank, or if a
shorter shaft is required to alleviate mechanical stresses (e.g., when mixing viscous fluids),
the stirrer shaft may be designed to enter through the base of the vessel. The vessel config-
uration for a bottom-entering stirrer is shown in Figure 8.5. The main disadvantage of bot-
tom-entering stirrers is the increased risk of fluid leaks due to failure or wear of the seals
between the rotating stirrer shaft and the vessel floor.
8.3 FLOW PATTERNS IN STIRRED TANKS 261
FIGURE 8.5 Stirred tank configuration for a
bottom-entry stirrer.
Impeller
Sparger
Air
Stirrer motor and
gearbox drive unit
8.3 FLOW PATTERNS IN STIRRED TANKS
The liquid flow pattern established in stirred tanks depends on the impeller design, the
size and geometry of the vessel, and the properties of the fluid. There are three directional
elements to the flow: rotational flow (i.e., around the stirrer shaft), radial flow (i.e., from the
central axis out to the sides of the tank and back again), and axial flow (i.e., up and down
the height of the vessel). An effective mixer will cause motion in all three directions; how-
ever, radial and axial flows generated at the impeller are primarily responsible for bulk
mixing. Impellers are broadly classified as radial-flow impellers or axial-flow impellers
depending on the direction of the flow leaving the impeller; some impellers have both
radial- and axial-flow characteristics. When gas is sparged into the tank, two-phase flow
patterns are created. The characteristics of this flow depend mainly on the stirrer speed,
the gas flow rate, and the fluid properties.
262 8. MIXING
FIGURE 8.6 (a) Circular flow in an unbaffled stirred tank

viewed from above. (b) Vortex formation during circular flow.
Circular flow
around stirrer
shaft
(a)
Vortex
(b)
8.3.1 Rotational Flow

Because stirring is rotational in action, all impellers generate circular flow of liquid
around the stirrer shaft as illustrated in Figure 8.6(a). However, simple rotational flow is
disadvantageous for mixing and should be avoided as much as possible. In circular flow,
liquid moves in a streamline fashion and there is little top-to-bottom mixing between dif-
ferent heights in the vessel. Circular flow also leads to vortex development as shown in
Figure 8.6(b). At high impeller speeds, the vortex may reach down to the impeller so that
gas from the surrounding atmosphere is drawn into the liquid. This is undesirable as it
produces very high mechanical stresses in the stirrer shaft, bearings, and seals. Attenuating
circular flow has a high priority in design of mixing systems. It is usually minimised by
installing baffles to interrupt the rotational flow pattern and create turbulence in the fluid.
8.3.2 Radial Flow

Radial or horizontal flow is generated by impellers with blades aligned parallel to the
stirrer shaft. A typical liquid circulation pattern set up by a high-speed radial-flow impel-
ler is shown in Figure 8.7. Liquid is driven radially from the impeller towards the walls of
the tank where it divides into two streams, one flowing up to the top of the tank and the
other flowing down to the bottom. These streams eventually reach the central axis of the
tank and are drawn back to the impeller. Radial-flow impellers also set up circular flow
that must be reduced by baffles.
8.3 FLOW PATTERNS IN STIRRED TANKS 263
FIGURE 8.7 Flow pattern produced by a radial-flow impel-
ler in a baffled tank.
Baffle
FIGURE 8.8 Flow pattern produced by an axial-flow impel-

ler in a baffled tank.
Baffle
8.3.3 Axial Flow

Axial flow is necessary for top-to-bottom mixing in stirred tanks. Axial flow is gener-
ated by impellers with inclined or pitched blades that make an angle of less than 90" with
the plane of rotation. Axial flow is particularly useful when strong vertical currents are
required. For example, if the fluid contains solids, a strong downward flow of liquid leav-
ing the impeller will discourage settling at the bottom of the tank. The type of flow pattern
set up by a typical axial-flow impeller is illustrated in Figure 8.8. Fluid leaving the impel-
ler is driven at a downward angle until it is deflected from the floor of the vessel. It then
spreads out over the floor and flows up along the walls before being drawn back to the
impeller. As rotational flow is also generated by axial-flow impellers, baffles are required.
264 8. MIXING
Axial-flow impellers are most commonly operated to generate downward flow of fluid
leaving the impeller blades, as illustrated in Figure 8.8. However, if the direction of impel-
ler rotation is reversed, axial-flow impellers may be applied for upward pumping, but this
mode of operation is not often used.
8.3.4 Gas Flow Patterns

Sparging stirred tanks with gas creates two-phase flow patterns of bubbles in liquid.
Different bubble distributions develop depending on the relative rates of gas flow and stirring.
At high gassing rates or low stirrer speeds, the impeller is surrounded by gas and is unable to
pump effectively, indicating that the gas-handling capacity of the impeller has been exceeded.
This condition is called impeller flooding. Flooding should be avoided because an impeller blan-
keted by gas no longer contacts the liquid properly, resulting in poor mixing and gas disper-
sion. As the stirrer speed is increased or the gas flow rate reduced, the impeller blades start
to process the gas and bubbles are dispersed towards the walls of the tank. This condition is
known as impeller loading. At even higher stirrer speeds or lower gas flow rates, complete gas
dispersion is achieved below as well as above the impeller. Complete gas dispersion with
homogeneous distribution of gas to all parts of the vessel is the desirable operating condition.
As the stirrer speed is raised above that required to prevent impeller flooding, gas recir-
culation occurs. As well as rising upwards from the sparger to the liquid surface, gas bub-
bles are increasingly recirculated around the tank via the impeller. The extent of gas
recirculation depends on the type of impeller and the stirrer speed, but can become very
high. Under these conditions, the total flow rate of gas entering the impeller region at any
given time is significantly greater than that supplied directly from the sparger.
Different impellers vary considerably in their ability to handle high gas flow rates with-
out flooding. A parameter giving a rough indication of the tendency of impellers to flood
is the impeller solidity ratio. The solidity ratio is equal to the projected blade area divided
by the area of the circle swept out by the impeller during rotation. This is illustrated in
Figure 8.9 for an axial-flow impeller with a relatively low solidity ratio. Impellers with low
Projected
blade area
Circle enclosing the

(a) (b) impeller sweep area
FIGURE 8.9 Illustration of the solidity ratio for an axial-flow impeller.
(a) Low-solidity-ratio Lightnin A310 hydrofoil impeller. Photograph provided courtesy of Lightnin Mixers, Australia.
(b) Diagram showing the projected blade area and impeller sweep area. Adapted from J.Y. Oldshue, 1989, Fluid mix-
ing in 1989. Chem. Eng. Prog. 85(5), 33#42.
8.4 IMPELLERS 265
solidity ratio tend to flood at lower gas velocities than impellers with high solidity ratio
operated at the same stirrer speed. Impellers with solidity ratios greater than 90% have
been developed for improved gas handling.
8.4 IMPELLERS
Many different types of impeller are available for mixing applications. A small selection
is illustrated in Figure 8.10. The choice of impeller depends on several factors, including
the liquid viscosity, the need for turbulent shear flows (e.g., for bubble break-up and gas
dispersion), and whether strong liquid currents are required. The recommended viscosity
ranges for a number of common impellers are indicated in Figure 8.11. Impellers can also
be classified broadly depending on whether they produce high levels of turbulence, or
whether they have a strong pumping capacity for generation of large-scale flow currents.
Both functions are required for good mixing but they usually do not work together. The
characteristics of several impellers in terms of their turbulence- and flow-generating prop-
erties are indicated in Figure 8.12.
Typically, mixing in fermenters is carried out using turbines or propellers. These impel-
lers are described in detail in Sections 8.4.1 through 8.4.4. Turbines and propellers are
remote-clearance impellers; this means they have diameters of 1/4 to 2/3 the tank diameter,
thus allowing considerable clearance between the rotating impeller and the vessel walls.
They are operated at relatively high speeds to generate impeller tip velocities of the order
of 3 m s21. As indicated in Figure 8.11, turbines and propellers are recommended for mix-
ing liquids with viscosities between 1 and about 104 centipoise, which includes most fer-
mentation broths. From Figure 8.12, compared with the broad variety of other stirring
devices available, turbines and propellers generate moderate-to-high levels of turbulence
while retaining significant pumping capacity.
When remote-clearance agitators are applied in low-viscosity fluids, a turbulent region
of high shear and rapid mixing is generated near the impeller. This high-shear region is
responsible for bubble break-up in sparged systems. Because the mixing process should
FIGURE 8.10 Selected impeller designs.
Six-flat-blade disc turbine Propeller Pitched-blade turbine
Anchor Gate anchor Helical screw
266 8. MIXING
FIGURE 8.11 Viscosity ranges for different

107 impellers.
From F.A. Holland and F.S. Chapman, 1966, Liquid
106 Mixing and Processing in Stirred Tanks, Reinhold,
Helical ribbons
New York.
Helical screws
105
Viscosity (centipoise)
Gate anchors
104
Paddles
Flat-blade turbines
103
Anchors
Propellers
102
10
1
Impeller type
Turbulence generation Pumping/flow capacity

Low High Low
Gates, rakes
Helical ribbons, anchors
Hydrofoils, propellers
Pitched-blade turbines
Rushton turbines
Bladeless impellers, disc dispersers
FIGURE 8.12 Characteristics of different impellers for generation of turbulence and liquid pumping.
involve fluid from all parts of the vessel, the impeller must also generate circulatory cur-
rents with sufficient velocity to carry material from the impeller to the furthermost regions
of the tank and back again. In viscous fluids, it is often impossible for mechanical or
economic reasons to rotate the impeller fast enough to generate turbulence; instead, impel-
lers for viscous mixing are designed to provide maximum bulk movement or turnover of
material. As indicated in Figures 8.11 and 8.12, impellers with high flow capacity
suitable for mixing high-viscosity fluids include gate, anchor, and helical stirrers. Examples
of these impellers are shown in Figure 8.10. All are large agitators installed with small
wall clearance (around 1%#5% of the tank diameter) and operated at low stirrer speeds
(5220 rpm) to generate bulk fluid currents. For viscous fluids or when high shear
rates must be avoided, slow-speed, low-turbulence, high-flow impellers are preferred to
8.4 IMPELLERS 267
high-speed, high-turbulence, low-flow impellers. Although gate, anchor, and helical stir-
rers are not often applied for mixing in bioreactors because they do not disperse air bub-
bles adequately for oxygen supply to the cells, they may be used for other applications in
bioprocessing such as blending viscous slurries, pastes, or gums.
Several impellers used in industrial fermentations are described in the following
sections.
8.4.1 Rushton Turbine

The most frequently used impeller in the fermentation industry is the six-flat-blade
disc-mounted turbine shown in Figures 8.10 and 8.13. This impeller is also known as the
Rushton turbine. The Rushton turbine has been the impeller of choice for fermentations
since the 1950s, largely because it has been well studied and characterised, and because it
is very effective for gas dispersion. Although Rushton turbines of diameter one-third the
tank diameter have long been used as standard hardware for aerobic fermentations, in
recent years it has been recognised that larger impellers of size up to one-half the tank
diameter provide considerable benefits for improved mixing and gas distribution.
Without Gassing
A typical mean velocity vector plot for a Rushton turbine is shown in Figure 8.14. In
velocity vector plots, the length and direction of the arrows indicate the magnitude and
direction of the velocities at discrete locations in the fluid. The velocities in Figure 8.14
were measured using laser Doppler velocimetry—see Section 7.9.3 (Laser Doppler
Velocimetry subsection). Figure 8.14 represents the mean liquid flow pattern; as described
in Section 7.9.1 (Mean and Fluctuating Velocities subsection), turbulent flow in agitated
tanks is characterised by highly chaotic secondary fluid motion and fluctuating velocity
fields that disappear when instantaneous flow velocities are averaged over time.
FIGURE 8.13 Rushton turbine.
268 8. MIXING
FIGURE 8.14 Typical mean velocity vector plot for a Rush-

ton turbine. The velocities were measured using laser Doppler
velocimetry.
Adapted from M. Schäfer, M. Höfken, and F. Durst, 1997, Detailed
LDV measurements for visualization of the flow field within a stir-
red-tank reactor equipped with a Rushton turbine. Trans. IChemE
75A, 729#736.
As indicated in Figure 8.14, the Rushton turbine is a radial-flow impeller. It generates a jet
of high-speed flow directed radially outwards from the impeller; this jet entrains the sur-
rounding fluid to form the impeller discharge stream. The discharge stream slows down as it
approaches the tank wall and splits into two sections to create upper and lower circulatory
flows. These circulatory currents traverse the remainder of the tank before returning directly
to the impeller or becoming entrained again in the impeller outflow. The bulk flow in the ves-
sel therefore comprises two large ring vortices, one above and one below the impeller. For
impeller off-bottom clearances of less than one-half the liquid height, liquid velocities in the
lower ring vortex are somewhat stronger than those in the upper circulatory stream, which
traverses a greater distance into the upper reaches of the vessel during each circuit. Under
these conditions, the return axial flow from beneath the impeller can be sufficiently strong to
affect the radial discharge pattern, with the result that the outflow issuing from the impeller
blades can be inclined slightly upwards instead of purely horizontal, as illustrated in
Figure 8.14.
Compared with other types of turbine, Rushton impellers have a low pumping or circu-
latory capacity per unit power consumed. This is due mainly to a relatively high power
requirement, as described in Section 8.5.1. The pumping capacity of impellers is discussed
further in Section 8.7.
Operation of the Rushton turbine is characterised by the formation of two high-speed roll
and trailing vortices in the liquid behind the horizontal edges of each flat blade of the impel-
ler, as shown in Figure 8.15. These vortices play a critical role in determining the perfor-
mance of the impeller. Most of the mixing in stirred vessels takes place near the vortices
issuing from the impeller blades. As discussed further in the next section, the trailing vor-
tex system is responsible for gas dispersion in aerated vessels. It is also the most important
flow mechanism for turbulence generation. Steep velocity gradients are associated with
trailing vortices; however, as the vortices move out from the blades they lose their identi-
ties and break down, thus providing a major source of turbulence in the fluid. The forma-
tion of trailing vortices affects the distribution of power dissipation in stirred tanks, which
has consequences for phenomena such as cell damage in bioreactors. Fluid entrained by
the vortices comprises much of the radial discharge stream generated by Rushton turbines.
8.4 IMPELLERS 269
Direction of FIGURE 8.15 Roll and trailing vortices
rotation generated behind the blade of a Rushton
turbine.
From K. van’t Riet and J.M. Smith, 1975, The
Blade trailing vortex system produced by Rushton
Disc Trailing turbine agitators. Chem. Eng. Sci. 30,
vortices 1093#1105.
FIGURE 8.16 Distribution of k, the turbulence kinetic energy per unit mass of fluid, in a tank stirred by
a Rushton turbine. The data were determined using computational fluid dynamics (CFD) modelling.
(a) Distribution of k in the vertical plane; (b) distribution of k in the horizontal plane at impeller height,
with the impeller rotating clockwise.
Adapted from K. Ng, N.J. Fentiman, K.C. Lee, and M. Yianneskis, 1998, Assessment of sliding mesh CFD predictions and
LDA measurements of the flow in a tank stirred by a Rushton impeller. Trans. IChemE 76A, 737#747. Images provided
courtesy of M. Yianneskis, King’s College, London.
Even if the impeller Reynolds number Rei (Section 7.2.3) is high, indicating that flow is
in the turbulent regime, the turbulence intensity in stirred vessels is far from uniform or
randomly distributed. Figure 8.16 shows the distribution of turbulence kinetic energy per
unit mass of fluid, k (Section 7.9.2, Turbulence Kinetic Energy subsection), in the vertical
and horizontal planes of a tank stirred with a Rushton turbine. As illustrated in
Figure 8.16(a), the highest values of k occur in the outwardly flowing liquid jet leaving the
impeller blades: in this zone, k is at least an order of magnitude greater than in the remain-
der of the vessel. Figure 8.16(b) shows the turbulence kinetic energy distribution in the
horizontal plane of the tank at the height of the impeller. The six plumes of elevated
kinetic energy stretching out from the blades identify the trailing vortices generated by the
impeller and indicate the extent of their radial spread. The vortices and associated turbu-
lence dominate the horizontal plane of fluid near the impeller; however, these high levels
of turbulence kinetic energy are not transmitted very far above or below the impeller, as
indicated in Figure 8.16(a).
The distribution of the rate of dissipation of turbulence kinetic energy per unit mass, ε
(Section 7.9.2, Rate of Dissipation of Turbulence Kinetic Energy subsection), in a tank
270 8. MIXING
FIGURE 8.17 Distribution of ε, the rate of dissipation of turbulence kinetic energy per unit mass, in a tank
stirred by a Rushton turbine. The data were determined using computational fluid dynamics (CFD) modelling.
(a) Distribution of ε in the vertical plane; (b) distribution of ε in the horizontal plane at impeller height, with the
impeller rotating clockwise.
Adapted from K. Ng and M. Yianneskis, 2000, Observations on the distribution of energy dissipation in stirred vessels.
Trans. IChemE 78A, 334#341. Images provided courtesy of M. Yianneskis, King’s College, London.
stirred by a Rushton turbine is shown in Figure 8.17. The results are qualitatively similar
to those for k in Figure 8.16. The values of ε are highest near the blades and in the region
dominated by the trailing vortices; in most of the rest of the vessel, ε is 1 to 2 orders of
magnitude lower. The intense turbulence generated by the trailing vortices is contained
within a relatively small region and dissipates quickly away from the impeller.
With Gassing
Rushton turbines are very effective for gas dispersion. To some extent, this can be
attributed to the way the rotating disc on the turbine captures gas released below the
impeller and channels it into the regions of high turbulence near the blades. Rushton tur-
bines are often chosen for their gas-handling capacity, as they can be operated with rela-
tively high gas flow rates without impeller flooding.
Typical gas flow patterns generated by a Rushton turbine in a low-viscosity fluid are
shown in Figure 8.18. At high gassing rates or low stirrer speeds, the impeller is blanketed
by gas, indicating impeller flooding. Under these conditions, as shown in Figure 8.18(a),
the flow pattern is dominated by buoyant gas#liquid flow up the middle of the vessel. At
higher stirrer speeds or lower gas flow rates, the impeller is loaded as gas is captured
behind the impeller blades and dispersed towards the vessel walls, as indicated in
Figure 8.18(b). Further increase in stirrer speed or reduction of the gas flow rate produces
complete gas dispersion, as illustrated in Figure 8.18(c).
Photographs demonstrating the transition from impeller flooding to complete gas dis-
persion in a transparent tank stirred by a Rushton turbine are shown in Figure 8.19.
In these experiments, the volumetric gas flow rate was held constant as the stirrer speed
was increased from 100 rpm to 400 rpm. At the two lowest stirrer speeds in Figure 8.19(a)
8.4 IMPELLERS 271
FIGURE 8.18 Patterns of gas distribu-
tion in an aerated tank stirred with a
Rushton turbine as a function of the impel-
ler speed Ni and gas flow rate Fg. (a)
Impeller flooding; (b) impeller loading; (c)
complete gas dispersion.
Adapted from A.W. Nienow, M. Konno, and
W. Bujalski, 1986, Studies on three-phase mix-
ing: a review and recent results. Chem. Eng.
Res. Des. 64, 35#42.
Gas Gas Gas
(a) (b) (c)
Constant Fg, increasing Ni

Constant Ni, increasing Fg
and (b), the impeller remains flooded as the stirrer is surrounded by gas and there is little
outward dispersion of bubbles towards the vessel wall. Liquid motion is weak as blanketing
by the gas prevents the impeller from pumping effectively. As indicated in Figure 8.19(c),
at higher stirrer speeds the impeller becomes loaded and gas is distributed throughout the
upper part of the vessel above the stirrer. Further increases in stirrer speed allow complete
dispersion of bubbles below as well as above the impeller, as shown in Figure 8.19(d).
Correlations have been developed for predicting the operating conditions under which
impeller loading and complete gas dispersion are achieved using Rushton turbines. These
relationships are expressed using two dimensionless variables, the gas flow number Flg:
Fg
Flg 5 ð8:1Þ
Ni D3i
and the Froude number Fr:

Ni2 Di
Fr 5 ð8:2Þ
g
where Fg is the volumetric gas flow rate, Ni is stirrer speed, Di is impeller diameter, and g
is gravitational acceleration. Conceptually, Flg is the ratio of the gas flow rate to the pump-
ing capacity of the impeller; Fr is the ratio of inertial to gravitational or buoyancy forces.
Conditions at the flooding#loading transition for Rushton turbines are represented by the
following equation [1]:
! "3:5
Di
Flg 5 30 Fr at the flooding#loading transition ð8:3Þ
DT
The conditions for complete dispersion of gas by Rushton turbines are represented by
another equation [1]:
! "0:5
Di
Flg 5 0:2 Fr0:5 for complete gas dispersion ð8:4Þ
DT
272 8. MIXING
(a) 100 rpm (b) 200 rpm
(c) 300 rpm (d) 400 rpm

FIGURE 8.19 Gas#liquid flow patterns in a tank stirred by a Rushton turbine. The sparger positioned below
the impeller is a horizontal steel tube with eight holes on its upper surface.
From K.L. Man, 1985, A study of local heat-transfer coefficients in mechanically agitated gas#liquid vessels. In: Mixing,
Proc. 5th Eur. Conf. on Mixing, Würzburg, Germany, pp. 221#231, BHRA: The Fluid Engineering Centre, Cranfield,
U.K.
Equations (8.3) and (8.4) apply to low-viscosity fluids and were determined using a variety
of point and ring (diameter , Di) spargers. The flooding#loading transition is not affected
by the impeller off-bottom clearance; however complete gas dispersion is, and Eq. (8.4)
was developed for an impeller clearance of a one-quarter the liquid height.
Equations (8.1) through (8.4) demonstrate the strong dependence of gas-handling capacity
on the impeller diameter. The volumetric gas flow rate Fg for loading is proportional to
Di7.5; Fg for complete gas dispersion is proportional to Di4. This means that for a 10% increase
in Di, the gas flow rate that can be handled without flooding increases by about 100%, while
the gas flow rate for complete dispersion increases by about 50%. The dependence on stirrer
speed is not as strong: a 10% increase in Ni increases Fg for flooding and complete gas disper-
sion by about 30% and 20%, respectively.
8.4 IMPELLERS 273
EXAMPLE 8.1 GAS HANDLING WITH A RUSHTON TURBINE

A fermenter of diameter and liquid height 1.4 m is fitted with a Rushton impeller of diameter
0.5 m and off-bottom clearance 0.35 m operated at 75 rpm. The fermentation broth is sparged
with air at a volumetric flow rate of 0.28 m3 min21. Half-way through the culture some bearings
in the stirrer drive begin to fail and the stirrer speed must be reduced to a maximum of 45 rpm
for the remainder of the process.
(a) Under normal operating conditions, is the gas completely dispersed?
(b) After the stirrer speed is reduced, is the impeller flooded or loaded?
Solution # #
#1 min#
(a) Ni 5 75 min21 ## & # 5 1:25 s21
60 s #
From Eq. (2.16), g 5 9.81 m s22. Under normal operating conditions, from Eq. (8.2):
ð1:25 s21 Þ2 0:5 m

Fr 5 5 0:0796
9:81 m s22
Applying Eq. (8.4) for complete gas dispersion:

! "
0:5 m 0:5
Flg 5 0:2 ð0:0796Þ0:5 5 0:0337
1:4 m
Therefore, from Eq. (8.1):
Fg 5 Flg Ni D3i 5 0:0337ð1:25 s21 Þ ð0:5 mÞ3 5 0:00527 m3 s21 5 0:32 m3 min21
As the air flow rate that can be completely dispersed by the impeller (0.32 m3 min21) is
greater than the operating flow rate (0.28 m3 min21), we can conclude that the air provided
is completely dispersed under normal operating conditions.
(b) After the stirrer speed reduction:
# #
#1 min#
Ni 5 45 min21 ## & # 5 0:75 s21
60 s #
From Eq. (8.2):
ð0:75 s21 Þ2 0:5 m

Fr 5 5 0:0287
9:81 m s22
Applying Eq. (8.3) for the flooding#loading transition:
! "
0:5 m 3:5
Flg 5 30 0:0287 5 0:0234
1:4 m
Therefore, from Eq. (8.1):
Fg 5 Flg Ni D3i 5 0:0234 ð0:75 s21 Þ ð0:5 mÞ3 5 0:00219 m3 s21 5 0:13 m3 min21
At the reduced stirrer speed, the maximum air flow rate that can be handled without
impeller flooding is 0.13 m3 min21. As the operating air flow rate (0.28 m3 min21) is greater
than this, the impeller is flooded.
274 8. MIXING
Gas dispersion by Rushton turbines is related directly to the trailing vortices that
develop behind the impeller blades (refer to Figure 8.15). Rolling or rotation of each vortex
results in centrifugal acceleration of the liquid and a reduction in pressure along the vortex
axis. When the liquid is sparged with gas, the gas accumulates readily in these low-pres-
sure regions, producing ventilated cavities behind the blades. A photograph of ventilated
cavities behind the blades of a Rushton turbine is shown in Figure 8.20. Bubble dispersion
occurs primarily at the outer tails of the cavities, where small bubbles are shed to balance
the rate of gas flow into the cavities from under the disc.
The effectiveness of gas dispersion in stirred vessels is controlled by the size and struc-
ture of the ventilated cavities behind the impeller blades. The types of cavity formed with
Rushton turbines in low-viscosity fluids have been well documented and are illustrated in
Figure 8.21. With increasing gas flow rate Fg at constant stirrer speed Ni, vortex cavities at
each of the blades give rise to clinging cavities, after which there is a transition to a very
stable ‘3#3’ structure characterised by large cavities behind three of the blades and smaller
cavities behind the remaining alternate three blades. If the gas flow rate is increased
beyond the gas-handling capacity of the impeller, six equal-sized ragged cavities are
formed; these cavities are unstable and oscillate violently. The formation of ragged cavities
occurs at the point represented in Figure 8.18 as the transition from Figure 8.18(b) to (a): at
these high gas flow rates, the impeller ceases to function effectively and is flooded. No
particular change in cavity structure has been associated with the transition shown in
Figure 8.18 from loading (b) to complete gas dispersion (c). Cavity formation in fluids
with viscosity greater than about 20 centipoise is more complex than that represented in
Figure 8.21; cavities of different shape and greater stability are produced. Cavities in
FIGURE 8.20 Gas cavities that are formed

behind the blades of a 7.6-cm Rushton tur-
bine in water. The impeller was rotated
counterclockwise; the camera was located
below the impeller and ring sparger.
From W. Bruijn, K. van’t Riet, and J.M. Smith,
1974, Power consumption with aerated Rushton
turbines. Trans. IChemE 52, 88#104.
8.4 IMPELLERS 275
Vortex Clinging ‘3 – 3’ structure ‘3 – 3’ structure Ragged

cavities cavities with 3 clinging with alternate cavities
and 3 large large cavities
cavities of different size
Constant Ni
Increasing Fg
Vortex cavities Clinging cavity Large cavity
Blade Blade Blade

Disc Disc Disc
Direction of Direction of Direction of

rotation rotation rotation
FIGURE 8.21 Changes in ventilated cavity structure for a Rushton turbine at constant stirrer speed Ni and
increasing gas flow rate Fg.
Adapted from J.M. Smith and M.M.C.G. Warmoeskerken, 1985, The dispersion of gases in liquids with turbines. In:
Mixing, Proc. 5th Eur. Conf. on Mixing, Würzburg, Germany, pp. 115#126, BHRA: The Fluid Engineering Centre,
Cranfield, U.K.
viscous fluids may be so stable that they persist behind the impeller blades for several
hours even after the gas supply is stopped [2].
Solids Suspension
Rushton turbines are effective for solids suspension, including in three-phase
(solid#liquid#gas) systems. Suspension of solids is generally improved by reducing the
impeller off-bottom clearance, but this can cause flow instabilities when the system is aer-
ated. For three-phase mixing, an impeller clearance of a 1/4 the tank diameter has been
recommended for Rushton turbines, as this allows effective solids suspension, gas disper-
sion under the impeller, and adequate agitation in the upper parts of the vessel [3].
8.4.2 Propellers
A typical three-blade marine-type propeller is illustrated in Figure 8.22. The slope of the
individual blades varies continuously from the outer tip to the inner hub. The pitch of a
propeller is a measure of the angle of the propeller blades. It refers to the properties of the
propeller as a segment of a screw: pitch is the advance per revolution, or the distance that
276 8. MIXING
FIGURE 8.22 Propeller.
liquid is displaced along the impeller axis during one full turn. Propellers with square
pitch, that is, pitch equal to the impeller diameter, are often used.
Propellers are axial-flow impellers. They may be operated for either downward or
upward pumping of the fluid; downward pumping is more common. Propellers have
high flow capacity and produce mean flow patterns similar to that shown in Figure 8.8.
They are used with low-to-medium viscosity fluids and are usually installed with diame-
ter around one-third the tank diameter. With gassing, propellers operated at high speed
can generate flow and torque instabilities. However, propellers are very effective for sus-
pending solids, outperforming Rushton turbines in that respect.
8.4.3 Pitched-Blade Turbines

Pitched-blade turbines have flat inclined blades, as shown in Figure 8.10. Although com-
monly referred to as axial-flow impellers, pitched-blade turbines generate discharge streams
with significant radial as well as axial velocity components. Pitched-blade turbines produce
strong liquid flows and have a much higher pumping efficiency than Rushton turbines,
making them very effective for blending applications. With gassing, ventilated cavities form
behind the blades as a result of underpressure in the trailing vortices in much the same way as
for Rushton turbines. However, for pitched-blade turbines, there is only one vortex per blade.
Pitched-blade turbines can be operated in either downward or upward pumping
modes, downward being the more common.
Downward Pumping
Figure 8.8 shows the liquid flow pattern typical of downward-pumping pitched-blade
turbines. Performance of these impellers is sensitive to several aspects of tank geometry,
such as the impeller off-bottom clearance and the sparger size and location. This sensitiv-
ity is greater than is found normally with Rushton turbines. Depending on the impeller
off-bottom clearance, secondary circulation loops may develop in the lower regions of the
tank in addition to the primary flow pattern shown in Figure 8.8. Because the strength of
these secondary currents determines the angle of fluid discharge from the impeller, which
8.4 IMPELLERS 277
in turn determines whether the primary circulation currents reach the vessel floor, the
flows set up by downward-pumping turbines can be considerably more complex than that
shown in Figure 8.8 [4]. Unlike with Rushton turbines, the fluid currents generated by
pitched-blade turbines are not compartmentalised into upper and lower circulatory loops.
However, because the flow velocity becomes progressively weaker away from the impel-
ler, the primary circulation currents generated by downward-pumping turbines may not
reach the upper parts of the tank, even when the liquid height does not exceed the tank
diameter [5].
The distribution of turbulence kinetic energy per unit mass of fluid, k (Section 7.9.2,
Turbulence Kinetic Energy subsection), in a tank stirred with a downward-pumping
pitched-blade turbine is shown in Figure 8.23. Turbulence kinetic energy is not distributed
uniformly throughout the tank. The highest values of k are concentrated near the impeller
where the turbulence is most intense. As the discharge streams move downwards away
from the impeller blades, the turbulence kinetic energy decreases accordingly. Levels of
turbulence kinetic energy in the remainder of the tank away from the impeller are up to
an order of magnitude lower than the maximum values measured.
With gassing, downward-pumping turbines are prone to flooding, especially if the
impeller diameter and solidity ratio are small. The hydrodynamic changes that occur as
the stirrer speed is increased at constant gas flow rate are similar to but more complex
than those represented in Figure 8.18 for a Rushton turbine; for example, asymmetrical
flow patterns may be generated. At low stirrer speeds when the impeller is flooded, liquid
circulation is weak and the flow is dominated by gas bubbling up the stirrer shaft. With
increasing stirrer speed, streaming ventilated cavities form behind the impeller blades and
bubbles breaking away from the cavities are dispersed downwards; however, the large
flow of bubbles up the centre of the vessel remains the primary flow pattern. Further
increases in speed allow the pumping action of the impeller to dominate, so that gas is
vigorously dispersed throughout the vessel. However, this condition is unstable; the flow
FIGURE 8.23 Distribution of k, the tur-

bulence kinetic energy per unit mass of
fluid, in a tank stirred by a downward-
pumping pitched-blade turbine located
k (m2 s−2)
near the centre of the tank. The data were
0.00150
0.00135 measured using particle image velocimetry.
0.00120 From J. Sheng, H. Meng, and R.O. Fox, 1998,
0.00105
0.00090
Validation of CFD simulations of a stirred tank
0.00075 using particle image velocimetry data. Can. J.
0.00060 Chem. Eng. 76, 611#625. Image provided
0.00045
0.00030 courtesy of the authors.
0.00015
278 8. MIXING
pattern periodically reverts to a nondispersed state and large flow oscillations and torque
and power instabilities can occur. Eventually, at high enough stirrer speeds, large gas-
filled cavities are formed behind the impeller blades, the instabilities disappear, and the
gas remains fully dispersed. Because instability and flow oscillations prior to stable gas
dispersion can lead to mechanical problems including vessel vibration, operation in this
regime is not recommended. Thus, in contrast with Rushton impellers, for downward-
pumping pitched-blade turbines there is no practical range of operating conditions
between flooding and complete gas dispersion. As neither flow instability nor incomplete
gas dispersion is desirable, the impeller should always be operated at speeds high enough
to fully distribute the gas.
The instabilities associated with downward-pumping turbines are generally thought to
occur because of the inherent opposition of flow directions generated by the impeller and
by the gas; that is, fluid driven downward by the impeller is opposed by bubble flow ris-
ing up from the sparger. As shown in Figure 8.24, two flow regimes have been identified
for downward-pumping impellers. Direct loading, which occurs at low stirrer speeds or
high gas flow rates, is characterised by gas entering the impeller region directly from the
sparger. In contrast, during indirect loading at high stirrer speeds or low gas flows, gas
approaching the impeller is swept away by the downward thrust of liquid from the stirrer
and only enters the impeller region by recirculation. Flow instabilities are associated with
the transition from direct to indirect loading, and coincide with the formation of large gas-
filled cavities behind the impeller blades. Problems with instability are greater for impellers
with four blades rather than six, and when small impeller-to-tank diameter ratios are used.
Sparger geometry and position have significant effects on the performance of
downward-pumping impellers. The use of point spargers increases the likelihood of flow
instabilities with gassing. Although pitched-blade impellers are more prone to flooding
than Rushton turbines, their gas-handling ability can be improved by replacing point spar-
gers with ring spargers of diameter about 0.8 times the impeller diameter, and optimising
the separation between the sparger and impeller [6, 7].
At low gassing rates, suspension of solids is achieved by downward-pumping agitators
with very high energy efficiency. However, severe loss of suspension capacity can occur
FIGURE 8.24 Direct and indirect gas loading regimes

for a downward-pumping pitched-blade turbine.
From M.M.C.G. Warmoeskerken, J. Speur, and J.M. Smith, 1984,
Gas#liquid dispersion with pitched blade turbines. Chem. Eng.
Commun. 25, 11#29.
Gas Gas
(a) Direct loading (b) Indirect loading
8.4 IMPELLERS 279
with gassing, especially when there are flow instabilities associated with the direct#
indirect loading transition.
Upward Pumping
Many of the problems associated with downward-pumping pitched-blade impellers
under gassed conditions can be avoided by reversing the direction of rotation so that the
impeller pumps upwards. The upward flow generated is then cocurrent with that of the
sparged gas, and the resulting flow patterns are inherently more stable than with down-
ward pumping. Yet, the use of turbine impellers in upward-pumping mode is much less
common than downward-pumping operation.
The gas#liquid flow patterns developed by upward-pumping turbines are very differ-
ent from those produced during downward flow operation. At low stirrer speeds, there is
negligible gas dispersion, the impeller is flooded, and there are no gas cavities behind the
blades. As the impeller speed is increased, there is a change from flooding to loading as
more and more gas is dispersed towards the vessel walls. With loading, streaming vortex
gas cavities are formed behind the impeller blades. Clinging cavities develop with further
increases in stirrer speed and gas bubbles start to be dispersed below the impeller. At
even higher speeds, large clinging cavities are formed and bubbles shed from the tails of
these cavities are dispersed throughout the vessel. No significant flow instabilities occur
with upward-pumping turbines. Compared with downward-pumping impellers, complete
gas dispersion is achieved at lower stirrer speeds and with less energy consumption, and
more gas can be handled before flooding occurs.
Because upward-pumping impellers generate relatively small velocities beneath the
impeller, the energy required for solids suspension is significantly greater than for
downward-pumping impellers. However, an advantage in aerated systems is that both the
agitation speed and power required for solids suspension are almost independent of the
gassing rate [8].
8.4.4 Alternative Impeller Designs

So far, we have considered the characteristics of several traditional impellers that have
been used in the chemical and bioprocessing industries for many decades. More recently,
a variety of new agitator configurations has been developed commercially. These modern
impellers have a range of technical features aimed at improving mixing in stirred tanks.
Curved-Blade Disc Turbines

Curved-blade disc turbines such as that shown in Figure 8.25 generate primarily radial
flow, similar to the Rushton turbine. However, changing the shape of the blades has a sig-
nificant effect on the impeller power requirements and gas-handling characteristics.
Rotation with the concave side forward greatly discourages the development of trailing
vortices behind the blades; therefore, with sparging, no large ventilated cavities form on
the convex surfaces. A major advantage is that the impeller is more difficult to flood, being
able to handle gas flow rates several times higher than those that cause flooding of
Rushton turbines [9].
280 8. MIXING
FIGURE 8.25 Scaba 6SRGT six-curved-blade disc

turbine.
Photograph courtesy of Scaba AB, Sweden.
FIGURE 8.26 (a) Lightnin

A315 hydrofoil impeller.
Photograph courtesy of Lightnin
Mixers, Australia.
(b) Prochem Maxflo T hydro-
foil impeller.
Photograph courtesy of Chemi-
neer Inc., Dayton, OH.
Hydrofoil Impellers
Two different hydrofoil impellers are shown in Figure 8.26. The blade angle and width
are varied along the length of hydrofoil blades, and the leading edges are rounded like an
airplane wing to reduce form drag and generate a positive lift. The shape of hydrofoil
impellers allows for effective pumping and bulk mixing with strong axial velocities and
low power consumption. Most hydrofoils are operated for downward pumping, but
upward flow is also possible. A typical mean velocity vector plot for a downward-
pumping hydrofoil impeller is shown in Figure 8.27. Downward-flowing currents with
very strong axial velocity components leave the impeller. Fluid in these currents sweeps
the vessel floor in an outward radial direction, then moves up alongside the vessel walls
before returning back to the impeller. Liquid velocities in regions of the tank above the
main circulation loops are considerably lower than below the impeller.
With aeration, downward-pumping hydrofoil impellers exhibit many of the hydrody-
namic properties of downward-pumping pitched-blade turbines. They remain more prone
to flooding than Rushton turbines, even when the impeller solidity ratio is large. The types
8.4 IMPELLERS 281
FIGURE 8.27 Mean velocity vector plot for a Prochem Maxflo
T hydrofoil impeller. The velocities were measured using laser
Doppler velocimetry.
Adapted from Z. Jaworski, A.W. Nienow, and K.N. Dyster, 1996, An
LDA study of the turbulent flow field in a baffled vessel agitated by an
axial, down-pumping hydrofoil impeller. Can. J. Chem. Eng. 74,
3#15.
Constant Ni
Increasing Fg
(a) Vortex cavity (b) Growing cavity (c) Large cavity

FIGURE 8.28 Changes in ventilated cavity structure for a Lightnin A315 hydrofoil impeller at constant stirrer
speed Ni and increasing gas flow rate Fg.
Adapted from A. Bakker and H.E.A. van den Akker, 1994, Gas#liquid contacting with axial flow impellers. Trans. IChemE
72A, 573#582.
of ventilated cavity formed behind the blades of a downward-pumping hydrofoil impeller

in low-viscosity fluid are illustrated in Figure 8.28. The cavities in Figure 8.28(a) and (b)
occur during indirect loading (Section 8.4.3, Downward Pumping subsection); at relatively
low gas flow rates, only recirculating gas reaches the impeller to form vortex or growing
cavities. With increasing gas flow, the growing cavities can exhibit various types of insta-
bility. At higher gassing rates when the impeller is no longer capable of deflecting the ris-
ing gas flow, direct loading occurs as gas enters the impeller region directly from the
sparger to produce large cavities as shown in Figure 8.28(c). Further increases in gas flow
lead to impeller flooding.
The flow instabilities generated in gassed systems by downward-pumping hydrofoil
impellers can be eliminated using wide-blade, upward-pumping hydrofoils. These impel-
lers also have high gas-handling capacity and a limited tendency to flood at low stirrer
speeds.
282 8. MIXING
8.5 STIRRER POWER REQUIREMENTS

Electrical power is used to drive impellers in stirred vessels. The average power con-
sumption per unit volume for industrial bioreactors ranges from 10 kW m23 for small ves-
sels (approx. 0.1 m3) to 1 to 2 kW m23 for large vessels (approx. 100 m3). Friction in the
stirrer motor gearbox and seals reduces the energy transmitted to the fluid; therefore, the
electrical power consumed by stirrer motors is always greater than the mixing power by
an amount depending on the efficiency of the drive. Energy costs for operation of stirrers
in bioreactors are a major ongoing financial commitment and an important consideration
in process economics.
The power required to achieve a given stirrer speed depends on the magnitude of the
frictional forces and form drag that resist rotation of the impeller. Friction and form drag
give rise to torque on the stirrer shaft; experimentally, the power input for stirring can be
determined from measurements of the induced torque M:
P 5 2πNi M ð8:5Þ
where P is power and Ni is the stirrer speed.
General guidelines for estimating the power requirements in stirred tanks are outlined
in the following sections.
8.5.1 Ungassed Newtonian Fluids

The power required to mix nonaerated fluids depends on the stirrer speed, the impeller
shape and size, the tank geometry, and the density and viscosity of the fluid. The relation-
ship between these variables is usually expressed in terms of dimensionless numbers such
as the impeller Reynolds number Rei:
Ni D2i ρ
Rei 5 ð7:2Þ
µ
and the power number NP:
P
NP 5 ð8:6Þ
ρNi3 D5i
In Eqs. (7.2) and (8.6), Ni is stirrer speed, Di is impeller diameter, ρ is fluid density, µ is
fluid viscosity, and P is power. The power number NP can be considered analogous to a
drag coefficient for the stirrer system.
The relationship between Rei and NP has been determined experimentally for a range of
impeller and tank configurations. The results for five impeller designs—Rushton turbine,
downward-pumping pitched-blade turbine, marine propeller, anchor, and helical ribbon—
are shown in Figures 8.29 and 8.30. Once the value of NP is known, the power is calculated
from Eq. (8.6) as:
P 5 NP ρNi3 D5i ð8:7Þ
For a given impeller, the general relationship between power number and Reynolds
number depends on the flow regime in the tank. The following three flow regimes can be
identified in Figures 8.29 and 8.30.
8.5 STIRRER POWER REQUIREMENTS 283
102
Laminar Transition Turbulent
10
1
ρ Ni3 Di5
P
2
NP =
1
3
0.1
1 10 102 103 104 105 106
Ni Di2 ρ
Rei =
µ
FIGURE 8.29 Correlations between the Reynolds number and power number for Rushton turbines, down-
ward-pumping pitched-blade turbines, and marine propellers in fluids without gassing.
Data from J.H. Rushton, E.W. Costich, and H.J. Everett, 1950, Power characteristics of mixing impellers. Parts I and II.
Chem. Eng. Prog. 46, 395#404, 467#476; and R.L. Bates, P.L. Fondy, and R.R. Corpstein, 1963, An examination of some
geometric parameters of impeller power. Ind. Eng. Chem. Process Des. Dev. 2, 310#314.
1. Laminar regime. The laminar regime corresponds to Rei , 10 for many impellers, including
turbines and propellers. For stirrers with small wall-clearance such as anchor and helical
ribbon mixers, laminar flow persists until Rei 5 100 or greater. In the laminar regime:
1
NP ~ or P 5 k1 µNi2 D3i ð8:8Þ
Rei
284 8. MIXING
103
Laminar
102
2
ρ Ni3 Di5 1
P
10
NP =
0.1
1 10 102 103 104 105
Ni Di2 ρ
Rei =
µ
WB WB
C C
Di Di
HL Hi HL Hi
DT DT
1 2
Impeller DT C Hi WB
Di Di Di Di
1. Anchor 1.02 0.01 1 0.1
2. Helical ribbon 1.02 0.01 1 0.1
FIGURE 8.30 Correlations between the Reynolds number and power number for anchor and helical ribbon
impellers in fluids without gassing.
From M. Zlokarnik and H. Judat, 1988, Stirring. In: W. Gerhartz, Ed., Ullmann’s Encyclopedia of Industrial
Chemistry, 5th ed., vol. B2, pp. 25-1#25-33, VCH, Weinheim, Germany.
where k1 is a proportionality constant. Values of k1 for the impellers represented in

Figures 8.29 and 8.30 are listed in Table 8.1 [10, 11]. The power required for laminar
flow is independent of the density of the fluid but directly proportional to the fluid
viscosity.
2. Turbulent regime. The power number is independent of the Reynolds number in
turbulent flow. Therefore:
P 5 N 0P ρNi3 D5i ð8:9Þ
where N 0P is the constant value of the power number in the turbulent regime.
Approximate values of N 0P for the impellers in Figures 8.29 and 8.30 are listed in
TABLE 8.1 Values of the Constants in Eqs. (8.8) and (8.9) for the
Stirred Tank Geometries Defined in Figures 8.29 and 8.30
Impeller type k1 (Rei 5 1) N 0P (Rei 5 105)
Rushton turbine 70 5.0
Pitched-blade turbine 50 1.3

Marine propeller 40 0.35
Anchor 420 0.35
Helical ribbon 1000 0.35
Table 8.1 [10, 11]. N 0P for Rushton turbines is significantly higher than for most other
impellers, indicating that the Rushton turbine has strong form drag, generates high
levels of torque, and transmits more power at the same operating speed than other
designs. Values of N 0P for a selection of other impellers and system geometries are
listed in Table 8.2. Depending on the number of blades and solidity ratio, hydrofoil
impellers have relatively low N 0P ; this reflects their aerodynamic blade design, which
effectively minimises form drag.
As indicated in Eq. (8.9), the power required for turbulent flow is independent of
the viscosity of the fluid but proportional to the density. The turbulent regime is fully
developed at Rei . 104 for most small impellers in baffled vessels. For the same
impellers in vessels without baffles, the power curves are somewhat different from
those shown in Figure 8.29. Without baffles, turbulence may not be fully developed
until Rei . 105; the value of N 0P is also reduced to as little as 10 to 50% of that with
baffles [18, 19].
3. Transition regime. Between laminar and turbulent flow lies the transition region.
Although the viscosity of fermentation broths is not usually sufficient for stirrers in
industrial bioreactors to operate in the laminar regime, many broths become sufficiently
viscous to give Reynolds numbers in the transition region. Both density and viscosity
affect the power requirements in this regime. There is usually a gradual transition from
laminar to fully developed turbulent flow in stirred tanks; the Reynolds-number range
for transition depends on the system geometry.
Equations (8.8) and (8.9) express the strong dependence of power consumption on stir-
rer diameter and, to a lesser extent, stirrer speed. Small changes in impeller size have a
large effect on power requirements, as would be expected from dependency on impeller
diameter raised to the third or fifth power. In the turbulent regime, a 10% increase in
impeller diameter increases the power requirements by more than 60%; a 10% increase in
stirrer speed raises the power required by over 30%.
Friction and form drag, and therefore the power required for stirring, are sensitive to
the detailed geometry of the impeller and configuration of the tank. The curves of
Figures 8.29 and 8.30 and the values of N 0P in Tables 8.1 and 8.2 refer to the particular
geometries specified and are subject to change if the number, size, or position of the baf-
fles, the number, length, width, pitch, or angle of the impeller blades, the liquid height,
286 8. MIXING
0
TABLE 8.2 Values of the Turbulent Ungassed Power Number NP for a Selection of Impellers
and System Geometries
0
Impeller System geometry NP Reference
Rushton turbine Flat-bottom tank 5.9 [12, 13]

Di/DT 5 0.50 HL/DT 5 1.0
WB/Di 5 0.20 Number of baffles 5 4
LB/Di 5 0.25 WBF/DT 5 0.1
Ci/DT 5 0.25
Pitched-blade turbine Flat-bottom tank 1.8 [14]
Downward pumping HL/DT 5 1.0
6 blades, 45" Number of baffles 5 4
Di/DT 5 0.40 WBF/DT 5 0.1
Ci/DT 5 0.25
Pitched-blade turbine Flat-bottom tank 1.6 [13]
Di/DT 5 0.50 WBF/DT 5 0.1
WB/Di 5 0.20 Ci/DT 5 0.25
Pitched-blade turbine Flat-bottom tank 1.6 [13, 15]
Upward pumping HL/DT 5 1.0
Di/DT 5 0.5 WBF/DT 5 0.1
WB/Di 5 0.20 Ci/DT 5 0.25
Curved-blade disc turbine (Scaba 6SRGT) Flat-bottom tank 1.5 [9]

6 blades HL/DT 5 1.0
Di/DT 5 0.33 Number of baffles 5 4
WB/Di 5 0.15 WBF/DT 5 0.1
LB/Di 5 0.28 Ci/DT 5 0.25

Hydrofoil (Lightnin A315) Flat-bottom tank 0.84 [16]
4 blades Number of baffles 5 4
Di/DT 5 0.40 WBF/DT 5 0.1
Ci/DT 5 0.25
Hydrofoil (Prochem Maxflo T) Flat-bottom tank 1.6 [17]

6 blades Number of baffles 5 4
Di/DT 5 0.35 WBF/DT 5 0.1

Ci/DT 5 0.45
the impeller clearance from the bottom of the tank, and so on, are altered. For a Rushton
turbine under fully turbulent conditions (Rei . 104), N 0P lies between about 1 and 7 depend-
ing on these parameters [19]. For axial-flow impellers, blade angle has a major influence
on power requirements. For example, N 0P for a pitched-blade turbine with six blades set at
an angle of 60" to the horizontal is more than fivefold that for the same impeller with
blade angle 30" [20]. Impeller pitch has a similarly significant effect on the power number
for propellers [19, 21]. Experimental studies have shown that blade thickness and vessel
scale can also affect N 0P [13, 22].
EXAMPLE 8.2 CALCULATION OF POWER REQU IREMENTS

A fermentation broth with viscosity 1022 Pa s and density 1000 kg m23 is agitated in a 50-m3
baffled tank using a marine propeller 1.3 m in diameter. The tank geometry is as specified in
Figure 8.29. Calculate the power required for a stirrer speed of 4 s21.
Solution
From Eq. (7.2):
4 s21 ð1:3 mÞ2 1000 kg m23

Rei 5 5 6:76 3 105
1022 kg m21 s21
From Figure 8.29, flow at this Rei is fully turbulent. From Table 8.1, N 0P is 0.35. Therefore, from
Eq. (8.9):
P 5 ð0:35Þ 1000 kg m23 ð4 s21 Þ3 ð1:3 mÞ5 5 8:3 3 104 kg m2 s23
From Table A.8 in Appendix A, 1 kg m2 s23 5 1 W. Therefore:

P 5 83 kW
8.5.2 Ungassed Non-Newtonian Fluids

Estimation of the power requirements for non-Newtonian fluids is more difficult. It is
often impossible with highly viscous fluids to achieve fully developed turbulence; under
these conditions NP is always dependent on Rei and we cannot use the constant N 0P
value in power calculations. In addition, because the viscosity of non-Newtonian liquids
varies with shear conditions, the impeller Reynolds number used to correlate power
288 8. MIXING
requirements must be redefined. Some power correlations have been developed using an
impeller Reynolds number based on the apparent viscosity µa (Section 7.5):
Ni D2i ρ
Rei 5 ð8:10Þ
µa
Therefore, from Eq. (7.12) for power-law fluids:

Ni D2i ρ
Rei 5 ð8:11Þ
Kγ_ n21
where n is the flow behaviour index and K is the consistency index. A problem with appli-
cation of Eq. (8.11) is the evaluation of γ.
_ For stirred tanks, an approximate relationship is
often used:
γ_ 5 k Ni ð7:15Þ
where the value of the constant k depends on the geometry of the impeller. The relation-
ship of Eq. (7.15) is discussed further in Section 8.15; however, for turbine impellers k is
about 10. Substituting Eq. (7.15) into Eq. (8.11) gives an appropriate Reynolds number for
pseudoplastic fluids:
Ni22n D2i ρ
Rei 5 ð8:12Þ
K kn21
The relationship between the Reynolds number Rei and the power number NP for a
Rushton turbine in a baffled tank containing pseudoplastic fluid is shown in Figure 8.31.
The upper line was measured using Newtonian fluids for which Rei is defined by
Eq. (7.2); this line corresponds to part of the curve already shown in Figure 8.29. The lower
line gives the Rei#NP relationship for pseudoplastic fluids with Rei defined by Eq. (8.12).
The laminar region extends to higher Reynolds numbers in pseudoplastic fluids than
in Newtonian systems. At Rei below 10 and above 200, the results for Newtonian and
50 FIGURE 8.31 The correlation between the

Reynolds number and power number for a
Newtonian Rushton turbine in ungassed non-Newtonian
Non-Newtonian
fluid in a baffled tank.
From A.B. Metzner, R.H. Feehs, H. Lopez Ramos,
ρNi3Di5
10 R.E. Otto, and J.D. Tuthill, 1961, Agitation of viscous

P
Newtonian and non-Newtonian fluids. AIChE J. 7,

3#9.
NP =
1
2 10 102 103
N D2 ρ N D2 ρ
Rei = i µ i or Rei = iµ i
a
non-Newtonian fluids are essentially the same. In the intermediate range, pseudoplastic
liquids require less power than Newtonian fluids to achieve the same Reynolds number.
There are several practical difficulties with application of Figure 8.31 to bioreactors. As
discussed further in Section 8.15, flow patterns in pseudoplastic and Newtonian fluids dif-
fer significantly. Even if there is high turbulence near the impeller in pseudoplastic sys-
tems, the bulk liquid may be moving very slowly and consuming relatively little power.
Another problem is that, as illustrated in Figure 7.11, the non-Newtonian parameters K
and n, and therefore µa, can vary substantially during the course of fermentation.
8.5.3 Gassed Fluids

Liquids into which gas is sparged have reduced power requirements for stirring. The
presence of gas bubbles decreases the density of the fluid; however, the influence of den-
sity as expressed in Eq. (8.9) does not explain adequately all the power characteristics of
gas#liquid systems. The main reason that gassing affects power consumption is that bub-
bles have a profound impact on the hydrodynamic behaviour of fluid around the impeller.
As described for Rushton turbines and other impellers in Section 8.4, gas-filled cavities
develop behind the stirrer blades in aerated liquids. These cavities decrease the drag forces
generated at the impeller and significantly reduce the resistance to impeller rotation, thus
causing a substantial drop in the power required to operate the stirrer compared with
nonaerated conditions.
The relationship between the power drop with gassing and operating conditions such
as the gas flow rate and stirrer speed is often represented using graphs such as those
shown in Figure 8.32. In these graphs, Pg is the power required with gassing and P0 is the
power required without gassing; P0 is the same as the power evaluated using the methods
described in Sections 8.5.1 and 8.5.2. The operating conditions are represented by the
dimensionless gas flow number Flg defined in Eq. (8.1). Figure 8.32 shows reductions in
power as a function of the gas flow number for three different impellers in low-viscosity
fluid. Each curve represents experimental data for a given impeller operated at constant
stirrer speed (i.e., for each curve, the value of Flg changes only because of changes in the
gas flow rate).
As indicated in Figure 8.32(a), power reductions of up to 70% can occur with Rushton
turbines in low-viscosity fluids, although most practical operating conditions give a 40 to
50% loss of power. The rheological properties of the fluid exert a considerable influence:
power losses of more than 80% occur with Rushton turbines in viscous or non-Newtonian
fluids sparged with gas. The fall in power with increasing gas flow rate is due to the
increased size and changes in structure of the ventilated cavities behind the blades of
Rushton turbines as illustrated in Figure 8.21. The power demand at low gas flows (low
Flg) corresponding to the formation of vortex cavities differs little from that without gas-
sing. With the formation of six clinging cavities at higher gas flow rates, the power
requirement is reduced by at most 10% compared with ungassed conditions. Development
of the ‘3#3’ cavity structure is associated with a significant reduction in power of up to
60% of the ungassed value. If the gas supply is extended beyond the gas-handling capacity
290 8. MIXING
1.2 (a) FIGURE 8.32 Variation in
Pg
P0
Vortex cavities power consumption with gas-
Ratio of gassed to ungassed power,

Clinging cavities sing Pg relative to ungassed
1.0
Formation of Transition from power consumption P0 at
‘3–3’ cavities ‘3 – 3’ to ragged constant stirrer speed. Flg is
0.8 cavities: flooding dimensionless gas flow num-
ber; Fg is volumetric gas flow
rate; Ni is stirrer speed; Di is
0.6 impeller diameter.
(a) Rushton turbine: (x) Ni 5
0.4 120 rpm; (K) Ni 5 150 rpm;
(’) Ni 5 360 rpm.
(b) Downward-pumping pitched-
0.2 blade turbine, six blades are
angled 45" : (K) Ni 5 180 rpm;
1.2 (b) (x) Ni 5 300 rpm.
Pg
P0
(c) Upward-pumping pitched-

1.0 Indirect loading: gas recirculation blade turbine, six blades are
Direct loading: formation angled 45" : (K) Ni 5 190 rpm;
of large cavities (x) Ni 5 236 rpm.
0.8
0.6
0.4
0.2
1.2 (c)
Pg
P0
1.0
0.8
0.6
0.4
0.2
0.00 0.02 0.04 0.06 0.08 0.10 0.12 0.14
Fg
Gas flow number, Flg =
NiD3i
(a) Data from M.M.C.G. Warmoeskerken, J.M. Smith, and M. Konno, 1985, On the flooding/loading transition and the com-
plete dispersal condition in aerated vessels agitated by a Rushton-turbine. In: Mixing, Proc. 5th Eur. Conf. on Mixing,
pp. 143#154, The Fluid Engineering Centre; and M.M.C.G. Warmoeskerken, J. Feijen, and J.M. Smith, 1981, Hydrodynamics
and power consumption in stirred gas#liquid dispersions. In: Fluid Mixing, IChemE Symp. Ser. 64, J1#J14.
(b) Data from M.M.C.G. Warmoeskerken, J. Speur, and J.M. Smith, 1984, Gas#liquid dispersion with pitched blade
turbines. Chem. Eng. Commun. 25, 11#29.
(c) Data from W. Bujalski, M. Konno, and A.W. Nienow, 1988, Scale-up of 45" pitch blade agitators for gas dispersion and
solid suspension. In: Mixing, Proc. 6th Eur. Conf. on Mixing, pp. 389#398, The Fluid Engineering Centre.
of the impeller, the formation of six ragged cavities coincides with impeller flooding and a
rise in power consumption as indicated in Figure 8.32(a).
Changes in power consumption with gassing for a downward-pumping pitched-blade
turbine are shown in Figure 8.32(b). At low gas flow rates corresponding to the indirect
loading regime (Section 8.4.3, Downward Pumping subsection), gas flow to the impeller
relies on gas recirculation. Under these conditions, the loss of power is relatively small
at #10%. Depending on the stirrer speed, at higher gas flows there may be a transition to
direct loading; this coincides with a sharp decrease in power consumption as large venti-
lated cavities form behind the impeller blades. The power drawn by downward-pumping
pitched-blade turbines can be reduced to as little as 30 to 40% of the power requirements
without gassing.
Whereas the power consumption by ungassed upward-pumping pitched-blade turbines is
similar to that for downward-pumping turbines of the same geometry (Table 8.2), as indicated
in Figure 8.32(c) there is a much smaller reduction in power with aeration for upward-pump-
ing impellers. Power consumption is reduced by a maximum of only about 20% during
upward-pumping operation, even though ventilated cavities form behind the impeller blades.
Power characteristics with aeration have also been measured for the impellers described
in Section 8.4.4. When the curved-blade disc turbine shown in Figure 8.25 is rotated clock-
wise with the concave sides of the blades forward, the curvature of the blades ensures that
no large ventilated cavities can form on the convex surfaces. As a consequence, in low-
viscosity fluids the power consumption with gassing remains close to that without gassing
until impeller flooding occurs. In non-Newtonian or viscous fluids, power losses may be
greater at up to about 20% [9]. For the hydrofoil impellers shown in Figure 8.26 operated
for downward pumping, depending on the stirrer speed, abrupt reductions in power can
accompany the transition from indirect to direct loading as large cavities form behind the
blades. However, this drop in power is usually less that with Rushton turbines under simi-
lar conditions [7]. In contrast, for upward-pumping hydrofoils, there is virtually no reduc-
tion in power draw with aeration over a wide range of gas flow rates [23].
As indicated in Figure 8.32, there is a strong correlation between the power consump-
tion in gassed liquids and changes in the structure of the ventilated cavities behind the
impeller blades. However, although data such as those shown in Figure 8.32 have been
measured for different impellers and illustrate how aeration affects power consumption,
these graphs cannot be used to predict impeller power requirements with gassing. This
limitation is related to the inadequacy of the gas flow number Flg to fully define the
hydrodynamic conditions affecting the development of ventilated cavities. Whereas the
value of Flg reflects the flow rate of gas from the sparger, the amount of gas actually enter-
ing the impeller region depends also on gas recirculation within the fluid. As shown in
Figure 8.32, the size of the cavities and therefore the power draw vary with stirrer speed
at constant Flg. This indicates that the results for Pg/P0 are sensitive to aspects of the mix-
ing system not represented by Flg. The exact extent to which power requirements are
reduced by sparging is a complex function of the stirrer speed, air flow rate, vessel size,
fluid properties, and the geometry of the impeller and tank including the impeller off-bot-
tom clearance and sparger size. Because all of the changes in hydrodynamic behaviour
due to gassing are not completely understood, prediction of the power requirements in
aerated systems remains difficult and cannot yet be achieved with accuracy.
292 8. MIXING
Although the power requirements in aerated systems depend strongly on the size of the
ventilated cavities behind the impeller blades, differences in the power draw with gassing
between coalescing and noncoalescing liquids (Section 10.6.2) are small, both for Rushton
turbines and for axial-flow impellers such as hydrofoils [2, 24].
Reduction in stirrer power consumption with gassing may seem a desirable feature
because of the potential for energy and cost savings during operation of the impeller.
However, when all the relevant factors are considered, stirrers with power requirements
that are relatively insensitive to gassing are preferred. In the design of fermentation equip-
ment, the stirrer motor is usually sized to allow operation under nonaerated conditions.
This is necessary to prevent motor burn-out if there is a failure of air supply during opera-
tion of the fermenter: the stirrer motor must be large enough to provide sufficient power
during any abrupt change from gassed to ungassed conditions. In addition, medium in
fermenters is often mixed without aeration during the heating and cooling cycles of in situ
batch sterilisation (Section 14.6.1). Therefore, the decrease in impeller power consumption
with gassing represents an under-utilisation of the capacity of the stirrer motor. As out-
lined in Chapter 10 (Section 10.9), the rate of oxygen transfer from gas to liquid in aerated
systems depends on the power input to the fluid; therefore, any reduction in power
diminishes the effectiveness of mass transfer in the system with potential deleterious con-
sequences for culture performance. Power losses may also reduce the ability of the stirrer
to maintain complete suspension of solids. For example, the sudden reductions in power
shown in Figure 8.32(a) and (b) can result in severe loss of suspension capacity, with the
result that cells begin to settle out on the vessel floor. An additional problem with stirrers
that lose power with aeration is our inability to predict the exact extent of the power loss
and the conditions under which it will occur. This creates some degree of uncertainty in
the operation of gassed stirrer systems. All of these factors have promoted interest in the
development of impellers such as the curved-blade disc turbine and upward-pumping
hydrofoils, for which there is minimal reduction in power draw with gassing.
8.6 POWER INPUT BY GASSING

In addition to mechanical stirring, gas sparging itself contributes to the total power
input to bioreactors during operation under aerated conditions. The power input from
sparging, Pv, can be calculated using the equation:
Pv 5 Fg ρ gHL ð8:13Þ
where Fg is the volumetric flow rate of gas at the temperature and average pressure of the
liquid in the tank, ρ is the liquid density, g is gravitational acceleration, and HL is the liq-
uid height. For aerated vessels stirred with an impeller, Pv is usually only a small fraction
of the total power input and is often neglected. However, if high gas flow rates are used at
low stirrer speeds, for example in reactors that rely mainly on gas sparging for mixing
with the stirrer playing a relatively minor role, the contribution of Pv to the total power
input can be more substantial.
8.7 IMPELLER PUMPING CAPACITY 293
8.7 IMPELLER PUMPING CAPACITY

Fluid is pumped by the blades of rotating impellers. The volumetric flow rate of fluid
leaving the blades varies with operating parameters such as the stirrer speed and size of
the impeller, but is also a characteristic of the impeller type or design. The effectiveness of
impellers for pumping fluid is represented by a dimensionless number called the flow
number:
Q
Fl 5 ð8:14Þ
Ni D3i
where Fl is the flow number, Q is the volumetric flow rate of fluid leaving the impeller
blades, Ni is the stirrer speed, and Di is the impeller diameter. The flow number is a mea-
sure of the ability of the impeller to generate strong circulatory flows, such as those neces-
sary for blending operations and solids suspension.
The value of Q during impeller operation can be obtained by measuring local fluid
velocities near the impeller blades using techniques such as laser Doppler velocimetry
(Section 7.9.3). Typical fluid discharge velocity profiles for radial- and axial-flow turbines
are shown in Figure 8.33. For radial-flow impellers, the mean discharge velocity is maxi-
mum at the centre line of the blade and decays above and below the centre line to form a
bell-shaped curve, as illustrated in Figure 8.33(a). As well as fluid leaving the edge of the
blade directly, from which Q is evaluated, there is also significant additional flow of
entrained fluid above and below the blade that is swept along by the direct discharge
stream. The total discharge flow from the impeller region including the entrained fluid
can be several times greater than Q. The discharge velocity profile for an axial-flow turbine
is shown in Figure 8.33(b). In this case, Q is the volumetric flow rate of fluid leaving
directly from the lower edges of the blades, excluding entrained flow from the surround-
ing region.
Values of the flow number Fl for several impellers operating in the turbulent regime in
low-viscosity fluids without gassing are listed in Table 8.3. Fl is dependent on the vessel
and blade geometry; however this dependence and the variation of Fl between impeller
types is not as great as for the turbulent power number N 0P .
FIGURE 8.33 Graphic of discharge

velocity profiles from the blades of: (a)
Velocity a radial-flow impeller and (b) an axial-
profile Impeller flow impeller.
blades
Q
Impeller Velocity
blade profile
Q
(a) (b)
294 8. MIXING
TABLE 8.3 Values of the Turbulent Ungassed Flow Number Fl for a Selection
of Different Impellers in Baffled Tanks Containing Low-Viscosity Fluid
Impeller Fl Reference
Rushton turbine 0.78 [25]
Propeller 0.73 [26]

3 blades
Pitched-blade turbine 0.75 [4]
Downward pumping
4 blades, 45"
Pitched-blade turbine 0.81 [27]

Downward pumping
6 blades, 45"
Hydrofoil (Lightnin A315) 0.74 [27]
Downward pumping
4 blades
Hydrofoil (Prochem Maxflo T) 0.82 [17]
Downward pumping
6 blades
The effectiveness of different impellers for generating flow can be compared relative to
their power requirements. Combination of Eqs. (8.14) and (8.9) yields an expression for the
impeller discharge flow rate per unit power consumed:
Q Fl
5 0 ð8:15Þ
P N P ρNi2 D2i
Therefore, if two different impellers of the same size are operated in the same fluid at the
same stirrer speed, their pumping efficiencies can be compared using the ratio:
! " ! "
Q Fl
N 0
P impeller 1 P impeller 1
! " 5! " ð8:16Þ
Q Fl
P impeller 2 N 0P impeller 2
We can compare the pumping efficiencies of Rushton and pitched-blade turbines using
Eq. (8.16) and values of N 0P and Fl from Tables 8.2 and 8.3. For a Rushton turbine with
Di/DT 5 0.50, N 0P 5 5:9 and Fl 5 0.78. For a six-blade downward-pumping pitched-blade
turbine of the same size, N 0P 5 1:6 and Fl 5 0.81. Therefore:
! " ! "
Q 0:81
P pitched-blade 1:6
! " 5! " 5 3:8 ð8:17Þ
Q 0:78
P Rushton 5:9
8.8 SUSPENSION OF SOLIDS 295
This result indicates that pitched-blade turbines produce almost four times the flow for
the same power input as Rushton turbines. The analysis provides an explanation for why
Rushton turbines are considered to have relatively low pumping efficiency (Section 8.4.1,
Without Gassing subsection), while pitched-blade turbines are recognised for their high
pumping capacity and effectiveness for blending operations (Section 8.4.3). A comparison
of Rushton turbines with hydrofoil impellers yields similar results. The analysis applies
only for ungassed liquids; the effect of gassing on liquid pumping rates and power con-
sumption varies considerably between different impellers.
8.8 SUSPENSION OF SOLIDS

Bioreactors used for cell culture contain biomass solids of varying size. Microorganisms
such as single-cell bacteria and yeast are small and finely divided; other cells such as
mycelia and plant cells form macroscopic aggregates or clumps depending on the culture
conditions. Because cells contain a high percentage of water, the difference in density
between the solid phase and the suspending liquid is generally very small. However, in
some bioprocesses, cells are immobilised on or in solid matrices of varying material den-
sity. These systems include anchorage-dependent animal cells cultured on microcarrier
beads, and bacteria attached to sand grains for waste water treatment.
One of the functions of mixing in stirred fermenters is to maintain the cells in suspen-
sion. Accumulation of biomass at the bottom of the vessel is highly undesirable, as cells
within settled layers have poor access to nutrients and oxygen and can become starved of
these components. It is important, therefore, to know what operating conditions are
required to completely suspend solids in stirred tanks.
8.8.1 Without Gassing

A common criterion used to define complete suspension of solids is that no particle
should remain motionless on the bottom of the vessel for more than 1 to 2 seconds.
Applying this criterion, the Zwietering equation [28]:
h i0:45
S ν 0:1
L D 0:2
p g ðρ p 2 ρ L Þ=ρL X0:13
NJS 5 ð8:18Þ
D0:85
i
is generally accepted as the best correlation for NJS, the stirrer speed required for just com-
plete suspension of solids in the absence of gassing. In Eq. (8.18), S is a dimensionless
parameter dependent on the impeller and tank geometry, ν L is the liquid kinematic viscos-
ity (Section 7.3), Dp is the diameter of the solid particles, g is gravitational acceleration, ρp
is the particle density, ρL is the liquid density, X is the weight percentage of particles in
the suspension, and Di is the impeller diameter.
Zwietering’s equation has been subjected to extensive testing over many years using a
wide range of system properties. The exponents in Eq. (8.18) are independent of the tank
size, impeller type, impeller-to-tank diameter ratio, and impeller off-bottom clearance;
296 8. MIXING
TABLE 8.4 Values of the Geometric Parameter S in Eq. (8.18) for Flat-Bottom Tanks
Impeller Di/DT Ci/DT S Reference
Rushton 0.25 0.25 12 [12]

0.33 0.17 5.8 [12]
0.33 0.25 6.7 [12]
0.33 0.50 8.0 [12]
0.50 0.25 4.25 [12]
0.50 0.17 3.9 [29]
Propeller 0.33 0.25 6.6 [29]
Pitched-blade turbine 0.33 0.20 5.7 [30]
Downward pumping, 4 blades, 45" 0.33 0.25 6.2 [30]
0.33 0.33 6.8 [30]
0.33 0.50 11.5 [30]
0.50 0.25 5.8 [12]
Downward pumping, 6 blades, 45"
Upward pumping, 6 blades, 45"
these geometric factors are reflected in the value of S. Table 8.4 lists some values of S for
different impeller geometries; these data were obtained using flat-bottomed cylindrical
vessels with four baffles of width 1/10 the tank diameter and liquid height equal to the
tank diameter.
NJS decreases significantly as the size of the impeller increases, not only because of the
direct effect of Di in Eq. (8.18) but also because the Di/DT ratio changes the value of S. For
a fixed impeller off-bottom clearance, a general relationship is:
! "α
DT
S~ ð8:19Þ
Di
where α is approximately 1.5 for Rushton turbines and 0.82 for propellers [28]. For many
impellers, S is sensitive to the impeller off-bottom clearance ratio Ci/DT. As shown in
Table 8.4 for Rushton and pitched-blade turbines, S at constant Di/DT decreases as the
impeller clearance is reduced, so that lower stirrer speeds are required for complete sus-
pension. The shape of the base of the vessel (Figure 8.2) also influences the efficiency of
solids suspension, with dished, contoured, and cone-and-fillet bases offering advantages
in some cases for reducing S. The extent of this effect depends, however, on the type of
flow pattern generated by the impeller [14, 29, 32]. The presence of obstructions near the
vessel floor, such as large sparger pipes or the bearing and seal housing for bottom-entry
stirrers, can significantly impede solids suspension. Under these conditions, application of
Eq. (8.18) will result in substantial underestimation of NJS.
8.8 SUSPENSION OF SOLIDS 297
Even if a stirrer is operated at speeds equal to or above NJS to obtain complete particle
suspension, this does not guarantee that the suspension is homogeneous throughout the
tank. Although, in general, speeds considerably higher than NJS are required to achieve
uniform particle concentration, for small particles such as dispersed cells with density sim-
ilar to that of the suspending liquid, a reasonable degree of homogeneity can be expected
at NJS. In some respects, the criterion of complete solids suspension is a severe one, as sus-
pension of the final few particles can require a disproportionately large increase in
power—for example, as much as 100% [6]. In some systems, an adequate level of solids
suspension may be achieved at stirrer speeds lower than NJS.
EXAMPLE 8.3 SOLIDS SUSPENSION

Clump-forming fungal cells are cultured in a flat-bottomed 10-m3 fermenter of diameter 2.4 m
equipped with a Rushton turbine of diameter 1.2 m operated at 50 rpm. The impeller off-bottom
clearance is 0.6 m. The density of the fermentation medium is 1000 kg m23 and the viscosity is
0.055 Pa s. The density and diameter of the cell clumps are 1035 kg m23 and 600 µm, respectively.
The concentration of cells in the fermenter reaches 40% w/w. Are the cells suspended under
these conditions?
Solution
Di/DT 5 0.50 and Ci/DT 5 0.25. Therefore, from Table 8.4, S 5 4.25. From Eq. (2.16),
g 5 9.81 m s22. From Table A.9 in Appendix A, 1 Pa s 5 1 kg m21 s21. Therefore, using Eq. (7.9) to
calculate the kinematic viscosity:
0:055 kg m21 s21

νL 5 5 5:5 3 1025 m2 s21
1000 kg m23
Substituting values into Eq. (8.18) gives:
!0:45
25 2 21 0:1 26 0:2 ð1035 2 1000Þkg m23
4:25 ð5:5 3 10 m s Þ ð600 3 10 mÞ 9:81 m s22 400:13
1000 kg m23
NJS 5
ð1:2 mÞ0:85
NJS 5 0:31 s21 5 18:6 rpm
The operating stirrer speed of 50 rpm is well above the stirrer speed required for solids suspen-
sion. The cells are therefore completely suspended.
As illustrated in Example 8.3, complete suspension of cells and small cell clumps is
generally achieved at low to moderate stirrer speeds. This is due mainly to the small
size and almost neutral density (i.e., density close to that of the suspending liquid) of the
cells.
298 8. MIXING
8.8.2 With Gassing

The formation of ventilated cavities behind impeller blades reduces the power draw
and liquid pumping capacity of the impeller, with the result that higher stirrer speeds are
often required for solids suspension in aerated systems. If the fall in power with gassing is
severe—for example, as shown in Figure 8.32(b) for downward-pumping pitched-blade
turbines—there may be a corresponding sudden loss of solids suspension capacity.
Equations relating NJSg, the stirrer speed required for just complete suspension of solids
in the presence of gassing, to NJS have been developed for various impellers [3, 31]. For
Rushton turbines of two different sizes with an impeller off-bottom clearance of one-
quarter the tank diameter [3]:
NJSg 5 NJS 1 2:4 Fgv for Di =DT 5 0:33 ð8:20Þ
NJSg 5 NJS 1 0:94 Fgv for Di =DT 5 0:50 ð8:21Þ
where Fgv is the gas flow rate in units of vvm or volume of gas per volume of liquid per
minute, and NJS and NJSg both have units of s21. The above equations apply when the
impeller is not flooded; however, they may not apply for suspension of low-density parti-
cles with (ρp 2 ρL) less than about 50 kg m23 [33], or at very low gas flow rates [34]. In
both these cases, gassing has been shown to aid rather than hinder suspension. Solids sus-
pension in aerated systems is sensitive to several aspects of tank geometry such as the
impeller off-bottom clearance and the impeller#sparger separation [3, 14].
8.9 MECHANISMS OF MIXING
In this section we consider the mechanisms controlling the rate of mixing in stirred
tanks containing a single liquid phase. As illustrated schematically in Figures 8.7 and 8.8,
large liquid circulation loops develop in stirred vessels. For mixing to be effective, the
velocity of fluid leaving the impeller must be sufficient to carry material into the most
remote regions of the tank; fluid circulated by the impeller must also sweep the entire ves-
sel in a reasonable time. In addition, turbulence must be developed in the fluid as mixing
is certain to be poor unless flow is turbulent. All these factors are important in mixing,
which can be described as a combination of three physical processes:
• Distribution
• Dispersion
• Diffusion
Distribution is sometimes called macromixing; diffusion is also called micromixing. Dispersion
can be classified as either micro- or macromixing depending on the scale of fluid motion.
The pattern of bulk fluid flow in a baffled vessel stirred by a centrally located radial-
flow impeller is shown in detail in Figure 8.34. Due to the periodic pumping action of the
impeller, the contents of the vessel are recirculated through the mixing zone in a very reg-
ular manner. Near the impeller there is a zone of intense turbulence where fluid currents
converge and exchange material. However, as fluid moves away from the impeller, flow
8.9 MECHANISMS OF MIXING 299
FIGURE 8.34 Flow pattern developed
by a centrally located radial-flow impeller.
From R.M. Voncken, J.W. Rotte, and A.Th. ten
Houten, 1965, Circulation model for continu-
ous-flow, turbine-stirred, baffled tanks. In:
Mixing—Theory Related to Practice, Proc.
Symp. 10, AIChE#IChemE Joint Meeting,
London.
becomes progressively slower and less turbulent. In large tanks, streamline or laminar
flow may develop in these local regions. Under these conditions, because fluid elements
move mostly parallel to each other in streamline flow (Section 7.2.1), mixing is not very
effective away from the impeller zone.
Let us consider what happens when a small amount of liquid dye is dropped onto the
top of the fluid in Figure 8.34. First, the dye is swept by circulating currents down to the
impeller. At the impeller there is vigorous and turbulent motion of fluid; the dye is mechan-
ically dispersed into smaller volumes and distributed between the large circulation loops.
These smaller parcels of dye are then carried around the tank, dispersing all the while into
those parts of the system not yet containing dye. Returning again to the impeller, the dye
aliquots are broken up into even smaller volumes for further distribution. After a time, dye
is homogeneously distributed throughout the tank and achieves a uniform concentration.
The process whereby dye is transported to all regions of the vessel by bulk circulation
currents is called distribution. Distribution is an important process in mixing but can be rel-
atively slow. In large tanks, the size of the circulation paths is large and the time taken to
traverse them is long; this, together with the regularity of fluid pumping at the impeller,
inhibits rapid mixing. Accordingly, distribution is often the slowest step in the mixing process.
If the stirrer speed is sufficiently high, superimposed on the distribution process is turbu-
lence. In turbulent flow, the fluid no longer travels along streamlines but moves erratically
in the form of cross-currents; this enhances the mixing process at scales much smaller than
300 8. MIXING
the scale of bulk circulation. As described in Section 7.9.1 (Eddies and Scales of
Turbulence subsection), the kinetic energy of turbulent fluid is directed into regions of
rotational flow called eddies; masses of eddies of various size coexist in turbulent flow.
Large eddies are continuously formed from the bulk flow generated by the stirrer; these
break down into small eddies that produce even smaller eddies. Eddies, like spinning tops,
possess kinetic energy that is transferred to eddies of decreasing size. When the eddies
become very small they can no longer sustain rotational motion and their kinetic energy is
dissipated as heat. At steady state in a mixed tank, most of the energy from the stirrer is
dissipated through the eddies as heat; energy lost in other processes (e.g., fluid collision
with the tank walls) is generally negligible.
The process of breaking up the bulk flow into smaller and smaller eddies is called dis-
persion. Dispersion facilitates rapid transfer of material throughout the vessel. The degree
of homogeneity possible as a result of dispersion is limited by the size of the smallest
eddies that may be formed in a particular fluid. This size is given approximately as the
Kolmogorov scale of mixing or scale of turbulence, λ, defined in Eq. (7.36) as:
! 3 "1=4
ν
λ5 ð7:36Þ
ε
where λ is the characteristic dimension of the smallest eddies, ν is the kinematic viscosity
of the fluid (Section 7.3), and ε is the local rate of turbulence energy dissipation per unit
mass of fluid. At steady state, the average rate of energy dissipation by turbulence over
the entire tank is equal to the power input to the fluid by the impeller; this power input is
the same as that estimated using the methods of Section 8.5. According to Eq. (7.36), the
greater the power input to the fluid, the smaller are the eddies. λ is also dependent on vis-
cosity: at a given power input, smaller eddies are produced in low-viscosity fluids. For
low-viscosity liquids such as water, λ is usually in the range 30 to 100 µm. For such fluids,
this is the smallest scale of mixing achievable by dispersion.
Within eddies, flow of fluid is rotational and occurs in streamlines. Because streamline
flow does not facilitate mixing, to achieve mixing on a scale smaller than the Kolmogorov
scale, we must rely on diffusion. Molecular diffusion is generally regarded as a slow pro-
cess; however, over small distances it can be accomplished quite rapidly. Within eddies of
diameter 30 to 100 µm, homogeneity is achieved in about 1 s for low-viscosity fluids.
Consequently, if the power input to a stirred vessel produces eddies of this dimension,
mixing on a molecular scale is accomplished virtually simultaneously.
8.10 ASSESSING MIXING EFFECTIVENESS

As explained in the last section, to achieve rapid mixing in a stirred tank, the agitator
must provide good bulk circulation or macromixing. Micromixing at or near the molecular
scale is also important, but occurs relatively quickly compared with macromixing. Mixing
effectiveness is therefore usually a reflection of the rate of bulk flow.
Mixing time is a useful parameter for assessing the overall speed of mixing in stirred
vessels. The mixing time tm is the time required to achieve a given degree of homogeneity
8.10 ASSESSING MIXING EFFECTIVENESS 301
starting from the completely segregated state. It can be measured by injecting a tracer into
the vessel and following its concentration at a fixed point in the tank. Tracers in common
use include acids, bases, and concentrated salt solutions; corresponding detectors are pH
probes and conductivity cells. Mixing time can also be determined by measuring the tem-
perature response after addition of a small quantity of heated liquid.
Let us assume that a small pulse of tracer is added to fluid in a stirred tank already con-
taining tracer material at concentration Ci. When flow in the system is circulatory, the
tracer concentration measured at some fixed point in the tank can be expected to follow a
pattern similar to that shown in Figure 8.35. Before mixing is complete, a relatively high
concentration will be detected every time the bulk flow brings tracer to the measurement
point. The peaks in concentration will be separated by a period approximately equal to the
average time required for fluid to traverse one bulk circulation loop. In stirred vessels this
period is called the circulation time, tc. After several circulations the desired degree of
homogeneity is reached.
Definition of the mixing time tm depends on the degree of homogeneity required.
Usually, mixing time is defined as the time after which the concentration of tracer differs
from the final concentration Cf by less than 10% of the total concentration difference
(Cf 2 Ci). However, there is no single, universally applied definition of mixing time; some-
times deviations greater or less than 10% are specified. Nevertheless, at tm the tracer con-
centration is relatively steady and the fluid composition approaches uniformity.
Industrial-scale stirred vessels with working volumes between 1 and 100 m3 have mixing
times between about 30 and 120 s, depending on conditions. For rapid and effective mix-
ing, tm should be as small as possible.
Intuitively, we can predict that the mixing time in stirred tanks will depend on vari-
ables such as the size of the tank and impeller, the fluid properties, and the stirrer speed.
The relationship between mixing time and several of these variables has been determined
experimentally for different impellers: results for a Rushton turbine in a baffled tank are
shown in Figure 8.36. The dimensionless product Nitm, which is also known as the homoge-
nisation number or dimensionless mixing time, is plotted as a function of the impeller
FIGURE 8.35 Concentration

response after dye is injected
into a stirred tank.
tc tc tc
Tracer concentration
0.1(Cf – Ci)
Cf
Ci
tm
0 Time
302 8. MIXING
4 FIGURE 8.36 Variation of mixing time with

Reynolds number for a Rushton turbine in a baf-
fled tank. The impeller is located one-third the
2 tank diameter off the floor of the vessel, the
impeller diameter is one-third the tank diameter,
the liquid height is equal to the tank diameter,
103
8 and the tank has four baffles of width one-tenth
the tank diameter. Several measurement techni-
6
ques and tank sizes were used: (K) thermal
4 method, vessel diameter 1.8 m; (x) thermal
method, vessel diameter 0.24 m; (∆) decoloura-
tion method, vessel diameter 0.24 m.
Ni tm 2 Reprinted from C.J. Hoogendoorn and A.P. den
Hartog, Model studies on mixers in the viscous flow
region, Chem. Eng. Sci. 22, 1689#1699. Copyright
102 1967, with permission from Pergamon Press Ltd,
8 Oxford.
6
10
102 2 4 6 8 103 2 4 6 8 104
Rei
Reynolds number Rei; tm is the mixing time based on a 10% deviation from the total
change in conditions, and Ni is the rotational speed of the stirrer. Conceptually, Ni tm
represents the number of stirrer rotations required to homogenise the liquid after addition
of a small pulse of tracer. At relatively low Reynolds numbers in the laminar#transition
regime, Nitm increases significantly with decreasing Rei. However, as the Reynolds number
is increased, Nitm approaches a constant value that persists into the turbulent regime at Rei
above about 5 3 103. The relationship between Nitm and Rei for most other impellers is
qualitatively similar to that shown in Figure 8.36 [10]; in practice, therefore, we can
assume that Nitm reaches a constant value in turbulent flow. With Nitm constant, mixing
time reduces in direct proportion to increase in stirrer speed.
An equation has been developed for estimating the mixing time in stirred vessels under
turbulent flow conditions. This expression for tm can be applied irrespective of the type of
impeller used [35, 36]:
! "1=3 ! "1=3
2=3 ρVL DT
tm 5 5:9 DT ð8:22Þ
P Di
where tm is the mixing time, DT is the tank diameter, ρ is the liquid density, VL is the liq-
uid volume, P is the power input, and Di is the impeller diameter. Equation (8.22) applies
to baffled vessels stirred with a single impeller and with liquid height equal to the tank
diameter. The relationship has been verified using a range of different impellers with
8.10 ASSESSING MIXING EFFECTIVENESS 303
0.2 # Di/DT # 0.7 in vessels of diameter up to 2.7 m. The equation is also valid under aer-
ated conditions provided the impeller disperses the gas effectively (i.e., is not flooded) and
P is the power drawn with gassing [36]. As Eq. (8.22) does not depend on the type of
impeller, we can deduce that all impellers are equally energy efficient with respect to mix-
ing time. Equation (8.22) indicates that, for a tank of fixed diameter and liquid volume,
mixing time is reduced if we use a large impeller and a high power input.
For a cylindrical tank with liquid height equal to the tank diameter, the geometric for-
mula for the volume of a cylinder is:
π
VL 5 D3T ð8:23Þ
4
Also, as Eq. (8.22) applies under turbulent conditions, we can express P in terms of the tur-
bulent power number N 0P using Eq. (8.9). Substituting Eqs. (8.9) and (8.23) into Eq. (8.22)
gives:
! " ! "
5:4 1 1=3 DT 2
tm 5 ð8:24Þ
Ni N 0P Di
Equation (8.24) indicates that mixing time reduces in direct proportion to stirrer speed.
This is the same as saying that Nitm for a given impeller and tank geometry is constant for
turbulent flow, as discussed earlier with reference to Figure 8.36. At constant Ni, tm is
directly proportional to (DT/Di)2, showing that mixing times can be reduced significantly
using impellers with large Di/DT ratio. However, because of the strong influence of impel-
ler diameter on power requirements (Section 8.5.1), increasing Di also raises the power
consumption, so there will be a cost associated with using this strategy to improve mixing.
EXAMPLE 8.4 ESTIMATION OF MIXING TIME

A baffled fermenter with tank diameter and liquid height equal to 1.2 m is stirred using a six-
blade downward-pumping Prochem Maxflo T hydrofoil impeller. The impeller diameter is
0.42 m and the stirrer speed is 1.5 s21. The viscosity of the fermentation broth is 1022 Pa s and the
density is 1000 kg m23. Estimate the mixing time under nonaerated conditions.
Solution
From Eq. (7.2):
1:5 s21 ð0:42 mÞ2 1000 kg m23

Rei 5 5 2:6 3 104
1022 kg m21 s21
Flow is turbulent for remote-clearance impellers at this Reynolds number; therefore Eq. (8.24)
can be used to calculate tm. From Table 8.2, N 0P for this hydrofoil impeller is equal to 1.6.
Therefore:
! "1=3 ! "
5:4 1 1:2 m 2
tm 5 5 25:1 s
1:5 s21 1:6 0:42 m
The mixing time is 25 s.
304 8. MIXING
Because the parameters affecting mixing efficiency also affect stirrer power require-
ments, it is not always possible to achieve small mixing times without consuming enor-
mous amounts of energy, especially in large vessels. Relationships between power
requirements, mixing time, and tank size are explored further in the following section.
8.11 SCALE-UP OF MIXING SYSTEMS

Design of industrial-scale bioprocesses is usually based on the performance of small-scale
prototypes. It is always better to know whether a particular process will work properly
before it is constructed in full size: determining optimum operating conditions at production
scale is expensive and time-consuming. Ideally, scale-up should be carried out so that condi-
tions in the large vessel are as close as possible to those producing good results at the smaller
scale. As mixing is of critical importance in bioreactors, it would seem desirable to keep the
mixing time constant on scale-up. Unfortunately, as explained in this section, the relationship
between mixing time and power consumption makes this rarely possible in practice.
Suppose a cylindrical 1-m3 pilot-scale stirred tank is scaled up to 100 m3. Let us con-
sider the power required to maintain the same mixing time in the large and small vessels.
From Eq. (8.22), equating tm values gives:
! "1=3 ! " ! "1=3 ! "
2=3 ρ1 VL1 DT1 1=3 2=3 ρ2 VL2 DT2 1=3
5:9 DT1 5 5:9 DT2 ð8:25Þ
P1 Di1 P2 Di2
where subscript 1 refers to the small-scale system and subscript 2 refers to the large-scale
system. If the tanks are geometrically similar, the ratio DT/Di is the same at both scales.
Similarly, the fluid density will be the same before and after scale-up, and we can also
cancel the constant multiplier from both sides. Therefore, Eq. (8.25) reduces to:
! " ! "
VL1 1=3 VL2 1=3
DT1 2=3 5 DT2 2=3 ð8:26Þ
P1 P2
Cubing both sides of Eq. (8.26) and rearranging gives:

! "
DT2 2 VL2
P 2 5 P1 ð8:27Þ
DT1 VL1
The geometric relationship between VL and DT is given by Eq. (8.23) for cylindrical tanks
with liquid height equal to the tank diameter. Solving Eq. (8.23) for DT gives:
! "
4VL 1=3
DT 5 ð8:28Þ
π
Substituting this expression into Eq. (8.27) for both scales gives:
! "
VL2 5=3
P 2 5 P1 ð8:29Þ
VL1
8.12 IMPROVING MIXING IN FERMENTERS 305
In our example of scale-up from 1 m3 to 100 m3, VL2 5 100 VL1. Therefore, the result
from Eq. (8.29) is that P2 5 B2000 P1; that is, the power required to achieve equal mixing
time in the 100-m3 tank is B2000 times greater than in the 1-m3 vessel. This represents an
extremely large increase in power, much greater than is economically or technically feasi-
ble with most equipment used for stirring. This example illustrates why the criterion of
constant mixing time can hardly ever be applied for scale-up. Because the implications for
power consumption are impractical, it is inevitable that mixing times increase with scale.
Reduced culture performance and productivity often accompany scale-up of bioreactors
as a result of lower mixing efficiencies and consequent alteration of the physical environ-
ment. One way to improve the design procedure is to use scale-down methods. The general
idea behind scale-down is that small-scale experiments to determine operating parameters
are carried out under conditions that can actually be realised, physically and economically,
at the production scale. For example, if we decide that power input to a large-scale vessel
cannot exceed a certain limit, we can calculate the corresponding mixing time in the larger
vessel, then use an appropriate power input to a small-scale reactor to simulate the mixing
conditions that are in the large-scale system. Using this approach, as long as the flow
regime (e.g., turbulent flow) is the same in the small- and large-scale fermenters, there is a
better chance that the results achieved in the small-scale unit will be reproducible in the
larger system.
8.12 IMPROVING MIXING IN FERMENTERS
Because of the limitations outlined in the previous section, longer mixing times are
often unavoidable when stirred vessels are scaled up in size. In these circumstances, it is
not possible to reduce mixing times sufficiently by simply raising the power input to the
stirrer. In this section we consider methods for improving mixing in stirred tanks that do
not involve consumption of significantly greater amounts of energy.
8.12.1 Impeller and Vessel Geometry

Mixing can sometimes be improved by changing the system’s physical configuration.
• Baffles should be installed; this is routine for stirred fermenters and produces greater
turbulence.
• For efficient mixing, the impeller should be mounted below the geometric centre of the
vessel. For example, mixing by radial impellers such as Rushton turbines is facilitated
when the circulation currents below the impeller are smaller than those above
(as shown in Figure 8.7), as this makes the upper and lower circulation loops
asynchronous. Under these conditions, fluid particles leaving the impeller at the same
instant but entering different circulation paths take different periods of time to return
and exchange material. The rate of distribution throughout the vessel is increased when
the same fluid particles from different circulation loops do not meet each other every
time they return to the impeller region.
306 8. MIXING
• For two- and three-phase mixing (i.e., in gas#liquid and gas#liquid#solid systems),
good mixing includes achieving complete gas dispersion and solids suspension. These
stirrer functions are sensitive to various aspects of tank geometry, including the
impeller off-bottom clearance, type of sparger, clearance between the sparger and the
impeller, and base profile of the tank. Optimisation of these features of the system can
yield considerable improvements in mixing effectiveness without necessarily requiring
large amounts of extra power.
• Although installation of multiple impellers on the same stirrer shaft may seem a
solution to problems of poor mixing, as discussed further in Section 8.13, the power
required increases substantially when extra impellers are fitted. Furthermore,
depending on the impeller design and the separation allowed between the impellers,
mixing efficiency can actually be lower with multiple impellers than in single-impeller
systems.
8.12.2 Feed Points

Severe mixing problems can occur in industrial-scale fermenters when material is fed
into the vessel during operation. Concentrated acid or alkali and antifoam agents are often
pumped automatically into the broth for pH and foam control; fermenters operated with
continuous flow or in fed-batch mode also have fresh medium and nutrients added during
the culture. If mixing and bulk distribution are slow, very high local concentrations of
added material develop near the feed point. This problem has been observed in many
types of culture but is particularly acute during large-scale production of single-cell pro-
tein from methanol. Because high levels of methanol are toxic to cells, biomass yields
decrease significantly when mixing of feed material into the broth is slow. Such problems
can be alleviated by installing multiple injection points to aid the distribution of substrate
throughout the vessel. It is much less expensive to do this than to increase the stirrer
speed and power input.
Location of the feed point (or feed points) is also important. In most commercial opera-
tions, material is fed into bioreactors using a single inlet delivering to the top surface of
the liquid. However, mixing can be improved substantially by feeding directly into the
impeller zone. This ensures rapid distribution and dispersion as convective currents and
turbulence are strongest in this region. In many respects, use of surface-entry feeding
represents the worst choice of feed-point location; as illustrated in Figures 8.14 and 8.27,
fluid velocities in the upper reaches of stirred vessels can be very weak. Under these con-
ditions, flow may be virtually stagnant in the regions where feeding is taking place, result-
ing in very poor rates of blending.
8.13 MULTIPLE IMPELLERS
Typical bioreactors used for aerobic fermentations do not conform to the standard con-
figuration for stirred tanks illustrated in Figure 8.1, where the liquid height is approxi-
mately equal to the tank diameter. Instead, aerobic cultures are carried out in tall vessels
8.13 MULTIPLE IMPELLERS 307
with liquid heights 2 to 5 times the tank diameter, an aspect ratio of 3:1 being common.
The reasons for using this geometry are that relatively high hydrostatic pressures are pro-
duced in tall vessels filled with liquid, thus increasing the solubility of oxygen, while ris-
ing air bubbles have longer contact time with the liquid, thus improving oxygen transfer
from the gas phase.
Mixing in tall fermenters is carried out using more than one impeller mounted on the
stirrer shaft. Each impeller generates its own circulation currents, but interaction between
the fluid streams from different impellers can produce very complex flow patterns. An
important parameter affecting the performance of multiple impellers is the spacing
between them.
8.13.1 Multiple Rushton Turbines without Gassing

Because Rushton turbines have been the standard impeller in the bioprocessing indus-
try for many years, the agitation system in many fermentation vessels consists of multiple
Rushton turbines. In low-viscosity fluids, if Rushton turbines are spaced adequately apart,
they each produce a radial discharge stream and generate independent large-scale circula-
tion loops as illustrated in Figure 8.37. In this case, a vessel equipped with three Rushton
turbines is mixed as if three separate stirred tanks were stacked one on top of the other.
Under these conditions without gassing, the power required by multiple impellers can be
estimated using the following equation:
ðPÞn 5 n ðPÞ1 ð8:30Þ
where (P)n is the power required by n impellers and (P)1 is the power required by a single
impeller. The minimum spacing between the impellers for the flow pattern of Figure 8.37
FIGURE 8.37 Independent circulation loops generated by multiple Rushton tur-

bines in a tall fermenter when the spacing between the impellers is relatively large.
308 8. MIXING
to occur, and for Eq. (8.30) to be valid, is not well defined, being reported variously as one
to three impeller diameters or one tank diameter [37]. Equation (8.30) indicates that, at this
spacing, each impeller draws the same power as if it were operating alone.
An important drawback associated with operation of multiple Rushton turbines is com-
partmentation of the fluid. As indicated in Figure 8.37, the radial flow pattern generated
by Rushton turbines creates separate circulation currents above and below each impeller,
providing little opportunity for interaction between the fluid streams emanating from dif-
ferent impellers. The resulting segregation creates a strong barrier to axial flow and top-to-
bottom mixing of the contents of the tank over the entire height of the vessel. As a conse-
quence, at a fixed stirrer speed, the overall rate of mixing is lower with multiple Rushton
turbines than in a standard single-impeller system. In other words, installation of two
Rushton impellers in a vessel with liquid height twice the tank diameter does not achieve
the same mixing time as one Rushton impeller in liquid with height equal to the tank
diameter. This result reflects the slow speed of fluid exchange between the separate circu-
lation loops induced by multiple Rushton turbines compared with the rate of convective
flow developed by a single impeller. In multiple impeller systems, the rate of exchange flow
between the fluid compartments generated by each impeller determines the rate of overall mixing in
the vessel. Therefore, improving the exchange between compartments has a high priority in
mixing operations using multiple impellers.
When material is added to fermenters with multiple Rushton turbines, the location of
the feed point has a significant effect on mixing time. This is illustrated in Figure 8.38, in
which mixing time is plotted as a function of the height of the injection point in a vessel
stirred with dual Rushton turbines. In this experiment, the clearance between the impellers
was two times the impeller diameter or one tank diameter. The mixing time was lowest
when the tracer was injected at the height where the circulation loops from the upper and
lower impellers came together. Using this strategy, the barriers to mixing between the cir-
culation loops were minimised as tracer became available to both impellers relatively
quickly. The mixing time under these conditions was less than half that when material
FIGURE 8.38 Variation in mixing time

with height of the tracer injection point for
1.38 dual Rushton turbines with clearance
between the impellers equal to the tank
1.14
Injection height (m)
diameter.
From D.G. Cronin, A.W. Nienow, and G.W.
0.90 Moody, 1994, An experimental study of mixing
in a proto-fermenter agitated by dual Rushton
0.66 turbines. Trans. IChemE 72C, 35#40.
0.42
0.18
0
5 10 15 20
Mixing time (s)
8.13 MULTIPLE IMPELLERS 309
was fed to the top surface of the liquid. If more than two impellers are present, similar
advantage could be obtained by injecting feed material separately into each impeller zone.
Impeller spacings other than those giving the flow conditions illustrated in Figure 8.37
have also been tested [38]. If the impellers are moved further apart, the chances of material
being exchanged between the circulation loops is reduced further, resulting in poorer
overall mixing. On the other hand, if the impellers are moved closer together, the circula-
tion loops from each impeller impinge on each other, allowing better fluid interchange.
However, depending on the liquid height and the position of the impellers within the
tank, placing impellers close together may leave large volumes of liquid in the upper or
lower regions of the vessel less well mixed. Therefore, although inter-impeller mixing is
improved, overall mixing is reduced. Additional impellers could be installed to overcome
this problem but this would require a substantial increase in the total power input to the
system.
8.13.2 Other Impeller Combinations without Gassing

Various combinations of impellers, including Rushton turbines, curved-blade disc tur-
bines, propellers, and pitched-blade and hydrofoil impellers in both downward-pumping
and upward-pumping modes, have been tested for multiple-impeller mixing. Combining
radial- and axial-flow impellers, or combining different axial-flow turbines, significantly
reduces fluid compartmentation and enhances mixing compared with multiple Rushton
turbines. However, the flow patterns generated can be complex and show strong sensitiv-
ity to impeller geometry, diameter, and clearance.
A typical mean velocity vector plot for a dual-impeller system comprising a Rushton
turbine in the lower position and a downward-pumping pitched-blade turbine above is
shown in Figure 8.39. The upper impeller generates strong downward flow that enters the
lower impeller axially from the top. The lower Rushton turbine generates an outward
FIGURE 8.39 Mean velocity vector plot for a dual-impeller combination

of a Rushton turbine in the lower position and a downward-pumping
pitched-blade turbine above. The velocities were measured using laser
Doppler velocimetry.
From V.P. Mishra and J.B. Joshi, 1994, Flow generated by a disc turbine. Part IV:
Multiple impellers. Trans. IChemE 72A, 657#668.
310 8. MIXING
radial jet, which divides into two streams near the wall. The fluid flowing downwards at
the wall forms a large-scale circulation loop below the Rushton impeller, while the fluid
flowing upwards joins the major circulation current generated by the pitched-blade tur-
bine. In this way, in contrast to the independent circulation loops developed by multiple
Rushton turbines illustrated in Figure 8.37, the radial-flow#axial-flow combination in
Figure 8.39 allows a greater degree of interaction between the fluid currents originating
from each impeller. Only two major fluid compartments are generated, in contrast to the
four separate circulation loops (i.e., one above and one below each impeller) that would be
created with dual Rushton turbines. Generally, the smaller the number of separate large-
scale flow loops in the vessel, the better is the mixing.
As discussed in Section 8.10, for single impellers under turbulent flow conditions, the
mixing time tm for a given power input is independent of impeller type. Therefore,
Eqs. (8.22) and (8.24) can be applied irrespective of the impeller design. In contrast, in
multiple impeller systems, the mixing time depends strongly on the type (or types) of
impeller used. This is because the extent of interaction between the flow currents gener-
ated by each impeller, which has a significant effect on overall mixing, varies with impel-
ler design.
For operation without gassing and with individual impellers spaced at least one impel-
ler diameter apart, Eq. (8.30) provides a reasonable basis for estimating the power required
by combinations of radial- and axial-flow impellers or multiple axial-flow turbines [39, 40].
The total power for the multiple impeller system is approximately equal to the sum of the
individual power requirements for each impeller.
8.13.3 Multiple-Phase Operation

When gassing is required in vessels with multiple impellers, Rushton or curved-blade
disc turbines are often used in the lowest position closest to the sparger because of their
superior ability to handle gas compared with axial-flow turbines. Combining hydrofoil or
pitched-blade impellers above a Rushton turbine is very effective in aerated systems: the
lower impeller breaks up the gas flow while the upper, high-flow impeller/s distribute the
dispersed gas throughout the tank.
With gassing, the power relationship for multiple impellers is not as simple as that in
Eq. (8.30). The main reason is that the power required by the individual impellers may be
affected by the presence of gas, but each impeller is not affected to the same extent. At
the lowest impeller, the formation of ventilated cavities at the impeller blades reduces the
power drawn as described in Section 8.5.3. However, as the quantity of gas passing
through the upper impellers is much smaller than that handled by the lowest impeller, the
reduction in power at the upper impellers is less significant. The effect of gassing on
the power required by the upper impellers depends on the distance of the impeller from
the sparger and the level of gas recirculation in the system. Although the lower impeller
can be flooded if the gas flow rate is too high relative to the stirrer speed, the upper impel-
lers are rarely flooded and continue to pump liquid effectively.
Solids suspension in multiple impeller systems has been studied comparatively little. In
general, irrespective of the liquid height, there appears to be no advantage associated with
8.14 RETROFITTING 311
using more than one impeller to suspend solids, with respect to either the stirrer speed or
power required. At the same time, however, the use of multiple impellers may improve
the uniformity of particle concentration throughout the tank [3, 41].
8.14 RETROFITTING
It is sometimes necessary or desirable to change the type of impeller used in existing

fermentation vessels. For many years, Rushton turbines with diameter one-third the tank
diameter were used almost exclusively for bioprocessing applications, and most fermenta-
tion facilities were constructed for operation with this impeller geometry. However, with
increasing recognition of the importance of mixing in bioprocesses, and as modern impel-
lers with improved features are being developed, the idea of replacing Rushton turbines
with new impellers in retrofitting operations is now being pursued.
An important requirement for retrofitting is that neither the agitator drive nor the drive
assembly be modified. This is because the expense associated with replacing the shaft,
seals, bearings, gearbox, and possibly also the motor itself adds considerably to the overall
cost. Therefore, ideally, retrofitting is carried out so that the power draw and torque in the
new stirrer system are the same as for the old stirrer. From Eq. (8.5), maintaining the same
power and torque implies that the operating stirrer speed Ni will also be the same.
For operation in the turbulent regime, we can calculate the size of the new impeller
using Eq. (8.9). If the new and old impellers consume the same power, from Eq. (8.9):
ðN 0P ρ Ni3 D5i Þnew 5 ðN 0P ρ Ni3 D5i Þold ð8:31Þ
As the liquid density and stirrer speed will be the same:

ðN 0P Þold
ðDi Þ5new 5 ðDi Þ5old ð8:32Þ
ðN 0P Þnew
or
! "1=5
ðN 0P Þold
ðDi Þnew 5 ðDi Þold ð8:33Þ
ðN 0P Þnew
Because Rushton turbines have a relatively high turbulent power number N 0P , if a Rushton
turbine is replaced with a low-N 0P impeller such as a hydrofoil or curved-blade disc tur-
bine, Eq. (8.33) indicates that the diameter of the new impeller will be larger than the old.
This has implications especially for improved gas handling and dispersion, as discussed
in Section 8.4.1 (With Gassing subsection). Additional benefits may also apply: for exam-
ple, if a curved-blade disc turbine is chosen as the replacement impeller, the loss of power
with aeration will be much smaller compared with a Rushton turbine, thus allowing
improved rates of oxygen transfer. In multiple impeller systems, replacing upper Rushton
turbines with larger-diameter upward- or downward-pumping axial-flow turbines signifi-
cantly reduces fluid compartmentation and enhances bulk mixing.
312 8. MIXING
While there are advantages associated with retrofitting, there are also potential difficul-
ties. Because of the different directions of fluid discharge from radial- and axial-flow
impellers, different stresses are exerted by these types of impeller on the stirrer shaft, gear
drive, and tank. The vessel and stirrer assembly must be able to withstand any change in
external load after retrofitting. Mechanical as well as hydrodynamic instabilities, including
increased vessel vibration, may occur in gassed systems after retrofitting of downward-
pumping axial-flow impellers [42]; these problems are largely absent with Rushton tur-
bines. Other factors may also affect retrofitting of larger-diameter impellers: for example,
the presence of cooling coils and other fittings inside the vessel could limit the extent to
which the impeller size can be increased, or the stirrer assembly may not be capable of
supporting the increased impeller weight.
8.15 EFFECT OF RHEOLOGICAL PROPERTIES ON MIXING
As discussed in Sections 7.7 and 7.8, many fermentation broths have high viscosity or
exhibit non-Newtonian flow behaviour. These properties have a profound influence on
mixing, making it more difficult to achieve small mixing times and homogeneous broth
composition. The principal deleterious effects of high fluid viscosity and non-Newtonian
rheology are reduced turbulence and the formation of stagnant zones in the vessel.
For effective mixing, flow must be turbulent. As described in Section 8.9, turbulence is
responsible for dispersing material at the scale of the smallest eddies. The existence of tur-
bulence is indicated by the value of the impeller Reynolds number Rei. Turbulence is
damped at Rei below about 104 for remote-clearance impellers; as a consequence, mixing
times increase significantly as shown in Figure 8.36. Rei as defined in Eq. (7.2) is inversely
proportional to viscosity. Accordingly, nonturbulent flow and poor mixing are likely to
occur during agitation of highly viscous fluids. Increasing the power input is an obvious
solution; however, raising the power sufficiently to achieve turbulence is often impractical.
Most non-Newtonian fluids in bioprocessing are pseudoplastic. Because the apparent
viscosity of these fluids depends on the shear rate, their rheological behaviour varies with
the shear conditions in the fermenter. Metzner and Otto [43] proposed that the average
shear rate γ_ av in a stirred vessel is a linear function of the stirrer speed Ni:
γ_ av 5 k Ni ð8:34Þ
where k is a constant dependent on the type of impeller. Experimentally determined

values of k are listed in Table 8.5. The validity of Eq. (8.34) was established in studies by
Metzner et al. [44]. However, like other properties such as fluid velocity and turbulence
kinetic energy, the shear rate in stirred vessels is far from uniform, being strongly depen-
dent on distance from the impeller. Figure 8.40 shows estimated values of the shear rate in
a pseudoplastic fluid as a function of radial distance from the tip of a Rushton turbine.
The maximum shear rate close to the impeller is much higher than the average calculated
using Eq. (8.34).
Pseudoplastic fluids are shear-thinning; that is, their apparent viscosity decreases with
increasing shear. Accordingly, in stirred vessels, pseudoplastic fluids have relatively low
8.15 EFFECT OF RHEOLOGICAL PROPERTIES ON MIXING 313
TABLE 8.5 Observed Values of k in Eq. (8.34)
Impeller type k
Rushton turbine 10#13

Propeller 10
Anchor 20#25
Helical ribbon 30
From S. Nagata, 1975, Mixing: Principles and Applications, Kodansha, Tokyo.
FIGURE 8.40 Shear rates in a pseudoplastic fluid as a

function of the stirrer speed and radial distance from the
impeller: (x) impeller tip; (¢) 0.10 in.; (’) 0.20 in.; (r)
240 0.34 in.; (K) 0.50 in.; (&) 1.00 in.
From A.B. Metzner and J.S. Taylor, 1960, Flow patterns in
agitated vessels. AIChE J. 6, 109#114.
Shear rate, γ (s–1)
160
80
0
0 200 400
Rotational speed, Ni (min–1)
apparent viscosity in the high-shear zone near the impeller, and relatively high apparent
viscosity away from the impeller. As a result, flow patterns similar to that shown in
Figure 8.41 can develop. Highly shear-thinning fluids with flow behaviour index n
(Section 7.5.1) less than 0.2 to 0.3 form caverns when subjected to agitation. Caverns are cir-
culating pools of fluid surrounding the impeller; outside the caverns, the bulk liquid
scarcely moves at all. Caverns also develop in liquids that exhibit a yield stress
(Section 7.5.1); in this case, fluid remains stagnant in regions away from the impeller
where the yield stress is not exceeded. Mixing inside caverns is intense, but there is very
little exchange of material between the cavern and the rest of the tank. The effect of stirrer
speed on cavern size is shown in Figure 8.42. Although the cavern expands as a propor-
tion of the tank volume with increasing stirrer speed, dead zones persist in the peripheral
314 8. MIXING
FIGURE 8.41 Mixing pattern for pseudoplastic fluid in a stirred

fermenter.
Stagnant
zones
FIGURE 8.42 Effect of stirrer speed on

cavern size in non-Newtonian, shear-thin-
ning fluid with a yield stress.
From J. Solomon, T.P Elson, A.W. Nienow, and
G.C. Pace, 1981, Cavern sizes in agitated fluids
with a yield stress. Chem. Eng. Commun. 11,
143#164. Reproduced with permission from
Taylor & Francis.
(a) 258 rpm (b) 468 rpm
(c) 600 rpm (d) 1020 rpm
8.16 ROLE OF SHEAR IN STIRRED FERMENTERS 315
regions of the vessel even with vigorous agitation and high power input. Therefore, in fer-
menters containing highly non-Newtonian broths (e.g., for production of biological gums
such as xanthan) stagnant zones are very likely to develop in regions of the vessel away
from the impeller caverns.
The effects of local fluid thinning in pseudoplastic fluids can be countered by modifying
the geometry of the system and/or the impeller design. The use of multiple impellers,
even when the liquid height is no greater than the tank diameter, improves the mixing of
pseudoplastic fluids significantly. The combination of a Rushton turbine in the lower posi-
tion and downward-pumping pitched-blade turbines above is particularly effective.
Because the size of the caverns increases with impeller diameter, large impellers with dia-
meters greater than one-half the tank diameter are also recommended. Different impeller
designs that sweep the entire volume of the vessel may also be beneficial. As discussed in
Section 8.4, the most common impellers used for viscous mixing are gate, anchor, and heli-
cal stirrers mounted with small clearance between the impeller and tank wall. However,
application of these impellers in fermenters is only possible when the culture oxygen
demand is low. Although small-clearance impellers operating at relatively slow speed give
superior bulk mixing in viscous fluids, high-shear systems with high-speed, remote-clear-
ance impellers are preferable for breaking up gas bubbles and promoting oxygen transfer
to the liquid.
8.16 ROLE OF SHEAR IN STIRRED FERMENTERS

As outlined in Section 7.2.2, the development of shear stresses in fluids is related to the
presence of velocity gradients. In turbulent flow, unsteady velocity components associated
with eddies give rise to additional fluctuating velocity gradients that are superimposed on
the mean flow. As a consequence, turbulent shear stresses are much greater than those
developed in laminar flow (Section 7.9.2, Reynolds Stresses subsection). Turbulent shear
stress also varies considerably with time and position in the fluid.
Mixing in bioreactors used for aerobic cell culture must provide the shear conditions
necessary to disperse gas bubbles and, if appropriate, break up liquid droplets or cell flocs.
The dispersion of gas bubbles by impellers involves a balance between opposing forces.
The interaction of bubbles with turbulent velocity fluctuations in the fluid causes the bub-
bles to stretch and deform. Bubble break-up occurs if the induced stresses exceed the stabi-
lising forces due to surface tension at the gas#liquid interface, which tend to restore the
bubble to its spherical shape. For droplets, the droplet viscosity is an additional stabilising
influence opposing break-up. In the case of solid material such as cells or aggregates,
break-up occurs if the stresses induced by the turbulent velocity gradients are greater than
the mechanical strength of the particles.
Whereas bubble break-up is required in fermenters to facilitate oxygen transfer, disrup-
tion of individual cells is undesirable. As indicated in Table 8.6, different cell types have
different susceptibilities to damage in the bioreactor environment. This susceptibility is
usually referred to as shear sensitivity, although the damage observed may not arise neces-
sarily from the effects of fluid velocity gradients. In this context, the term ‘shear’ is used
316 8. MIXING
TABLE 8.6 Cell Shear Sensitivity

Cell type Size Shear sensitivity
Microbial cells 1210 µm Low

Microbial pellets/clumps Up to 1 cm Moderate
Plant cells 100 µm Moderate/high
Plant cell aggregates Up to 122 cm High
Animal cells 20 µm High
Animal cells on microcarriers 802200 µm Very high
imprecisely to mean any mechanism dependent on the hydrodynamic conditions in the

vessel that results in cell damage.
Because of their relatively small size, bacteria and yeast are considered generally not to
be shear sensitive. The main effect of agitation on filamentous fungi and actinomycetes is
a change in morphology between pelleted and dispersed forms; the mean size of cell
clumps also varies with hydrodynamic conditions. Insect, mammalian, and plant cells are
considered to be particularly sensitive to hydrodynamic forces. Bioreactors used for cul-
ture of these cells must take this sensitivity into account while still providing adequate
mixing and oxygen transfer. At the present time, the effects of hydrodynamic forces on
cells are not understood completely. Cell disruption is an obvious outcome; however more
subtle sublytic effects such as slower growth or product synthesis, denaturation of extracel-
lular proteins, and thickening of the cell walls may also occur. In some cases, cellular
metabolism is stimulated by exposure to hydrodynamic forces. Because there are many
gaps in our knowledge of how cells are affected, it is not possible in most cases to predict
with confidence what bioreactor operating conditions will or will not damage shear-sensi-
tive organisms. Some of the findings from research into various culture systems, and some
of the mechanisms postulated to explain cell damage in bioreactors, are outlined next.
8.16.1 Studies with Animal Cell Cultures

Studies of cell damage in fermenters have been carried out mostly with animal cells. As
animal cells lack a protective cell wall, cell injury and loss are significant problems in
large-scale culture. Some types of animal cell such as insect, hybridoma, and blood cells
grow readily when freely suspended in liquid medium. However, many transformed ani-
mal cell lines are anchorage dependent: this means that the cells must be attached to a solid
surface for survival. In bioreactors, the surface area required for cell attachment is often
provided by microcarrier beads, which are plastic or polymer beads of diameter 80 to
200 µm. As shown in Figure 8.43, the animal cells cover the surface of the beads, which
are then suspended in nutrient medium. Many benefits are associated with the use of
microcarriers; however, a disadvantage is that the cells cannot change position or rotate
FIGURE 8.43 Chinese hamster ovary (CHO)
cells attached to microcarrier beads.
Photograph courtesy of J. Crowley.
20 µm
easily in response to hydrodynamic forces in the fluid. This makes animal cells on micro-
carriers particularly susceptible to shear damage.
Several mechanisms have been postulated to cause cell damage in bioreactors:
• Interaction between cells and turbulent eddies
• Collisions between cells or between microcarrier beads, collision of cells with the
impeller, and collision of cells with stationary surfaces in the vessel
• Interaction between cells and shear forces in the boundary layers and wakes near solid
objects in the reactor, especially the impeller
• Interaction between cells and the mechanical forces associated with formation of
bubbles at the sparger and bubble rise through the liquid
• Interaction between cells and bursting bubbles at the liquid surface
Investigation of these effects has so far resulted in only a partial understanding of ani-
mal cell damage in bioreactors. Industrial practice in this area is based largely on assump-
tions and ‘rules of thumb’ rather than theoretical or scientific principles. For cells attached
to microcarrier beads, bead#bead interactions and interactions between the microcarriers
and small turbulent eddies are considered most likely to damage cells. However, it has
also been established that if the vessel is sparged with air, severe damage of suspended
cells can occur during gas bubble burst at the surface of the liquid.
Interaction between Microcarriers and Turbulent Eddies

Interactions between microcarriers and eddies in turbulent flow have the potential to
cause mechanical damage to cells. The intensity of the forces associated with these interac-
tions is dependent on the size of the eddies relative to the microcarrier particles. If the par-
ticles are small, they tend to be captured or entrained in the eddies as shown in
Figure 8.44(a). As fluid motion within eddies is laminar, if the density of the microcarriers
is about the same as the suspending fluid, there is little relative motion of the particles.
Accordingly, the velocity difference between the fluid streamlines and the microcarriers is
small, except for brief periods of acceleration when the bead enters a new eddy. On
318 8. MIXING
High shear FIGURE 8.44 Eddy#microcarrier inter-

zone actions. (a) Microcarriers are captured in
large eddies and move within the stream-
Eddy
streamlines line flow. (b) When several small eddies
with opposing rotation interact with the
Microcarrier microcarrier simultaneously, high levels of
shear develop on the bead surface.
Microcarrier From R.S. Cherry and E.T. Papoutsakis, 1986,
paths Hydrodynamic effects on cells in agitated tissue
(a) (b) culture reactors. Bioprocess Eng. 1, 29#41.
average, therefore, if the particles are smaller than the eddies, the shear effects of
eddy#cell interactions are minimal.
If the stirrer speed is increased and the average eddy size reduced, interactions between
eddies and microcarriers can occur in two possible ways. A single eddy that cannot fully
engulf the particle may act on part of its surface and cause the particle to rotate in the
fluid; this will result in a relatively low level of shear at the surface of the bead. However,
much higher shear stresses develop when several eddies with opposing rotation interact
with the particle and dissipate their energies on its surface simultaneously, as illustrated
in Figure 8.44(b).
Experimental data for cell damage on microcarriers has been correlated by comparing
the microcarrier diameter with the eddy size represented by the Kolmogorov scale as
defined in Eq. (7.36):
! 3 "1=4
ν
λ5 ð7:36Þ
ε
In this equation, λ is the characteristic size of eddies in the dissipative range of the turbu-
lence spectrum, ν is the kinematic viscosity of the fluid, and ε is the local rate of dissipa-
tion of turbulence kinetic energy per unit mass of fluid. For ε calculated as the average
rate of energy dissipation over the entire tank volume, it has been found that cells on
microcarriers suffer detrimental effects when the Kolmogorov scale drops below 2/3 to
1/2 the diameter of the microcarrier beads [45, 46]. Under these conditions, excessive agi-
tation is considered to generate eddies of small enough size but sufficient energy to cause
damage to the cells. The recommendation, therefore, is to operate the bioreactor so that
the Kolmogorov scale remains greater than the microcarrier diameter.
As indicated in Eq. (7.36), if the viscosity of the fluid is increased, the size of the smal-
lest eddies also increases. Raising the fluid viscosity should, therefore, reduce cell damage
in bioreactors. This effect has been demonstrated by adding thickening agents to animal
cell culture medium: Moderate increases in viscosity led to significant reductions in turbu-
lent cell death [46].
Although this approach to predicting turbulent cell damage has been found to apply
reasonably well in small-scale vessels, its broader application at larger scales raises several
difficulties. Conceptually, the method is based on comparing the size of the microcarriers
to the size of the dissipative eddies in the fluid. However, determining the size of the
dissipative eddies can be problematical. As discussed in Section 7.9.2 (Isotropic
Turbulence subsection), Kolmogorov’s theories about scales of turbulence apply only to
isotropic turbulence, which does not exist within the entire volume of bioreactors and may
not exist even in the region of most intense turbulence near the impeller. Moreover,
Kolmogorov’s equation gives only an order-of-magnitude estimate of the size of eddies in
the dissipative range. On a practical level, the value of λ estimated using Eq. (7.36) is
dependent directly on the volume used to evaluate the rate of turbulence energy dissipa-
tion per unit mass of fluid, ε. One approach is to assume that the power input by the
impeller is dissipated uniformly over the entire tank volume. In this case:
P
ε5 ð8:35Þ
ρVL
where P is the power input, ρ is the fluid density, and VL is the volume of fluid in the ves-
sel. Yet, this is clearly an inaccurate method for estimating ε: as illustrated in Figure 8.17,
the rate of turbulence kinetic energy dissipation is far from uniformly distributed in stirred
tanks, being much greater near the impeller than in the remainder of the fluid. Therefore,
use of the total liquid volume to evaluate ε is questionable and becomes increasingly so at
large scales where rates of energy dissipation in most parts of the tank are very low.
Application of Eq. (8.35) gives relatively high values of λ for a given power input, thus
potentially underestimating the damaging effects of eddies. In contrast, if we assume that
the power is dissipated mostly in the region of intense turbulence around the impeller, an
alternative expression for ε is [47]:
P
ε5 ð8:36Þ
ρ D3i
where Di is the impeller diameter and D3i represents the approximate volume near the
impeller where most of the energy dissipation occurs. Compared with Eq. (8.35), the calcu-
lated value of λ is smaller if Eq. (8.36) is used to estimate ε, so that damaging effects will
appear more likely. In practice, it is very difficult to know the precise distribution of
energy dissipation rates under particular operating conditions, and this affects accurate
estimation of ε and thus the size of the dissipative eddies.
Cell Damage from Bursting Bubbles

Because of the much smaller size of individual animal cells (B20 µm) compared with
microcarriers (80#200 µm), individual animal cells in suspension are relatively insensitive
to the hydrodynamic forces generated by turbulence unless the agitation intensity is
extremely high. However, when liquid containing shear-sensitive cells is sparged with
gas, damaging mechanisms other than those associated with turbulence come into play.
From research carried out so far, damage of suspended animal cells in bioreactors appears
to be caused mainly by bubbles bursting as they disengage from the liquid at the surface
of the culture. If the same cells are grown in the absence of air sparging, and the stirrer is
prevented from entraining gas from the atmosphere, very high agitation rates can be toler-
ated without significant cell damage [48, 49].
320 8. MIXING
Figure 8.45 shows the sequence of events associated with rupture of a bubble at a
gas#liquid surface. The rising bubble just below the surface pushes liquid at the interface
into a hemispherical film cap of thickness 1 to 10 µm. The cap drains under gravity until a
critical thickness (typically , 0.1 µm) is reached, then ruptures at the thinnest point at the
apex. The hole formed expands rapidly in size as the liquid drains away and flows under-
neath the bubble cavity. When this flow from all sides meets at the bottom of the cavity, if
the bubbles are small enough, the resulting fluid pressure creates an upward jet that rises
above the surface before disintegrating into droplets and disappearing back into the
liquid.
Very high levels of energy dissipation are associated with bubble rupture. The liquid
draining back from the film cap reaches velocities of 1 to 50 m s21, generating shear forces
several orders of magnitude greater than known tolerance levels for animal cells. To make
matters worse, in typical animal cell cultures, cells attach to rising bubbles and are retained
in the thin film of the liquid cap [50], thus being present at the most damaging location as
the bubbles burst. After bubble rupture, the cells are subjected to high liquid velocities
within the draining film and, as a consequence, suffer severe damage. Measurements
with insect cells have shown that, for each 3.5-mm bubble ruptured at the culture
(a) Bubble at the liquid surface FIGURE 8.45 Sequence of events when a bubble bursts at a
Film cap gas#liquid interface.
Air bubble
Liquid
(b) Film cap rupture
(c) Rapidly receding film
(d) Upward liquid jet
surface, about 103 cells are killed from a suspension initially containing approximately
106 cells ml21 [51]. At this rate, most of the cells in an average bioreactor would be killed in
only a few hours.
The severe damaging effects of bursting bubbles on suspended animal cells can be
attenuated by adding protective agents to the medium. Additives such as Pluronic F-68
lower the gas#liquid interfacial tension significantly and prevent the attachment of cells to
bubbles, thus removing cells from the damaging zone at the time of bubble rupture.
Stable foam layers on the surface of animal cell cultures have also been found to offer
some protection from bursting bubbles [52].
8.16.2 Other Studies and Approaches

A common concept in studies of cell damage is that the potential for damage or the
extent to which it occurs is related to the rate of dissipation of turbulence kinetic energy in
the fluid. For dissipation of a given amount of power, cell damage is likely to depend on
whether the power is dissipated within a small volume, in which case the damaging
effects will probably be severe, or whether it is dissipated more uniformly within a larger
volume at a lower turbulence intensity. The way in which turbulence kinetic energy and
rates of energy dissipation are distributed in stirred vessels depends largely on the type of
impeller used. Therefore, if the distribution of energy dissipation in the tank is important in
addition to the overall rate of energy dissipated (i.e., the power), we can predict that differ-
ent degrees of cell damage will occur using different impellers, even if each impeller is
dissipating the same amount of power.
The turbulence measurement techniques described in Section 7.9.3 can be used to charac-
terise the performance of impellers in terms of turbulence generation and the distribution
of energy dissipation rates. To date, however, mainly because of the practical difficulties
associated with evaluating rates of energy dissipation as outlined in Section 7.9.2 (Rate of
Dissipation of Turbulence Kinetic Energy subsection) we do not have an unambiguous pic-
ture of how different impellers utilise and distribute the power supplied to them.
Nevertheless, some success in predicting cell damage has been achieved using empirical
correlations derived from our broader understanding of impeller function. In work with
fungal hyphae in stirred fermenters, damage measured as a reduction in total hyphal
length or cell clump size depended not only on the power input, but also on the size of the
impeller, the number of trailing vortices generated per impeller blade, and the frequency
with which different impellers circulated fluid through the impeller region [53]. By taking
all these factors into account, the extent of damage could be predicted irrespective of the
type of impeller used. In these experiments, Rushton turbines were found to be less damag-
ing than pitched-blade turbines, and cell damage increased as the ratio of impeller diameter
to tank diameter was reduced. An implication from this work is that the potential for cell
damage decreases with scale-up of mixing systems. Although successful with fungal cul-
tures, the applicability of this approach to other cell types remains to be tested.
It is possible that the response of cells to hydrodynamic conditions depends not only on
the intensity of the damaging forces generated by turbulent flow, but also on their dura-
tion. The cumulative amount of energy dissipated on cells has been used to correlate the
322 8. MIXING
damaging effects of turbulence in fungal and plant cell cultures [54#57]. The cumulative
energy dissipated on the cells per unit volume can be calculated as:
ð
1
E5 Pφ dt ð8:37Þ
V
where E is the cumulative energy dissipation per unit volume, V is the culture volume, P
is the power input, φ is the fraction of the culture volume occupied by the cells, and t is
time. Under steady-state conditions, for example, during continuous fermentations,
Eq. (8.37) reduces to:
Pφ τ
E5 ð8:38Þ
V
where τ is the average residence time in the vessel. In experiments with plant cells, several
properties have been measured as indictors of sublytic damage, including cell viability,
growth, membrane integrity, cell chain length, protein release into the medium, and cell
aggregate size. Depending on the species, cell damage was found at cumulative energy
dissipation levels of 105 to 109 J m23. The implication of this result is that no damage will
occur if cumulative energy dissipation is limited to less than these threshold values.
Another view of the damaging effects of hydrodynamic conditions on cells is related to
the fine-scale intermittency of turbulence and the fluctuations in instantaneous velocity,
energy dissipation rate, and other properties that are characteristic of turbulent flow.
Rather than the average properties of turbulent flow causing cell damage, as is assumed in
most other approaches to this problem, it is possible that rare, rapid, and very violent
events in turbulence are responsible for rupturing cells and aggregates. Although it has
been shown theoretically that strong bursts of energy dissipation have the potential to con-
trol cell damage in bioreactors [58], practical application to cell cultures has not yet been
demonstrated.
SUMMARY OF CHAPTER 8
This chapter covers various topics related to mixing and cell damage in bioreactors. At
the end of Chapter 8 you should:
• Be familiar with the broad range of equipment used for mixing in stirred vessels,
including different impeller designs, their operating characteristics, and their suitability
for particular mixing applications
• Be able to describe rotational-, radial-, and axial-flow patterns in stirred tanks and the
types of impeller that induce them
• Be able to describe the different two-phase flow patterns in stirred vessels with gassing,
using terms such as impeller flooding, impeller loading, complete gas dispersion, and gas
recirculation
• Understand the role of trailing vortices and ventilated cavities in impeller operation
• Know how impeller size, stirrer speed, tank geometry, liquid properties, and gas
sparging affect power consumption in stirred vessels
PROBLEMS 323
• Be able to determine the stirrer operating conditions for complete solids suspension
• Understand the mechanisms of mixing
• Know what is meant by mixing time and how it is measured
• Be able to describe the effects of scale-up on mixing, and options for improving mixing
without the input of extra power
• Understand the factors affecting the performance of multiple-impeller systems
• Know the problems associated with mixing highly pseudoplastic or yield-stress fluids
• Understand how cells can be damaged in stirred and aerated fermenters
PROBLEMS
8.1. Impeller loading and gas dispersion
An aerated bioreactor of diameter 1.2 m is used for batch culture of Brevibacterium flavum.
The reactor is stirred using a Rushton turbine of diameter 0.4 m. At the beginning of the
culture when the cell concentration and culture oxygen requirements are relatively small,
the stirrer operating speed of 110 rpm is just sufficient to completely disperse the air
bubbles. Towards the end of the culture when the cell density is high, it is decided to
double the volumetric flow rate of air to supply more oxygen.
(a) Will the impeller be flooded or loaded under these conditions?
(b) What stirrer speed is needed to achieve complete gas dispersion at the new aeration rate?
8.2. Hydrodynamic conditions for animal cell culture
Mouse hybridoma cells are grown in suspension culture in a stirred tank of diameter 1 m
and liquid volume 0.8 m3. The stirrer speed is 1.5 rps. The vessel has four 10% baffles and
is sparged with air at a flow rate of 0.3 vvm (volume of gas per volume of liquid per
minute). Pluronic F-68 is added to the medium to protect the cells against the effects of
bursting bubbles. The density of the medium is 1000 kg m23 and the viscosity is 1.4 cP. The
impeller is a Rushton turbine of diameter 0.4 m.
(a) Is the air dispersed effectively in this system?
(b) Is the flow turbulent?
(c) What proportion of the power input is from sparging?
8.3. Gas dispersion and power requirements
A stirred, baffled fermenter is used to culture Streptomyces cinnamonensis for production
of monensin. The tank diameter and liquid height are both 1.1 m. The broth density is
1000 kg m23 and the viscosity is 15 cP. The fermenter is mixed using a Rushton turbine of
diameter one-half the tank diameter. The air flow rate is 0.66 vvm (volume of gas per
volume of liquid per minute).
(a) What stirrer speed is required for complete gas dispersion?
(b) What are the power requirements for complete gas dispersion? Assume the power
draw with gassing is 50% of that without gassing.
8.4. Electrical power required for mixing
Laboratory-scale fermenters are usually mixed using small stirrers with electric motors
rated between 100 and 500 W. One such motor is used to drive a 7-cm Rushton turbine in
a small reactor containing fluid with the properties of water. The stirrer speed is 900 rpm.
Estimate the power requirements for this process. How do you explain the difference
324 8. MIXING
between the amount of electrical power consumed by the motor and the power required by
the stirrer?
8.5. Effect of viscosity on power requirements
A cylindrical bioreactor of diameter 3 m has four baffles. A Rushton turbine mounted in
the reactor has a diameter of one-third the tank diameter. The liquid height is equal to the
tank diameter and the density of the fluid is approximately 1 g cm23. The reactor is used to
culture an anaerobic organism that does not require gas sparging. The broth can be
assumed Newtonian. As the cells grow, the viscosity of the broth increases.
(a) The stirrer is operated at a constant speed of 90 rpm. Compare the power requirements
when the viscosity is:
(i) Approximately that of water
(ii) 100 times greater than water
(iii) 2 3 105 times greater than water
(b) The viscosity reaches a value 1000 times greater than water.
(i) What stirrer speed is required to achieve turbulence?
(ii) Estimate the power required to achieve turbulence.
(iii) What power per unit volume is required for turbulence? Is it reasonable to expect
to be able to provide this amount of power? Why or why not?
8.6. Power and scale-up
A pilot-scale fermenter of diameter and liquid height 0.5 m is fitted with four baffles of
width one-tenth the tank diameter. Stirring is provided using a Scaba 6SRGT curved-blade
disc turbine with diameter one-third the tank diameter. The density of the culture broth is
1000 kg m23 and the viscosity is 5 cP. Optimum culture conditions are provided in the
pilot-scale fermenter when the stirrer speed is 185 rpm. Following completion of the pilot
studies, a larger production-scale fermenter is constructed. The large fermenter has a
capacity of 6 m3, is geometrically similar to the pilot-scale vessel, and is also equipped with
a Scaba 6SRGT impeller of diameter one-third the tank diameter.
(a) What is the power consumption in the pilot-scale fermenter?
(b) If the production-scale fermenter is operated so that the power consumed per unit
volume is the same as in the pilot-scale vessel, what is the power requirement after
scale-up?
(c) For the conditions in (b), what is the stirrer speed after scale-up?
(d) If, instead of (b) and (c), the impeller tip speed (5 πNiDi) is kept the same in the pilot-
and production-scale fermenters, what is the stirrer speed after scale-up?
(e) For the conditions in (d), what power is required after scale-up?
8.7. Particle suspension and gas dispersion
Bacteria immobilised on particles of gravel are being studied for bioremediation of
polychlorinated biphenyls (PCBs). The cells are cultured in an aqueous solution of density
1000 kg m23 and viscosity 0.8 mPa s in a 400-litre bioreactor of diameter 0.8 m. The reactor
is stirred with a Rushton turbine of diameter one-third the tank diameter positioned with
an impeller off-bottom clearance of a one-quarter the tank diameter. The average particle
diameter is 250 µm, the density of gravel is 1.9 g cm23, and the particle concentration is
15% by weight. The bioreactor is supplied with air at a flow rate of 0.5 vvm (volume of gas
PROBLEMS 325
per volume of liquid per minute). If the stirrer speed is set so that the solids are just
completely suspended in the solution, will the gas be completely dispersed?
8.8. Impeller flooding and power requirements
You are asked to purchase a new Rushton turbine and drive assembly for a fermenter of
diameter 2 m equipped with four 10% baffles and with liquid height equal to the tank
diameter. The fermenter is used to culture organisms that require an air flow rate of
1.5 vvm (volume of gas per volume of liquid per minute). The broth density is close to
1.0 g cm23 and the viscosity is 0.9 mPa s. Your colleagues have suggested that you design
the system for an impeller of diameter one-third the tank diameter, as this is standard
company practice. However, after inspecting the vessel internals, you find that an impeller
diameter of one-half the tank diameter is also a possibility.
(a) What stirrer speed is required to prevent impeller flooding with Di 5 1/3 DT?
(b) What stirrer speed is required to prevent impeller flooding with Di 5 1/2 DT?
Your supervisor is worried about the operating costs associated with the new impeller,
and suggests you use the smaller impeller with Di 5 1/3 DT to reduce the factory power
bill.
(c) Under the operating conditions determined in (a) and (b) to avoid impeller flooding,
which impeller consumes the least power? Assume that the % drop in power with
gassing is the same for both impellers. Which of the two impellers would you
recommend?
8.9. Stirrer effectiveness with sparging
A baffled cylindrical tank of diameter and liquid height 1.15 m is stirred using a four-blade
pitched-blade turbine of diameter 0.36 m operated at 200 rpm. The vessel is sparged with
air at a volumetric flow rate of 0.036 m3 s21. Under these conditions, the turbulent power
number for the impeller with gassing is about 1.0. The liquid in the tank has a density of
1 g cm23 and viscosity 1 cP.
(a) Is the impeller likely to be flooded or loaded? What assumptions are involved in your
answer?
(b) What is the rate of energy input by the impeller?
(c) What is the rate of energy input by gassing?
(d) In your opinion, is this stirring system effective for mixing and gas dispersion?
8.10. Cell suspension and power requirements
A fermentation broth contains 40 wt% cells of average dimension 10 µm and density
1.04 g cm23. A marine propeller of diameter 30 cm is used for mixing. The density and
viscosity of the medium are approximately the same as water. The fermentation is carried
out without gas sparging in a vessel of diameter 75 cm.
(a) Estimate the stirrer speed required to just completely suspend the cells.
(b) What power is required for cell suspension?
You plan to improve this fermentation process by using a new cell strain immobilised in
porous plastic beads of diameter 2 mm and density 1.75 g cm23. The particle concentration
required for comparable rates of fermentation is 10% by weight.
(c) How does changing over to the immobilised cell system affect the stirrer speed and
power required for particle suspension?
326 8. MIXING
8.11. Particle suspension and scale-up

Bacteria attached to particles of clinker are being tested for treatment of industrial waste. In
the laboratory, the process is carried out under anaerobic conditions in a stirred bioreactor
with liquid height equal to the tank diameter. The system is then scaled up to a
geometrically similar vessel of volume 90 times that of the laboratory reactor. The
suspending fluid and the particle size, density, and concentration remain the same as in
the smaller vessel. The type of impeller is also unchanged, as is the impeller-to-tank
diameter ratio.
(a) How does the stirrer speed required for suspension of the particles change after scale-up?
(b) Assuming operation in the turbulent regime for both vessels, what effect does scale-up
have on:
(i) The power required for particle suspension?
(ii) The power per unit volume required for particle suspension?
8.12. Impeller diameter, mixing, and power requirements
Solids suspension and gas dispersion can both be achieved at lower stirrer speeds if the
impeller diameter is increased. However, the power requirements for stirring increase with
impeller size. In this exercise, the energy efficiencies of small and large impellers are
compared for solids suspension and gas dispersion.
(a) For ungassed Rushton turbines operating in the turbulent regime, how does the power
required for complete solids suspension vary with impeller diameter if all other
properties of the system remain unchanged?
(b) Using the result from (a), compare the power requirements for complete solids
suspension using Rushton turbines of diameters one-third and one-half the tank
diameter.
(c) All else being equal, how does the power required for complete gas dispersion by
Rushton turbines vary with impeller diameter under turbulent flow conditions?
(d) Compare the power requirements for complete gas dispersion using Rushton turbines
of diameters one-third and one-half the tank diameter.
8.13. Efficiency of different impellers for solids suspension
Compared with a Rushton turbine of diameter one-half the tank diameter, what are the
power requirements for solids suspension by a downward-pumping, six-blade pitched-
blade turbine of the same size?
8.14. Power and mixing time with aeration
A cylindrical stirred bioreactor of diameter and liquid height 2 m is equipped with a
Rushton turbine of diameter one-third the tank diameter. The bioreactor contains
Newtonian culture broth with the same density as water and viscosity 4 cP.
(a) If the specific power consumption must not exceed 1.5 kW m23, determine the
maximum allowable stirrer speed.
(b) What is the mixing time at the stirrer speed determined in (a)?
(c) The tank is now aerated. In the presence of gas bubbles, the approximate relationship
between the ungassed turbulent power number ðN 0P Þ0 and the gassed turbulent power
number ðN 0P Þg is: ðN 0P Þg 5 0:5 ðN 0P Þ0 . What maximum stirrer speed is now possible in the
sparged reactor?
(d) What is the mixing time with aeration at the stirrer speed determined in (c)?
PROBLEMS 327
8.15. Scale-up of mixing system

To ensure turbulent conditions during agitation with a turbine impeller, the Reynolds
number must be at least 104.
(a) A laboratory fermenter of diameter and liquid height 15 cm is equipped with a 5-cm-
diameter Rushton turbine operated at 800 rpm. If the density of the broth is close to
that of water, what is the upper limit for the broth viscosity if turbulence is to be
achieved?
(b) Estimate the mixing time in the laboratory fermenter.
(c) The fermenter is scaled up so the tank and impeller are 15 times the diameter of the
laboratory equipment. The stirrer in the large vessel is operated at the same impeller
tip speed (5 πNiDi) as in the laboratory apparatus. How does scale-up affect the
maximum viscosity allowable for maintenance of turbulent conditions?
(d) What effect does scale-up have on the mixing time?
8.16. Effect on mixing of scale-up at constant power per unit volume
A baffled pilot-scale fermenter with liquid height equal to the vessel diameter is scaled up
to a geometrically similar production vessel. The working volume of the production
fermenter is 50 times greater than that at the pilot scale. If the agitation system is scaled up
using the basis of constant power per unit volume, what effect will scale-up have on the
mixing time?
8.17. Alternative impellers
Escherichia coli cells are cultured in an industrial-scale fermenter for production of
supercoiled plasmid DNA used in gene therapy.
(a) A fermenter of diameter 2.3 m and working volume 10 m3 is equipped with a Rushton
turbine of diameter one-third the tank diameter. The impeller is operated at 60 rpm
and the vessel is sparged with air. The density and viscosity of the fermentation fluid
are close to those of water, that is, 1000 kg m23 and 1 cP, respectively. The power with
gassing is about 60% of the ungassed power.
(i) Calculate the power draw.
(ii) Estimate the mixing time.
(b) To satisfy the burgeoning demand for gene therapy vectors and DNA vaccines, the
fermentation factory is being expanded. Two new 10-m3 fermenters are being designed
and constructed. It is decided to investigate the use of different impellers in an effort to
reduce the power required but still achieve the same mixing time as that obtained with
the Rushton turbine described in (a).
(i) Compared with the Rushton turbine of diameter one-third the tank diameter,
what power savings can be made using a Rushton turbine of diameter one-half
the tank diameter?
(ii) Compared with the Rushton turbine of diameter one-third the tank diameter,
what power savings can be made using a Lightnin A315 hydrofoil impeller of
diameter 0.4 times the tank diameter?
(iii) The power with gassing is about 50% of the ungassed power for both the larger
Rushton turbine and the A315 hydrofoil. What stirrer speeds are required
with these impellers to achieve the same mixing time as that determined in
(a) (ii)?
328 8. MIXING
8.18. Retrofitting
A 1.8-m-diameter cylindrical fermenter with four baffles of width one-tenth the tank
diameter and working volume 4.6 m3 is used to produce amylase. The density of the culture
broth is 1000 kg m23 and the viscosity is 20 cP. The vessel is equipped with a 0.6-m Rushton
turbine. Under these conditions, the motor delivers a maximum stirrer speed of 150 rpm. A
sales representative from the local impeller manufacturing company has suggested that
better culture performance might be achieved by replacing the Rushton turbine with a more
modern impeller design. On offer at discounted prices are curved-blade disc turbines and a
downward-pumping hydrofoil with characteristics similar to those of the Lightnin A315.
(a) Estimate the impeller diameters appropriate for retrofitting of the alternative agitators.
What assumptions are involved in your answer?
(b) Compare the mixing times expected after retrofitting each of the alternative impellers
with the mixing time delivered by the existing Rushton impeller. Considering mixing
time only, do you think that the cost of retrofitting a new impeller is justified? Why or
why not?
(c) Which of the three impellers would you recommend for use in future fermentations:
(i) Under nonaerated conditions?
(ii) With aeration?
Explain your decision in each case.
8.19. Retrofitting multiple impellers
An industrial fermentation vessel with diameter 1.9 m and aspect ratio 3:1 is used for
production of leucine by aerobic cultures of Serratia marcescens. At present, the fermenter is
fitted with three Rushton turbines of diameter one-third the tank diameter. The impellers
are spaced far enough apart so there is no significant interaction between their flow
currents. The stirrer motor is rated for a maximum stirrer speed of 1.2 rps. It is proposed
to carry out an impeller retrofitting operation to improve the performance of the culture.
Although the culture is aerobic, the new stirring system will be designed for operation
under nonaerated conditions as a safety precaution against accidental blockage of the air
supply. It is decided to replace the two upper Rushton turbines with two identical
downward-pumping hydrofoil impellers. The turbulent power number for the hydrofoil
impellers is around 0.9. If the bottom Rushton impeller is replaced by a curved-blade disc
turbine of diameter one-third the tank diameter, what size hydrofoil impellers are
required? What assumptions are involved in your answer?
8.20. Impeller viscometer
An impeller viscometer is being developed to measure the rheological properties of
pseudoplastic fermentation broths. The broth density is 1000 kg m23. Typical rheological
parameters for the broth are flow behaviour index n 5 0.2 and consistency index
K 5 0.05 N sn m22. As outlined in Section 7.6.3, impeller viscometers must be operated
under laminar flow conditions.
(a) If a Rushton turbine of diameter 4 cm is used at speeds between 2.5 and 10 rpm, is
the flow laminar?
(b) What shear rate range does operation with the Rushton turbine provide?
(c) Approximately what range of shear stresses is induced in the Rushton turbine
viscometer?
REFERENCES 329
(d) If the Rushton turbine is replaced by a close-clearance helical ribbon impeller of

diameter 7.5 cm, what maximum stirrer speed can be used under laminar flow
conditions?
(e) If the minimum practical stirrer speed with the helical ribbon impeller is 2.5 rpm, what
ranges of shear rates and shear stresses can be investigated using this viscometer?
What advantages, if any, does the helical ribbon viscometer offer over the Rushton
turbine viscometer?
8.21. Turbulent shear damage
Microcarrier beads 120 µm in diameter are used to culture recombinant CHO cells for
production of growth hormone. It is proposed to use a 20-cm Rushton turbine to mix the
culture in a 200-litre bioreactor. Oxygen and carbon dioxide are supplied by gas flow
through the reactor headspace. The microcarrier suspension has a density of approximately
1010 kg m23 and a viscosity of 1023 Pa s.
(a) Assuming that the power input by the stirrer is dissipated uniformly in the vessel,
estimate the maximum allowable stirrer speed that avoids turbulent shear damage of
the cells.
(b) How is your estimate affected if the stirrer power is dissipated close to the impeller,
within a volume equal to the impeller diameter cubed?
8.22. Avoiding cell damage
Suspended plant cell cultures derived from lemon trees are being used to produce citrus
oil for the cosmetic industry. The cells are known to be sensitive to agitation conditions:
cell damage occurs and oil production is detrimentally affected if the cumulative energy
dissipation level exceeds 105 J m23. The cells are grown in a bioreactor with continuous
feeding of nutrient medium and withdrawal of culture broth. At the operating flow rate,
the average residence time in the bioreactor is 2.9 days. The diameter of the vessel is
0.73 m, the liquid height is equal to the tank diameter, four 10% baffles are fitted, and
stirring is carried out using a 25-cm curved-blade disc turbine with six blades. The vessel is
aerated but the effect of gassing on the impeller power draw is negligible. At steady state,
the cell concentration is 0.24 v/v, the broth density is 1 g cm23, and the viscosity is
3.3 mPa s. What is the maximum stirrer speed that can be used without damaging the
cells?
References
[1] A.W. Nienow, M.M.C.G. Warmoeskerken, J.M. Smith, M. Konno, On the flooding/loading transition and the
complete dispersal condition in aerated vessels agitated by a Rushton-turbine, in: Mixing, Proc. 5th Eur. Conf.
on Mixing, Würzburg, Germany, 1985, pp. 143#154; BHRA: The Fluid Engineering Centre, Cranfield, U.K.
[2] W. Bruijn, K. van’t Riet, J.M. Smith, Power consumption with aerated Rushton turbines, Trans. IChemE 52
(1974) 88#104.
[3] C.M. Chapman, A.W. Nienow, M. Cooke, J.C. Middleton, Particle#gas#liquid mixing in stirred vessels.
Part III: Three phase mixing, Chem. Eng. Res. Des. 61 (1983) 167#181.
[4] S.M. Kresta, P.E. Wood, The mean flow field produced by a 45" pitched blade turbine: changes in the circula-
tion pattern due to off bottom clearance, Can. J. Chem. Eng. 71 (1993) 42#53.
[5] K.J. Bittorf, S.M. Kresta, Active volume of mean circulation for stirred tanks agitated with axial impellers,
Chem. Eng. Sci. 55 (2000) 1325#1335.
330 8. MIXING
[6] J.J. Frijlink, M. Kolijn, J.M. Smith, Suspension of solids with aerated pitched blade turbines, in: Fluid Mixing
II, IChemE Symp. Ser. 89, U.K. Institution of Chemical Engineers, 1984, pp. 49#58.
[7] A.W. Nienow, Gas dispersion performance in fermenter operation, Chem. Eng. Prog. 86 (2) (1990) 61#71.
[8] A.W. Nienow, M. Konno, W. Bujalski, Studies on three-phase mixing: a review and recent results, Chem.
Eng. Res. Des. 64 (1986) 35#42.
[9] F. Saito, A.W. Nienow, S. Chatwin, I.P.T. Moore, Power, gas dispersion and homogenisation characteristics
of Scaba SRGT and Rushton turbine impellers, J. Chem. Eng. Japan 25 (1992) 281#287.
[10] M. Zlokarnik, H. Judat, Stirring, in: W. Gerhartz (Ed.), Ullmann’s Encyclopedia of Industrial Chemistry,
fifth ed., vol. B2, VCH, 1988, pp. 25-1#25-33.
[11] R.L. Bates, P.L. Fondy, R.R. Corpstein, An examination of some geometric parameters of impeller power,
Ind. Eng. Chem. Process Des. Dev. 2 (1963) 310#314.
[12] C.M. Chapman, A.W. Nienow, M. Cooke, J.C. Middleton, Particle#gas#liquid mixing in stirred vessels. Part
I: Particle#liquid mixing, Chem. Eng. Res. Des. 61 (1983) 71#81.
[13] W. Bujalski, A.W. Nienow, S. Chatwin, M. Cooke, The dependency on scale and material thickness of power
numbers of different impeller types, in: Proc. Int. Conf. on Mechanical Agitation, Toulouse, France, ENS
Association pour la Promotion du Genie Chimique, 1986, pp. 1-37#1-46 .
[14] J.J. Frijlink, A. Bakker, J.M. Smith, Suspension of solid particles with gassed impellers, Chem. Eng. Sci. 45
(1990) 1703#1718.
[15] S. Ibrahim, A.W. Nienow, Power curves and flow patterns for a range of impellers in Newtonian fluids:
40 , Re , 5 3 105, Trans. IChemE 73A (1995) 485#491.
[16] C.M. McFarlane, X.-M. Zhao, A.W. Nienow, Studies of high solidity ratio hydrofoil impellers for aerated
bioreactors. 2. Air#water studies, Biotechnol. Prog. 11 (1995) 608#618.
[17] Z. Jaworski, A.W. Nienow, K.N. Dyster, An LDA study of the turbulent flow field in a baffled vessel agitated
by an axial, down-pumping hydrofoil impeller, Can. J. Chem. Eng. 74 (1996) 3#15.
[18] J.H. Rushton, E.W. Costich, H.J. Everett, Power characteristics of mixing impellers, Part I. Chem. Eng. Prog.
46 (1950) 395#404.
[19] J.H. Rushton, E.W. Costich, H.J. Everett, Power characteristics of mixing impellers. Part II, Chem. Eng. Prog.
46 (1950) 467#476.
[20] M.M.C.G. Warmoeskerken, J. Speur, J.M. Smith, Gas#liquid dispersion with pitched blade turbines, Chem.
Eng. Commun. 25 (1984) 11#29.
[21] R.L. Bates, P.L. Fondy, J.G. Fenic, Impeller characteristics and power, in: V.W. Uhl, J.B. Gray (Eds.), Mixing:
Theory and Practice, vol. 1, Academic Press, 1966, pp. 111#178.
[22] K. Rutherford, S.M.S. Mahmoudi, K.C. Lee, M. Yianneskis, The influence of Rushton impeller blade and disk
thickness on the mixing characteristics of stirred vessels, Trans. IChemE 74A (1996) 369#378.
[23] A.W. Nienow, Gas#liquid mixing studies: a comparison of Rushton turbines with some modern impellers,
Trans. IChemE 74A (1996) 417#423.
[24] C.M. McFarlane, A.W. Nienow, Studies of high solidity ratio hydrofoil impellers for aerated bioreactors.
3. Fluids of enhanced viscosity and exhibiting coalescence repression, Biotechnol. Prog. 12 (1996) 1#8.
[25] K.N. Dyster, E. Koutsakos, Z. Jaworski, A.W. Nienow, An LDA study of the radial discharge velocities gener-
ated by a Rushton turbine: Newtonian fluids, Re $ 5, Trans. IChemE 71A (1993) 11#23.
[26] V.P. Mishra, K.N. Dyster, Z. Jaworski, A.W. Nienow, J. McKemmie, A study of an up- and a down-pumping
wide blade hydrofoil impeller: Part I. LDA measurements, Can. J. Chem. Eng. 76 (1998) 577#588.
[27] A. Bakker, H.E.A. van den Akker, A computational model for the gas#liquid flow in stirred reactors, Trans.
IChemE 72A (1994) 594#606.
[28] Th.N. Zwietering, Suspending of solid particles in liquid by agitators, Chem. Eng. Sci. 8 (1958) 244#253.
[29] A.W. Nienow, The suspension of solid particles, in: N. Harnby, M.F. Edwards, A.W. Nienow (Eds.), Mixing
in the Process Industries, second ed., Butterworth-Heinemann, 1992, pp. 364#393.
[30] C.W. Wong, J.P. Wang, S.T. Huang, Investigations of fluid dynamics in mechanically stirred aerated slurry
reactors, Can. J. Chem. Eng. 65 (1987) 412#419.
[31] W. Bujalski, M. Konno, A.W. Nienow, Scale-up of 45" pitch blade agitators for gas dispersion and solid sus-
pension, in: Mixing, Proc. 6th Eur. Conf. on Mixing, Pavia, Italy, 1988, pp. 389#398, BHRA The Fluid
Engineering Centre, Cranfield, U.K.
REFERENCES 331
[32] S. Ibrahim, A.W. Nienow, Particle suspension in the turbulent regime: the effect of impeller type and impel-
ler/vessel configuration, Trans. IChemE 74A (1996) 679#688.
[33] C.M. Chapman, A.W. Nienow, J.C. Middleton, Particle suspension in a gas sparged Rushton-turbine agitated
vessel, Trans. IChemE 59 (1981) 134#137.
[34] M.M.C.G. Warmoeskerken, M.C. van Houwelingen, J.J. Frijlink, J.M. Smith, Role of cavity formation in stir-
red gas#liquid#solid reactors, Chem. Eng. Res. Des. 62 (1984) 197#200.
[35] S. Ruszkowski, A rational method for measuring blending performance, and comparison of different impel-
ler types, in: Mixing, Proc. 8th Eur. Conf. on Mixing, Cambridge, U.K., IChemE Symp. Ser. 136, Institution of
Chemical Engineers, 1994, pp. 283#291.
[36] A.W. Nienow, On impeller circulation and mixing effectiveness in the turbulent flow regime, Chem. Eng.
Sci. 52 (1997) 2557#2565.
[37] V. Hudcova, V. Machon, A.W. Nienow, Gas#liquid dispersion with dual Rushton turbine impellers,
Biotechnol. Bioeng. 34 (1989) 617#628.
[38] D.G. Cronin, A.W. Nienow, G.W. Moody, An experimental study of mixing in a proto-fermenter agitated by
dual Rushton turbines, Trans. IChemE 72C (1994) 35#40.
[39] R. Kuboi, A.W. Nienow, The power drawn by dual impeller systems under gassed and ungassed conditions,
in: Mixing, Proc. 4th Eur. Conf. on Mixing, Noordwijkerhout, The Netherlands, 1982, pp. 247#261, BHRA
Fluid Engineering, Cranfield, U.K.
[40] D. Hari-Prajitno, V.P. Mishra, K. Takenaka, W. Bujalski, A.W. Nienow, J. McKemmie, Gas#liquid mixing
studies with multiple up- and down-pumping hydrofoil impellers: power characteristics and mixing time,
Can. J. Chem. Eng. 76 (1998) 1056#1068.
[41] A.T.C. Mak, S. Ruszkowski, Solids suspension with multiple impellers. Proc. 1993 IChemE Research Event,
Birmingham, U.K., 1993, pp. 756#758, Institution of Chemical Engineers, Rugby, U.K.
[42] A.W. Nienow, G. Hunt, B.C. Buckland, A fluid dynamic study of the retrofitting of large agitated bioreactors:
turbulent flow, Biotechnol. Bioeng. 44 (1994) 1177#1185.
[43] A.B. Metzner, R.E. Otto, Agitation of non-Newtonian fluids, AIChE J. 3 (1957) 3#10.
[44] A.B. Metzner, R.H. Feehs, H. Lopez Ramos, R.E. Otto, J.D. Tuthill, Agitation of viscous Newtonian and non-
Newtonian fluids, AIChE J. 7 (1961) 3#9.
[45] M.S. Croughan, J.-F. Hamel, D.I.C. Wang, Hydrodynamic effects on animal cells grown in microcarrier
cultures, Biotechnol. Bioeng. 29 (1987) 130#141.
[46] M.S. Croughan, E.S. Sayre, D.I.C. Wang, Viscous reduction of turbulent damage in animal cell culture,
[47] R.S. Cherry, E.T. Papoutsakis, Physical mechanisms of cell damage in microcarrier cell culture bioreactors,
[48] K.T. Kunas, E.T. Papoutsakis, Damage mechanisms of suspended animal cells in agitated bioreactors with
and without bubble entrainment, Biotechnol. Bioeng. 36 (1990) 476#483.
[49] S.K.W. Oh, A.W. Nienow, M. Al-Rubeai, A.N. Emery, The effects of agitation intensity with and without con-
tinuous sparging on the growth and antibody production of hybridoma cells, J. Biotechnol. 12 (1989) 45#62.
[50] J.J. Chalmers, F. Bavarian, Microscopic visualization of insect cell#bubble interactions. II. The bubble film
and bubble rupture, Biotechnol. Prog. 7 (1991) 151#158.
[51] K. Trinh, M. Garcia-Briones, F. Hink, J.J. Chalmers, Quantification of damage to suspended insect cells as a
result of bubble rupture, Biotechnol. Bioeng. 43 (1994) 37#45.
[52] J.D. Michaels, J.E. Nowak, A.K. Mallik, K. Koczo, D.T. Wasan, E.T. Papoutsakis, Interfacial properties of cell
culture media with cell-protecting additives, Biotechnol. Bioeng. 47 (1995) 420#430.
[53] P. Jüsten, G.C. Paul, A.W. Nienow, C.R. Thomas, Dependence of mycelial morphology on impeller type and
agitation intensity, Biotechnol. Bioeng. 52 (1996) 672#684.
[54] M. Reuß, Influence of mechanical stress on the growth of Rhizopus nigricans in stirred bioreactors, Chem.
Eng. Technol. 11 (1988) 178#187.
[55] E.H. Dunlop, P.K. Namdev, M.Z. Rosenberg, Effect of fluid shear forces on plant cell suspensions, Chem.
Eng. Sci. 49 (1994) 2263#2276.
[56] R. Wongsamuth, P.M. Doran, The filtration properties of Atropa belladonna plant cell suspensions; effects of
hydrodynamic shear and elevated carbon dioxide levels on culture and filtration parameters, J. Chem. Tech.
Biotechnol. 69 (1997) 15#26.
332 8. MIXING
[57] P.F. MacLoughlin, D.M. Malone, J.T. Murtagh, P.M. Kieran, The effects of turbulent jet flows on plant cell
suspension cultures, Biotechnol. Bioeng. 58 (1998) 595#604.
[58] J. Bałdyga, J.R. Bourne, Interpretation of turbulent mixing using fractals and multifractals, Chem. Eng. Sci. 50
(1995) 381#400.
Suggestions for Further Reading

Mixing Equipment
Oldshue, J. Y. (1983). Fluid Mixing Technology. McGraw-Hill.
Power Requirements and Hydrodynamics in Stirred Vessels

See also references 7 and 10.
Harnby, N., Edwards, M. F., & Nienow, A. W. (Eds.), (1992). Mixing in the Process Industries (2nd ed.).
Butterworth-Heinemann.
McFarlane, C. M., & Nienow, A. W. (1995, 1996). Studies of high solidity ratio hydrofoil impellers for aerated bior-
eactors: Parts 1#4. Biotechnol. Prog., 11, 601#607, 608#618; 12, 1#8, 9#15.
Nienow, A. W. (1998). Hydrodynamics of stirred bioreactors. Appl. Mech. Rev., 51, 3#32.
Tatterson, G. B. (1991). Fluid Mixing and Gas Dispersion in Agitated Tanks. McGraw-Hill.
Tatterson, G. B. (1994). Scaleup and Design of Industrial Mixing Processes. McGraw-Hill.
Multiple Impeller Systems

Gogate, P. R., Beenackers, A.A.C.M., & Pandit, A. B. (2000). Multiple-impeller systems with a special emphasis on
bioreactors: a critical review. Biochem. Eng. J., 6, 109#144.
Cell Damage in Fermenters

Chalmers, J. (2000). Animal cell culture, effects of agitation and aeration on cell adaptation. In R. E. Spier (Ed.),
Encyclopedia of Cell Technology (vol. 1, pp. 41#51). John Wiley.
Namdev, P. K., & Dunlop, E. H. (1995). Shear sensitivity of plant cells in suspensions. Appl. Biochem. Biotechnol.,
54, 109#131.
Papoutsakis, E. T. (1991). Fluid-mechanical damage of animal cells in bioreactors. Trends in Biotechnol, 9, 427#437.
Thomas, C. R., & Zhang, Z. (1998). The effect of hydrodynamics on biological materials. In E. Galindo & O. T.
Ramı́rez (Eds.), Advances in Bioprocess Engineering II (pp. 137#170). Kluwer Academic.
Wu, J. (1995). Mechanisms of animal cell damage associated with gas bubbles and cell protection by medium
additives. J. Biotechnol., 43, 81#94.

Doran (2013) - Mixing

Uploaded by

Copyright:

Available Formats

Doran (2013) - Mixing

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Doran (2013) - Mixing

Uploaded by

Copyright:

Available Formats

C H A P T E R

8.1 FUNCTIONS OF MIXING

8.2 MIXING EQUIPMENT

8.2.1 Vessel Geometry and Liquid Height

FIGURE 8.2 Different profiles for the base of stirred vessels:

CBF = baffle clearance HL = liquid height

8.2.4 Stirrer Shaft

8.3 FLOW PATTERNS IN STIRRED TANKS

FIGURE 8.6 (a) Circular flow in an unbaffled stirred tank

8.3.1 Rotational Flow

8.3.2 Radial Flow

FIGURE 8.8 Flow pattern produced by an axial-flow impel-

8.3.3 Axial Flow

8.3.4 Gas Flow Patterns

Circle enclosing the

FIGURE 8.10 Selected impeller designs.

Six-flat-blade disc turbine Propeller Pitched-blade turbine

Anchor Gate anchor Helical screw

FIGURE 8.11 Viscosity ranges for different

Turbulence generation Pumping/flow capacity

Helical ribbons, anchors

Bladeless impellers, disc dispersers

8.4.1 Rushton Turbine

FIGURE 8.13 Rushton turbine.

FIGURE 8.14 Typical mean velocity vector plot for a Rush-

Constant Fg, increasing Ni

and the Froude number Fr:

(a) 100 rpm (b) 200 rpm

(c) 300 rpm (d) 400 rpm

EXAMPLE 8.1 GAS HANDLING WITH A RUSHTON TURBINE

ð1:25 s21 Þ2 0:5 m

Applying Eq. (8.4) for complete gas dispersion:

ð0:75 s21 Þ2 0:5 m

Therefore, from Eq. (8.1):

FIGURE 8.20 Gas cavities that are formed

Vortex Clinging ‘3 – 3’ structure ‘3 – 3’ structure Ragged

Vortex cavities Clinging cavity Large cavity

Blade Blade Blade

Direction of Direction of Direction of

FIGURE 8.22 Propeller.

8.4.3 Pitched-Blade Turbines

FIGURE 8.23 Distribution of k, the tur-

FIGURE 8.24 Direct and indirect gas loading regimes

8.4.4 Alternative Impeller Designs

Curved-Blade Disc Turbines

FIGURE 8.25 Scaba 6SRGT six-curved-blade disc

FIGURE 8.26 (a) Lightnin

(a) Vortex cavity (b) Growing cavity (c) Large cavity

of ventilated cavity formed behind the blades of a downward-pumping hydrofoil impeller

8.5 STIRRER POWER REQUIREMENTS

8.5.1 Ungassed Newtonian Fluids

Laminar Transition Turbulent

where k1 is a proportionality constant. Values of k1 for the impellers represented in

Impeller type k1 (Rei 5 1) N 0P (Rei 5 105)

Rushton turbine 70 5.0

Pitched-blade turbine 50 1.3

Rushton turbine Flat-bottom tank 5.9 [12, 13]

Curved-blade disc turbine (Scaba 6SRGT) Flat-bottom tank 1.5 [9]

LB/Di 5 0.28 Ci/DT 5 0.25

Hydrofoil (Prochem Maxflo T) Flat-bottom tank 1.6 [17]