Biotechnol. J. 2011, 6, 911–925 DOI 10.1002/biot.201000414 www.biotechnology-journal.

com

Review

Scale-down/scale-up studies leading to improved commercial

beer fermentation

Alvin W Nienow1, Mikkel Nordkvist2 and Christopher A. Boulton3

1 Centre for Bioprocess Engineering, University of Birmingham, UK
2 AlfaLaval Copenhagen A/S, Denmark
3 Division of Food Sciences, University of Nottingham, UK

Scale-up/scale-down techniques are vital for successful and safe commercial-scale bioprocess Received 8 March 2011
design and operation. An example is given in this review of recent studies related to beer pro- Revised 20 May 2011
duction. Work at the bench scale shows that brewing yeast is not compromised by mechanical Accepted 23 May 2011
agitation up to 4.5 W/kg; and that compared with fermentations mixed by CO2 evolution, agitation
≥ 0.04 W/kg is able to reduce fermentation time by about 20%. Work at the commercial scale in
cylindroconical fermenters shows that, without mechanical agitation, most of the yeast sediments
into the cone for about 50% of the fermentation time, leading to poor temperature control. Stirrer
mixing overcomes these problems and leads to a similar reduction in batch time as the bench-
scale tests and greatly reduces its variability, but is difficult to install in extant fermenters. The mix-
ing characteristics of a new jet mixer, a rotary jet mixer, which overcomes these difficulties, are
reported, based on pilot-scale studies. This change enables the advantages of stirring to be
achieved at the commercial scale without the problems. In addition, more of the fermentable
sugars are converted into ethanol. This review shows the effectiveness of scale-up/scale-down
studies for improving commercial operations. Suggestions for further studies are made: one
concerning the impact of homogenization on the removal of vicinal diketones and the other on the
location of bubble formation at the commercial scale.

Keywords: Beer fermentation · Flavor · Jet mixing · Mechanical stress · Yeast

1 Introduction microorganism being considered. Such studies

1.1 Scale-down/scale-up concepts
Scale-down studies are critical for successful and
efficient scale-up of bioreactors regardless of the
Correspondence: Dr Alvin W. Nienow, Emeritus Professor of Biochemical
Engineering, Centre for Bioprocess Engineering, Department of Chemical
Engineering, University of Birmingham, Edgbaston, Birmingham,
B15 2TT, UK.
E-mail: A.W.Nienow@bham.ac.uk
Abbreviations: CIP, cleaning in place; dO2, dissolved oxygen concentration;
pCO2, partial pressure of carbon dioxide; PFR, plug flow reactor; PG, pres-
ent gravity (= wort density (kg/m3) – 1000) as an indication of the concen-
tration of fermentable sugars; RJM, rotary jet mixer; VDK, vicinal diketones;
vvm, volumetric flow rate of gas per minute/volume of media (min–1)
must consider two aspects: Firstly, they must ad-
dress basic fluid dynamics, generally multi-phase,
and transfer processes appropriate to the specific
type of bioreactor, for example, mechanically
stirred, airlift or bubble column. Such studies can
be quite generic and in many respects, the under-
standing of such phenomena is applicable even to
the scale-up of similar non-bioprocesses. For ex-
ample, knowledge of the two-phase, gas–liquid flow
and transfer parameters is equally applicable to
the scale-up of aerated bioreactors as it is to chem-
ical reactors for hydrogenation or oxidation. In
many bioreactors, for example, those handling fila-
mentous or polysaccharide fermentations, the
broth often becomes very viscous and non-New-
tonian and these fluid dynamic studies must cover
such properties too. For scale-up to commercial

© 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim 911

Biotechnology
Journal
scale, information must be obtained over a range of
scales from bench to pilot for the task to be com-
pleted with confidence. This requirement arises
because the phenomena are very complex and still
cannot be predicted by computational fluid dynam-
ics, but have to be based on hard-earned experi-
mentally based correlations [1].
The second aspect relates to the characteristics
of the specific organism. In particular, this relates
to its specific oxygen uptake rate and the cell den-
sity that can be achieved and the associated carbon
dioxide and heat evolution rate. These features de-
termine the demands on the bioreactor to provide
a sufficient mass transfer rate to maintain the dis-
solved oxygen concentration, dO2, above its critical
value; heat transfer rate to maintain the desired op-
erating temperature; and rate of CO2 stripping to
prevent build up of partial pressure of carbon diox-
ide (pCO2) and difficulties of pH and osmolality
control. Lack of control of all of these factors can be
damaging to the viability of an organism. Measure-
ment of such parameters to determine acceptable
levels of dO2, pCO2, pH and osmolality can often be
made at the very small scale (shaken microwell or
shake flask) provided suitable instrumentation and
control are available [2,3]. Another parameter of
importance is the impact of mechanical stress, of-
ten loosely called ‘shear’, on organisms. The impact
depends generally on the fluid dynamic stresses
(and in the case of animal cells, the stresses from
bursting bubbles [4]) and on the size and shape of
the organism. In the case of mycelia, damage lead-
ing to breakage of the organism may or may not
lead to loss of productivity [5]. Such studies can be
undertaken at the bench (and often even smaller)
scale because the organisms are all very small rel-
ative to the scale of the equipment.
A particularly important issue that affects scale-
up is the loss of homogeneity inevitably found with
increasing scale [6]. As a result, the organism is ex-
posed to different levels of dO2, pCO2, osmolality,
pH and nutrient concentration throughout the re-
actor. In the case of the last two parameters, the ef-
fect is particularly amplified near the feed pipe, es-
pecially with fed-batch processes and top feeding
[4,6]. The study of the behavior of cells under such
conditions is very limited, and inevitably, relatively
crude. However, such studies have shown that
bench-scale measurements with fed-batch Esche-
richia coli fermentations can reproduce the results
found at the 20 m3 scale, which are quite different
from those found under the usual well-mixed
bench conditions [7]. Much of this earlier work by
Nienow was summarized in a keynote address
“Bench-scale studies for understanding large-scale
stirred bioreactors” at the BioProScale Symposium
912
Biotechnol. J. 2011, 6, 911–925
in Berlin in November 2009. In this review, an ad-
ditional example of recent complementary studies
covering these various aspects of scale-down/
scale-up related to beer brewing, which are leading
to significant changes in practice at the industrial
scale, is presented.
1.2 The traditional brewing process
Typically today commercial beer fermenters are
cylindroconical vessels up to 500 m3 (and occasion-
ally more) with aspect ratios of 2.5:1 to 4:1, as
shown diagrammatically in Fig. 1. The actual beer
fermentations have traditionally been undertaken
(and still are) without the use of mechanical agita-
tion, with mixing being provided mainly by the CO2
evolved during the batch process plus a small con-
tribution from convection as a result of jacket cool-
ing [8, 9]. This approach has largely been main-
tained because of the belief in the industry [10] that
rotating agitators would damage the yeast. In addi-
tion, despite the fact that there is very little CO2
evolution at the start of the batch run, since a lag
phase may occur and initially CO2 has not yet ac-
cumulated to the point of saturation and therefore
bubbles are not formed; or at the end, as the nutri-
ents become depleted, it has been traditional to
model such fermenters as though they were well
mixed [11]. This assumption also means that auto-
matic temperature control based on a single-point
measurement assumes a spatially constant tem-
perature even at the largest scales. Apart from
modeling and temperature control, the well-mixed
assumption has a major impact on the ability of
samples withdrawn from the fermenter to repre-
sent the progress of the batch fermentation and to
enable accurate sequencing decisions to be made
based on them. In this review, three complementa-
ry studies are presented that were undertaken to
improve the understanding of this very traditional
process. They involve fermentation work at the
bench scale and at the commercial scale, and phys-
ical studies at the pilot scale of a novel mixing tech-
nology; the latter of these is eventually being used
to give a much enhanced process in a range of ac-
tual beer production plants.
2 Individual research topics
2.1 Bioreactor scale-down studies
2.1.1 Fluid dynamic stress
One of the traditional beliefs of the brewing indus-
try has been that mechanical agitation during fer-
mentation would damage the yeast. A measure of
© 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Biotechnol. J. 2011, 6, 911–925 www.biotechnology-journal.com

this attitude was the comment of one referee on an
earlier paper [10]; “Do the authors realize that
Schlitz introduced stirrers into their fermenters
and they closed down?” Both statements are true
and Schlitz, who were once one of the largest beer
producers in Milwaukee (“The beer that made Mil-
waukee famous” was their slogan) introduced me-
chanical agitation in the early 1970s. Later, they
made massive losses and were finally sold to the
Stroh Brewing Co. in 1982. However, these events
do not show a causal linkage. As a result of such at-
titudes, mechanical mixing has generally not been
used commercially. The first scale-down study,
therefore, addressed the mechanical stress issue by

Figure 1. A diagrammatic representation of a typical cylindroconical beer fermenter: (A) the 160 m3 fermenter used for detailed mixing studies [The four
cooling jackets and the arrangement of the Biomass probes with their location and the method of data handling are also indicated (10 hl ≡ 1 m3)] (modi-
fied from [8]); (B) the impeller type and its position for mixing studies.

© 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim 913

Biotechnology
Journal
using chemostat culture, as previously performed
for E. coli [12] and Corynebacterium glutamicum
[13]. However, since beer fermentation is essential-
ly an anaerobic batch process (except for the nec-
essary presence of dissolved oxygen at the start of
the batch, which impacts on the formation of flavor
compounds), it is difficult to use the technique that
way. Therefore, it was decided, as a compromise, to
begin by assessing the sensitivity to mechanical
stress under aerobic conditions.
Studies were undertaken to determine the im-
pact of fluid mechanical stress on a brewing yeast
strain (Saccharomyces cerevisiae (NCYC 1324) ob-
tained from the National Collection of Yeast Cul-
tures (Institute of Food Research, Norwich, UK))
[14]. Experiments were conducted in a stirred
chemostat (essentially a continuous stirred tank
bioreactor) culture 1 or 2.5 L in volume at 25°C and
pH 5.5 under air-sparged conditions for up to
15 days. During the run, both nutrient feed rate and
dissolved oxygen concentration, dO2, were held
constant, the latter by gas blending. In a particular
example at the 1L scale at 40% saturation, agitation
was at either 100 or 475 rpm with a Rushton turbine
under turbulent flow conditions (Re > ≈ 2 × 104).
These agitation conditions gave mean specific en-
ergy dissipation rates, ε–T (equivalent to power input
per mass of media in the bioreactor) of either 0.045
or 4.5 W/kg. This latter value is at the upper limit of
that used at an industrial scale for aerobic fermen-
tations and the former is similar to that of about
0.035 W/kg generated by the maximum rate of CO2
release during a beer fermentation of 300 m3 [15].
In addition, to check on the possibility of damage
from bursting bubbles, as found in animal cell cul-
ture, sparging was undertaken at a typical rate for
aerobic fermentations of 1 vvm and a high rate of
3 vvm. Both are much higher than the typical max-
imum CO2 evolution rates in anaerobic beer fer-
mentations of about 7 × 10–3 vvm [15]. During the
run, samples were taken for analysis by flow cy-
tometry as well as cell density and glucose concen-
tration, while dO2 and exit gas composition were
monitored continuously, the latter of these by mass
spectrometry.
At ε–T = 4.5 W/kg, a transient effect on the cells,
as measured by multi-parameter flow cytometry,
was found. In particular, for a short time, there was
no indication of cell budding, but, even at this high
level, within a few hours, budding returned (Fig. 2).
All other measured parameters, including dO2, bio-
mass, and exit gas concentration, showed no statis-
tically significant variation throughout the run re-
gardless of the level of agitation or aeration inten-
sity. While not necessarily conclusive with respect
to anaerobic beer fermentation, these results
914
Biotechnol. J. 2011, 6, 911–925
strongly suggest that gentle agitation should not
damage yeast during that process.
The mechanical stress on the cells in turbulent
flow is now generally analyzed by applying Kol-
mogoroff’s theory of locally homogeneous isotrop-
ic turbulence. The microscale of turbulence, λK, is
given by the relationship shown in Eq. (1):
λK = (ν3/εT)1/4 (1)
in which εT is the local specific energy dissipation
rate and ν is the kinematic viscosity. For cells with
size dc > λK, the mechanical stress on them due to
turbulence ∝ dc2/3 and for dc < λK is ∝ dc2. Thus, for
entities < λK, such stress falls very rapidly, so that
if a cell is of a size < λK, it is generally considered
that it will not be damaged by turbulence. At ε–T =
4.5 W/kg (or allowing for the maximum value εT of
the order of 30ε–T [6]), dc < λK, so that damage would
not be expected.
2.1.2 Improving homogeneity by agitation
In a paper published in 1994 [16], inhomogeneities
in both temperature and yeast concentration were
reported at certain times during batch beer fer-
mentations at the commercial scale and this vari-
ability was ascribed to poor mixing during periods
of low and zero CO2 evolution rate. The value of ε–T
from CO2 evolution can be estimated [17] from Eq.
(2):
ε–T = νsg (2)
in which νs = QCO2 (V/A) and V and A are the vol-
ume and cross-sectional area of the fermenter, re-
spectively, and a typical maximum CO2 evolution
rate, QCO2, at commercial yeast and sugar concen-
trations is 1.2 × 10–4 (m3/s)/m3 [16]. In addition, V/A
≈ αT, in which α is the fermenter aspect ratio
(= H/T) and H and T are its height and diameter, re-
spectively. Equation (2) as defined this way is for a
gas being introduced at the base of a vessel. There-
fore, since here CO2 is generated in situ, but at a lo-
cation that is not actually known, it has been sug-
gested [11] that a value of H/(2T) should be as-
sumed for the aspect ratio. Thus, it can be seen that
ε–T increases with scale due to both increasing size
and, in general, the larger-scale aspect ratio. For
typical industrial fermenters of the order of 400 to
500 m3 with aspect ratios about 4 to 5, from Eq. (2),
(ε–T)max values are of the order of 0.045 W/kg.
Scale-down studies on the impact of agitation
leading to improved homogeneity were undertaken
in 500 mL working volume fermenters at 12°C with
S. cerevisiae NCYC 1324 and a wort used to make
Munton’s Pale Ale. Experiments were run without
© 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Biotechnol. J. 2011, 6, 911–925
Figure 2. Forward scatter data from flow cytometric analysis of S. cerevisiae
cells, showing yeast budding disappearing and then returning at high in-
tensity agitation and being completed with a return to low intensity
(adapted from [14] and published with permission from Food and Bio-
products Processing).
agitation and at 4 agitator speeds from 150 to
600 rpm (to give ε–T values from 0.004 to 0.26 W/kg,
thus straddling the maximum found due to CO2
evolution for a short time at the large scale). One of
the parameters measured was wort specific gravi-
ty. Essentially, wort is the brewer’s term for the me-
dia and specific gravity is an indirect measure of
the fermentable sugars available as the nutrient
source for fermentation. In addition, the dry cell
weight, the actual concentration of fermentable
sugars and flavor compounds, was also monitored.
The results could be divided into two sets, one cor-
responding to unagitated and 150 rpm conditions
and the others at 300, 450, and 600 rpm. Of the latter
set, the lowest speed gives an ε–T value of 0.03 W/kg
(of the order of the maximum at the large commer-
cial scale due to CO2 evolution). Comparing stirred
with unstirred, the overall batch fermentation time
was significantly reduced from about 160 to 100 h,
with increased dry cell weight, an increase in high-
er alcohols, a reduction in esters (Fig. 3), and the
percentage of ethanol produced was unchanged.
These changes can mainly be ascribed to the full
suspension of yeast that could be observed at
speeds ≥ 300 rpm, so that all of the surface of the
cells was available for mass transfer and a small
enhancement of the latter due to the faster slip ve-
locity of the cells relative to the fluid [18]. Multi-pa-
rameter flow cytometry was used to assess the per-
centage of dead cells at the end of fermentation (as
indicated by the utilization of all sugars). Without
agitation and at 0.03 W/kg, there were about 6%
© 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.biotechnology-journal.com
dead cells and at higher agitation intensities it was
about 9%. For all of these values of ε–T, dc << λK.
In the 1994 report of poor homogeneity at
the industrial scale [16], a low temperature of about
5°C had been noted in the cone of the cylindrocon-
ical vessel being used for lager fermentation nom-
inally controlled to operate at 12°C. Such inhomo-
geneities have recently been investigated for the
first time [19] by an experimental scale-down tech-
nique previously developed to study spatial and
temporal concentration variations in large-scale
agitated fermenters [5,7]. In this technique, a well-
stirred bench-scale fermenter (STR), with con-
trolled dO2 and pH, and homogeneous with respect
to both of these parameters and to nutrient con-
centration even when used in the fed-batch mode,
has attached to it a flow loop (PFR). In this loop,
variations in such parameters, such as those found
on the large scale, can be imposed.
To mimic the large-scale batch beer fermenta-
tion, the reported temperature homogeneities were
simulated by using a well-mixed 5 L Electrolab fer-
menter (working volume 4 L) controlled to 12°C
and a PFR of 0.4 L, in which temperature variations
from 12 to 5°C as a function of time were imposed,
matching those observed industrially. A Grolsch
lager wort (Coors Brewery Ltd, Burton-upon-
Trent, UK) was used for these studies. Circulation
around the loop was done by a peristaltic pump to
give circulation times between a minimum of about
10 min and a maximum of 25 min.These times were
based on Eq. (3) for the circulation time, tC [19]:
tC = 2.75(H/T)(gvsT–2)–0.33 (3)
in which νS was based on QCO2 values, as a function
of time, and the fermenter aspect ratio given in the
1994 study [15, 16]. The longer time assumed some
mixing even without CO2 evolution due to convec-
tion currents from the cooling of the cone.
When comparing the results with circulation
through the low-temperature PFR with those with-
out it, the fermentation time was longer (110–120 h
versus ≈ 95 h) and the amount of fermentable sug-
ars converted to ethanol and flavor compounds and
the dry cell weight were all less. In addition, the
amount of CO2 evolved was greater (≈ 50% com-
pared with ≈ 40%). Overall, this simulation showed
that the presence of the low temperature in the
base of the cone at the industrial scale led to a poor-
er performance than that obtained when it was
eliminated by improved mixing throughout the
batch time.
In addition, experiments were undertaken in
which unagitated conditions were compared with
agitated conditions with the Grolsch Lager at the
915
Biotechnology
Journal
Figure 3. Impact of agitation compared with unagitated conditions in pale
ale fermentation: (A) fermentation time; (B) the change of certain flavor
compounds (filled symbols, unagitated conditions; unfilled symbols, agi-
– concentration of cells at various points were meas-
tated with a Rushton turbine at ε T at ≈ 0.25 W/kg) (adapted from [10] and
published with permission from Journal of the American Society of
Brewing Chemists).
5L scale at 12°C without the PFR. Thus, just the
impact of maintaining a full yeast suspension
throughout the batch could be assessed.The results
obtained essentially showed the same trends as
those with Munton’s Pale Ale.Agitation reduced the
overall batch fermentation time from about 104 to
80 h, increased the dry cell weight of yeast and
higher alcohols, suppressed ester formation, and
left the percentage of ethanol unchanged.
916
Biotechnol. J. 2011, 6, 911–925
2.1.3 Conclusions
The above studies show that at the bench scale, in-
tense agitation under aerobic conditions did not
damage the yeast, as found recently for other mi-
croorganisms [12,13]. Even animal cells, which do
not have a cell wall, have been shown to be able to
be agitated at ε–T values up to 0.25 W/kg without a
reduction in cell viability or productivity [4]. The
work also showed that under anaerobic conditions,
mechanical agitation up to about 0.25 W/kg does
not increase the amount of dead cells produced rel-
ative to unagitated conditions. In addition, agitation
intensities from 0.03 to about 0.25 W/kg improved
beer productivity by reducing the fermentation
time by 1–2 days in both Munton’s Pale Ale and
Grolsch lager fermentation compared with unagi-
tated fermentations with a modest change in the
distribution of flavor compounds. Also, a scale-
down simulation of the use of agitation to eliminate
spatial temperature variations previously reported
in the literature for commercial-scale lager beer
fermenters was undertaken. It showed that the
performance was worse (less ethanol) when tem-
perature variations were present. This scale-down
work suggests such improvements should be
achievable at the large scale by the introduction of
mechanical agitation.
2.2 Bioreactor large-scale studies
2.2.1 Advanced measurement of inhomogeneity
during fermentation
Two related studies of heterogeneities [8,20] were
undertaken in a commercial cylindroconical fer-
menter installed at the Coors Brewery, Alton,
Hampshire, UK, with a 160 m3 working volume, a
diameter of 4.25 m, an aspect ratio of 3.9:1, and an
internal cone angle of 72°. Cooling was provided by
external jackets (three on the cylinder and one on
the cone) (Fig. 1A). In it, the local temperature and
ured by an array of eight Aber biomass probes
(Aber Instruments, Science Park,Aberystwyth, UK)
inserted in waterproof enclosures positioned
throughout the fermenter.
An additional thermometer was placed about
1 m above the cone for use as the measuring device
in a feedback control loop for controlling the tem-
perature during the fermentation to 16°C for the
first 45h, ramping to 20°C over the next 25h, and
then holding it for the remaining fermentation time
(120 h total) plus about 20 h to allow time to remove
the unpleasant flavors associated with vicinal dike-
tones (VDK). During this time, the flavor com-
pounds associated with VDK are readsorbed by the
yeast cell and converted into acetoin and butanedi-
© 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Biotechnol. J. 2011, 6, 911–925
ol to a level that does not affect flavor. Finally, the
feedback loop is also used to control the decrease
in the temperature to 5°C at the end of the fermen-
tation. Sometimes the yeast is removed before cool-
ing and sometimes after, but at the end of this pe-
riod the beer is sent to storage.
Malt wort was pumped in at 35 m3/h and after
5 m3 yeast was added to give, when the wort addi-
tion was complete, a target yeast concentration of
about 5 × 106 cells/mL. Fig. 4A shows the change in
yeast concentration as a function of time during
batch fermentation as measured at each probe
when using a moderately flocculating lager yeast
strain [8]. It also shows the concentration of fer-
mentable sugars [expressed as Present Gravity, PG,
(= density (kg/m3) – 1000]. It was found that during
the first 10–12 h, the measured yeast concentration
varied significantly both spatially and temporally.
For the next 50–60 h, the concentration rose and
was essentially the same at probes 1–5 (Fig. 4A) in
the cylindrical portion of the vessel with a maxi-
mum value of about 10 × 106 cells/mL. After that,
the concentration fell sharply, remaining approxi-
mately the same from 80 to 120 h at ≈ 5 × 106
cells/mL. For the upper probe in the cone, when the
wort PG had fallen halfway to the final concentra-
tion (indicating that all of the fermentable sugars
had been utilized), the measured yeast concentra-
tion suddenly rose to about 30 × 106 cells/mL before
falling to a value similar to that in the cylinder. For
the two lower ones, the concentration rose rapidly
to a maximum of about 120 × 106 cells/mL just as
that in the cylinder fell. Thus, well before all of the
fermentable sugars were used up after 120 h and
when there was still significant CO2 evolution, so
that the assumption of good mixing might have
seemed reasonable, there was an intense concen-
tration gradient of yeast cells from about 5 × 106
cells/ml in the cylinder to about 100 × 106 cells/mL
in the cone; a ratio of 20:1. Although not measured,
it was likely that the concentration at the very bot-
tom of the cone was even higher and, in addition,
other dissolved chemical species were probably
present in different concentrations spatially.
The temperatures measured by the Aber probes
told a similar story (Fig. 4B). Small variations in
temperature from probe to probe during the first
10 h were observed followed by constant values
matching the desired profile until the end of the
ramp to 20°C. However, although temperature ho-
mogeneity was observed during the ramp, its com-
mencement corresponded with the start of the
yeast buildup in the cone.After the end of the ramp,
the temperatures in the cylinder followed the con-
trol value, but that in the cone rose above it to about
22°C and remained there during the holding peri-
© 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.biotechnology-journal.com
od for VDK removal. Temperature heterogeneity
was even worse during the cooling period. For ex-
ample, at its end, when the temperature throughout
should have been 5°C, it was still aroud 16°C in the
cone and also in a band at the top (probe 1) (Fig.
4C).
It can be concluded that, with just CO2 evolution
and natural convection, the mixing was very poor.
As a result, at no time during the fermentation was
the yeast concentration constant throughout the
vessel and for the last 50 to 60 h of the 120 h fer-
mentation, some 70% of yeast was in the cone at a
cell concentration about 20 times greater than that
in the cylinder. In addition, the temperature in the
cone was 2°C higher than that in the cylinder. That
difference remained during the period for the re-
moval of VDK. During cooling, the increase in tem-
perature was up to 10°C in the cone and at the top
to give a total time of 60 h with temperatures at
many locations higher than optimum. Overall, for
all but the first 70 h of a full batch run lasting 180 h,
temperature control and homogeneity were poor.
2.2.2 Impact of mechanical agitation on homogeneity
and fermentation
The second study [20] used three lager yeast
strains with different flocculation characteristics
(non-, moderately, and heavily flocculating). Al-
though there were differences in detail, the results
for all three strains was the same as in the first
study, namely, very poor homogeneity with respect
to both yeast concentration and temperature
throughout the latter part of fermentation and dur-
ing subsequent stages of the total batch time. An
example of the spatial temperature variations
throughout a whole batch run for the moderately
flocculating yeast is shown in Fig. 4B, which also
shows the fermentable sugars being utilized in
120 h.
In addition, fermentations were undertaken
with mechanical agitation by using a three-bladed
propeller, D = 0.7 m, mounted at the junction of the
cylinder side wall and the cone (Fig. 1B), pointing
down at an angle of 6°. This orientation of the im-
peller was chosen because it was desired to induce
a downward flow in the region where the yeast was
sedimenting to help maintain it in suspension.This
impeller would input a power, P, given by Eq. (4):
P = PoρN3 D5 (4)
By assuming that the power number for such an
impeller is 0.5 [1], it would draw about 5 kW at the
chosen impeller speed of 230 rpm, or an average
specific energy dissipation rate, ε–T for the whole
fermenter contents of about 0.03 W/kg; a similar
917
Biotechnology
Journal
Figure 4. Measurements at different points in a 160 m3 beer fermenter containing a moderately flocculating yeast. (A) Yeast concentration and PG profile
(adapted from [8]); (B) PG profile and temperatures (adapted from [20]; (C) temperature during the holding period for VDK removal and programmed
cooling (adapted from [20]); and (D) impact of agitation on the PG profile and on the distribution of the cells (adapted from [20]).
value to that found to be effective at reducing batch
time and suspending cells at the bench scale.
Figure 4D gives the change of concentration of
fermentable sugars as a PG profile and the viable
yeast concentration at the various probes for the
duration of the fermentation and for the time to re-
duce VDK. Agitation was stopped after 82 h when
all of the fermentable sugars had been used up. Un-
til this time, even at this gentle agitation intensity,
the probes all gave similar readings for the yeast
concentration even during the start-up period and
at the end of fermentation when CO2 evolution is
low or zero. Immediately after agitation ceased, the
concentration in the cone increased dramatically
as the yeast flocculated and sank into it, and that in
the cylinder fell.
Overall, the time for the fermentable sugars to
be depleted fell from an average of 83 h without ag-
itation to 67 h with it (based on 6 runs in each case).
In addition, the variation in this time was reduced
by about a third, thus giving a more consistent per-
formance. Somewhat surprisingly, the time to re-
move the VDK was about 50 h after sugar depletion
with or without agitation. To assess the impact of
agitation on the whole batch time, agitation was
918
Biotechnol. J. 2011, 6, 911–925
also stopped after 50% of fermentable sugars had
been utilized. In this case, the time to sugar deple-
tion was a little less at an average of around 60 h
(but within the range of those that were agitated
throughout), but the time for VDK removal was sig-
nificantly reduced to about 30 h to give a total batch
time of about 90 h with much reduced variability.
The precise reasons for the improved VDK re-
moval are beyond the scope of this paper. The im-
portant point is that the use of agitation gives the
opportunity to tune and control the level of homo-
geneity with respect to the concentration of yeast
and soluble species and of temperature to give a
much reduced and more consistent total batch
time, independent of CO2 evolution. Indeed, it was
found that slightly different strategies should be
used depending on the flocculating characteristics
of the yeast [20]. After more runs, it was also shown
that during 17 tests with agitation at the large scale,
there was a 7% increase in the viability and 5% in-
crease in mass of the removed yeast compared with
that found from 22 standard unagitated runs as
controls [21]. In the case of the controls, the yeast
was removed after the vessel was cooled to 4oC. For
the trials, to maximize yeast quality and because
© 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Biotechnol. J. 2011, 6, 911–925
the yeast flocculated so well, it was removed 24h af-
ter agitation was discontinued without prior cool-
ing of the beer.
2.2.3 Conclusions
This large-scale study confirmed in more detail the
invalidity of the assumption that mixing due to con-
vection currents and CO2 generation is sufficient to
ensure homogeneity during batch beer production.
It was also shown that the use of mechanical agita-
tion essentially eliminated the measured spatial in-
homogeneities in yeast concentration and in wort
temperature; the latter during both fermentation
when heat is evolved and the temperature is held
constant for many days and also when a cooling
profile is being imposed after VDK removal. As a
result, the fermentation time and the total batch
time is significantly reduced and also more consis-
tent. In addition, the yeast did not deteriorate with
multiple reuse. All of these results fit in well with
the scale-down studies presented previously
[10,14,19]. In addition, depending on the character-
istics of the yeast, the fermenter configuration, and
the particular malt, the agitation can be tuned to
optimize the batch performance, independent of
the CO2 evolution.
However, although these results suggest that the
addition of an agitator with a low specific energy
dissipation rate offers distinct operating advan-
tages when installed as indicated in Fig. 1B, such an
installation is difficult and expensive in many ex-
tant fermenters. It requires breaking into the body
of the fermenter, which is difficult both because of
its shape and especially due to the presence of the
cooling jackets. Furthermore most breweries are,
due to cleaning issues and the presence of a me-
chanical seal penetrating the tank wall, hesitant to
install agitators.
Mixing in a vessel can also be achieved by using
a jet of the fluid in the tank, which is easier to in-
troduce than an impeller because it can be done
much less invasively. The efficiency and range of
applications of jet devices has recently been in-
creased by the use of a jet head that rotates in two
orthogonal planes [21]: the rotary jet mixer (RJM)
(Alfa Laval, Copenhagen, Denmark). The RJM is a
modification of the jet heads commonly used for
cleaning-in-place (CIP) in breweries. It can there-
fore be installed essentially by using the same
pipework and be used for both mixing and CIP.
However, while the physical aspects of agitators
and simple jets have been well studied over many
years, the RJM is still being characterized; an es-
sential aspect of scale-up/scale-down studies in
bioprocessing. A brief summary of that work con-
stitutes the next part of this paper.
© 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.biotechnology-journal.com
2.3 Physical studies of the rotary jet mixer (RJM)
2.3.1 The concept
The principle behind an RJM (see Fig. 5A) is that
the pressure of a liquid entering at the top causes it
to flow past a guide, 1, to a turbine, 2, whereby the
head is set into a slow horizontal rotation by a gear-
ing system (not shown). The four distributor arms,
3, are simultaneously set into a slow rotation in the
vertical plane, and the liquid jets that leave the ma-
chine through the nozzles, 4, sweep the tank vol-
ume (http://www.alfalaval.com/solution-finder/
products/rotary-jet-mixer/pages/howitworks.
aspx). The RJM is placed in the liquid to be mixed
in the tank and the liquid is withdrawn from the
bottom, circulated through a loop, re-injected into
the tank through the four nozzles, and the jets mix
the contents (Fig. 5B). After 45 rotations in the ver-
tical plane and 43 in the horizontal plane, the
sweeping motion is repeated. As a result of these
two orthogonal movements, a pseudo-chaotic flow
pattern is introduced. With impeller mixing, intro-
ducing a chaotic element in some way to the flow
the impeller generates significantly reduces the
mixing time in laminar and transitional flow [22].
Thus, it might be anticipated that such an improve-
ment would be obtained when using jet mixers.
Two studies on the basic physical features of the
RJM have been reported: the first mainly concen-
trated on mass transfer, considering its application
in aerobic fermentation [23] and the second con-
sidered its ability to blend [22]. The latter is most
relevant to this paper.
2.3.2 Homogenization by RJMs
It is important that physical studies are undertak-
en in a system of appropriate size to give confi-
dence in scale-up to commercial equipment and to
guide scale-down studies. In [22], homogenization
was studied using a decolorization technique in a
cylindrical tank 0.75 m in diameter with a liquid
height of about 1.9 m (Fig. 5B). In this method, the
mixing time, θ, is determined by how long it takes
for the color associated with a solution of iodine
and starch to disappear following the rapid addi-
tion of sodium thiosulfate. To compare the results
with the mixing achievable by impellers, the mean
specific energy dissipation rate, ε–T, can be calculat-
ed from Eqs. (5) and (6)
ε–T = P/ρV = QΔp/ρV (5)
1
Δp = ρQ2d–4 (6)
2π 2
919
Biotechnology
Journal
In Eqs. (5) and (6), Q is the volumetric flow rate, Δp
is the pressure drop, d is the pipe size, and V is the
volume of the vessel. Equation (6) applies in low
viscosity liquid where pressure losses can be neg-
lected. In addition to mixing time experiments by
using the standard RJM, runs were also undertak-
en without the head rotating to indicate the effec-
tiveness of the chaotic element that the head move-
ment introduces into the flow.
In summary, the results of particular relevance
to this review (Fig. 5C) showed that the chaotic mo-
tion of the head improved mixing compared with
that without head motion even in the turbulent re-
gion. In addition, the functionality between mixing
920
Biotechnol. J. 2011, 6, 911–925
Figure 5. The RJM: (A) a schematic view; (B) the basic
mode of operation with details added for the measure-
ment of mixing time: A: centrifugal pump, B: positive
pump for fast dosing of sodium thiosulfate, C: rotary
jet head, 1: flow meter, 2 & 4: pressure transmitters,
3: temperature sensor; (C) the mixing time, θ, versus
mean specific energy dissipation rate, P/ρV (adapted
from [22] and reproduced with permission from
Chemical Engineering Research and Design).
time and mean specific energy dissipation rate was
the same as that in stirred vessels, namely, θ ∝
(P/ρV)–1/3. When compared with the mixing time
achieved with dual axial flow impellers, which are
recommended for vessels of aspect ratio of about 2
[6], it was shown that the RJM gave very similar val-
ues when compared at the same P/ρV [22].
2.3.3 Conclusions
It has been shown that the RJM is able to provide
homogenization essentially as effectively as im-
pellers in high aspect ratio vessels at the same
mean specific energy dissipation rate, ε–T. Thus, it
should be possible to improve homogeneity in
© 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Biotechnol. J. 2011, 6, 911–925
commercial cylindroconical beer fermenters as ef-
fectively with an RJM as with stirrers at the same
low ε–T values. Also, since damage to cells is related
to ε–T and the use of agitators does not damage
them, then at the same ε–T the RJM should not ei-
ther. Finally, the RJM is easy to install in cylindro-
conical beer fermenters. The last part of this paper
considers their use for such an application.
2.4 The use of the RJM in commercial
cylindroconical beer fermenters
RJMs have now been used for a large number of
commercial fermentations covering a wide range of
scales, vessel shapes, and types of beer. The results
achieved have matched well with predictions from
earlier scale-down/scale-up tests. Some examples
are given here.
2.4.1 Batch times and their variation
The significant reduction in batch time is illustrat-
ed in Table 1. Obtaining data in an industrial envi-
ronment in a consistent way is rarely possible.
Thus, for the four fermentations reported in Table
1, the assessment of the batch time varies from the
time at which the temperature is held for VDK re-
moval to the time required for the whole batch pro-
duction process to be completed together with the
time for CIP. In addition, in each case, the vessel
geometry, wort concentrations, and type and yeast
strain were different. In every case, the reduction is
quite significant in relation to productivity of the
operation.
Fig. 6A and 6B show, for a different 500 m3 high
alcohol lager fermentation, the variation in batch
times and the average up to the start of cooling with
and without the use of an RJM. Of course, a range
of RJM sizes are available and the size and the re-
circulation rate chosen must be matched to the size
of the vessel and the mean specific energy dissipa-
tion rate required for effective operation based on
the previous fermentation and physical studies. In
this case, an IM20 RJM with 10 mm nozzles was
used at a flow rate of 25 m3/h corresponding to an
energy input [from Eqs. (5) and (6)] of 1.7 kW or as
low as about 0.0034 W/kg. Again, there is a marked
Table 1. Effect on different process times at four breweries for standard conditions and with mixing by RJH
Brewery Period of process
A Start filling to end of CIP
B Time to VDK
C Time to end of cooling
D Start filling to end of CIP
© 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.biotechnology-journal.com
reduction in the time (≈ 50%) required when the
RJM is used and even more impressive is the con-
sistency of the batch time.
Figure 6C shows the temperature–time profile
for the period of programmed cooling from 16 to
4°C (in this case in a 180 m3 fermenter) for which
the cooling time was reduced from 25 to 13.5 h.This
reduction arises from the very significant increase
in velocity over the internal surface of the fer-
menter in the vicinity of the cooling jackets be-
cause, by this time in the batch, there is no motion
generated by CO2 evolution. As a result, the inter-
nal heat transfer coefficient, which is the rate lim-
iting one, is markedly increased and so is the over-
all heat transfer rate.
2.4.2 Other process parameters
When producing such huge batches, there is some
variation in the initial amount of fermentable sug-
ars in the wort at the start of the fermentation. Nev-
ertheless, in addition to the consistently shorter cy-
cle times, for all the runs illustrated in Fig. 6A, the
remaining unutilized fermentable sugars at the end
of a batch was much more consistent and lower
when the RJM was used compared with the stan-
dard runs. In addition, when the amount of ethanol
normalized by the original fermentable sugars was
determined, the stirred runs gave more consistent
and higher values, averaging about 3%.This level of
consistency on all of these factors also implies that
the yeast viability and quality remained undimin-
ished, as would be expected from the earlier small-
scale studies.
2.4.3 Beer quality
It is vital that the beer, following such modified
treatment, still appeals to the consumer. At the
commercial scale, this is usually assessed by taste
panels. Here, about 400 consumers were used to
taste the beers from the unstirred fermentations
and compare them with those in which the RJM
was used. No consistent preference or differences
were detected; the beers were ‘true-to-type’. Again,
the consistent beer quality also implied that the
yeast quality remained undiminished by the use of
the RJM.
Standard batch (days) Batch with RJM (days) Reduction (%)
13.0 9.5 27
7.0 6.0 15
11.0 10.0 10
20.0 16.0 20
921
Biotechnology
Journal
Figure 6. Impact of the RJM on batch performance: (A) range of times; (B) statistical average and spread; (C) cooling profile during standard operation
with one mixed by an RJM.
2.4.4 Conclusions
The reduced fermentation time and the robustness
of the yeast when using the RJM matched the out-
come expected following the stirred bench-scale
experiments as well as those when commercial-
scale cylindroconical beer fermenters were stirred
by an impeller at similar low mean specific energy
dissipation rates. In addition, the more consistent
batch times found in the latter case were also re-
peated. Other consistent benefits noted were the
lower levels of residual fermentable sugars and
the higher normalized yield of ethanol. The use of
the RJM after the bulk of the yeast had been re-
moved to cool to aid flocculation for further yeast
removal and for storage greatly reduced the cool-
ing time.
922
Biotechnol. J. 2011, 6, 911–925
3 Overall conclusions
Overall, the problems of poor mixing at the com-
mercial scale have been shown to be greater than
previously thought as measured by nine Aber bio-
mass probes spaced throughout an approximately
160 m3 working volume cylindroconical beer fer-
menter to assess the local yeast concentration and
temperature. In particular, it was shown that a
range of yeasts with different flocculating proper-
ties started to sediment to the cone when about 50%
of fermentable sugars still remained; and more
than about 90% had sedimented when some 25%
fermentable sugars still remained. In addition, for
much of the total batch time, the temperature in the
cone was significantly higher than that in the cylin-
der.
Studies in standard bench-scale stirred fer-
menters to improve homogeneity showed that
yeast could be agitated at relatively high agitation
© 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Biotechnol. J. 2011, 6, 911–925
intensities under aerobic conditions without dam-
aging its viability or fermentation performance.
With anaerobic fermentations, agitation reduced
the time to use up fermentable sugars by some 25%
both for lager and pale ale fermentation. Using
similar agitation intensities at the 160 m3 scale was
shown to maintain the yeast concentration and
temperature uniform spatially. It also gave more
consistent and reduced fermentation times by a
similar amount to the bench-scale work; the exact
value depending on the yeast type. Following this
change, yeast viability was also slightly improved.
However, while the cylindroconical vessel has a
shape with many advantages for beer fermentation
by the traditional route, it is far from ideal for mix-
ing by rotating stirrers. In addition, the retrofitting
of a mechanical stirrer in one is expensive and may
compromise sterility.
Recently, a new mixing system based on circu-
lating the content of the vessel by using a modified
CIP head, the Alfa Laval RJM, has been developed.
Pilot-scale tests showed that the randomized mo-
tion produced by the RJM gave mixing characteris-
tics better than static jets. Also, at the same mean
specific energy dissipation rate, the required time
to mix the contents of a vessel of 2:1 aspect ratio
was similar to that of the preferred agitation sys-
tem, namely, dual axial flow impellers.
Given this quality of homogenization and the
ease with which an RJM could be installed in a
commercial fermentation (CIP heads are usually
present anyway), its use was tested in a number of
fermenters under conditions implied by the earlier
work. Similar results as those obtained for small-
scale fermentations, namely, reduced and more
consistent fermentation times, were obtained for a
wide range of different yeasts, worts, and configu-
rations. In addition, the yield of ethanol produced
based on the amount of fermentable sugars was
also higher and more consistent. The use of the
RJM also led to a significant reduction in the time
for the cooling stage of the batch process. Thus, the
use of the RJM should lead to improved productiv-
ity by shortening cycle times and also making batch
scheduling easier.
In passing, it is worth noting that the loop of the
RJM may also be used for other purposes. Normal-
ly, at the start of the batch, the wort is saturated with
oxygen, which gets utilized in an uncontrolled way
by the yeast, beginning when yeast addition is
started followed by a gradual fall in dO2 to zero over
about 15–20 h. The addition of wort and yeast (as a
mixture) and with a controlled level of dO2 could
also be conducted in the loop. In addition, a plate
heat exchanger could be used to control the tem-
perature during fermentation and to give even
© 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.biotechnology-journal.com
faster cooling. Finally, the RJM can, of course, be
used for CIP.
In general, the biological and physical results
found from the experiments at different scales are
comparable. One aspect that is clearly different be-
tween the bench and commercial scale relates to
flavor; a critical aspect of beer production. At the
bench scale, there is a notable change in flavor
compounds without agitation compared with those
found under agitated conditions. On the other
hand, at the commercial scale, tasting panels could
not detect any difference between the beer pro-
duced by the standard method, for which mixing
was just due to CO2 evolution, and that mixed by the
RJM. There are a number of possible explanations
for that. First, without mechanical agitation, the
mean specific energy dissipation rates generated
by CO2 and its linear superficial rise velocity are
much greater at the commercial scale because of
the greater height and aspect ratio than at the
bench scale. Thus, the yeast motion under only
bubbling agitated conditions is worse on the small
scale than the large. Second, the addition of me-
chanical agitation, whether by an impeller or the
RJM, leads to improved homogeneity than that
achieved by CO2 evolution alone. However, since
the mean specific energy dissipation rate is similar
at the bench and commercial scale, the level of ho-
mogeneity on the large scale will still be less than
at the bench [6].
In addition, at the production scale, there are ex-
tra steps in the process. There is the initial period
where the level of dO2 falls and the time and details
of this drop depend very critically on how the fer-
menter is filled and the yeast added, and the filling
step may take up to 24 h while at the bench it is es-
sentially instantaneous. Thus, the change of dO2
with time at the start, which is also known to impact
on flavor compounds, such as esters [21], is almost
certainly different at the two scales. It may also ac-
count for the different flavor compounds obtained
under agitated conditions compared with unagitat-
ed at the bench scale shown in Fig. 3. Finally, at the
end, there are extra steps associated with the hold-
ing period for the removal of VDK and the cooling
time to encourage flocculation and yeast removal.
Thus, with respect to flavor, it is almost impossible
to produce a totally representative biological scale-
down test.
In general, the difficulty of producing totally
representative biological scale-down tests is true
for many bioprocesses. Such tests should be used
as a guide for large-scale process improvement and
for highlighting potential scale-up problems. Al-
though the scale-down model is a crude one, it is
still very important because of the almost total lack
923
Biotechnology Biotechnol. J. 2011, 6, 911–925
Journal

of fundamental studies on the way cells behave in
even a well-controlled dynamic environment, that
is, one continuously fluctuating spatially and tem-
porally. Scale-up/scale-down studies should also
be used in an iterative mode. For example, the fact
that VDK is removed more rapidly if mechanical
agitation ceases before all of the fermentable sug-
ars have been removed is not easy to explain. From
this work, it is now known from the commercial
scale tests that such action allows early flocculation
and separation of the yeast into the cone. This ob-
servation suggests a possible further scale-down
study to ascertain the optimum agitation strategy
for VDK removal.
The work also suggests another physical study.
The very poor yeast suspension, even when there
are 50% fermentable sugars left and the yeast con-
centration is high leading to a significant CO2 evo-
lution rate, is surprising. As pointed out earlier, the
frequent assumption is that CO2 bubbles are nucle-
ated at or close to the base of the cylindroconical
vessel and therefore lift the cells in their wake.
However, given the huge static head, evolution at
that level seems to be very unlikely. Recent work on
boiling reactor systems clearly shows that bubbles
only form in the top meter or even less, even though
the size of the bubbles and the vigor of their release
at the surface might easily lead to the conclusion
that they come from much deeper in the vessel [24].
A study of the precise location at which CO2 bub-
bles are generated seems very important if better
physical models, leading to further improvements
in brewing technology, are to be developed.
The authors would like to thank Coors Brewers Ltd,
UK by whom Dr. Boulton was previously employed,
for providing the data presented in Figure 4.
The authors have declared no conflict of interest.

Nomenclature

A [m2] Cross-sectional area of a cylindroconical vessel

d [m] Pipe diameter
dc [m] Cell diameter
D [m] Impeller diameter
g [m2/s] Gravitational constant (9.81)
H [m] Height of liquid in a cylindroconical vessel
N [rev/s] Impeller speed
P [W] Power input into the liquid
Po Power number of impeller
Q [m3/s] Flow rate through RJM
QCO2 [m3 CO2/m3 liquid/s] Specific CO2 production rate
Re Reynolds number
tc [s] Circulation time
T [m] Tank diameter
vs [m/s] Superficial gas velocity in a cylindroconical vessel
V [m3] Volume of liquid
α Aspect ratio
Δp [N/m2] Pressure drop
εT [W/kg] Local specific energy dissipation rate
ε–T [W/kg] Mean specific energy dissipation rate
λK [m] Kolmogoroff microscale of turbulence
ν [m2/s] Kinematic viscosity
ρ [kg/m3] Liquid density
θ [s] Mixing time
max Maximum

924 © 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Biotechnol. J. 2011, 6, 911–925
4 References
[1] Nienow, A. W., Scale-up, stirred tank reactors. in: Flickinger,
M. C. (Ed.), Encyclopedia of Industrial Biotechnology, Vol. 7,
Wiley, Hoboken, NJ, 2010, pp. 4328–4341.
[2] Buchs, J., Introduction to advantages and problems of shak-
en cultures. Biochem. Eng. J. 2001, 7, 91–98.
[3] Amanullah, A., Otero, J. M., Mikola, M., Hsu, A. et al., Novel
micro-bioreactor high throughput technology for cell cul-
ture process development: reproducibility and scalability
assessment of fed-batch CHO cultures. Biotechnol. Bioeng.
2010, 106, 57–67.
[4] Nienow, A. W. Reactor engineering in large scale animal cell
culture. Cytotechnology 2006, 50, 9–33.
[5] Nienow, A. W. Scale-up considerations based on studies at
the bench scale in stirred bioreactors. J. Chem. Eng. Jpn.
2009, 42, 789–796.
[6] Nienow, A. W., Hydrodynamics of stirred bioreactors. Appl.
Mech. Rev. 1998, 51, 3–32.
[7] Onyeaka, H., Nienow, A. W., Hewitt, C. J., Further studies re-
lated to the scale-up of high cell density Escherichia coli
fed-batch fermentations: the additional effect of a changing
micro-environment when using aqueous ammonia to con-
trol pH. Biotechnol. Bioeng. 2003, 84, 474–484.
[8] Boulton, C. A., Wilson, C., Peters, A., Wright, J., An investiga-
tion into the distribution of viable yeast mass and tempera-
ture variation in cylindroconical vessels during fermenta-
tion, in: Proceedings of the 30th European Brewing Conven-
tion Congress, Hans Carl Fachverlag, Nurnberg, 2005, pp.
695–702.
[9] Boulton C. A., Quain, D., Brewing Yeast and Fermentation,
Blackwell Science, Oxford, 2001.
[10] Boswell, C. D., Nienow, A. W., Hewitt, C. J., Studies on the ef-
fect of mechanical agitation on the performance of brewing
fermentations; fermentation rate, yeast physiology, and de-
velopment of flavor. J. Am. Soc. Brew. Chem., 2002, 60,
101–106.
[11] Hind A., The application of modelling to the brewing
process in: Proc. 26th Convention Inst. Brew. Asia-Pacific
Section, Singapore, 2000, 104–109.
[12] Hewitt, C. J., Boon, L. A., McFarlane, C. M., Nienow, A. W., The
use of flow cytometry to study the impact of fluid mechani-
cal stress on Escherichia coli W3110 during continuous cul-
tivation in an agitated bioreactor. Biotechnol. Bioeng. 1998,
59, 612–620.
[13] Chamsartra S., Nienow, A.W., Hewitt, C. J.,The impact of flu-
id mechanical stress on Corynebacterium glutamicum dur-
ing continuous cultivation in an agitated bioreactor. Biotech-
nol. Lett. 2005, 27, 693–700.
[14] Boswell, C. D., Nienow, A. W., Gill, N., Kocharunchitt, S., He-
witt, C. J., The impact of fluid mechanical stress on Saccha-
romyces cerevisiae cells during continuous cultivation in an
agitated, aerated bioreactor; its implications for mixing in
the brewing process and aerobic fermentations. Food Bio-
product. Proc. 2003, 81, 23–32.
[15] Garcia, A. I., Garcia, L. A., Diaz, M., Analysis of internal con-
centration profiles in industrial beer fermentation columns.
Master Brewers Association of the Americas Technical Quar-
terly 1995, 32, 201–207.
[16] Garcia, A. I., Pandiella, S. S., Garcia, L. A., Diaz, M, Mecha-
nism for mixing and homogenization in beer fermentation.
Bioprocess Eng. 1994, 10, 179–184.
© 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.biotechnology-journal.com
Dr. Alvin W. Nienow is Emeritus Pro-
fessor of Biochemical Engineering at
the University of Birmingham, Visiting
Professor in the Centre for Regenera-
tive Medicine at Loughborough Univer-
sity, and a Fellow of the Royal Academy
of Engineering. He has published over
400 papers on mixing, covering the full
spectrum of the process industries. For
bioprocessing, these papers have cov-
ered the physical and biological aspects of bioreactor scale-up/scale-
down for bacteria (including the production of polysaccharides), yeast,
animal cells, and mycelia plus cell lysis for plasmid DNA production.
His awards include The Moulton Medal and Donald Medal from The
Institution of Chemical Engineers, The Jan. E. Purkyne Medal from the
Czech Academy of Sciences, a Lifetime Recognition Award from The
European Federation of Chemical Engineers, and a Dhc from the West
Pomeranian University of Technology. He is an editor of Mixing in the
Process Industries and also consults for companies in the UK, USA, and
Europe.
[17] Boswell, C. D., Varley, J., Boon, L., Hewitt, C. J., Nienow, A. W.,
Studies on the impact of mixing in brewing fermentation:
comparison of methods of effecting enhanced liquid circu-
lation. Food Bioproduct. Proc. 2003, 81, 33–39.
[18] Nienow, A. W., The mixer as a reactor—liquid/solid systems,
in: Harnby, N, Edwards M. F., Nienow, A. W., (Eds), Mixing in
the Process Industries, 2nd Edn, Butterworth Heinemann,
London, 1997, pp. 394–411.
[19] Nienow, A. W., McLeod, G., Hewitt, C. J., Studies supporting
the use of mechanical mixing in large scale beer fermenta-
tions. Biotechnol. Lett. 2010, 32, 623–633.
[20] Boulton, C. A., Price, S. G., Peters, A., Yeast distribution in
cylindro-conical vessels, new insights into fermentation
performance and management, in: Proceedings of the 31st
European Brewing Convention Congress, Hans Carl Fachver-
lag, Nurnberg, 2007, pp. 267–279.
[21] Boulton, C. A., Nordkvist, M., The design of large capacity
conical fermentation vessels and their management, in:
Proceedings of the 33rd European Brewing Convention Con-
gress, Hans Carl Fachverlag, Nurnberg, 2009 (CD ROM).
[22] Nordkvist, M., Vognsen, M., Nienow, A. W., Villadsen, J., Ger-
naey, K.V. Mixing by rotary jet heads: Indications of the ben-
efits of head rotation under turbulent and transitional flow
conditions. Chem. Eng. Res. Des. 2008, 86, 1454–1461.
[23] Nordkvist, M., Grotkjar, T., Hummer, J. S., Villadsen, J., Ap-
plying rotary jet heads for mixing and mass transfer in a
forced recirculation tank reactor system. Chem. Eng. Sci.
2003, 58, 3877–3890.
[24] Gao, Z., Smith, J. M., Muller-Steinhagen, H.Void fraction dis-
tribution in sparged and boiling reactors with modern im-
peller configuration. Chem. Eng. Process. 2001, 40, 489–497.
925