ARTICLE

A New Dynamic Model for Highly Efcient Mass

Transfer in Aerated Bioreactors and Consequences
for k
L
a Identication
Stefan Mu ller,
1
Douglas B. Murray,
2
Rainer Machne
3
1
Johann Radon Institute for Computational and Applied Mathematics, Austrian Academy of
Sciences, Altenberger Strae 69, 4040 Linz, Austria; telephone: 43-1-51581-6553;
fax: 43-1-51581-6550; e-mail: stefan.mueller@ricam.oeaw.ac.at
2
Institute for Advanced Biosciences, Keio University, Tsuruoka, Japan
3
Institute for Theoretical Chemistry, University of Vienna, Vienna, Austria
ABSTRACT: Gasliquid mass transfer is often rate-limiting
in laboratory and industrial cultures of aerobic or autotro-
phic organisms. The volumetric mass transfer coefcient k
L
a
is a crucial characteristic for comparing, optimizing, and
upscaling mass transfer efciency of bioreactors. Reliable
dynamic models and resulting methods for parameter iden-
tication are needed for quantitative modeling of microbial
growth dynamics. We describe a laboratory-scale stirred
tank reactor (STR) with a highly efcient aeration system
(k
L
a ~570 h
1
). The reactor can sustain yeast culture with
high cell density and high oxygen uptake rate, leading to a
signicant drop in gas concentration from inow to outow
(by 21%). Standard models fail to predict the observed mass
transfer dynamics and to identify k
L
a correctly. In order to
capture the concentration gradient in the gas phase, we
rene a standard ordinary differential equation (ODE)
model and obtain a system of partial integro-differential
equations (PIDE), for which we derive an approximate
analytical solution. Specic reactor congurations, in par-
ticular a relatively short bubble residence time, allow a quasi
steady-state approximation of the PIDE system by a simpler
ODE model which still accounts for the concentration
gradient. Moreover, we perform an appropriate scaling of
all variables and parameters. In particular, we introduce the
dimensionless overall efciency k, which is more informa-
tive than k
L
a since it combines the effects of gas inow,
exchange, and solution. Current standard models of mass
transfer in laboratory-scale aerated STRs neglect the gradient
in the gas concentration, which arises from highly efcient
bubbling systems and high cellular exchange rates. The
resulting error in the identication of k (and hence k
L
a)
increases dramatically with increasing mass transfer efcien-
cy. Notably, the error differs between cell-free and culture-
based methods of parameter identication, potentially con-
founding the determination of the biological enhance-
ment of mass transfer. Our new model provides an
improved theoretical framework that can be readily applied
to aerated bioreactors in research and biotechnology.
Biotechnol. Bioeng. 2012;109: 29973006.
2012 Wiley Periodicals, Inc.
KEYWORDS: gasliquid mass transfer; gas gradient; partial
integro-differential equations; volumetric mass transfer
coefcient; dimensionless reactor characteristics; parameter
identication
Introduction
Stirred tank reactors (STR) are well-established laboratory
and industrial tools which allow quantitative modeling of
microbial growth dynamics (Monod, 1949; Agrawal et al.,
1982; Heinzle et al., 1983; Strassle et al., 1988) and in vivo
enzyme kinetics (Wu et al., 2003). Gasliquid mass transfer
is often rate-limiting for cultures of aerobic or autotrophic
organisms that require gaseous substrates, such as molecular
oxygen or carbon dioxide. According to the two lm model
(Whitman, 1923), the net mass transfer is determined by
the deviation from equilibrium as given by Henrys law.
More specically, one has V
l
q = k
L
A(k
H
c
g
c
l
), where q is
the mass transfer rate (in concentration/time) and c
g
and c
l
are the concentrations in the gas and liquid phases.
The (dimensionless) Henry constant k
H
depends on the
physicochemical properties of the system, whereas the
specic mass transfer coefcient k
L
additionally depends
on its hydrodynamic properties (Alves et al., 2004). Usually
Correspondence to: S. Mu ller
Contract grant sponsor: Vienna Science and Technology Fund WWTF
Contract grant number: MA07-30
Contract grant sponsor: FP7 Marie Curie Actions
Contract grant number: 238017
Contract grant sponsor: Deutsche Forschungsgemeinschaft DFG
Contract grant number: SFB 618
Contract grant sponsor: Japan Science and Technology Agency JST
Contract grant sponsor: Yamagata Prefecture
Contract grant sponsor: Tsuruoka City
Additional supporting information may be found in the online version of this article.
Received 2 April 2012; Revision received 12 June 2012; Accepted 20 June 2012
Accepted manuscript online 5 July 2012;
Article rst published online 25 July 2012 in Wiley Online Library
(http://onlinelibrary.wiley.com/doi/10.1002/bit.24594/abstract)
DOI 10.1002/bit.24594
2012 Wiley Periodicals, Inc. Biotechnology and Bioengineering, Vol. 109, No. 12, December, 2012 2997
k
L
is lumped with the gasliquid interface area A and the
liquid volume V
l
into the volumetric mass transfer
coefcient k
L
a = k
L
(A=V
l
).
Efcient mass transfer can be achieved by an aeration
system, where gas enters the liquid through a sparger at the
bottom of the reactor, and the resulting bubbles exchange
gaseous substances with the liquid on their way up to the
headspace. For such aerated STRs, the volumetric mass
transfer coefcient depends on the bubbling/stirring
geometry, the aeration rate, and the agitation speed which
makes it difcult to capture by theoretical models (Alves
et al., 2004). The parameter k
L
a is crucial for (i) ensuring
equal conditions between reactors; (ii) optimizing labora-
tory-scale reactors; and (iii) upscaling to industrial reactors
(Garcia-Ochoa and Gomez, 2009; Marques et al., 2010). Its
value is identied using standard experimental protocols
(Garcia-Ochoa and Gomez, 2009). However, several
problems of parameter identication have been noted.
For example, it was observed that k
L
a can increase with
cell density, a phenomenon that is usually captured by
a biological enhancement factor (Garcia-Ochoa and
Gomez, 2009; Ju and Sundarajan, 1992). A technical
problem arises when the time scale of mass transfer is
smaller than the response time of the dissolved gas probe
(Merchuk et al., 1990).
The aeration process in (small) laboratory-scale bioreactors
is commonly modeled by assuming a constant gas concen-
tration in the bubbles, namely the concentration at inow,
leading to an ordinary differential equation (ODE) for the
concentration in the liquid phase (Garcia-Ochoa and Gomez,
2009), or by assuming an effective time-dependent bubble
concentration, leading to an additional ODE (Wu et al.,
2003). The dynamics of the bubble concentration becomes
especially important, when mass transfer is efcient and
metabolic activity is high enough to considerably change the
gas concentration between inow and outow into head-
space. For tall industrial-scale reactors (bubble column) or
mixed aeration/agitation systems (airlift), concentration
gradients in both gas and liquid phases have been accounted
for by partial differential equation (PDE) models (Dhaouadi
et al., 1997, 2001, 2008; Gourich et al., 2008; Han and
Al-Dahhan, 2007). Recently, a gradient in gas phase has been
considered in an ODE model of carbon dioxide uxes in a
laboratory-scale photobioreactor (Nedbal et al., 2010).
In this work, we model a laboratory-scale continuous STR
with a highly efcient bubbling system. The experimental
setup allows high cell density and high oxygen uptake rate
(OUR) causing a signicant drop in gas concentration from
inow to outow. We propose a new partial integro-
differential equation (PIDE) model to account for the
concentration gradient in the bubbling system. The standard
models from literature turn out to be approximations of the
new model for certain reactor congurations.
Materials and Methods
Saccharomyces cerevisiae (strain IFO 2033) were grown
continuously on dened glucose medium in an aerated STR
with total volume V=2.667 L, liquid volume V
l
=0.650 L,
and dilution rate D=0.085 h
1
. Growth rate and cell density
were constant. Aeration with a ow rate F
g
=0.150 L min
1
and stirring with 750 rpm caused a total bubble volume
V
b
~0.090 L. Temperature was controlled at 308C and
pH at 3.4. Partial pressure of offgas oxygen and relative
concentration of dissolved oxygen were measured online.
Details are given in Supplement S1.
Results
When budding yeast is grown continuously at high density,
the cells enter a state of synchronized respiratory oscillation
(Murray et al., 2007). In Figure 1, we show oxygen
measurements in the liquid and offgas phases from a
periodically oscillating culture (left) and the results of a
cell-free gassing out experiment (right). From the two
Figure 1. Data for k
L
a identication. Left: Respiratory oscillations in continuous culture of budding yeast. Dimensionless oxygen concentrations in liquid phase and
headspace, ^ c
l
= c
l
=(k
H
c
g;in
) (thick line) and ^ c
h
= c
h
=c
g;in
(thin line), over three oscillation periods. Deviations of average concentrations (dotted lines) from saturation and ingas
values (=1), as indicated by the arrows, are used to identify k
L
a for different dynamic models, cf. Steady-State/Continuous Culture Section. Right: Cell-free gassing out experiment.
Dimensionless oxygen concentrations in the liquid phase. The measured concentration ^ c
l;ms
(circles) lags behind the actual concentration ^ c
l
(line) due to probe respones time
t
p
=40 s. To identify k
L
a, different dynamic models are t to the time-series data by using an optimization approach and by accounting for probe response time, cf. Dynamics/
Gassing Out Section.
2998 Biotechnology and Bioengineering, Vol. 109, No. 12, December, 2012
datasets, we obtain drastically different k
L
a values, if we
use the textbook model for gasliquid mass transfer,
dc
l
=dt = k
L
a(c
+
l
c
l
) OUR (Finn, 1954; Garcia-Ochoa
and Gomez, 2009). More specically, k
L
a values are much
higher for the culture-based method (334 h
1
), where OUR
has been estimated from offgas measurements, than for the
gassing out experiment (94 h
1
), where the model has been
t to the measured time-series by using an optimization
approach and by accounting for probe response time. Such
an increase may be explained by a biological enhancement of
mass transfer (Garcia-Ochoa and Gomez, 2009; Ju and
Sundarajan, 1992). However, we also nd that cellular
uptake causes a signicant drop in oxygen concentrations
between ingas (atmospheric air) and offgas. We conclude
that there is a concentration gradient in the bubbling system,
and that the standard model is over-simplifying since
it assumes a constant gas concentration. A model that
accounts for an effective concentration in the bubbles has
been proposed to study in vivo enzyme kinetics from pulse
experiments (Wu et al., 2003), cf. Equations (1). If we use
this model, we obtain much higher k
L
a values both for
the culture-based method (1,156 h
1
) and the cell-free
experiment (738 h
1
). The existence of the concentration
gradient in the bubbling system and the drastically different
k
L
a values obtained from different models and identication
methods raise the question whether the literature models are
applicable to the given reactor conguration. Hence, we set
out to explicitly account for the concentration gradient.
Dynamic Models
We present a series of dynamic models for gasliquid mass
transfer in laboratory-scale STRs and discuss their validity
for given reactor congurations. In particular, we introduce
a new model which contains the literature models as special
cases. We start with a straight-forward model capturing
mass ow through the gas phases (bubbles and headspace),
the actual mass transfer between gas and liquid phases, and
mass exchange between cells and liquid phase:
dc
l
dt
= k
L;b
A
b
V
l
(k
H
c
b
c
l
) k
L;h
A
h
V
l
(k
H
c
h
c
l
) q
cell
(1a)
dc
b
dt
=
F
g
V
b
(c
g; in
c
b
) k
L;b
A
b
V
b
(k
H
c
b
c
l
) (1b)
dc
h
dt
=
F
g
V
h
(c
b
c
h
) k
L;h
A
h
V
h
(k
H
c
h
c
l
) (1c)
We note that q
cell
>0 means excretion by the cells,
whereas q
cell
<0 means uptake; in the case of oxygen uptake
by the cells, one has q
cell
=OUR. An additional term
(F
l
=V
l
)(c
l;in
c
l
) in the rst ODE, representing a possible
mass ow through the liquid phase due to dilution, can
be omitted, if the dilution rate D = F
l
=V
l
is negligible
compared to k
L
a. In this case, mass conservation involves gas
inow and outow and cellular mass exchange:
d
dt
(V
l
c
l
V
b
c
b
V
h
c
h
) = F
g
(c
g;in
c
h
) V
l
q
cell
(2)
In steady-state, mass conservation can be used to determine
the cellular exchange rate from the headspace concentration:
q
cell
= (F
g
=V
l
)(c
g;in
c
h
). Since Equations (1) involve one
effective concentration in the bubbles, we will refer to this
model as M
eff
. Two simplications are commonly made: If
headspaceliquid mass transfer can be ignored, the bubble
liquid dynamics becomes independent from the headspace
dynamics. In this form, the model has been used to study in
vivo kinetics of Saccharomyces cerevisiae (Wu et al., 2003). If
additionally the gas concentrations are approximately
constant, namely the concentration at inow, one obtains
the textbook model for the liquid dynamics (Finn, 1954;
Garcia-Ochoa and Gomez, 2009):
dc
l
dt
= k
L
A
b
V
l
(k
H
c
g;in
c
l
) q
cell
= k
L
a(c
+
l
c
l
) q
cell
(3)
with k
L
a = k
L;b
(A
b
=V
l
) and c
+
l
= k
H
c
g;in
. We will refer to
this model as M
con
.
Since the concentration in the bubbles can vary
signicantly from inow to outow (into headspace), we
will present a new model in order to capture this gradient.
Before that, we simplify M
eff
in order to facilitate the
subsequent model renement and analysis. We non-
dimensionalize all variables and introduce dimensionless
parameters which characterize a given reactor conguration.
The scaled version of M
eff
amounts to
d^c
l
d
^
t
= k(^c
b
^c
l
) ak(^c
h
^c
l
) ^ q
cell
(4a)
d^c
b
d
^
t
=
1
t
(1 ^c
b
k(^c
b
^c
l
)) (4b)
d^c
h
d
^
t
=
1
nt
(^c
b
^c
h
ak(^c
h
^c
l
)) (4c)
with the following non-dimensional variables and parameters:
^
t =
t
t
s
(5a)
^c
b
=
c
b
c
g;in
(5b)
^c
h
=
c
h
c
g;in
(5c)
^c
l
=
c
l
k
H
c
g;in
(5d)
Muller et al.: Highly Efcient Mass Transfer in Aerated Bioreactors 2999
Biotechnology and Bioengineering
^ q
cell
=
q
cell
t
s
k
H
c
g;in
=
V
l
q
cell
F
g
c
g;in
(5e)
a =
k
L;h
k
L;b
A
h
A
b
(6a)
n =
V
h
V
b
(6b)
t =
V
b
=F
g
t
s
=
V
b
=V
l
k
H
(6c)
k = k
L
a t
s
= k
L
a k
H
V
l
F
g
(6d)
Thereby, we have performed the following steps: in order
to scale time, we have dened a minimal saturation time,
t
s
=
V
l
k
H
c
g;in
F
g
c
g;in
= k
H
V
l
F
g
; (7)
which is the (hypothetical) time needed to saturate the
liquid in the (ideal) case of instantaneous mass transfer.
Further, we have scaled all gas concentrations by the inow
concentration and the liquid concentration by the satura-
tion concentration. This makes all concentrations easily
comparable. (For example, the steady-state of Equations (4)
for ^q
cell
= 0 is given by ^c
l
= ^c
b
= ^c
h
= 1.) Finally, we have
dened the ratios between headspace and bubble para-
meters, a and n, as well as the scaled bubble residence time t
and the overall efciency k, which comprehends the effects
of gas ow, gas exchange, and gas solution, and which plays
the role of the volumetric mass transfer coefcient k
L
a in
the scaled models. In total, we have reduced the number of
parameters from 9 to 4.
For completeness, we state the scaled version of M
con
:
d^c
l
d
^
t
= k(1 ^c
l
) ^ q
cell
(8)
Now, we present a new model which captures the gradient
in the bubble concentration. We start from M
eff
, and group
the bubbles into (virtual) compartments according to their
current residence time in the liquid (or equivalently to the
path they have already traveled). This yields an ODE model
with one bubble concentration for each compartment. In the
limit of innitely many compartments, we obtain a PIDE
model, where the bubble concentration ^c
b
(
^
t; x) depends on
the additional independent variable x[0; 1[ which repre-
sents the (relative) current bubble residence time (x =0 at
inow and x =1 at outow into headspace):
d^c
l
d
^
t
= k
_
1
0
^c
b
dx ^c
l
_
_
_
_
ak(^c
h
^c
l
) ^ q
cell
(9a)
@^c
b
@
^
t
=
1
t

@^c
b
@x
k(^c
b
^c
l
)
_ _
(9b)
^c
b
(
^
t; 0) = 1 (9c)
d^c
h
d
^
t
=
1
nt
(^c
b
(t; 1) ^c
h
ak(^c
h
^c
l
)) (9d)
The details of the derivation can be found in Supplement
S2. We will refer to Equations (9) as M
PIDE
grad
. If the bubble
residence time is much shorter than the time scale for mass
transfer, that is, V
b
=F
g
1=k
L
a or t 1=k, then we can
approximate the PIDE model by another ODE model. In
this case, the bubble concentration is in quasi steady-state,
that is, @^c
b
=@
^
t = 0, and we can express the bubble by the
liquid concentration. We obtain:
d^c
l
d
^
t
= (1 e
k
)(1 ^c
l
) ak(^c
h
^c
l
) ^ q
cell
(10a)
d^c
h
d
^
t
=
1
nt
(e
k
(1 ^c
l
) (1 ak)(^c
h
^c
l
)) (10b)
The details of the derivation can be found in Supplement
S2a. We will refer to Equations (10) as M
ODE
grad
. This model has
been derived in a different way by Nedbal et al. (2010). We
note that Equation (10a) reduces to M
con
, if headspace
liquid mass transfer is negligible, that is, a 1, and mass
transfer efciency is low, that is, k 1 and hence
1 e
k
~ k.
In Figure 2, we summarize our presentation of the
different dynamic models. More specically, we draw a
Figure 2. Hierarchy of dynamic models for mass transfer between gas and
liquid phases. The PIDE model, which captures the gradient in the bubble concentra-
tion, can be simplied in two ways: (left) by assuming an effective bubble concentra-
tion, or (right) by assuming a short bubble residence time. The resulting ODE models
can be further simplied, nally arriving at the model for constant bubble concentra-
tion and negligible mass transfer between headspace and liquid.
3000 Biotechnology and Bioengineering, Vol. 109, No. 12, December, 2012
directed graph with the models as nodes and the simplifying
assumptions as edges. Crucial parameters are: k, the overall
mass transfer efciency, t, the scaled bubble residence time,
and a, the ratio of headspace and bubble k
L
a.
Steady-State/Continuous Culture
The equilibria of the new PIDE model and the ODE models
from literature are studied in Supplements S3S6. Here, we
summarize the results for M
PIDE
grad
. We note that our analysis
is also relevant for periodic oscillations: in this case, the
integrals of the left-hand sides of the ODEs over one period
are zero,
_
T
0
(d^c=d
^
t)d
^
t = ^c(
^
t)[
T
0
= ^c(T) ^c(0) = 0, and
since the right-hand sides of the ODEs are linear in the
variables, the dening equations for the equilibrium values
also hold for the average values. Further, we will address the
inverse problem of identifying parameters from steady-state
measurements of the liquid and the headspace concentra-
tion. In particular, we are interested in the overall efciency
k (which contains k
L
a). We will compare the results for the
new PIDE model and the ODE models from literature.
At steady-state (or for periodic oscillations), M
PIDE
grad
(9)
yields the following equilibrium (or average) values of the
liquid, bubble, and headspace concentrations:
^c
l
= 1 ^ q
cell
1 ak
1 e
k
ak
(11a)
^c
b
(x) = (1 ^c
l
)e
kx
^c
l
(11b)
^c
h
= 1 ^ q
cell
(11c)
For details, see Supplement S3. Since the parameters n
and t do not appear in Equations (11), the equilibria are
determined solely by a and k (and a given ^q
cell
). In Figure 3,
the concentration proles are shown for both uptake and
excretion by the cells. Note that ^c
h
is always located between
^c
l
and ^c
b
(1). If headspaceliquid mass transfer can be
ignored, that is, a 1, we nd ^c
h
= ^c
b
(1).
Clearly, the gradient in the gas concentration equals the
cellular exchange rate, c
h
1 = ^q
cell
, cf. Equations (11). In
turn, ^q
cell
is determined by the biology of the culture and the
reactor conguration: In the case of uptake by the cells, the
rate increases with liquid concentration and eventually
saturates; hence it can be modeled by an increasing function
[^q
cell
[ = f (^c
l
). On the other hand, the new PIDE model for
gasliquid mass transfer yields [^q
cell
[ = (1 e
k
)(1 ^c
l
), cf.
Equation (11a) assuming a 1. In Figure 4 (left), we
intersect the graphs of the two functions and obtain
the cellular uptake rate at steady-state. We note that the
resulting value of [^q
cell
[ increases with k. In other words,
a bioreactor with a higher efciency of gasliquid mass
transfer allows for a higher cellular uptake rate.
Parameter Identication From Steady-State Data
In order to identify parameters from steady-state measure-
ments of ^c
l
and ^c
h
, we eliminate ^q
cell
from Equations (11ac).
We may write the result as
1 e
k
ak
1 ak
=
1 ^c
h
1 ^c
l
: (12)
This means that, in general, the parameters a and k
cannot be identied independently. However, if headspace
liquid mass transfer can be ignored, that is, a 1, then
Equation (12) can be solved for k:
k = k
PIDE
grad
= ln 1
1 ^c
h
1 ^c
l
_ _
(13)
Now, we can compare the parameter value k
PIDE
grad
identied from the new PIDE model, that is,
Equation (13), with k
con
and k
eff
identied from the
respective ODE models, that is, Equations (68) and (75) in
Figure 3. Equilibriumvalues of the scaled concentrations in liquid, bubbles, and headspace: ^ c
l
(solid line), ^ c
b
(x) (dashed line), and ^ c
h
(dotted line). The bubble concentration
varies between inow (x =0) and outow into headspace (x =1). The two plots differ in the sign of the mass transfer from the cells to the liquid. Left: ^ q
cell
< 0, uptake.
Right: ^ q
cell
> 0, excretion. Parameters: a=1, k =1, and ^ q
cell
= ((1=2).
Muller et al.: Highly Efcient Mass Transfer in Aerated Bioreactors 3001
Biotechnology and Bioengineering
Supplements S4 and S5:
k
con
=
1 ^c
h
1 ^c
l
(14a)
k
eff
=
1
1^c
h
1^c
l
_ _
1
1
(14b)
In Figure 4 (right), we plot k
PIDE
grad
, k
con
, and k
eff
over the
steady-state ratio (1 ^c
h
)=(1 ^c
l
). With increasing ratio,
the textbook model M
con
increasingly underestimates k,
whereas M
eff
overestimates it. For (1 ^c
h
)=(1 ^c
l
) 1,
the new model M
PIDE
grad
shows that k becomes essentially
unidentiable.
In the laboratory-scale STR under consideration, we
observe periodic oscillations in continuous culture of
budding yeast (Murray et al., 2007), cf. Figure 1 (left).
We determine the average values of the measured ^c
l
and ^c
h
as 0.704 and 0.790 and obtain (1 ^c
h
)=(1 ^c
l
) = 0:71.
Using Equations (13) and (14) and thereby assuming a 1,
we get k
PIDE
grad
= 1:24, k
con
=0.71, and k
eff
=2.46. In other
words, M
con
underestimates k (and hence k
L
a) by 43%,
whereas M
eff
overestimates it by 98%.
Dynamics/Gassing Out
We analyze the dynamics of mass transfer in the cell-free
case, that is, for ^q
cell
= 0. In particular, we are interested in
the gassing out experiment for the identication of the
overall efciency k (which contains k
L
a). The solutions of
the new PIDE model and the ODE models from literature
are derived in Supplements S3S6. Here, we summarize
these results. It turns out that the literature models are
approximations of the new model for certain reactor
congurations, characterized by k, and the scaled bubble
residence time t. Further, we will address the inverse
problem of identifying parameters from time-series mea-
surements of the liquid concentration. For each ODE model,
we identify k from data which have been measured in the
laboratory-scale STR under consideration. As we will see, the
identied parameter values differ considerably.
First, we study M
PIDE
grad
(9) in the cell-free case, that is,
^q
cell
= 0, and for negligible headspaceliquid mass transfer,
that is, a 1. As a consequence, the bubbleliquid
dynamics (9ac), becomes independent of the headspace
dynamics (9d). The solution of the PIDE system (9ac) with
the initial conditions ^c
l
(0) = 0 and ^c
b
(0; x) = 0 involves the
following steps: (i) Laplace transformation of the system:
transformation of differential equations in the time variable
t into algebraic equations in the new variable s. (ii) Solution
of an initial value problem for the transformed bubble
concentration in the spatial variable x; thereby, expression of
the bubble by the liquid concentration. (iii) Inverse Laplace
transformation of the liquid concentration: integration in
the complex plane using the Cauchy Residue Theorem;
solution as an innite series of residues. (iv) Calculation
of the rst three residues, that is, approximation of the
solution. The details of the comprehensive derivation can be
found in Supplement S3. We obtain the following analytical
approximation of the liquid concentration:
^c
l
(
^
t) ~
1
1 e
k

1
k(1=t 1 k=2)
e

k
t
^t
ge
l^t
;
^
t < t
1 g(1 e
ltk
)e
l^t
;
^
t > t
_

_
(15)
For times smaller than the bubble residence time,
^
t < t,
the solution involves two terms of exponential decay with
rates k=t and l, whereas for
^
t > t, the saturation dynamics is
determined by l only. The initial phase corresponds to the
establishment of the concentration gradient in the bubbling
Figure 4. Left: Cellular uptake rate and overall efciency: [ ^ q
cell
[ is determined by a biologically motivated increasing function [ ^ q
cell
[ = f (^ c
l
) (thin line) and the function
[ ^ q
cell
[ = (1 e
k
)(1 ^ c
l
) from the new PIDE model for gasliquid mass transfer (thick line). The resulting value of the cellular uptake rate (circle) increases with k.
Right: Identication of the overall efciency k from steady-state measurements of the scaled headspace and liquid concentrations, ^ c
h
and ^ c
l
. Parameter values identied from the
new model M
PIDE
grad
(solid line) and the literature models M
con
(dotted line) and M
eff
(dashed line) assuming a=0. Values for the STR under consideration (circle):
(1 ^ c
h
)=(1 ^ c
l
) = 0:71, k
PIDE
grad
= 1:24.
3002 Biotechnology and Bioengineering, Vol. 109, No. 12, December, 2012
system: The rst full bubbles enter and release gas to the
liquid, while empty bubbles are still present and take up
gas from the liquid. It remains to explain the (scaled)
saturation rate l and the coefcient g, which depend on the
parameters k and t: in general, the function
p(s) = s(st k)(st k kt) k
2
(1 e
(stk)
) (16)
has two real zeroes, s
1
= (k=t) and s
2
<0, the latter of
which can be determined only numerically. Now, l =[s
2
[
and g = k(lt k)=(lp
/
(l)). We note that l(0; 1) for all
k and t, cf. Figure 6 (top-left). As a consequence, even for
highly efcient mass transfer, that is, k , the saturation
rate l is nite, since in this case mass ow is the limiting
process.
Now, we can compare the solution for the liquid
concentration predicted by the new model M
PIDE
grad
with
the solutions predicted by M
con
, M
eff
, and M
ODE
grad
. That is, we
compare Equation (15) above with Equations (71), (77), and
(82) in Supplements S4S6:
M
con
: ^c
l
(
^
t) = 1 e
k^t
(17a)
M
eff
: ^c
l
(
^
t) = 1
l

e
l

^t

e
l

^t
(17b)
M
ODE
grad
: ^c
l
(
^
t) = 1 e
l
0
^t
(17c)
where
l

=
1 k kt
2t

1 k kt
2t
_ _
2

k
t

(18a)
l
0
= l[
t=0
= 1 e
k
(18b)
From Figure 5, we see that the solutions differ
considerably for the parameter values we have chosen.
(The laboratory-scale STR under consideration is charac-
terized by a 1, k ~1, and t ~5.) Obviously, M
con
is not a
good approximation for small ^c
l
, since this model assumes a
constant bubble concentration throughout the dynamics,
that is, ^c
b
= 1, which in particular implies d^c
l
=d
^
t(0) > 0.
Hence, this model overestimates the increase in ^c
l
in the
initial phase. On the other hand, M
eff
predicts d^c
l
=d
^
t(0) = 0
correctly such that it is a good approximation in the initial
phase. However, this model uses a single effective bubble con-
centration, that is, ^c
b
(t), which does not cover the gradient
from inow to outow. In particular, it underestimates the
increase in ^c
l
in the later phase since it overestimates the
outow into headspace. For the parameters chosen, M
ODE
grad
is
only a slightly better approximation than the structurally
equivalent M
con
(since t 1=k is not fullled). This model
considers the gradient in the bubble concentration ^c
b
(t; x),
however, it assumes that the gradient is established
instantaneously, which causes an initial overestimation.
Finally, we study the validity of the ODE models as
approximations of the PIDE model for wide ranges of the
crucial parameters k and t. To this end, we contrast the
exponential term e
l^t
, which determines the long-term
dynamics of M
PIDE
grad
with the corresponding terms e
k ^t
, e
l ^t
,
and e
l
0
^t
in the solutions of M
con
, M
eff
, and M
ODE
grad
. This boils
down to a comparison of the saturation rate l with k, l

,
and l
0
. In Figure 6, we plot the saturation rate l as well as
the resulting relative errors (k l)=l, (l

l)=l, and
(l
0
l)=l over the (k, t)-plane. The most simple ODE
model M
con
is a valid approximation of the PIDE model only
if k _0.1, that is, in a small region of the (k, t)-plane. On the
other hand, M
eff
is a valid approximation only if k _0.1 or
k _10. Hence, M
eff
is valid in a larger region of the (k, t)-
plane than M
con
. Finally, M
ODE
grad
is a valid approximation
only if kt _0.1. This shows that t 1=k is not only a
sufcient, but also a necessary condition for the quasi
steady-state approximation of the PIDE model.
Parameter Identication from Time-Series Data
After predicting the liquid concentration over time for given
parameters k and t (and a 1), we solve the inverse
problem of identifying parameters from time-series mea-
surements of the liquid concentration. In particular, we are
interested in the overall efciency k (which contains k
L
a).
For the laboratory-scale STR under consideration, k is not
the only unknown parameter. For example, also t (which
contains the gas holdup V
b
) and the probe response time t
p
have to be identied. For a consistency check, we determine
the gas holdup and the probe response time from indepen-
dent experiments and nd t ~4.7 (from V
b
~90 mL) and
t
p
~40 s. We note that probe response is typically modeled
as a rst order process (Merchuk et al., 1990), where the
measured liquid concentration ^c
l;ms
depends on the actual
value ^c
l
via an additional ODE:
d^c
l;ms
dt
=
1
t
p
(^c
l
(t) ^c
l;ms
(t)) (19)
Figure 5. The liquid concentration as a function of time in the cell-free case
(gassing out experiment). Predictions by the new model M
PIDE
grad
(thick line), and the
literature models M
con
(dotted line), M
eff
(dashed line), and M
ODE
grad
(thin line).
Parameter values: a=0, k =1, t =5.
Muller et al.: Highly Efcient Mass Transfer in Aerated Bioreactors 3003
Biotechnology and Bioengineering
In the standard method for parameter identication
from gassing out data, linear regression is applied to the
transformed data ln(1 ^c
l
(t)). Thereby, one makes the
following assumptions: (i) the actual data ^c
l
can be obtained
from the measured data ^c
l;ms
by means of numerical
differentiation, cf. Equation (19); (ii) the transformed data
follow a linear model, that is, ln(1 ^c
l
(t)) = d k t; and
(iii) for a given dynamic model, the unknown parameter k
can be calculated from the (scaled) rate k t
s
. For the
laboratory-scale STR under consideration, all three assump-
tions are problematic: (i) since probe response time and
bubble residence time are of comparable size, the
establishment of the concentration gradient can be observed
only with a large uncertainty arising from numerical
differentiation of noisy data; (ii) the transformed data do
not exhibit one particular rate k, rather there is a fast increase
from zero to a maximum rate within one bubble residence
time followed by a slower decrease (to the reciprocal of the
headspace residence time); and (iii) not all parameters
which determine the rate constant in an individual model
are known, for example, t is needed to solve l(k; t) = k t
s
in
the new PIDE model. For comparison with other methods,
we take the maximum rate k =0.04 s
1
and use t =4.7.
By equating k t
s
with l in the PIDE model and k, l , and l
0
in the ODE models, we identify the values for k listed
in Table I (Gassing out/Linear regression). We note that
l

(k; t) = k t
s
cannot be solved for k; hence, there is no
result for M
eff
.
Figure 6. Saturation rate l as function of parameters k and t (top-left); relative errors of k, l , and l
0
with respect to l (top-right and bottom); errors cut off at 10% and view
points chosen for best visibility. In the cell-free case with negligible feedback from the headspace, that is, for ^ q
cell
= 0 and a 1, the long-term dynamics of mass transfer is
determined by l in M
PIDE
grad
, k in M
con
, l in M
eff
, and l
0
in M
ODE
grad
.
Table I. Overall efciency k identied for different dynamic models and from different data/with different methods.
Gassing out Continuous culture
Linear regression Optimization a=0 a=0.025
M
PIDE
grad
(0.51) 1.08 1.24 1.21
M
con
(0.31) 0.20 0.71 0.71
M
eff
() 1.57 2.46 2.27
M
ODE
grad
(0.38) 0.24 1.24 1.21
The new PIDE model yields the most consistent identication, whereas the ODE models from literature considerably under- or overestimate k. The linear
regression approach for gassing out data is problematic, cf. main text following Equation (19), and its results are put into brackets. When data from
continuous culture are used, k is identied for negligible headspaceliquid mass transfer a=0 and the realistic value a=0.025.
3004 Biotechnology and Bioengineering, Vol. 109, No. 12, December, 2012
Alternatively, we take an optimization approach to
parameter identication: We vary k (and the other
unknowns a, n, t, and t
p
) and minimize the least-squares
distance between model predictions and data. This method
is preferable, since numerical differentiation is avoided, each
dynamic model is directly t to the data, and all unknowns
are accounted for. On the other hand, it is computationally
more costly, since it involves the numerical solution of the
respective model and the evaluation of the data mismatch in
each optimization step. In Figure 7, we plot the optimal
solutions predicted by the new model M
PIDE
grad
(9) and the
literature models M
con
(8), M
eff
(4), and M
ODE
grad
(10). Clearly,
the new PIDE model gives the best t. In Table I (Gassing
out/Optimization), the identied values for k are listed. We
have already mentioned that M
con
overestimates the initial
increase in ^c
l
; this effect is compensated by an underesti-
mation of k. The same argument holds for M
ODE
grad
. On the
other hand, M
eff
underestimates the long-term increase in ^c
l
which is compensated by an overestimation of k.
Finally, we compare the parameter values identied from
the cell-free gassing out experiment and the (periodically
oscillating) continuous culture, cf. Table I. For the culture-
based method, we use a=0 as in Steady-State/Continuous
Culture Section as well as a =0.025, which has been
consistently identied from the gassing out data using the
optimization approach. Clearly, the new PIDE model yields
the most consistent identication of k, and we conclude
that k =1.21 in continuous culture. It remains to compute
k
L
a = k=t
s
using the minimal saturation time t
s
=7.65 s, cf.
Equation (7). The k
L
a values identied for the different
dynamic models and from the two datasets are summarized
in Table II. The new PIDE model yields k
L
a =569 h
1
in
continuous culture.
Biological Enhancement of Mass Transfer
The concentration gradient in the bubbling system is
neglected in standard dynamic models of gasliquid mass
transfer, which can result in a signicant under- or
overestimation of the volumetric mass transfer coefcient
k
L
a, depending on whether the model assumes a constant or
effective bubble concentration. Notably, the underestima-
tion by the textbook model M
con
(and the overestimation by
M
eff
) also depend on the method of parameter identica-
tion. The method based on measurements of cellular
exchange rates in continuous culture typically yields a higher
k
L
a than the gassing-out method performed in cell-free
medium thus confounding the determination of the
biological enhancement of mass transfer (Garcia-Ochoa
and Gomez, 2009; Ju and Sundarajan, 1992). In order to
estimate the biological enhancement factor E, we use the k
L
a
values identied from continuous culture and the gassing
out experiment:
E =
k
L
a
continuous culture
k
L
a
gassing out
(20)
The resulting values are given in Table II. Since standard
models lead to errors in the identication of k
L
a, these errors
propagate to the estimation of E, which in these cases must
be called an apparent biological enhancement factor. For
the culture under consideration, the textbook model M
con
yields an apparent enhancement factor of E =3.55, whereas
the new dynamic model M
PIDE
grad
yields a realistic value of
E =1.12.
Conclusions
This study provides a framework for rened modeling of
efcient aeration designs. Its main impact is the correct
identication of the volumetric mass transfer coefcent k
L
a,
which plays a crucial role in process monitoring and
development. In the presence of a concentration gradient in
the bubbling system, we advise to use the new PIDE model
Figure 7. Identication of the overall efciency k from time-series measure-
ments of the liquid concentration (gassing out experiment) by minimization of the
mismatch between model predictions and data (circles). Optimal solutions predicted
by the new model M
PIDE
grad
(solid line) and the literature models M
con
(dotted line), M
eff
(dashed line), and M
ODE
grad
(thin line). Parameters values identied from the individual
models listed in Table I (Gassing out/Optimization).
Table II. Volumetric mass transfer coefcient k
L
a (h
1
) identied for
different dynamic models and from different data/with different methods
(gassing out experiment using the optimization approach and continuous
culture using a=0.025, cf. Table I).
Gassing out Continuous culture
M
PIDE
grad
508 569 E =1.12
M
con
94 334 E =3.55
M
eff
739 1,068 E =1.45
M
ODE
grad
113 569 E =5.04
The resulting (apparent) biological enhancement factor E is given in the
rightmost column, cf. Biological Enhancement of Mass Transfer Section.
Dened glucose medium as described in Supplement S1. Experimental
conditions: liquid volume V
l
=0.650 L, aeration rate F
g
=0.150 L min
1
,
stirring with 750 rpm; temperature controlled at 308C, pH at 3.4. Average
dissolved oxygen in culture: 70%.
Muller et al.: Highly Efcient Mass Transfer in Aerated Bioreactors 3005
Biotechnology and Bioengineering
(9) or, depending on the reactor conguration, its ODE
approximation (10). To characterize an aerated STR,
we advise to use the dimensionless overall efciency
k = k
L
a k
H
(V
l
=F
g
) and the scaled bubble residence time
t. If k 1, then the gradient is negligible, and the standard
dynamic models (4) and (8) can be used. If t 1=k, that is,
the bubble residence time is much smaller than the time
scale for mass transfer, then the gradient in the bubble
concentration is established instantaneously, and the full
PIDE system can be approximated by an ODE system for the
liquid and headspace concentrations only.
Recently, a gradient has been considered in a model of
carbon dioxide uxes in a laboratory-scale photobioreactor
(Nedbal et al., 2010). The model has been derived based on
the assumption of a negligible bubble residence time, and we
note that it is equivalent to the ODE approximation of the
full PIDE model. While our considerations have been based
on experimental data for oxygen transfer, we note that
carbon dioxide has a much higher solubility (k
CO
2
H
= 0:74 vs.
k
O
2
H
= 0:029 at T=303 K), which implies that a gradient will
arise already in much less efcient bubbling systems, where
oxygen transfer may still be accounted for with standard
models.
Nomenclature
Variables:
c
b
concentration in the bubbles
c
h
concentration in the headspace
c
l
concentration in the liquid phase
q
cell
cellular mass exchange rate (conc./time)
Constants:
k
H
Henry constant (in conc./conc.)
k
L,b
mass transfer coefcient (between bubbles and liquid)
k
L,h
mass transfer coefcient (between headspace and liquid)
A
b
mass transfer area (between bubbles and liquid)
A
h
mass transfer area (between headspace and liquid)
V
b
bubble volume (gas holdup)
V
h
headspace volume
V
l
liquid volume
F
g
gas ow rate (aeration rate)
F
l
liquid ow rate
c
g,in
gas concentration at inow
c
l,in
liquid concentration at inow
t
s
minimal saturation time (reference time scale)
t
p
probe response time
Dimensionless Parameters:
a ratio of headspace and bubble k
L
a
n ratio of headspace and bubble volume
t scaled bubble residence time
k overall mass transfer efciency
E enhancement factor
We thank James Lu and Clemens Zarzer for fruitful discussions. R.M.
acknowledges nancial support by the Vienna Science and Technolo-
gy Fund WWTF, project MA07-30, the Marie Curie Actions Network
HARVEST, project 238017, and the research program Theoretical
Biology, SFB 618, HU Berlin. D.M. is grateful for funding from the
Japan Science and Technology Agency JST, Yamagata prefecture and
Tsuruoka city.
References
Agrawal P, Lee C, Lim H, Ramkrishna D. 1982. Theoretical investigations of
dynamic behavior of isothermal continuous stirred tank biological
reactors. Chem Eng Sci 37:453462.
Alves S, Maia C, Vasconcelos J. 2004. Gasliquid mass transfer coefcient in
stirred tanks interpreted through bubble contamination kinetics. Chem
Eng Process 43:823830.
Dhaouadi H, Poncin S, Hornut J, Wild G, Oinas P, Korpijarvi J. 1997. Mass
transfer in an external-loop airlift reactor: Experiments and modeling.
Chem Eng Sci 52:39093917.
Dhaouadi H, Poncin S, Midoux N, Wild G. 2001. Gasliquid mass transfer
in an airlift reactorAnalytical solution and experimental conrma-
tion. Chem Eng Process 40:129133.
Dhaouadi H, Poncin S, Hornut J, Midoux N. 2008. Gasliquid mass transfer
in bubble column reactor: Analytical solution and experimental con-
rmation. Chem Eng Process 47:548556.
Finn R. 1954. Agitationaeration in the laboratory and in industry. Bacteriol
Rev 18:254274.
Garcia-Ochoa F, Gomez E. 2009. Bioreactor scale-up and oxygen transfer
rate in microbial processes: An overview. Biotechnol Adv 27:153176.
Gourich B, Vial C, Azher NE, Soulami MB, Ziyad M. 2008. Inuence of
hydrodynamics and probe response on oxygen mass transfer measure-
ments in a high aspect ratio bubble column reactor: Effect of the
coalescence behaviour of the liquid phase. Biochem Eng J 39:114.
Han L, Al-Dahhan MH. 2007. Gasliquid mass transfer in a high pressure
bubble column reactor with different sparger designs. Chem Eng Sci
62:131139.
Heinzle E, Dunn I, Furukawa K, Tanner R. 1983. Modelling of sustained
oscillations observed in continuous cutlure of Saccharomyces cerevisiae.
In: Halme A, editor. Modelling and control of biotechnical processes.
London: Pergamon Press. p 5765.
Ju LK, Sundarajan A. 1992. Model analysis of biological oxygen transfer
enhancement in surface-aerated bioreactors. Biotechnol Bioeng 40:
13431352.
Marques M, Cabral J, Fernandes P. 2010. Bioprocess scale-up: Quest for the
parameters to be used as criterion to move from microreactors to lab-
scale. J Chem Technol Biotechnol 85:11841198.
Merchuk J, Yona S, Siegel M, Zvi A. 1990. On the rst-order approximation
to the response of dissolved oxygen electrodes for dynamic K
L
a
estimation. Biotechnol Bioeng 35:11611163.
Monod J. 1949. The growth of bacterial cultures. Annu Rev Microbiol
3:371394.
Murray D, Beckmann M, Kitano H. 2007. Regulation of yeast oscillatory
dynamics. Proc Natl Acad Sci USA 104:22412246.
Nedbal L, Cerveny J, Keren N, Kaplan A. 2010. Experimental validation of a
nonequilibrium model of CO
2
uxes between gas, liquid medium, and
algae in a at-panel photobioreactor. J Ind Microbiol Biotechnol
37:13191326.
Strassle C, Sonnleitner B, Fiechter A. 1988. A predictive model for the
spontaneous synchronization of Saccharomyces cerevisiae grown in
continuous culture. I. Concept. J Biotechnol 7:299317.
Whitman W. 1923. Preliminary experimental conrmation of the two-lm
theory of gas absorption. Chem Metall Eng 29:146149.
Wu L, Lange H, Van Gulik W, Heijnen J. 2003. Determination of in vivo
oxygen uptake and carbon dioxide evolution rates from off-gas mea-
surements under highly dynamic conditions. Biotechnol Bioeng 81:
448458.
3006 Biotechnology and Bioengineering, Vol. 109, No. 12, December, 2012