*Corresponding author

Tel:+447413541769
Email:myrtozacharof1981@yahoo.com
_____________________________________________________________________________
Modelling and Simulation of Cell growth Dynamics, Substrate
Consumption and Lactic acid production Kinetics of Lactococcus lactis
Myrto-Panagiota Zacharof
1*
and Robert W. Lovitt
2

1*
College of Engineering, Multidisciplinary Nanotechnology Centre, Swansea University, Swansea, SA2 8PP, UK

2
College of Engineering,Center of Complex Fluids Processing, Multidisciplinary Nanotechnology Centre, Swansea
University, Swansea, SA2 8PP, UK
___________________________________________________________________________________________
Abstract: Lactococcus lactis species have been and still are extensively investigated due to their significant
commercial importance. Current scientific research focuses on strains utilized in food industry, due to their multiple
uses in food and beverages fabrication. Biomass of Lactococcus lactis is of great interest as well as the end products
of its metabolism such as lactic acid and nisin. However their production is constantly challenged due to end
product inhibition occurring during intensive propagation of the coccus in reactor systems. To successfully predict
the behavior of the culture, the approach of combining mathematics with biology, ergo the development of an
unstructured mathematical model, was taken. Although Luedeking and Piret is the model that has been extensively
used to demonstrate growth in end-product inhibition cultures, its applicability is limited due to its dependance on
the specific growth and product coefficients, particularly related to the culturing conditions used. To overcome these
hurdles, a combination of the non competitive single product end inhibition Taylor and Hinselwood models was
used, with the significance of this model laying in the fact that it offers a feasible alternative to the commonly used
model of Luedeking and Piret for describing fermentation kinetics governed by end-product inhibitions. The fitting
with the experimental values, in batch mode, was tested in terms of the coefficient of determination (R), having
values 0.97 to 0.99 and suggesting a very good fitting with the experimental data. The model was further developed
to achieve theoretical predictions of volumetric cell productivity in continuous and fed-batch mode of substrate feed
in different culturring systems.
Keywords: end product inhibition, growth kinetics, lactic acid, Lactococcoi, mathematical model
INTRODUCTION
[2]

Lactococcus Lactis (L.lactis) belongs to the genera of Lactic Acid Bacteria (LAB), which are microorganisms
whose distinctive ability is to produce lactic acid, as a major end product of their metabolism, through fermentation,
that is the anaerobic catabolism of carbon compounds existing into several types of substrates. L.lactis has been
graded as a non pathogenic bacterium for humans (Generally Regarded As Safe, GRAS) [1-4], enabling its wide use
in the food industry, as a natural acidproduction bioreactor. This bacterium is of particular interest to the
alimentary industry, due to its use as a starter culture, it can be inoculated in a variety of food products, assuring the
production of enough lactic acid [5-7], which will lower the pH point acidic point [8], inhibiting the development of
other pathogenic microbial strains and helping the preservation for longer periods of the products. At this point a
secondary starter culture of different microbial strains can be added so as to produce the desirable flavor, taste,
texture and odor [9-11].However, for the evaluation of productivity, optimization, and prediction of behavior and
control of a fermentation process mathematical modeling remains a valuable asset [12-15]. A mathematical model
can be defined as a set of correlations between the parameters of interest of a studied system. This set of relations
can be represented by mathematical equations which evaluate the dependence between the variables (agitation and
aeration, heat transfer and temperature maintenance, viability and preservation of microorganism, product recovery,
feed rate of nutrients, energy and power requirements in order to improve the fermentation process [16-19].
Each of the above parameters can be mathematically represented and be differentiated according to the mode of
operation of the fermentation system [20- 21]. Mathematical models, depending on their complexity, can be divided
into primary, secondary and tertiary, segregated and structured, and unsegregated and unstructured. They can be
evaluated, according to the bias and the accuracy factor, especially if used for performance prediction. Bias factor
can be used to judge the response of the model to the growth of microorganism and the accuracy factor measures the
overall model error [22-24]. The unstructured models are based on the assumption of a fixed cell composition which
is treated as a sole component, irrespective of the individual cellular structure, while biomass growth is balanced.
They are based on experimental observations (empirical models) that so not take into account the theoretical
stoichiometry of reactions [25-27]. On the other hand, segregated structured models use a multicomponent complex
description of cellular population based on its theoretical heterogeneity, including internal kinetics and cellular
regulation [28]. Due to the extensive demand of experimental data for their structure as well as the complexity of
their structure, they demand extended analytical equations they are not easily applicable [29-31].
The simplicity and applicability of unstructured models has led to the majority of fermentation processes being
evaluated according to these models, since a single state of culturing is used throughout the process, with the most
prominent example being the extensive use of the Monod equation to describe microbial kinetics [33-35].
Fermentation procedures are expected to operate under the restriction of a substrate limited growth model. The
relation between substrate concentration and specific growth rate is the basic parameter for the formation of kinetic
models [36]. For the majority of cellular fermentation processes the kinetics can be described by the Monod
[3]

equation, while several other rate expressions (Table 1) have been proposed to cover different conditions
parameters.
Due to the significant industrial importance of Lactococcus lactis, its fermentation on various sources has been
thoroughly investigated [37] and numerous kinetic studies have been published [38]. Regardless the great variety of
growth and production studies, mathematical modelling of the process remains limited to the use of the Luedeking
and Piret model [39], associated with the experimental definition of a growth-associated and of a non growth-
associated constant, limiting its applicability due to its dependence on the specific growth and product coefficients,
particularly related to the culturing conditions used [40-42]. To overcome these hurdles, a combination of the non
competitive single product end inhibition models of Taylor and Hinselwood were used in this study. To further
evaluate the predictive ability and the significance of estimating parameters, the model was developed into three
different culturing conditions: batch, continuous and fed-batch and two growth development systems: CSTR and
MBR.
MATERIALS AND METHODS
Bacterial strains
Lactococcus lactis NCIMB 700185 was provided in a lyophilised form by National Collection of Industrial Food
and Marine bacteria (NCIMB), Aberdeen, Scotland, United Kingdom.
Culturing Conditions
The strain was cultured in modified optimised liquid medium containing 20 g L
-1
glucose, yeast extract 10 g L
-1
,
bacteriological peptone 4 g L
-1
, sodium acetate 10 g L
-1
, sodium chloride 5 g L
-1
, potassium hydrogen phosphate 5 g
L
-1
. The mediums pH was set at 6.5.
Continuously Stirring Tank Reactor (CSTR)
A 2L capacity continuously stirring tank (CSTR) pyrex glass reactor (Figure 1) for the procedure. The fermenter
was designed and equipped accordingly to achieve a constant temperature of 30C and pH control, as well as
constant anaerobic conditions through constant flow of gaseous nitrogen. It was equipped with a hydrargiric
thermometer (temperature control) a pH probe (Fischer Scientific, UK) connected with a pH controller automation
apparatus (Electrolab FerMac 260, UK), a magnetic stir bar for agitation (350 r/m), a glass aeration port, a sampling
and inoculation port, a gas flow stainless steel port connected with a filter (Polyvent filter, 0.2m, Whatman Filters,
UK), a port for alkali/acid feed and stainless steel coils for heat emission. The sampling was performed on an hourly
basis on a 10 hours circle via the sampling port with a 10 ml sterile syringe and all the samples were measured for
biomass concentration, via optical density measurements.
Measurement of cellular growth and biomass
[4]

Determination of cell growth was monitored as an increase of turbidity in terms of optical density (O.D.) at 660 nm
wavelength into a spectrophotometer (PU 8625 UV/VIS Philips, France). The light path of the tube was 1.8 cm.
Measuring the O.D. was carried out on an hourly basis until it reached the late stationary phase. The growth curves
were obtained by plotting the O.D. against time. The maximum and specific growth rates (
max
,
1
h

and ,
1
h

)
of bacteria were calculated from the logarithmic plots of the O.D. versus time during the exponential growth phase
[5], according to the formula:
(
1
h )=
DT
2 ln
dt
) x (ln d
dt
dx
x
1
= = (1)
where DT (h) =
x
)
1
t
2
t (
(O.D. at 660nm hourly basis) (2)
Determination of End Products through High Performance Liquid Chromatography
L.lactis was reported to produce mainly lactic acid and a very small amount of acetic acid during the exponential
growth phase of its metabolism. In order to quantify the amount of the produced lactic acid and also to confirm the
fact that the strain was homofermentative all the samples were analysed by the High Performance Liquid
Chromatography (HPLC) method. The HPLC system was connected with a conductivity detector (ED 40
Electrochemical Detector, Dionex, UK) and fitted with an ion exchange column (Dionex, Canada). The solvent
(mobile phase) delivery system was constructed by 2 pumps (pumps A and B) (Varian Co. Canada), with a pressure
operation range between 100 to 1900 mbar. Temperature control of the solvents was maintained with a hotplate
(Millipore Co., UK) at 27C. All the samples were injected in the system via a sterile HPLC plastic syringe at a 20
l injection loop connected with the HPLC system.
The mobile phase in gradient flow, with a flow rate of 1 ml min
-1
, was represented by two solution of NaOH of
0.05mM and 100mM concentration. The solutions were delivered to the pumps via rubber tubes and valves with
each run lasting 38 min. The mobile phase was organised as following: NaOH 0.05 mM was poured for 5min., then
the concentration of the solution increased to 5.5mM for 10min. and then to 100mM for 5min. At the final phase the
NaOH concentration was kept at 50.5 mM for 15min and then at 0.5 mM for 3 min. The operation of the system
was controlled automatically using Prostar Workstation Data analysis software package (Varian Co., Canada).
Determination of Carbohydrate consumption rate
In order to investigate the amount of carbohydrate accumulated and to define the carbohydrate consumption rate
during the fermentation performed by the coccus glucose, concentration was measured using GOD-POD enzymical
method. The samples, which were taken from the fermentation, were transferred into 10ml conical plastic tubes
(Fisherbrand, UK) and centrifuged twice (5000 r/ m*10 m.) (Biofuge Stratos Sorall, Kendro Products, Germany) for
complete biomass removal.

[5]



Mathematical Model Development
The mathematical algorithm was based on the Monod model for microbial kinetics and was represented by three
differential equations each one of them representing product formation (
dt
dP
), substrate consumption (
dt
ds
) and
biomass formation (
dt
dx
). As it was confirmed experimentally L.lactis is a homolactic strain, its growth is inhibited
by constantly augmenting concentration of the metabolic end product. A mathematical term was incorporated in the
model based on Taylor and Hinselwood models so as to describe inhibition effects. Maintenance of the coccus into
stationary phase was represented by a maintenance coefficient (K
d
X) defined experimental, in addition to the
parameters representing the constants for product formation (K
P
) and substrate consumption rate (K
S
), which were
considered to be pH depended constants. All the equations were developed in Excel 2003 Microsoft office software
written on a VBA language spreadsheet, with the use of numerical methods, using the primary estimates of the
coefficients in parallel with the initial values of the parameters to create a series of time course curves that were
compared to the experimental data. Prior to correlating the model formula simultaneously and validating the
goodness of fit of the model, using the Linewear-Burke fitting plot, initial estimates of all constants were performed.
Each model formula was taken per term of development and the respective experimental data were used, to achieve
the best estimates of all the coefficients used. The basic estimates and the standardisation of all the model constants
were performed on every set of experimental data for the analysis of each set of batch of cultures. Several trials
were also done for the calibration of the model using different values for time base, constants, coefficients and
parameters. The best fitting was used in order to avoid any confusion in the results.
Numerical Analysis of the Experimental Data
Each differential parameter was triplicated to obtain the average data. The data were statistically analysed for
accuracy and precision, calculating standard deviation, standard error, experimental error, regression factor and
reading error (Microsoft Excel software Version 2003). All the numerical data were proven to be highly accurate
and reproducible, having standard deviation of mean below 5% and experimental error below 5%, offering highly
significant results.
RESULTS AND DISCUSSION

Batch Culture
The model simulation was done with the experimental result of the pH control fermentations, performed on a 2L
CSTR reactor, using optimised medium for growth of L.lactis. The numerical accuracy of the numerical predictions
given by the models were evaluated in terms of standard deviation (
-2
), assuming that the numerical values follow
[6]

a normal distribution. The data were analysed used the chi-test goodness of fit as well as the Shapiro-Wilk test,
indicating that the data had a close to zero skewness and a close to three kurtosis. In the case of batch growth
modelling, the fitting with the experimental values was evaluated in terms of the coefficient of determination (R).
R has values 0 to 1, which become larger as the theoretical predictions given from the model fit to the experimental
values [51]. The R values for the batch growth model were 0.97to 0.99 suggesting a very good fit with the
experimental data.
The estimation of parameters used in the model was allied to the experimental results available. Constants such as
the biomass coefficient (K
d
X), the substrate consumption coefficient (K
S
) and the product formation coefficient (K
P
)
(Table 2) were calculated experimentally and were used to give the fit for the batch culture performed on a pH
control fermentation on a CSTR.
Lactococcus lactis has complex nutritional needs due to its limited biosynthetic capacity [56]. Several researchers
[17,57-58], have highlighted its growth dependence over amino acids and vitamins that can stimulate significantly
biomass formation and cellular growth. A strong correlation has also been proved to exist between the nitrogen
sources within the substrate sources with L.lactis growth, however this dependence has been found being significant
in the case of other metabolites produced during its growth such as nisin [59-61].
Product yields achieved during L.lactis growth at a yield of 1.6 g g
-1
in batch culture. This is due to the strong
dependence of the strain on the pH conditions which were kept constant throughout the course of fermentation, but
as well as to the utilisation of the nutrient media, specifically designed and formulated to support intensive biomass
and lactic acid production. Glucose was the principal carbohydrate source provided in the nutrient media is mainly
responsible for lactic acid production (and energy generation) by L.lactis [64]. The carbon nitrogen sources such as
yeast extract, peptone, beef extracts although stimulate growth influence bacteriocin production such as nisin
[2,38,55].
Theoretically L.lactis converts 1 mole glucose to 2 moles of lactic acid [6]. Neverthelless, during culturing of
L.lactis observed product yields have been well below though of the theoretical prediction [51,54,62]. mainly due to
strain used, the environmental conditions and the fact that some of the carbon must be fixed as biomass. Lactic acid
production is also not only dependent on the carbohydrate sources composition but also of the culturing conditions
where environmental and physical conditions, i.e. oxygen levels, pH, temperature, sugar concentration, other media
components, and end products concentrations all affect the biomass and product yields [62], particularly pH and
oxygen. As a consequence observed yields of lactic acid to glucose have been found in the range 0.4 to 1.6 g g
-1

[54,62] with strain dependent yields between 1.0 to 1.6 g g
-1
when is cultured in optimum nutritional and
physicochemical conditions [63-68].
L.lactis growth is inhibited by the major end product of its metabolism, lactic acid; the threshold value for complete
growth inhibition due to lactic acid is 366 mM while growth kinetics are affected down to about 30 mM. During
[7]

growth of L.lactis on pH control fermentation, complete growth inhibition does not occur, but there is a strong
effect of lactic acid on biomass formation. A product inhibition term was incorporated in the model following the
Taylor and Hinselwood end product inhibition model, to describe the growth rate. The biomass formation (
dt
dx
) rate
is described as follows,
X
d
k X *
dt
dx
= (3)
where
|
|
|
|
|

\
|
|
|

\
|
+
|
|

\
|
+
=
) (
) (
) (
) * (
max
P K
P
S K
S
P
S

(4)
On further integrating the equation becomes:
( ) ( )
( ) ( )
|
|

\
|
+
+
=
P S K
P K S
S
P
*
* *
max

(5)
If the biomass formation rate (
dt
dx
) is on stationary phase, the maintenance coefficient (
d
k X), is included the
formula becomes:

( ) ( )
( ) ( )
|
|

\
|

+
+
= X k
P S K
P K S
dt
dx
d
S
P
*
* *
max

(6)
As biomass formation kinetics are strongly related to substrate consumption the equation for substrate consumption
rate (
dt
ds
) becomes:

dt
dx
s
x
Y
dt
ds
*
1
= (7)
On further integration the equation becomes

( ) ( )
( ) ( )
(

|
|

\
|

+
+
= X k
P S K
P K S
s
x
Y
dt
ds
d
S
P
*
* *
*
1
max

(8)
The product formation rate equation (
dt
dP
) proposed is based on the simplified assumption that the rate of product
formation is directly related to the rate of substrate consumption, through a yield coefficient
s
d pLacticaci
Y
[8]

For lactic acid formation the proposed equation is the following:
dt
ds
*
s
) lacticacid ( p
Y
dt
) Lacticacid ( dP
= (9)
The results of the investigation were developed and compared to the experimental to simulated data in a range of pH
values (5.5 to 7.0) (Figure 2 [a, b, c, d]). When growth reached stationary phase (in the experiment, 6 hours of
culturing the biomass concentration was 4.5 g L
-1
and using the model simulation 4.25 g L
-1
was achieved 6.25
hours) As it can be seen, the simulated data did fit to the experimental well in terms of predicting product formation
and growth but required a biomass maintenance coefficient term is used to model the growth accurately. In these
cases the product and cell formation developed in relation to the actual data. The end product inhibition term was
required in the mathematical model to describe the growth, as according to the experimental data available, it is
limited most probably, due to end product inhibition.
The theoretical predictions are in much better accordance with the experimental values for biomass formation and
lactic acid production (R
2
0.98), especially in pH 5.5, 6.0 and 7.0. However, there was still some poor fitting at the
end of the growth phase of pH 6.5 where some inhibition, most probably due to the pre-existing amount of lactic
acid in the inoculums, could not be described by the model, and the product and substrate were not modelled very
accurately.
In order to investigate the effect of substrate feed on different concentrations, over growth and cell productivity of
L.lactis on a continuous system, the model was tested against different substrate concentrations and theoretical
predictions were made. The efficiency of the system was evaluated in terms of volumetric cell productivity which is
given by the following equation:
Volumetric Cell Productivity (g L
-1
h
-1
) = Final X concentration / Total Fermentation Time (h) (10)
Using the model a series of simulation were made where the substrate concentrations varied between 2.5 g L
-1
to 20
g L
-1
. All the conditions of the system were kept steady on a 10 h fermentation circle. All the numerical values of
the parameters and constants were kept the same (Table 3). The results (Figure 3) show that the volumetric cell
productivity was higher in pH values of 6.5 and 7 and on substrate concentrations of 15 and 20 g L
-1
. At lower
substrate concentrations, the substrate becomes completely exhausted in a short period of time, with a lower
concentration of lactococcal biomass. At lower pH values the biomass formation rate is also lower with reduced
productivity.
The maximum volumetric cell productivity for batch culturing was 0.47 g h
-1
and was achieved on 20 g L
-1
substrate
on pH 6.5. While the mechanical construction and operation of a batch system is far less complicated compared to
other systems of culturing, the volumetric cell productivity is proven to be relatively small when compared to the
theoretical predictions made by modelling the intensive propagation of L.lactis on continuous and on the MBR
systems.
[9]

Culturing in bath mode in this time circle was insufficient for high amounts of lactococcal biomass. The batch
system, although is preferable especially in industrial scale practise due to the mechanical simplicity of the
apparatus used, results in simpler operation, also owing to the easier extraction methods of the desired end products.
Continuous Culture
Having investigated batch cultures, an unstructured kinetic mathematical model was developed for theoretical
predictions of growth of L.lactis on a continuous culture system operating on constant pH control. The mathematical
algorithm was based on the previously developed equations over biomass formation, substrate consumption and
lactic acid production for batch growth of L.lactis. Maintenance and an end product inhibition term were also
incorporated. In order to evaluate the end product inhibition effects during growth the model was tested against
different substrate concentrations, between 2.5 to 20 g L
-1
. The efficiency and feasibility of the system was
evaluated in terms of volumetric cell productivity according to the equation
Volumetric Cell Productivity (g L
-1
h
-1
) = Final X in the system *D (11)
The terms describing the biomass formation rate and the biomass maintained in the system were kept unchanged.
However several new terms were developed especially those needed to demonstrate the changes in the system. The
biomass leaving the system (X
out
) was given by
t * F *
in
X
out
X = (12)

while biomass maintained in the system (X
in
) is given by the formula:

out
X
dt
dx
0
X
in
X + = (13)
In the case of substrate, the substrate inflow rate (S
in
) in the system is given by the following equation:
t * F *
0
S
in
S = (14)
The substrate outflow rate (S
out
) in the system is given by the following equation:
S
out
= S
in
*F*t (15)
The substrate existing initially in the system (S
ins
) which is accumulated during growth is given by the following
equation:
Sin Sout
dt
ds
So
ins
S + + = (16)
On further analysing the above equation, it becomes:
[10]


in out d
P
S
ins
S S X k
P K
P
S K
S
s
x
Y
So S + +
|
|
|
|
|

\
|
|
|

\
|
+
|
|

\
|
+
+ =
) (
) (
) (
) * (
*
1
max

(17)
On further analysing the above equation, it becomes:

in out d
S
P
ins
S S X k
P S K
P K S
s
x
Y
So S + +
|
|

\
|
+
+
+ =
) ( * ) (
) ( * ) * (
*
1
max

(18)
When describing the product formation rate (
dt
dp
), the following equations occur:

|
|
|
|
|

\
|
|
|
|
|
|

\
|
|
|

\
|
+
|
|

\
|
+
=
) (
) (
) (
) * (
*
1
*
) (
) (
max
P K
P
S K
S
s
x
Y
s
lacticacid p
Y
dt
Lacticacid dP
P
S

(19)

On further analysis, it becomes:

|
|
|

\
|
|
|

\
| +
=
) ( * ) * (
) ( * ) * (
*
1
*
) (
) (
max
P S K
P K S
s
x
Y
s
lacticacid p
Y
dt
Lacticacid dP
S
P

(20)
The rate on the product maintained in the system (P
in
) is given by the formula:
|

\
|
=
out in in
P S S
s
p
Y P ) ( *
0
(21)
The rate of product outflow (P
out
) is given by the formula:
t F P P
in out
* * = (22)
The dilution rate (D) is given by the equation:
V
F
D =
(23)

Theoretical predictions were made over the volumetric cell productivity on every pH set using several dilution rates
and substrate concentrations. All the conditions of the system were kept steady and all the numerical values of the
parameters and the constants were kept the same using in the mathematical model for batch growth. Substrate feed
[11]

concentrations varied between 2.5g L
-1
to 20 g L
-1
and dilution rate between 0.1 to 0.5 L h
-1
. The efficiency of the
system was evaluated according to the volumetric cell productivity achieved on every test (Figure 4).
The volumetric cell productivity is higher than the one occurring in batch culture of the coccus on pH 6.5 and 7.0
and substrate concentrations of 15 and 20 g L
-1
(Figure 5). In addition, the productivity of the system is strongly
correlated with the substrate inflow rate which is dependent on the flow rate. For every set of pH and substrate
concentration the optimum flow rate is different.
A continuous system of intensive culture of L.lactis can offer higher yield of lactococcal biomass, as using a
continuous system for intensive culture ensures higher yields of biomass, due to simultaneous removal of end
products which are inhibitory for growth. However, its operation especially on a large scale is possibly more
complicated as there are numerous mechanical and physicochemical parameters that need to be considered.
Membrane Bioreactor Culture
Based on the unstructured kinetic mathematical algorithm developed for batch and continuous culture of L.lactis, an
unstructured kinetic mathematical model was developed for theoretical predictions over the growth of L.lactis on a
MBR, incorporating a maintenance coefficient term. The terms describing the biomass formation rate and the
biomass maintained in the system were kept unchanged. However several new terms were developed especially
those needed to demonstrate the changes in the system.
The efficiency of the system was evaluated in terms of volumetric cell productivity, using two methods for substrate
feed, continuous substrate feed and fed-batch, in order to compare the theoretical predictions done over the
volumetric cell productivity (Figure 6). The model was tested against different substrate concentrations and flow
rates varying between 2.5 to 20 g L
-1
of substrate feed and 1 to 10 L h
-1
flow rate in the optimum pH value for
growth selected (pH 6.5).
In order to describe the biomass maintained in the system (X
in
) the following equation is formed,
dt
dx
X X
in
+ =
0
(24)
The substrate inflow rate (S
in
) in the system is given by the following equation:
t F S S
s in
* *
0
= (25)

The substrate outflow rate (S
out
) in the system is given by the following equation:
dt
dF
S S
in out
* =

(26)

[12]

After further analysis of the above equation, it becomes

in out d
P
S
ins
S S X k
P K
P
S K
S
s
x
Y
So S + +
|
|
|
|
|

\
|
|
|

\
|
+
|
|

\
|
+
+ =
) (
) (
) (
) * (
*
1
max

(27)

On further analysing the above equation, it becomes:

in out d
S
ins
S S X k
P S K
P Kp S
s
x
Y
S S + +
|
|

\
|
+
+
+ =
) ( * ) (
) ( * ) * (
*
1
max
0

(28)
When describing the flow rate of nutrient in the system and the flow of nutrient in the system the following
equations are developed,
The flow rate of nutrient during time (
dt
dF
) is described by the following formula:
t F
dt
dF
* * = (29)
The flow of the substrate Fs, in the system is given by the following equation:

dt
dF
F Fs + =
0
(30)
When describing the rate of product formation, product maintained in the system and product outflow rate most of
the equations are based on the previously developed models, however to describe the additional characteristics of
the system new equations were developed.
The rate on product maintained in the system (P
in
) is given by the formula:
|

\
|
=
out in in
P S S
s
p
Y P ) ( *
0
(31)

The rate of product outflow (P
out
) is given by the formula:
t F P P
in out
* * = (32)
The volumetric cell productivity of the system was evaluated according to the following equation, both for fed-
batch and continuous substrate feed.
[13]

Volumetric Cell productivity (g L
-1
h
-1
) = Final X in the MBR(g) / Total Fermentation Time (h) (33)
The systems efficiency was evaluated. The volumetric cell productivity for fed-batch supply of substrate was 28 g
L
-1
h
-1
at a flux rate of 10 L h
-1
while for continuous substrate supply being 45 g L
-1
h
-1
at flux rate of 10 L h
-1
. When
comparing the volumetric cell productivity between the continuous and the fed-batch substrate support, the in
various substrate concentrations and flow rates continuous substrate support was proven to be superior.
The volumetric cell productivity of L.lactis intensively propagated on a MBR system is much higher when
compared with the ones predicted from the models for batch and continuous mode of culturing (Table 4).
Regardless the MBR system proven to be highly productive, it construction and its handling are far more
complicated, when compared with the handling of a batch system of culturing. Many mechanical parameters such as
heat transfer, flux rates, membrane fouling, membrane surface charge and resistance, sterilisation of the mechanical
apparatus and also of the growth media, cleaning and maintenance of the system have to be considered, resulting to
its limited applicability in the industry.
CONCLUSIONS
For the intensive propagation of L.lactis three different production systems were investigated. The coccus was
primarily grown in simple batch cultures without pH control, determining its nutritional needs. The inhibition
effects of the end products were also studied.
To explore further this system in more detail, the coccus was grown in a CSTR incorporating continuous pH
control. A simple mathematical model was developed for the deeper understanding of the growth kinetics of the
coccus when cultured on a batch system. Through this model, theoretical predictions were made and validated over
the volumetric cell productivity of the system when using different substrate concentrations and inoculum size.
Growth and lactic acid productivity of the strain was considered to be highly dependent on pH, temperature and
substrate composition. As growth conditions were defined experimentally, to optimise the culturing systems and
forecasting the potential productivity of the organism if grown. Using these batch data the computational simulation
models were developed validated and provided theoretical predictions for batch cultures. These were then used to
predict the theoretical behaviour of the continuous and the MBR culture under the assumption that product yields
would not vary significantly. Past work on this organism suggests that providing the environmental conditions and
initial medium composition are specified this should be the case.
Based on the model, the volumetric cell productivity of the batch system was relatively low and a continuous
culture and MBR systems were investigated. The model was tested against several substrate concentrations and flow
rates and theoretical predictions were made over the volumetric cell productivity. In both continuous and MBR
culture systems offered higher biomass yields when compared to batch culture. Comparative studies were conducted
between these systems. The selected parameter for comparison was the volumetric cell productivity achieved from
[14]

every system on different substrate concentrations and for the case of continuous culture and a membrane bioreactor
on different dilution and flow rates. These studies show that the MBR system was proven to be highly productive
offering higher yields of biomass in a shorter period of time, especially when operated with continuous substrate
feed.
The significance of this model lies in the fact that it offers an alternative to the commonly used model of Luedeking
and Piret, for describing fermentation kinetics govern by end-product inhibition. Unlike the Luedeking and Piret
model which is only limited to the only a specialised set of experimental conditions and microorganism used and
substrate source, the proposed model can be expanded to various culturing conditions and to a variety of
homofermentating microorganisms, including the commercially important Lactoccoci and Lactobacilli.

NOMECLATURE
Specific growth rate 1
h

max
Maximum specific growth rate 1
h

t
d
Doubling time h
dt
dP

Product formation rate g L
-1

dt
ds

Substrate accumulation rate g L
-1

dt
dx

Biomass formation rate g L
-1

dt
dF

Flow rate of nutrient during time
L
1
h

S Dissolved substrate concentration g L
-1
X Cell concentration g L
-1

P End product concentration g L
-1

K
S
Constant of substrate consumption dependent on pH maintenance g L
-1

K
P
Constant of end product production g L
-1

X
d
k

Biomass growth maintenance coefficient g substrate g
-1
cells h
-1
g g
-1
h
-1
[15]

s
x
Y
Yield of biomass from substrate g

g
-1
s
d pLacticaci
Y

Yield Productivity coefficient of lactic acid production from
substrate
g g
-1

s
p
Y
Yield coefficient of lactic acid production from substrate g g
-1

F Flux of the nutrient per hour
L
1
h

S
in
Substrate inflow rate per hour
g L
-1

1
h

S
0
Initial substrate concentration within the nutrient g L
-1

t Time step h
S
out
Substrate outflow rate per hour
g L
-1

1
h

S
ins
Substrate existing initially in the system g L
-1

X
0
Biomass inoculated initially in the system g L
-1

X
out
, Biomass outflow g L
-1

X
in
Biomass maintained in the system, g L
-1

D Dilution rate in the system during time 1
h

V Total Volume of Substrate in the system g L
-1

F
S
Flow of substrate in the system g L
-1

0
F Initial flow in the system g L
-1


[16]

REFERENCES
[1] Bai, D., Jia, M., Zhao, X., Ban, R., Shey, F., Li, X., et al. (2003). L (+) - lactic acid production by pellet
form Rhizopus oryzae R1021 in a stirred tank fermentor. Chem Eng Sci J. 58:785-791.
[2] Beasley, S.S., and Saris, P.E.J. (2004). Nisin-producing Lactococcus lactis strains isolated from human
milk. Appl Env Microb J.70: 5051-5053.
[3] Boonmee, M., Leksawasdi, N., Bridge, W., and Rogers, P.L. (2003). Batch and continuous culture of
Lactococcus lactis NZ133: Experimental data and model development. Biochem Eng J. 14: 127-135.
[4] Demain, L.A., and Davies, J.E. (1999). Washington D.C., ASM Press pp:29-45, 49-60, 61-75, 80-93, 183-
147, 151-163, 165-179 Manual of Industrial Microbiology and Biotechnology
[5] Bu'lock, J.D., and Kristiansen, B. (1978). pp:75-132,133-172, 371-374 New York, London, Academic
press Basic Biotechnology
[6] Carr, J.G., Cutting, C.V., and Whiting, G.C. (1975) pp:17-28, 233-266 Lactic Acid Bacteria in Beverage
and Food New York, London, Academic press.
[7] Glazier, A.N., and Nikaido, H. (1995) pp:358-390 New York, W.H. Freeman and Company Microbial
Biotechnology :Fundamentals of Applied Microbiology
[8] Dougherty, D.P., Breidt, F., McFeeters, R., and Lubuin, S.R. (2002). Energy-based dynamic model for
variable temperature batch fermentation by Lactococcus lactis. Biotech Bioeng J. 68:2468-2478.
[9] Kamoshita, Y., Ohashi, R., and Suzuki, R. (1998). Improvement of filtration performance of stirred
ceramic membrane reactor and its application to rapid fermentation of lactic acid by dense cell culture of
Lactococcus lactis. Ferm Bioeng J.85: 422-427.
[10] Lan, Q.C., Oddone, G., Mills, D.A., and Block , D.E. (2006). Kinetics of Lactococcus lactis growth and
metabolite formation under aerobic and anaerobic conditions in the presence or absence of hemin. . Biotech Bioeng
J. 94: 1-30.
[11] Moonchai, S., Madhoo, W., Jariyachadit, K., Shimizu, H., Shioya, S., and Chavatcharin, S. (2005).
Application of a mathematical model and differential evolution algorithm approach to optimization of bacteriocin
production by Lactococcus lactis C7. Bioprocess Biosys Eng. 28:15-26.
[12] Doran, P.M. (2004). pp:1-35, 66-77, 208-221, 622-647 New York, London, Academic press. Bioprocess
Engineering Principles
[13] Coulson, J.M., and Richardson, J.F. (1994). (3rd ed., Vol. 3)pp:2-69, 108-189, 196-249, 252-337,364-431
Oxford, Pergamon Press Chemical Engineering, Chemical and Biochemical Reactors and Process control
[17]

[14] DiSerio, M., Tesser, R., and Sanctaserria, E. (2001). A kinetic and mass transfer model to simulate the
growth of baker's yeast in industrial bioreactors. Chem Eng J. 82: 347-354.
[15] Goksunger, Y., and Guvenc, U. (1997). Batch and continuous production of lactic acid from beet molasses
by Lactobacillus delbrueckii IFO 32. Chem Tech Biotech J. 69: 399-404.
[16] Hietala, K.A., Lynch, M.L., Allhouse, J.C., Johns, C.J., and Roane, T.M. (2006). A mathematical model of
Saccharomyces cerevisiae growth in response to cadmium toxicity. Basic Microbiology J. 46: 196-202.
[17] Hoefnagel, M.H.N. (2002). Metabolic engineering of lactic acid bacteria, the combined approach: kinetic
modeling, metabolic control and experimental analysis. Microbio J. 148: 1003-1013.
[18] Kovarova, K., and Egli, T. (1998). Growth kinetics of suspended microbial cells: From single-substrate
controlled growth to mixed-substrate kinetics. Microb Mol Bio Rev 62:646-666.
[19] Kovarova, K., Zehnder, A.J.B., and Egli, T. (1996). Temperature dependent growth kinetics of E.coli MC
30 in glucose limited continuous culture. Bacter J. 178: 4530-4539.
[20] Narang, A., Konopka, A., and Ramkrishna, D. (1997). The dynamics of microbial growth on mixtures of
substrates in batch reactors. Theor Bio J. 184: 301-317.
[21] Okpokwasilli, G.C., and Nueke, C.O. (2005). Microbial growth and substrate utilization kinetics Afr J
Biotech 5:305-317.
[22] Panikov, N. (1995). (1st ed.):Dordrecht, Netherlands Chapman and Hall Editions pp:110-237.
[23] Pirt, S.J. (1985) pp:1-75,156-169 Principles of microbe and cell cultivation Oxford: Blackwell Scientific
Publications Microbial Growth Kinetics
[24] Stanbury, P.F., and Whitaker, A. (1993). pp:11-25, 108-191 London, Pergamon Press Principles of
Fermentation Technology
[25] Leroy, F., and Vuyst, L. (2001). Growth of the bacteriocin-producing Lactobacillus sakei strain CTC 494
in MRS broth is strongly reduced due to nutrient exhaustion: a nutrient depletion model for the growth of lactic acid
bacteria. Appl Env Microb J. 67: 4407-4413.
[26] Gregersen, L., and Jorgensen, S.B. (1999). Supervision of fed-batch fermentations Chem Eng J. 75: 69-76.
[27] Santago, A.B., Rodriguez, J., Carrasco, J., Gomez Gomez, E., Perkel, M., and Rojo, I. (1997). Growth
kinetics of L-aminoacylase-producing Pseudomonas sp.BA2. Chem Eng J. 52: 171-176.
[28] Vrabel, P., Van der Lan, G.M.J., Van der Schot, F.N., Luyden, K., Xu, B., and Enfors, S. (2001). CMA:
Integration of fluid dynamics and microbial kinetics in modeling of large scale fermentations. Chem Eng J. 84: 463-
474.
[18]

[29] Zhang, H., and Lennox, B. (2004). Integrated condition monitoring and control of fed-batch fermentation
process. Process Contr J. 14: 41-50.
[30] Zhang Y., and Henson, M.A. (2001). Bifurcation analysis of continuous biochemical reactor models.
Biotech Progr J. 17: 647-660.
[31] Patnaik, P.R. (2002). Comparative evaluation of batch and fed-batch bioreactors for GAPDH production by
recombinant Escherichia coli with distributed plasmid copy number. Chem Eng J. 87:357-366.
[32] Cicek, N. (2003). A review of membrane bioreactors and their potential application in the treatment of
agricultural wastewater. Canadian Biosyst Eng J. 45: 637-646.
[33] Kwon, S., Yoo, H., Lee, G., Chang, H., and Chang, K. (2000). High-rate continuous production of lactic
acid by Lactobacillus rhamnous in a two-stage membrane cell-recycle bioreactor. Biotech Bioeng J. 73: 25-34.
[34] Rose, A.H. (1982). pp:148-189. New York, London, Academic press Economic Microbiology :Fermented
Foods .
[35] Ross, R.P., Desmond, C., Fitzerald, G.F., and Stantch, C. (2005). Overcoming the technological hurdles in
the development of probiotic foods. App Microbiol J. 98:1410-1417.
[36] Martak, J., Schlosser, S., Sabalova, E., Kristofikova, L., and Rosenberg, M. (2003). Fermentation of lactic
acid with Rhizopus arrhizus in a stirred tank reactor with a periodical feed and bleed operation. Process Biochem J
38: 1573-1583.
[37] Pavan, S. (2000). Adaptation of the Nisin-Controlled Expression System in Lactobacillus plantarum: a tool
for study in vivo biological effects. Appl Env Microbiol, 66:4427-4432.
[38] Todorov, S.D., and Dicks, L.M.T. (2004). Influence of Growth conditions on the production of a
bacteriocin by Lactococcus lactis subp. lactis ST 34BR, a strain isolated from barley beer. Basic Microbiol J 44:
305-316.
[39] Luedeking, R., and Piret, E.L. (1959). A kinetic study of the lactic acid fermentation batch process at
controlled pH. Bioch Microbiol Tech Eng J. 1: 393-412.
[40] Wang, H., Seki, M., and Furusaki, S. (1995). Mass transfer behavior in lactic acid fermentation using
immobilized Lactobacillus delbrueckii. Chem Eng Science J. 28 : 480-482.
[41] Zhu, G. (2000). Model predictive control of continuous yeast bioreactors using cell population balance
models. Chem Eng J. 55:6155-6167.
[42] Yang, S. (2004). The growth kinetics of aerobic granules developed in sequencing batch reactors. Soc Appl
Microbiol J 38: 106-112.
[19]

[43] Thiel, M., Coster, D., Mavrangelos, O., Zola, H., and Williams, K.A. (2002). An economical 20 litre
bench-top fermenter. Protein Expr Purif J. 84:40-44.
[44] Webster-Miller, T. (2002). Influence of yeast culture on ruminal microbial metabolism in continuous
culture. Dairy Sci J. 85: 2009-2014.
[45] Takesono, S., Onodera, M., Yoshida, M., Yamagiwak, K., and Ohkawa, A. (2004). Performance
characteristics of a stirred tank reactor with a mechanical foam-breaker utilizing shear force. Chem Eng J. Japan
37: 1488-1496.
[46] Giorno, L., and Drioli, E. (2000). Biocatalytic membrane reactors: applications and perspectives.
TIBTECH J, 18:339-349.
[47] Wintgens, T., Gallenkemper, M., and Melin, T. (2002). Endocrine disrupter removal from wastewater
using membrane bioreactor and nanofiltration technology. Desal J. 146:387-391.
[48] Ye, K., Jin, S., and Shimizu, K. (1996). Cell recycle and broth reuse fermentation with cross-flow filtration
and ion exchange resin. Chem Tech Biotech J. 66: 223-226.
[49] Font, R., and Cabanes, J. (1995). Fermentation in fed-batch reactors: Application to the sewage sludge
anaerobic digestion. Chem Eng J. 50: 2117-2126.
[50] Navidi W. (2010) pp:608-772 McGraw-Hill Higher Education Statistics for Engineers and Scientists
[51] Hofvendahl K., Hahn-Hagerdal B.,(1997) L-lactic acid production from whole wheat flour hydrolysate
using strains of Lactobacilli and Lactococci, Enzyme Microb Technol 20:301-307
[52] Roble N.D, Ogbonna J.C., Tanaka H.,(2003) L-lactic acid production from raw cassava starch in a
circulating loop bioreactor with cell immobilized in loofa (Luffa cylindrica) Biotechnol Lett 25:1093-1098
[53] Nomura Y., IwaharaM., Hallsworth J.E., Tanaka T., Ishizaki A.,(1998) High-speed conversion of xylose to
L-lactate by electrodialysis bioprocess. J Biotechnol 60:131-135
[54] WeeY-J., Kim J-N., Ryu H-W., (2006) Biotechnological Production of Lactic acid and its recent
applications Food Technol Biotechnol 44:136-172
[55] Vazquez J.A., Gonzalez M.P., Murado M.A.,(2004) Nisin and Pediocin production by Lactococcus lactis
and Pediococcus acidilactici using waste protein sources from octopus E JEAF Che 3:648-657
[56] Skinner, F. A. and Quesnel, Louis B. 1978, Streptococci F. A. Academic Press, London ; New York :
[57] Dembczynski, R. and Jankowski, T.(2002) Growth characteristics and acidifying activity of Lactobacillus
rhamnosus in alginate/starch liquid-core capsules. Enzyme Microb Techn, 31:.111-115.
[20]

[58] Desjardins, P., Meghrous, J., Lacroix, C. (2001). Effect of aeration and dilution rate on nisin Z production
during continuous cultures with free and immobilized L. diacetylactis UL719 in supplemented whey permeate."
Intern. Dairy J. 11: 943-951.
[59] Konings W.N., Kok J., Kulipers O., Poolman B. (2000), Lactic Acid Bacteria : The bugs of the new
millennium. Curr Opinion Microbiol, 3: 276-282.
[60] Liew S.L., Ariff A. B., Racha A.R., Ho Y.W (2005) Optimization of medium composition for the
production of a probiotic microorganism Lactobacillus rhamnosus using response surface methodology. Int J Food
Microbiol 102: 137-142.
[61] Ostlie H.M., Helland M. H., Narvhus J.A., (2003) Growth and metabolism of selected strains of probiotic
bacteria in milk Int J Food Microbiol 87: 17-27.
[62] Hofvendahl K.and Brbel HahnHgerdal B. (2000) Factors affecting the fermentative lactic acid
production from renewable resources, Enzyme Microb Techn 26: 87-107
[63] Vlaemynck G, Traest P, De Vilder J. Fermentation of permeate for production of lactic acid recovered by
electrodialysis. Med Fac Landbouww Rijksuniv Gent 1989;54:1369 76.
[64] Axelsson LT. 1993:1 63 Lactic acid bacteria: classification and physiology. In: Salminen S, von Wright
A, editors. Lactic Acid Bacteria. New York: Marcel Dekker, Inc.
[65] Kwon, Y. J., Kaul, R. and Mattiasson, B. (1996), Extractive lactic acid fermentation in poly
(ethyleneimine)-based aqueous two-phase system. Biotechnol Bioeng 50: 280290
[66] kerberg C., Zacchi G. (2000) An economic evaluation of the fermentative production of lactic acid from
wheat flour, Bioresour Technol 75:119126.
[67] Dissing U, Mattiasson B. (1994) Cultivation of Lactococcus lactis in a polyelectrolyte-neutral polymer
aquaeous two-phase system. Biotechnol Lett 16:333 8.
[68] Hofvendahl K, kerberg C, Zacchi G, HahnHagerdal B. (1999) Simultaneous enzymatic wheat starch
saccharification and fermentation to lactic acid by Lactococcus lactis. Appl Microbiol Biotechnol 52:163169.





[21]

Growth Kinetics Models Developed by Comments
s K
S
s
+
= *
max


Monod Monod equation can
describe adequately process
of slow growth and low
population density
S
K
S
2
*
max

=
Blackman

Substrate limited equation
) 1 ( *
max
S
K
e

=
Tessier Substrate limited equation
n
S
n
s
S
S K
S
S K
+
= +

*
) * 1 ( *
max
1
max

Moser Substrate limited equation
S X K
S
S
+
=
*
*
max

Contois Growth inhibition due to
toxicity of the organic
compounds produced
) ( * ) (
* *
max
P K S K
K S
P S
P
+ +
=

Jerusalimsky Growth inhibition due to
toxicity of the organic
compounds produced
[ ] ) * 1 ( * ) ( P K S K S
P S
+ =
Hinselwood Growth inhibition due to
toxicity of the organic
compounds produced
(

+ = )
1
( * ) ( * *
max
P
S
K
P
S K S


Taylor Growth inhibition due to
toxicity of the organic
compounds produced
) * 1 ( * ) (
*
max
P K S K
S
P S
+
=


Taylor Growth inhibition due to
toxicity of the organic
compounds produced
N b
dt
dN
a
dt
dP
* * + =
Leudeking and Piret Growth inhibition due to
toxicity of the organic
compounds produced
Table 1: Models used to describe microbial fermentations [12,13,17,23]

[22]


Table 2: Coefficients and parameters used for mathematical modelling

Parameter Definition Dimension Values on
different pH

max
Maximum Specific growth rate 1
h

5.5 0.22
6 0.24
6.5 0.66
7 0.72
K
S
Constant g L
-1
5.5 0.40
6 0.435
6.5 0.471
7 0.750
K
P
Constant g L
-1
5.5 0.643
6 0.679
6.5 0.714
7 0.750
s
x
Y
Yield coefficient (cell growth) g g
-1
5.5 0.11
6 0.13
6.5 0.20
7 0.19
s
d pLacticaci
Y
Yield Productivity coefficient of lactic acid
production from substrate
g g
-1
1.6
X k
d

Maintenance coefficient g g
-1
h
-1
0.057
[23]


Table 3: Coefficients and Parameters used for theoretical prediction of volumetric cell productivity on
different substrate concentrations of batch growth of L.lactis on a STR on different pH











Parameter Definition Dimension pH Values

max
Maximum Specific growth rate 1
h

6.5
0.66
K
S
Constant g L
-1
0.471
K
P
Constant g L
-1
0.714

s
x
Y
Yield coefficient (cell growth) g g
-1
0.20
s
d pLacticaci
Y
Yield Productivity coefficient of lactic acid
production from substrate
g g
-1
1.6
X k
d

Maintenance coefficient g g
-1
h
-1
0.057
[24]

Table 4: Volumetric Cell Productivity of each system and improvement over batch growth

















Culturing Systems Volumetric Cell Productivity (g L
-1
h
-
1
)
Improvement over Batch system (times)
Batch 0.4 -
Continuous 1.3 3.25
MBR Fed-Batch feed 28 70
MBR Continuous feed 45 112.5

*Corresponding author
Tel:+447413541769
Email:myrtozacharof1981@yahoo.com

Fig. 1: Schematic representation of the CSTR used for mathematical modelling studies from the selected lactobacilli
[26]


Fig. 2. A comparison of the mathematical model predictions with the experimental data where () lactic acid production, () biomass formation and
carbohydrate ( ) on pH 6.5 on a STR including end product inhibition (a) pH 5.5 (b) pH 6.0 (c) pH 6.5 (d) pH 7.0.

[27]



Figure 3. Theoretical predictions of volumetric cell productivity on different substrate concentrations of
batch growth of L.lactis on a STR on different pH 5.5, () pH 6, () pH 6.5 (), pH 7()

0
0.05
0.1
0.15
0.2
0.25
0.3
0.35
0.4
0.45
0.5
0 5 10 15 20 25
Substrate Concentration (g/L)
V
o
l
u
m
e
t
r
i
c

C
e
l
l

P
r
o
d
u
c
t
i
v
i
t
y

(
g
/
L
/
h
)
[28]


Figure 4. Theoretical predictions of volumetric cell productivity on different substrate concentrations of
continuous growth of L.lactis on different pH 5.5 () pH 6 (), pH 6.5, () pH 7()
0
0.2
0.4
0.6
0.8
1
1.2
1.4
0 5 10 15 20 25
Substrate Concentration (g/L)
V
o
l
u
m
e
t
r
i
c

c
e
l
l

p
r
o
d
u
c
t
i
v
i
t
y

(
g
/
L
/
h
)
[29]


Figure 5.Comparison between the theoretical predictions of volumetric cell productivity on different
substrate concentrations of fed-batch and continuous substrate feed on growth of L.lactis on a MBR on
different flux rates. For continuous substrate feed the symbols are the following: 2.5 g/l(), 5 g/l()10 g/l
(), 15 g/l(+) , 20 g/l(-), for fed-batch substrate feed the symbols are the following: : 2.5()g/l, 5 g/l(*)10
g/l(),15 g/l(--) , 20 g/l ()


0
5
10
15
20
25
30
35
40
45
50
0 2 4 6 8 10 12
Flux (L/h)
V
o
l
u
m
e
t
r
i
c

c
e
l
l

P
r
o
d
u
c
t
i
v
i
t
y

(
g
/
L
/
h
)