ISSN 0040-5795, Theoretical Foundations of Chemical Engineering, 2017, Vol. 51, No. 3, pp. 282–298. © Pleiades Publishing, Ltd.

, 2017.
Original Russian Text © Yu.L. Gordeeva, E.G. Rudakovskaya, E.L. Gordeeva, A.G. Borodkin, 2017, published in Teoreticheskie Osnovy Khimicheskoi Tekhnologii, 2017, Vol. 51,
No. 3, pp. 270–287.

Mathematical Modeling of Biotechnological Process of Lactic Acid

Production by Batch Fermentation: A Review
Yu. L. Gordeevaa, *, E. G. Rudakovskayab, E. L. Gordeevab, and A. G. Borodkinb
aSkryabin Moscow State Academy of Veterinary Medicine and Biotechnology, Moscow, Russia
bMendeleeev University of Chemical Technology of Russia, Moscow, Russia
*e-mail: l.s.gordeev@yandex.ru
Received July 28, 2016

Abstract⎯The results of analyzing mathematical models of the process of batch fermentation for the produc-
tion of lactic acid have been presented. Three groups of mathematical models have been considered. The first
two groups are based exclusively on the notion of the specific growth rate, dependent or independent on sub-
strate concentration. The third group uses the notions of the specific rate of substrate consumption and specific
rate of product formation in additions to the notion of specific growth rate. The numeric estimates of constants
are presented for the majority of models as well as comparison with the experimental studies. Mathematical
models of the processes where the formation of side products is possible have been discussed, which may or
may not be of value on their own. The possible practical applicability of the models has been evaluated based
on the results of the analysis.

Keywords: lactic acid, batch fermentation, mathematical models

DOI: 10.1134/S0040579517030058

INTRODUCTION formation are used concurrently with the specific

growth rate.
Numerous studies of the kinetics of biotechnologi- The analysis of publications showed that modeling
cal process of lactic acid production based on a non- is most often based on equations of types (1)–(3) as
structured approach [1, 2] emphasize scientists’ sig- follows:
nificant interest in the process of using the strains that
produce lactic acid. Adequate kinetic expressions pro- for the growth rate,
vide a good basis for improving technological pro- dX = μ X ;
cesses and assessing their best parameters based on the (1)
dt
results of modeling. The latter is related to the need to for the rate of product formation,
consider both technological and biological limita-
tions, the underestimation of which can make it
impossible to conduct the process of synthesis in prac-
(
dP = α dX + β X = αμ + β X ;
) (2)
dt dt
tice. Mathematical modeling correlates with the for the rate of substrate consumption,
method of conducting the technological process. It
follows from the data on the methods of lactic acid fer- − dS = 1 dX + 1 dP + mS X . (3)
mentation presented in Table 1 [3] that both batch and dt Y X S dt Y P S dt
continuous fermentation modes are widely used. This
review based on the publications on modeling of the
processes of batch fermentation. MODELING OF BATCH-FERMENTATION
PROCESSES
Mathematical models are presented in three groups
of expressions. The first two groups include equations The simplest equation for the growth rate in which
based on the notion of specific growth rate. Moreover, the specific growth rate does not depend on the sub-
the expressions of the specific growth rate in the first strate concentration is [13–21]
group are independent of the substrate concentration, dX = μ ⎛1 − X ⎞ X ,
max ⎜ (4)
X max ⎟⎠
while they depend on the concentration of the sub-
strate in the second group. The third group includes dt ⎝
the expressions where the notions of the specific rate where X is the biomass concentration, g/L; Xmax is the
of substrate consumption and specific rate of product maximum biomass concentration reached for bacte-

282
 MATHEMATICAL MODELING OF BIOTECHNOLOGICAL PROCESS 283

Table 1. Data on fermentation procedures for producing lactic acid [4–12]

Lactic acid Productivity,
Microorganism Fermentation mode Reference
concentration, g/L g/(L h)
Lactobacillus casei SU no. 22+
Batch 47.0 2.0 4
Lactobacillus lactic WS 1042
Batch 95.7 4.0 5
Enteroccocus faecalis RKY1 Repeated batch with cell recy-
93.2 6.4 5
cling through membrane
Batch 120.0 2.1 6
Lactobacillus rhamnosus ATCC 10863 Continuous with cell recycling
92.0 57.0 6
through membrane
Lactobacillus casei ssp. rhamnosus
Continuous with cell recycling 22.4 9.0 7
ATCC 11443
Lactobacillus delbrueckii NRRL B445 Fed-batch 23.1 0.2 8
Lactococus lactis IO-1 JCM 7638 Batch 39.0 0.9 9
Batch 98.0 1.9 10
Lactobacillus rhamnosus IFO 3863
Continuous 20.0 8.2 10
Continuous with cell recycling
Lactobacillus helveticus CNRZ 303 55.0 7.1 11
through membrane
Lactobacillus delbrueckii CEST 286 Continuous 26.1 10.4 12

ria, g/L; and μmax is the maximum specific growth
rate, h–1.
The appeal of Eq. (4) is that it can be integrated inde-
pendently of the ratio of substrate S and product P.
The solutions for Eq. (4) are described in [15, 18,
20, 21] and take the form
and the equation for the rate of substrate consumption
(3). The designations in Eq. (7) are as follows: P0' is the
constant, h–1, and Pmax is the maximum concentration
of the product, g/L.
Equation (7) has an analytical solution as presented
in [15] and below:

X =
X 0 X max exp (μ maxt )
X max − X 0 + X 0 exp (μ maxt )
, (5)
P =
( ) .
P0Pmax exp P0't

− P + P exp ( P 't )
(8)
X max ⎛X ⎞ Pmax 0 0 0
X = ; c = ln ⎜ max − 1⎟ . (6)
1 + exp ( c − μ maxt ) ⎝ X0 ⎠ The estimation of the constants in Eqs. (3) and (8)
In Eqs. (5) and (6), X0 is the initial concentration of was conducted using the experimental data for cultiva-
biomass, g/L. tion on glucose. The estimates of kinetic constants at
pH 5.4, 6.0, 6.5, and 7.1 with an initial glucose con-
The equations for the rate of substrate consump- centration 20 g/L have been presented. A comparison
tion and product formation are different in different of the experimental results with the data of modeling
publications and their form depends on the used for all the presented examples demonstrated suffi-
strains of microorganisms and types of substrates used ciently good reproducibility. The data for pH 6.0 are
for cultivation. Glucose is a widely used substrate. The presented in Fig. 1 as an example.
list of kinetic models for cultivation with glucose is The fermentation of corn pulp at pH 6.2 by the
presented in [15] (the table is published in [2]). same type of organism was investigated in the same
A model of the process using cultivation of L. amy- study. The cultivation medium contained 45 g/L of
lophilus microorganisms in glucose [15] involves corn pulp. The following results were noted. The
Eq. (4) combined with the equation for the rate of the kinetics of lactic acid production using both substrates
product formation as follow: was similar. The ms value for all experimental results
was insignificant, i.e., 0.05, 0.001, 0.008, and 0.1,
dP = P ' ⎛1 − P ⎞ P (7) which allowed the simplified form of Eq. (3) to be
0 ⎜ ⎟
dt ⎝ Pmax ⎠ used, i.e., ms = 0.

THEORETICAL FOUNDATIONS OF CHEMICAL ENGINEERING Vol. 51 No. 3 2017

284 GORDEEVA et al.

25 Hence, the modeling of the process is conducted

Glucose Lactic acid using Eqs. (6), (12)–(14).
Concentrations, g dm–3

20 The experimental studies were performed using

glucose as a substrate (MRS is the broth, Pronadisa,
15 Lactic acid Hispanlab S. A., Span) at pH 7.0 and T = 30°C.
fermentation
The numeric values of the model parameters are
10 presented in Table 4. The results of modeling are pre-
sented in Figs. 2, 3. It is clear (Figs. 2, 3) that modeling
5 provides adequate estimates of X, P, and S. At the
Biomass
same time, the B values in the time region of t > 10 h
in Fig. 2 and of t > 20 h in Fig. 3 are not even qualita-
0 5 10 15 20 25 30 35 tively close to the values obtained from the model
Time, h equations.
It should be noted from Eq. (9) that the rate of bac-
Fig. 1. Dependences of process indicators on time: S (glu- teriocin production is determined by growing micro-
cose); P (lactic acid); X (biomass) at pH 6.0.
organisms (dX/dt), as well as by the nongrowing
microorganisms (X). By substituting expression (11)
The numeric values of the constants in the kinetic into Eq. (10), we obtain
equations changed with pH (Table 2). The corrected
values of YS/P and YX/S provided that ms = 0 in Eq. (3) dP = − 1 ⎡− 1 dX − m X ⎤
⎢ S ⎥
are presented. The simple form of the equation for the dt Y S P ⎣ Y X S dt ⎦
biomass growth rate (4) and its integral form (6) (15)
= 1 dX + 1 mS X .
allowed analytical expressions to be produced for
modeling the complex process of the simultaneous Y S PY X S dt Y S P
production of lactic acid and bacteriocin using L. lactis
and Pediococcus acidolactici bacterial species [20, 21]. It follows from this that the formation of the prod-
uct occurs due to the growing associates (dX/dt) with
The system of equations in [20] includes solutions coefficient αP = 1/(YS/PYX/S) and due to nongrowing
for the following expressions: associates (X) with coefficient βP = mS/YS/P; hence,
the equation of the rate of biomass formation (4); Eq. (15) can be rewritten as follows:
the equation for bacteriocin (B) formation rate
dP = α dX + β X . (16)
dB = α dX + β X ; P P
B B (9) dt dt
dt dt
The mathematical models considered in the indi-
the equation for lactic acid (P) formation rate cated publications [15, 20, 21] did not find practical
dP = − 1 dS ; applications. The logistic equation that defines the
(10) rate of biomass growth independent of the concentra-
dt Y S P dt tion of substrate and product was used to model the
the equation for the rate of substrate S consump- process of lactic acid production using bacteria of
tion: L. helveticus species [22].
dS = − 1 dX − m X . The equation for the rate of biomass formation is
S (11) presented using designations accepted in the publica-
dt Y X S dt tion with the respective units as follows:
The solutions of system of equations (4), (9)–(11)
are presented in Table 3. dC = μ 0C ⎛1 − C ⎞ . (17)
⎜ C ⎟
dt ⎝ max ⎠

Table 2. Numeric values of constants Despite that C and Cmax are identical to X and Xmax
(biomass concentration and maximum biomass con-
Corrected centration), they are presented in [22] in units of opti-
pH P0' Pmax P0 YS/P YX/S
YS/P YX/S cal density at 620 nm, i.e., OD620.
5.4 0.15 19.8 1.1 0.62 4.16 0.84 2.88 The constant μ0 (in the publication denoted as μ) is
essentially different from μmax in Eq. (4). The authors
6.0 0.29 21.4 0.5 0.62 4.19 0.62 4.16
define the value of μ0 (μ) as the initial specific growth
6.5 0.28 18.5 0.8 0.72 3.45 0.77 3.15
rate, h–1. The variables S and P are concentrations of
7.1 0.21 19.3 0.5 0.48 1.4 0.89 2.0 the substrate and product, g/L.

THEORETICAL FOUNDATIONS OF CHEMICAL ENGINEERING Vol. 51 No. 3 2017

 MATHEMATICAL MODELING OF BIOTECHNOLOGICAL PROCESS 285

Table 3. Solutions for Eq. (4), (9)–(11) [20]

No. Solution

X max ⎛X ⎞
1 X = c −μ maxt
; c = ln ⎜ max − 1⎟ (6)
1+ e ⎝ 0
X ⎠
α B X max
B = − αB X 0
⎛ X max ⎞ −μ maxt
1+⎜ − 1⎟ e
2 ⎝ X0 ⎠ (12)
β X
+ B max ln ⎢
⎡ X0 e μ maxt
(
− 1 + X max ⎤
⎥
)
μ max ⎢ X max ⎥
⎣ ⎦
X0 X max
S = S0 + − 1
YX S YX S ⎛X ⎞
1 + ⎜ max − 1⎟ e −μ maxt
3 ⎝ 0X ⎠ (13)
m X ⎡X 0
− S max ln ⎢
( )
e μ maxt − 1 + X max ⎤
⎥
μ max ⎢ X max ⎥
⎣ ⎦
X0 1 X max
P =− +
Y X SY S Y X SY S ⎛ X ⎞
P P
1 + ⎜ max − 1⎟ e −μ maxt
4 ⎝ X0 ⎠ (14)
m X
+ S max ln ⎢
⎡X 0 ( e μ maxt
)
− 1 + X max ⎤
⎥
Y S P μ max ⎢ X max ⎥
⎣ ⎦

Table 4. Values of parameters for Eqs. (6), (12)–(14)

Parameters
Microorganism species Xmax, μmax, αB, βB,
–1
X0, g/L S0, g/L YX/S, YS/P, ms, h–1
g/L h BU/mL BU/(mL h)
L. lactis 1.171 0.687 0.012 24.082 0.189 21.033 0.139 1.601 0.057
Pedicoccus acidilactici 1.274 0.230 0.060 185.546 0.001 19.929 0.149 1.432 0.086

To model the process, Eq. (17) is supplemented by quantity of the product obtained due to growing and
the equation for the rate of product formation accord- nongrowing associates
ing to the Luedeking–Piret expression as follows [23]:
⎡ ⎤
dP = α dC + β C . ⎢ eμ t
0
⎥
(18) Pm = α PC 0 ⎢ ⎥; (20)
( )
P P
dt dt
⎢1 − C 0 1 − e μ t
0
⎥
Here, αP, g of product /L OD620; βP, g of product /L ⎣⎢ C max ⎦⎥
OD620 h.
C max ⎡ C ⎤
(
β P ⎢1 − 0 1 − e μ t ⎥ , )
0
The consumption of substrate was presented by Pn = ln (21)
authors in the same manner as follows: μ 0
⎣ C max ⎦
dS = −α dC − β C . where Pm is the amount of lactic acid synthesized by
S S (19) the growing associates, g/L, and Pn is the amount of
dt dt lactic acid synthesized by nongrowing associates, g/L.
Here, αS, g of substrate(L OD620; βS, g of substrate/L Hence, the mathematical model allows one to not
OD620 h. only calculate the dependence of lactic acid concen-
The feature of the suggested mathematical model is tration P on fermentation time, but also the compo-
that, using the possibility to analytically solve Eq. (17), nents Pm and Pn generated by the growing and
the authors obtained separate solutions to estimate the nongrowing associates, which are difficult to evaluate

THEORETICAL FOUNDATIONS OF CHEMICAL ENGINEERING Vol. 51 No. 3 2017

286 GORDEEVA et al.

36 25 2.0 8
30
1.5 6

BT, BU/mL
24 20

L, g/L
X, g/L
G, g/L
18 1.0 4
12 15
0.5 2
6
0 0 0
0 5 10 15 20 25
t, h

Fig. 2. Dependences of process indicators on time for L. lactis: j indicates X; d indicates B; h indicates P(L); s indicates S(G).

250 25 2.0 9

200 20
1.5
BT, BU/mL

6
150 15

L, g/L
X, g/L
G, g/L

1.0
100 10
3
50 0.5
5

0 0 0
0 5 10 10 15 20 25
t, h

Fig. 3. Dependences of process indicators on time for Pedicoccus acidilactici: j indicates X; d indicates B; h indicates P(L);
s indicates S(G).

experimentally. Experimental studies were conducted conclusion is quite important for understanding the
with three types of substrate, i.e., glucose synthetic biological nature of synthesis. However, a noticeable
(GS), lactose synthetic (LS), and whey-yeast extract deviation of the calculations from the experimental
permeate (WYEP). An example of the modeling results is observed in the last version with prolonged
results is presented in Fig. 4. Calculations were carried fermentation (more than 18 h).
out using the data presented in Table 5 for T = 42°C
and pH 5.9. In order to take into account the effects of biomass
growth rate inhibition, the following modified logistic
Let us highlight the following important results of equation was used [16]:
modeling. The formation of lactic acid in GS and LS
media takes place predominantly due to the growing f h
associates. At the same time, in the case of WYEP, the dX = μ ⎛1 − X ⎞ ⎛1 − P ⎞ . (22)
max ⎜
contribution of growing associates in the process dt ⎝ X max ⎟⎠ ⎜⎝ Pmax ⎟⎠
remain constant for fermentation times longer than
approximately 12 h, and the increase in the concentra- Here, f defines the degree of inhibition by biomass and
tion of P occurs due to nongrowing associates. This h is the degree of inhibition by product.

Table 5. Numeric values of constants in Eqs. (17)–(21)

αP, βP, αS, βS,
Cultivation medium μ0, h–1
g/(l OD) g/(l OD h) g/(l OD) g/(l OD h)
GS 0.310 8.407 0.458 4.344 1.117
LS 0.479 11.460 0.881 8.257 1.269
WYEP 0.435 5.376 1.356 11.304 1.205

THEORETICAL FOUNDATIONS OF CHEMICAL ENGINEERING Vol. 51 No. 3 2017

 MATHEMATICAL MODELING OF BIOTECHNOLOGICAL PROCESS 287

50 50
GS WYEP
P P
40 Pm 40 Pm
Lactic acid, g/L

Lactic acid, g/L

Pn Pn
30 30

20 20

10 10

0 5 10 15 20 25 0 5 10 15 20 25
Fermentation time, h Fermentation time, h
50
P LS
Pm
40 Pn
Lactic acid, g/L

30

20

10

0 5 10 15 20 25
Fermentation time, h

Fig. 4. Results of modeling process with experimental data.

The following equations for the rate of formation of provided good agreement with the experiment for all
product P (2) and substrate S consumption were the trials. An example is presented for S0 = 77.1 g/L in
added to Eq. (22): Fig. 5. The largest deviation of the experimental and
calculated data was observed for prolonged fermenta-
dS = − 1 dP − m X . (23)
S tion of more than 10 h.
dt Y P S dt
The mathematical models with equations for the
By substituting (2) into (23), we obtain specific growth rate that include the substrate concen-
tration S and the product concentration P are most
dS = −α dX − β Х − m X , (24)
S S S
dt dt
β
Lactose and Lactic acid, g dm–3

where α S = − α ; β S = . 90 8
YP S YP S 80 7
Blomass, g dm–3

Equations of the mathematical model were evalu- 70 6

ated using whey as a fermentation substrate with 60 5
L. casei bacteria. The initial concentration of the sub- 50
4
strate S0 was 9.0, 21.4, 35.5, 48.1, 61.2, and 77.1 g/L. 40
30 3
The following estimates of the constants were
obtained as a result of processing the experimental 20 2
10 1
data: μmax = 0.265 h–1; α = 0.029S 0* + 2.686; β = 0
0.06 h–1; mS = 0.03 h–1; YP/S = 0.682; and S 0* = S 0, g/L. 0 3 6 9 12 15
Time, h
The exponent values for f and h changed as follows:
f(S0 = 9.0 g/L) = 0.1 and f(S0 = 77.1 g/L) = 0.7; for the
remaining values of S0, f = 0.5. For h, h(S0 = 9.0 g/L) = Fig. 5. Calculations according to the model (curves) and exper-
iment: + indicates biomass concentration, g/L; d indicates
0.3 and, for the remaining values of S0, h = 0.5. The product concentration, g/L; s indicates substrate concentra-
values f = h = 0.5 were used for calculations. Modeling tion, g/L. Initial substrate concentration 77.1 g/L.

THEORETICAL FOUNDATIONS OF CHEMICAL ENGINEERING Vol. 51 No. 3 2017

288 GORDEEVA et al.
Biomass/time, y(t), g/l h–1 0.4 The following constant values were obtained as a result
3% L0
of modeling: k1 = 0.556 h–1, KS = 0.0956 g/L, KP =
0.3 5% 68.8 g/L, YP = 0.88 g/L, k4 = 2.678 g/L, and k5 = 2.5 h–1.
4% The authors demonstrated an example of the prac-
0.2 2% tical application of the model equations. The notion of
biomass productivity ν was used, which was defined by
1% the following expression:
0.1
0.2%
0.5% ν (t ) = X (t ) t , (27)
0 where t is the time of fermentation, h, and ν(t) is the
2 4 6 8 10 12 14 productivity up to the time point t, g/(L h).
Time(t), h The results of estimating ν(t) from the model equa-
tions are presented in Fig. 6. The maximum produc-
Fig. 6. Dependence of biomass productivity on fermenta- tivity was obtained for an initial lactose concentration
tion time at different initial lactose concentrations at pH of 3% for a fermentation time of approximately t = 5 h.
6.0 and temperature 30°C. The consideration for inhibition by the substrate and
product are included in the equations [25].
often used to model processes with different species of However, contrary to the work reported in [24],
microorganisms and different substrates. A large num- inhibition by the substrate and product was only
ber of these expressions are available [2]. Moreover, related to the rate of biomass formation. This equation
the possibility often exists to select a different equation was presented as follows:
and obtain satisfactory results by fitting the process n1 n2
with different numerical values of coefficients. For dX = μ ⎛1 − S ⎞ ⎛1 − P ⎞ . (28)
max ⎜ ⎟ ⎜ ⎟
example, six different equations (three for the rate of dt ⎝ S max ⎠ ⎝ Pmax ⎠
product (lactic acid) formation and three for the rate
of lactose consumption) were tested in [24] when the The terms that define the level of inhibition include
process was modeled using the cultivation of Strepto- n1, n2, and Smax (maximum concentration of the sub-
coccus cremoris HP in lactose medium with a concen- strate above which the growth rate equals zero, g/L),
tration in the range of 0.2–5.0%. and Pmax (maximum concentration of the product,
above which the growth rate equals zero, g/L).
The authors preferred the following equations. The The equation for the rate of substrate consumption
equation for the rate of biomass formation (1) is as follows:
⎛ ⎞⎛ ⎞
μ = k1 ⎜ S ⎟ ⎜ P ⎟ , dS = − 1 dP . (29)
⎝KS + S ⎠⎝KP + P ⎠ dt Y P dt
where k1 is the constant, h–1; KS is the constant of sub- Equation (2) represents the rate of product forma-
strate limitation, g/g; and KP is the constant of product tion. Hence, the effect of inhibition is reflected on
inhibition, g/L. both dP and dS via dX .
The equation for the rate of product formation: dt dt dt
Evaluation of the constants in Eqs. (28), (29), (2)
was conducted using the experimental data obtained
dP = k dX + k ⎛ S ⎞ X ,
4 5⎜ ⎟ (25) with the L. delbrueckii NRRL B445 microorganism
⎝KS + S ⎠
'
dt dt cultivated with initial glucose concentrations of 5–
33%. Numeric values of the constants are as follows:
where k4, k5 are the constants (k4 dimensionless, k5 – h–1)
μmax = 0.58 h–1, Smax = 401.8 g/L, Pmax = 81.0 g/L,
and K S' is the constant of substrate inhibition, g/L. n1 = 0.71, n2 = 21, α = 2.36, and β = 0.816 h–1. The
authors emphasize satisfactory results of modeling
The equation of the rate of substrate consumption over the entire range of the initial concentrations of S0. A
is as follows: comparison of the experimental data with the curves cal-
culated using the model with the initial substrate concen-
− dL = 1 dP , (26) tration of 10% is presented in Fig. 7. It can be seen from
dt Y P dt Fig. 7 that the biomass growth ceases after 16 h of fer-
Here, L is the lactose concentration, g/L; S (L-LT) is mentation, and the increase in the lactic acid concentra-
lactose limitation; LT is the residual lactose concen- tion occurs due to the nongrowing associates.
tration, g/L; YP is the yield, in grams of lactic acid per Modeling of the process of glucose fermentation
gram of substrate. with microorganisms of the L. delbrueckii species was

THEORETICAL FOUNDATIONS OF CHEMICAL ENGINEERING Vol. 51 No. 3 2017

 MATHEMATICAL MODELING OF BIOTECHNOLOGICAL PROCESS 289
Clucose, cells, Lactic acid, g/L 110 Although it must be mentioned that processing the
100 experimental data according to [24] produced an
90 insignificant value of ms in (23), and it was not
80 reflected in the modeling results. A comparison of the
70 solutions for Eqs. (30), (2), and (23) with the experi-
60
mental data showed fairly good agreement for all pH
values. The numeric values of the constants are pre-
50 sented in Table 6 using the units used in [26].
40
30 The modeling of the process that used microorgan-
isms of the L. delbrueckii species was reported in [27]. A
20
feature of the expressions used for modeling was in the
10 fact that the authors introduced the effect of the inhibi-
0 4 8 12 16 20
tion of the product directly in the equation for the rate of
Time, h
product formation in addition to the effect of product
inhibition in the equation for the growth rate.
Fig. 7. Comparison of experimental data with calculations
The equations for this model are as follows:
according to the model: h indicates cells; e indicates glu-
cose; × indicates lactic acid. dX = μ ⎛ S ⎞ ⎛1 − P ⎞ X , (31)
max ⎜ ⎟⎜ ⎟
dt ⎝ K S + S ⎠ ⎝ Pmax ⎠

(
dP = α dX + β X ⎛⎜1 − P ⎞⎟
)
70
60 (32)
20 dt dt ⎜ P' ⎟
50 ⎝ max ⎠
P or S, g dm–3
X, g dm–3

40 and the equation for the rate of substrate consumption (3).

10
30 Here, Pmax is maximum concentration of the prod-
20 ' is
uct above which bacteria do not grow, g/L, and Pmax
10 the maximum concentration of the product above
0 which bacteria do not produce lactic acid. Hence,
0 nongrowing bacteria (at P > Pmax) can produce lactic
0 5 10 15 20 25 30 ' . Obviously, this conclusion is import-
acid at P < Pmax
Time, h ant for the practical application of the model equa-
tions.
Fig. 8. Mathematical modeling of the process: d indicates A comparison of the results of process modeling
B; s indicates substrate (succrose); h indicates lactic acid;
j indicates biomass.
with the experimental data is presented in Fig. 8. The
values of constants in the models are presented in
Table 7. The following can be concluded from the
conducted for pH 5.3, 5.5, 6.0 at 45°C with an initial results of modeling. The biomass growth ceases when
glucose concentration of 65 g/L [26]. P = Pmax = 45 g/L (the level predicted by the model);
the ability of bacteria to produce lactic acid was
The equation of the mathematical model are
arrested when the concentration reached P = Pmax ' =
n
dX = μ ⎛ S ⎞ ⎛1 − P ⎞ X , (30)
57 g/L. Hence, lactic acid was produced by nongrow-
max ⎜ ⎟⎜ ⎟ ing associates at P = 45–57 g/L.
dt ⎝ K S + S ⎠ ⎝ Pmax ⎠
The mathematical model created using the experi-
the equation of the rate of product formation (2), and the ments with microorganisms of the L. bulgaricus spe-
equation for the rate of substrate consumption (23). cies cultivated with three types of substrates (glucose, lac-

Table 6. Numeric constant values

pH μmax, 1/h 1/Pmax, mL/g Ks, g/mL α, g/A610u β, g/A610u n m, g/mL

5.30 0.136 22.5 0.0457 0.5960 0.003 0.88 3.4E-10

5.50 0.0696 30.23 0.0967 0.3853 0.0032 0.88 1.14E-04
6.00 0.0946 29.63 0.0358 0.9461 0.0173 0.88 2.56E-05

THEORETICAL FOUNDATIONS OF CHEMICAL ENGINEERING Vol. 51 No. 3 2017

290 GORDEEVA et al.

Table 7. Numeric constant values

μ = μ max S ⎛ K P ⎞;
⎜ ⎟ (33)
Parameter Value KS + S ⎝P + K p ⎠
A, g of lactic acid/g of biomass 0.235
μ = μ max S ⎛1 − P ⎞ .
B, g of lactic acid /h 0.087 ⎜ ⎟ (34)
mS, g of sucrose /(g of biomass h) 0.090 KS + S ⎝ K p ⎠
μmax, 1/h 0.831 The better results were obtained using the following
simplified version of Eq. (34) for the high initial sub-
YX/S, g of cells/g of sucrose 0.270
strate concentration of 43 g/L:
YP/S, g of lactic acid /g of sucrose 0.910
⎛ ⎞
μ = μ max ⎜1 − P ⎟ . (35)
⎝ Kp⎠
tose, and galactose) was tested in [28]. Equations of this
mathematical model are (1)–(3). In Eq. (3), ms = 0. The following values of the constants in Eqs. (1)–
(3) were obtained from the results of experiments: YX/S =
The authors concluded that the satisfactory (excel- 0.1, YP/S = 0.9, α = 9.0, KS = 3.36 g/L, and μmax =
lent) result of modeling were obtained for specific 1.14 h–1. The KP value was obtained from Eq. (35),
growth rates at initial lactose concentrations of 26 and
43 g/L as follows: which was 40.3 g/L. A comparison of the experimental
results with the solution of the mathematical model is
presented in Fig. 9.
30 3.0 A more complicated mathematical model of the
process using microorganisms of the Lactococcus lactis
25 2.5 ssp. lactis ATCC 19435 species cultivated in whole-
Concentration, g/L

wheat-flour-based medium was presented in [29] and

Cell Density, g/L

20 2.0 using microorganisms of the L. coriniformis subsp.

torquens DSM 20004 species (preparation of the strain
15 1.5 described in the paper) in [30]. The complexity of this
model lies in the need to introduce an equation that
10 1.0 takes into consideration formation of side products.
The model considers substrate and product inhibi-
5 0.5 tion, as well as the effect of pH and temperature.
The use of the whole-wheat flour-based medium
0 required certain substrate pretreatment, despite the
0 1 2 3 4 5 6 7 fact that this medium contained all components nec-
Time, h essary for fermentation. That is why the substrate for
this process was prepared by the pretreatment of the
Fig. 9. Comparison of modeling results of the batch pro- initial medium. The mechanism of the formation of
cess with L. bulgaricus and initial substrate (lactose) con- medium that facilitates fermentation is presented in
centration Sf = 26 g/L at pH 5.6 and temperature 42 ± 1°C.
Equation (33) was used for modeling. Fig. 10. As was mentioned by the authors of [29], the
process depended heavily on conditions such as tem-
perature and pH.
Liquefaction SSPHF In the general case, the formation of side products
(acetic acid, formic acid, ethanol), as well as of D-lac-
SSPH tic acid, is possible during fermentation. These pro-
Lactic cesses accelerate at temperatures above 30°C and pH
Starch Maltose Glucose below 6. The equations of the mathematical model
AMG Lactic Acid Acid
α-amylase Bacteria that take into consideration these processes are as fol-
lows [29]:
The equation of biomass growth rate
Starch Amino Bacteria K iS
2(
rx = μ max 1 − K pP ) ,
Protease n
proteines acids Growth (36)
K sK i + K iS + S
where Ki is the constant of substrate inhibition, g/L; Ks
Fig. 10. Schematic presentation of the process. Liquefac- is the saturation constant, g/L; Kp is the parameter
tion, simultaneous saccharification and protein hydrolysis
(SSPH) and simultaneous transformation into sugar, pro-
presenting pH in the product inhibition, g/L; and n is
tein hydrolysis, and fermentation (SSPHF). Starch, Malt- the dimensionless parameter used to decrease the
ose, Glucose, Protein, Lactic acid bacteria, Lactic acid. effect of product inhibition.

THEORETICAL FOUNDATIONS OF CHEMICAL ENGINEERING Vol. 51 No. 3 2017

 MATHEMATICAL MODELING OF BIOTECHNOLOGICAL PROCESS 291

Table 8. Numeric values of constants in Eqs. (47)–(52)

Parameter μmax, h–1 Ks, g/L Ki, g/L Kp, l/g n α β, h–1 m, h–1
Value 0.403 0.79 164 1.60 × 10–2 2.06 13.2 6.45 × 10–2 3.7 × 10–3

The equation of the rate of product (lactic acid) The values for A1, A2, E1, and E2 are presented in
formation rp and side products rap: Table 10.
r p = α rx + β X , (37) The temperature in Eq. (44) is calculated using the
equation
rap = α a rx + β a X , (38)
T = ( 273 ° + T1 ) − 300 °, (45)
where α is dimensionless constant of product forma-
tion due to the growing associates; β is the constant of where T1 is temperature in degrees Celsius at which
product formation due to the nongrowing associates, the process is realized; hence, for 37°C, we obtain T =
h–1; αа is the dimensionless constant of formation of side 10 K.
products due to the growing associates; and βа is the The values E1 and E2 are in J/mol, R are in J/(mol K).
constant of formation of side products due to the non- The values μmax and Kp are calculated according to
glowing associates, h–1. (42) and (43).
The Pap concentration (g/L) combines all side The process of producing lactic acid from wheat
products in the expression flour in continuous mode was considered in [30].
dPap The equation of biomass formation was the same as
rap =. (39) (30), which next allowed simplification at Ks ! S and,
dt consequently, the equation was
The equation of the rate of substrate consumption n
includes components rp, rap, rx, rst, and ms, where rst is dX = μ ⎛1 − P ⎞ X . (46)
the rate of maltose degradation with its value rst calcu- max ⎜ ⎟
dt ⎝ Pmax ⎠
lated with the equation
The equation for the product was
rst = −k st S. (40)
dP = Y P MX . (47)
The equation for the rate of substrate consumption
rS is dt Y X

rS = − 1 rP − 1 rap − 1 rx − rst − ms X . (41) Table 9. Numeric values of constants in Eqs. (42), (43)
YP Y aP YS
μmax Kp
The side products are not formed in medium based
on hydrolyzed wheat flour. μm, h–1 1.1 Kpm 0.154
The numeric values of the constants were obtained
during modeling based on experimental data with an kμ1 9.42 × 10–7 kp1 1.02 × 10–5
initial glucose concentration of 44–182 g/L, tempera- kμ2 1.27 × 105 kp2 3.82 × 10 4
ture of 30–40°C, and pH 6.0, 5.0, 4.0. The data for
standard conditions (pH 6.0, 30°C) are presented in
Table 8.
The effect of pH was modeled by the equations [31] Table 10. Numeric values of indicators A1, A2, E1, E2 for
calculations of parameters in (36)–(41)
μm
μ max = (42) A1 A2 E1 E2
( )
; Parameter
1 + k μ1 [H + ] + k μ2[H + ]
α 287 8.88 × 103 76.9 376
K pm β, h–1
Kp = (43) 1.77 82.4
( )
+ +
,
1 + k p1 [H ] + k p2[H ] αa 2.88 53.9
with numeric values of the constants taken from Table 9. βa, h–1 2.97 × 10–2 543
The value of [H+] = 10–pH. kst, h–1 7.52 × 10–2 0.59 151 343
The effect of temperature on some parameters of
model equations (36)–(41): Ya 169
Yp 1.0
parameter = A1 exp ⎛⎜ − 1 ⎞⎟ − A2 exp ⎛⎜ − 2 ⎞⎟ . (44)
E E
Yx 0.79
⎝ RT ⎠ ⎝ RT ⎠

THEORETICAL FOUNDATIONS OF CHEMICAL ENGINEERING Vol. 51 No. 3 2017

292 GORDEEVA et al.

Table 11. Numeric values of constants in Eqs. (30), The mathematical models are based on the use of
(46)–(49) notions of the specific rate of substrate consumption
μmax, Pmax, Km, K s, and the specific rate of product formation in addition
Parameter Yx, Yp, n to the notion of the specific growth rate. These notions
h−1 g/L g/L g/L
are introduced in [32–34] as follows:
Value 0.28 98.6 0.053 0.82 0.035 0.5 3 μ = [μ max S ( K SM + S )] [K PM ( K PM + P )] , (50)
ν = [ν max S ( K S ν + S )] [K P ν( K P ν + P )] .
(51)
The equation for the substrate includes a compo-
nent of the substrate formation due to maltose degra- The specific rate of product formation π is as fol-
dation is as follows: lows:
π = [π max S ( K S π + S )] [K P π ( K P π + P )] . (52)
dS = − 1 μ X + K M .
M (48) Although the equations for ν and π are seemingly
dt YX independent, it has been obtained by using (50)–(52)
The equation of maltose degradation is as shown below:
1 = y ⎛1⎞ + δ , (53)
dM = −K M . ν⎜ ⎟ ν
M (49) ν ⎝μ⎠
dt
1 = y ⎛1⎞ + δ . (54)
Hence, unlike in [29], the formation of a side prod- π⎜ ⎟ π
uct was not observed in this study. The numeric values π ⎝μ⎠
of the constants obtained from the experimental data Hence, in fact, ν and π are calculated via the spe-
are presented in Table 11. cific growth rate with the coefficients as follows:

yν = μ max K P ν ν max K P ν (g cells g glucose) ⎫

δν = ( K P ν − K P μ ) ν max K P ν (g cells h g glucose) ⎪⎪
⎬. (55)
yπ = μ max K P ν π max K P π (g cells g lactic acid) ⎪
νπ = ( K P π − K P μ ) π max K P π (g cells h g lactic acid)⎭⎪

Taking into consideration (50)–(52), the equation

15 1.5 of mathematical model for batch fermentation are as
follows:
dX = μ X ; dS = ν X ; dP = πX . (56)
dt dt dt
Experimental studies of the batch process were
P, g lactic acid/L
S, g glucose/L

10 1.0 conducted with Streptococcus faecalis microorganisms

X, g cell/L

5 0.5
0 10 20
Time, h
Fig. 11. Comparison of experimental results with calcula-
tions at T = 37°C and pH 7.0: s indicates lactic acid; h
indicates glucose; n indicates cells.
cultivated in glucose.
A comparison of the experimental results with the
calculations is presented in Fig. 11 [34]. The numeric
values of the constants obtained from the resulting
chemostatic cultivation are presented in Table 12 [33].
The use of notions of specific rates of growth, sub-
strate consumption, and product formation was
reported in [35, 36].
The difference of these publications from [32–34]
lies in the fact that the specific rates of product forma-
tion and substrate consumption are not expressed
using the specific growth rate.
The rate of biomass formation is as follows:
dX = μ S K ix ⎛ P − Pix ⎞
1− X. (57)
K sx + S K ix + S ⎜⎝ Pmx − Pix ⎟⎠
max
dt

THEORETICAL FOUNDATIONS OF CHEMICAL ENGINEERING Vol. 51 No. 3 2017

 MATHEMATICAL MODELING OF BIOTECHNOLOGICAL PROCESS 293

Table 12. Numeric values of constants

Parameter Ksμ Ksν Ksπ Kpμ Kpν Kpπ μm νm πm
Value 0.22 0.22 0.22 9.5 11.0 5.5 1.6 8.3 11.8

The rate of substrate consumption is of the mortality constant Kd = 0.00318 h–1 is rather low
and its effect can be ignored.
d S = −q S K is ⎛ P − Pis ⎞
1− X . (58)
K ss + S K is + S ⎝ Pms − Pis ⎟⎠
⎜
s max
dt The values of the coefficients vary because they are
determined from different kinetic models. However,
The rate of product formation is the fact that the maximum specific rate qs max equals
dP = α dX + q S 3.33 according to [36] and 3.42 according to [35] have
K sp + S
p max engaged our attention. The maximum specific rates of
dt dt
(59) product formation qр max are nearly the same, i.e., 3.00
K ip ⎛ P − Pip ⎞
× ⎜1 − ⎟ X , [36] and 3.02 [35]; the maximum specific rates of
K ip + S ⎝ Pmp − Pip ⎠ growth μp, max differ by approximately 32%.
where μpmax, qsmax, and qpmax are maximum specific Numeric values of the constants in Eqs. (57)–(59),
rates of biomass growth, h–1; substrate consumption, which are absent in Table 13, are Pmx = 49.9 g/L, Pms =
g/(g h); and product formation, g/(g h). Pmр = 95.5 g/L.
Let us note that the specific rate of lactic acid for-
mation is determined by both the specific rate of bio-
mass formation and product formation simultane- PRACTICAL APPLICATION
ously. The mathematical models in [36] differ from OF MODELING RESULTS
Eqs. (57)–(59) by introducing the coefficient of the
bacteria mortality Kd, as well as the exponential Data on the practical applications of mathematical
dependence is used to account for the inhibition of the models have been used in [16, 37, 38] for calculating
product instead of the linear dependence. The equa- the productivity indicator for the target component
tions are given as follows: (lactic acid) Qp.
The specific growth rate is
S K ix −P
μ = μ max
K px
e . (60) Table 13. Values of optimal parameters obtained in [35, 36]
K sx + S K ix + S
Kinetic parameters Reference [36] Reference [35]
The rate of biomass formation is
Model of biomass production
d X = (μ − K )X . (61)
dt
d
μmax, h–1 1.6 1.1
The rate of substrate consumption is Kiх, g/L 167.46 304
d S = −q S K is −P K ps Ksх, g/L 0.89 1.32
e X. (62)
K ss + S K is + S
s max
dt Kpх, g/L 17.07 –
The rate of product formation is Kd, h–1 0.00318 –
dP = α dX + q S K ip −P K pp Model of sugar processing
e X. (63)
K sp + S K ip + S
p max
dt dt Kis, g/L 303.17 140
Equations (57)–(59) contain 16 constants, while Kss, g/L 0.1 2.05
Eqs. (60)–(63) contain 14 constants. The correlation
of Eqs. (57)–(63) with this large number of constants Kps, g/L 29.17 –
to the experimental data is quite high. qs max, h–1 3.33 3.42
The authors introduce a number of assumptions
(simplifications) during the processing of experiemt- Model of lactic acid production
nal data, such as Kip, g/L 303.17 140
Kss = Ksp, Pms = Pmp, Pis = Pip, Kis = Kip [35] Kрp, g/L 29.17 –
and Kss = Ksp, Kis = Kip, Kps = Kpp [36].
qр max, h–1 3.00 3.02
The resulting estimates of the parameters are pre-
sented in Table 13. The authors [36] note that the value α, g/g 0.26 0.39

THEORETICAL FOUNDATIONS OF CHEMICAL ENGINEERING Vol. 51 No. 3 2017

294 GORDEEVA et al.

Qp, g/(L h) the calculations are presented in Fig. 12. Two practical
3.0 conclusions follow from Fig. 12. The model equations
make it possible to predict optimal conditions
2.5 (maxQp), for which the initial concentration equals
35.5 g/L. The second conclusion lies in the fact that
2.0
the same productivity can be obtained for two differ-
1.5 ent initial levels of Sf. Based on the figure, these values
1.0 are approximately S 1f = 21.4 g/L and S 2f = 61.2 g/L.
The same results are obtained in [37] and presented in
0.5 Fig. 13.
0 The time of fermentation that provides the Qp value
9.0 21.4 35.5 48.1 61.2 77.1 is also presented in the figure. Hence, the same pro-
Sf , g/L ductivity can be obtained for two different Sf values,
but reaches the same Qp over a different time period.
Fig. 12. Dependence of productivity Qp, g/(L h) on initial Figure 13 was obtained using Eqs. (1)–(3) of the
substrate concentration Sf, g/L. mathematical model. In (3) mS = 0 and dP/dt = 0. The
values of μ in (1) and (2) are found as follows:
⎛ ⎞
Qp, g/(L h) μ = μ max ⎜1 − P ⎟ S . (65)
⎝ Pmax ⎠ K S + S + S K i
2
2.0
The numeric values of constants in (1)–(3) and
(65) are presented in Table 14 [35]. Note that the pos-
(29.85) (29.85)
(30.45)
sibilities of applying mathematical models have been
(26.11) presented in the most detail in [38]. A description of
(31.43)
(23.94) (31.99) three algorithms for the realization of mathematical
1.5 (35.0) models has been provided.
(20.0)
(37.62) Algorithm 1 includes estimates of the process indi-
(17.15)
cators (dependences of concentrations on fermenta-
tion time); estimates of the time of process completion
tk, after which there is no product synthesis; and the
1.0 dependence of productivity and process indicators at
10 20 30 40 60 t = tk (Fig. 13 was obtained according to this algo-
Sf , g/L rithm).
Algorithm 2 includes calculations of the initial sub-
Fig. 13. Dependence of productivity for the target product strate concentration Sf, at which the maxQp is reached
P per unit of the reactor volume on the initial substrate and time of reaching this maximum topt, i.e., the opti-
concentration Sf. Time of fermetation (h) is presented in
brackets. mization problem is solved.
Algorithm 3 includes obtaining estimates of Sf, for
The value of productivity is calculated as the ratio which the same productivity Qp, S 1f , and S 2f can be
of the running concentration P (g/L) to the running reached, as well as the time of reaching this Qp value.
time point t (h), i.e., At the same time, all of the process indicators for
S 1f and S 2f are estimated.
Qp = P . (64) The results of numeric calculations with taking into
t consideration the technical limitations and ensuring
Using the mathematical model equations (20)– the possibility of realization the obtained solutions
(24), the authors in [16] obtained the dependence of have been presented for all algorithms.
productivity for the initial substrate concentrations The practical results of modeling nonstationary
9.0, 21.4, 35.5, 48.1, 61.2, and 77.1 g/L. The results of fermentation were obtained in [39, 40].

Table 14. Numeric values of constants in Eqs. (1)–(3) and (63)

YX/S, g/L α, g/g β, h–1 μmax, h–1 Pmax, g/L Km, g/L Ki, g/L

0.4 2.2 0.2 0.48 50 1.2 22

THEORETICAL FOUNDATIONS OF CHEMICAL ENGINEERING Vol. 51 No. 3 2017

 MATHEMATICAL MODELING OF BIOTECHNOLOGICAL PROCESS 295

δ, g/L process under perturbation that is in the nonstationary

0.4 state, which, to a certain degree, makes this regime
closer to the periodic batch fermentation.
0.3 The practical usefulness of modeling lies in the fact
that it is almost impossible to assess and predict stabil-
δ3 ity experimentally.
0.2
The following mathematical model equations are
presented in [39] (based on kintetic expressions [35]):
0.1 δ1
dX = μ K ix S
K sx K ix + ( K sx + K ix ) S + S
max 2
0 dt
5 10 15 (66)
t, h P −P
0.1 × mx X − DX ,
Pmx − Pix
δ2
0.2
( f
dS = D S − S − q
) s max
dt
0.3 K is S Pms − P (67)
× X,
K ss K is + ( K ss + K is ) S + S ms − Pis
2
P
0.4
dP = αμ K ix S
K sx K ix + ( K sx + K ix ) S + S 2
max
Fig. 14. Transient process in the vicinity of stationary state dt
1: δ1, δ 2, δ 3 are values of deviation from the value of sta- P −P
tionary state for X, S, and P, respectively; × mx X +
0 0 0
Pmx − Pis
δ1 = δ 2 = δ3 = 0.1 are initial deviations. (68)
K ipS
+ q p max
K sp K ip + ( K sp + K ip ) S + S 2
The publications [39, 40] have been devoted to
P −P
modeling continous process. However, estimating the × mp X − DP.
stability of stationary states requires considering the Pmp − Pip

δ3 × 10–3, g/L
1

10
3 δ1 × 10–3, g/L
δ2 × 10–3, g/L
5
2
10 10
5 5
2*
–5
–10 –5 –10

–10 3*

1*

Fig. 15. Phase trajectories in the transient process: (1) δ10 = 0.01; δ 02 = δ30 = 0.0; (1*) δ10 = − 0.01;δ 02 = δ30 = 0.0; (2) δ30 = 0.01;
δ10 = δ 20 = 0.0; (2*) δ30 = − 0.01; δ10 = δ 20 = 0.0; (3) δ 02 = 0.01; δ10 = δ30 = 0.0; (3*) δ 02 = − 0.01;δ10 = δ30 = 0.0.

THEORETICAL FOUNDATIONS OF CHEMICAL ENGINEERING Vol. 51 No. 3 2017

296 GORDEEVA et al.
The mathematical model equations in [39] are as
follows (based on kinetic expressions in [37]):
dX = −DX + μ X ,
(69)
dt
( f )
dS = D S − S − 1 μ X , (70)
dt YX S
(
dP = −DP + αμ + β X ,
) (71)
dt
where D = Q/V, h–1; V is the volume of the reactor, m3;
Q is the volumetric rate of feeding m3/h; μ is the spe-
cific biomass growth rate, h–1; YX/S is the stoichomet-
ric coefficient, g/g; X, S, P are the concentration of the
biomass, substrate, and product, respectively, at the
reactor output, g/L; Sf is the concentration of sub-
strate in the reactor input flow, g/L; and α, β are the
constants.
Note that Eqs. (69)–(71) were used to assess the
stability of the process in an optimal state, i.e., at Qp =
max, where Qp was calculated as a product of the flow
value D to the concentration of the product P (lactic
acid):
Qp = DP. (72)
The stability theory for ordinary differential equa-
tions was used for investigation [41, 42].
The expression for estimating the stability indica-
tors were taken according to Lyapunov [42, 43] as sta-
bility in small; hence, the system of linearized equa-
tions was obtained from (66)–(68) [39] and (69)–(71)
[40]. The characteristic equations were produced, i.e.,
algebraic cubic equations with constant coefficients,
and the Gurvits equations were derived from their
matrix. The coefficients of characteristic equations
were calculated using the data from Tables 12 and 13.
The results of modeling the transient process in
[39] were obtained for three stationary states.
The example for stationary state 1 is shown in
Fig. 14. The numeric indicators of the stationary state
1 are D = 0.5 h–1, X = 3.37 g/L, S = 20.032 g/L, P =
24.9 g/L, Sf = 46.8 g/L, and maxQp = 12.42 g/(L h).
The values of the stationary state indicators were
obtained as a result of solving of the stationary problem.
The results of modeling using Eqs. (69)–(71)
according to [41] are presented in Fig. 15. Indicators of
the stationary state under optimal conditions are
S opt
f = 23.4 g/L, Dopt = 0.1636 h–1, Sopt = 5.138 g/L,
Xopt = 7.304 g/L, Popt = 25.0 g/L, and QPopt = 4.09 g/(L h).
Analytical solutions were also obtained for the depen-
dence of deviations δ1, δ 2 , and δ 3 on time t, from
which the three-dimensional phase portrait was
obtained (Fig. 15), comprising a stable knot. The indi-
cators were calculated using data from Table 13.
THEORETICAL FOUNDATIONS OF CHEMICAL ENGINEERING
CONCLUSIONS
The analysis of publications devoted to the mathe-
matical modeling of the batch fermentaion process to
produce lactic acid revealed the following. Similar to
kinetic modeling [2], the nonstructured approach
forms a limited number of mathematical equations
that describe the process of synthesis. In the majority
of publications, this is a system of three ordinary first-
order differential equations, generally with variable
coefficients. The mathematical model contains four
ordinary first-order differential equations in the case
when a side product is synthesized in addition to the
main product, i.e., lactic acid.
Next, considering that microbial mass produces
lactic acid, in the majority of models, the rate of lactic
acid formation is determined by the rate of biomass
growth. The notions of specific rates of product for-
mation and substrate consumption are used in some
publications. When the inhibiting effects in the prod-
uct formation were observed, the indicators of bio-
mass, substrate, and product inhibition were included
in the equations. As a rule, the inhibition effects were
introduced into the equation for specific rates.
Modeling of the processes of continous fermenta-
tion has not been so common. It is likely related to the
fact that the continous fermentation attracts much less
attention from the practical point of view. A small
review of publications in Russian journals on model-
ing the process of lactic acid production using pre-
dominatly continous fermentation has been presented
in [43].
The main directions for the use of mathematical
models that involve estimates of the process indicators
considering biological and technological limitations
were presented in [43], as well as an estimate the opti-
mal conditions from the initial substrate concentration
Sf and flow D and an estimate of indicators of the mul-
tiplicity and stability of stationary states. Hence, the
indicated directions outline the conditions for the
practical implementation of the process of synthesiz-
ing lactic acid.
NOTATION
D = Q/V dilution rate, h–1;
Ki constant of substrate inhibition, g/L;
KS constant of substrate limitation, g/L;
KP constant of product inhibition, g/L;
constant of substrate inhibition, g/L;
K S'
k1 constant, h–1;
k4, k5 constants (k4 dimensionless, k5, h–1);
L lactose concentration, g/L;
LT residual lactose concentration, g/L;
Vol. 51 No. 3 2017
 MATHEMATICAL MODELING OF BIOTECHNOLOGICAL PROCESS
ms survival constant, h–1;
n dimensionless parameter used to decrease
the effect of product inhibition;
constant, h–1;
P0'
Pmax maximum product concentration g/L;
Pm amount of lactic acid synthesized by growing
associates, g/L;
maximum product concentration above which
'
Pmax
bacteria do not produce lactic acid;
Pn amount of lactic acid synthesized by nongrow-
ing associates, g/L;
Q volumetric feeding flow rate, m3/h;
rst maltose degradation rate, g/(L h);
S substrate concentration, g/L;
Sf substrate concentration in the reactor feeding
flow, g/L;
t fermentation time, h;
V reactor volume, m3;
X biomass concentration, g/L;
Xmax maximum biomass concentration above which
bacteria do not grow, g/L;
X0 initial biomass concentration, g/L;
YP product yield, g of lactic acid /g of substrate;
YP/S, YX/S stoichiometric coefficients, g/g;
α dimensionless constant of product formation
by growing associates;
αа dimensionless constant of side-product forma-
tion by growing associates;
β constant of product formation by nongrowing
associates, h–1;
βа constant of side product formation by nongrow-
ing associates, h–1;
μmax maximum specific growth rate, h–1;
ν(t) productivity at time point t, g/(L h).
REFERENCES
1. Bouguettoucha, A., Balannec, B., and Amrane, A.,
Unstructured models for lactic acid fermentation, Food
Technol. Biotechnol., 2011, vol. 49, no. 1, p. 3.
2. Gordeev, L.S., Koznov, A.V., Skichko, A.S., and
Gordeeva, Yu.L., Unstructured mathematical models
of lactic acid biosynthesis kinetics; a review, Theor.
Found. Chem. Eng. (in press).
3. Wee, Y.J., Kim, J.N., and Ryu, H.W., Biotechnological
production of lactic acid and its recent applications,
Food Technol. Biotechnol., 2006, vol. 44, no. 2, p. 163.
4. Roukas, T. and Kodzekidou, P., Lactic acid production
from deproteinized whey by mixed cultures of tree and
cammobilized Lactobacillus casei and Lactococcus lactis
THEORETICAL FOUNDATIONS OF CHEMICAL ENGINEERING
297
cells using fedbatch culture, Enzyme Microb. Technol.,
1998, vol. 22, no. 3, p. 199.
5. Oh, H., Wee, Y.J., Yun, J.S., and Ryu, H.W., Lactic
acid production through cell-recycle repeated-batch
bioreactor, Appl. Microbiol. Biotechnol., 2003, vol. 107,
nos. 1–3, p. 603.
6. Kwon, S., Yoo, I.K., Lee, W.G., Chang, H.N., and
Chang, Y.K., High-rate continuous production of lac-
tic acid by Lactobacillus rhamnosus in a two-stage mem-
brane cell-recycle bioreactor, Biotechnol. Bioeng., 2001,
vol. 73, no. 1, p. 25.
7. Cotton, J.C., Pometto, A.L., and Gvozdenovic-Jer-
mic, J., Continuous lactic acid fermentation using plas-
tic composite support biofilm reactor, Appl. Microbiol.
Biotechnol., 2001, vol. 57, nos. 5–6, p. 626.
8. Jyer, P.V. and Lee, Y.Y., Simultaneous saccharification
and extractive fermentation of lignocellulosic materials
into lactic acid in a two-zone fermentor–extractor sys-
tem, Appl. Biochem. Biotechnol., 1999, vol. 78, nos. 1–
3, p. 409.
9. Nomura, Y., Jwahara, M., Hallsworth, J.E., Tonaka, T.,
and Ishizaki, A., High-speed conversion of xylose to L-lac-
tate by electrodialysis bioprocess, J. Biotechnol., 1998,
vol. 60, nos. 1–2, p. 131.
10. Min-Tian, G., Koide, M., Gotou, R., Takanashi, H.,
Hirata, M., and Hano, T., Development of continuous
electrodialysis fermentation system for production of a
lactic acid by Lactobacillus rhamnosus, Process Bio-
chem., 2005, vol. 40, nos. 3–4, p. 1033.
11. Jeantet, R., Maubois, J.L., and Boyaval, R., Semicon-
tinuous production of lactic acid in bioreactor coupled
with nanofiltration membranes, Enzyme Microb. Tech-
nol., 1996, vol. 19, no. 8, p. 614.
12. Monteagudo, J.M. and Aldavero, M., Production of
L(−)-lactic acid by Lactobacillus delbrueckii in chemo-
stat culture using an ion exchange resins system, J.
Chem. Technol. Biotechnol., 1999, vol. 74, no. 7, p. 627.
13. Moraine, R.A. and Rogovin, P., Kinetic of polysaccha-
ride B-1459 fermentation, Biotechnol. Bioeng., 1966,
vol. 8, no. 4, p. 511.
14. Pandey, A., Soccol, C.R., and Mitchell, D., New
developments in solid state fermentation. Part 1. Bio-
processes and products, Process. Biochem., 2000,
vol. 35, no. 10, p. 1153.
15. Mercier, P., Yerushalmi, L., Rouleau, D., and
Dochain, D., Kinetics of lactic acid fermentation on
glucose and corn by Lactobacillus amylophilus, J. Chem.
Technol. Biotechnol., 1992, vol. 55, no. 2, p. 111.
16. Altiok, D., Tokatli, F., and Harsa, Ş., Kinetic modeling
of lactic acid production from whey by Lactobacillus
casei (NRRL B-441), J. Chem. Technol. Biotechnol.,
2006, vol. 81, no. 7, p. 1190.
17. Bouguettoucha, A., Balannec, B., Nacef, S., and
Amrane, A., A generalized unstructured model for
batch cultures of Lactobacillus helveticus, Enzyme
Microb. Technol., 2007, vol. 41, no. 3, p. 377.
18. Balannec, B., Bouguettoucha, A., and Amrane, A.,
Unstructured model for batch cultures without pH
control of Lactobacillus helveticus—Inhibitory effect of
the undissociated lactic acid, Biochem. Eng. J., 2007,
vol. 35, no. 3, p. 289.
Vol. 51 No. 3 2017
298 GORDEEVA et al.
19. Bouguettoucha, A., Balannec, B., and Amrane, A.,
Unstructured generalized models for the analysis of the
inhibitory and the nutritional limitation effects on Lac-
tobacillus helveticus growth—Models validation, Bio-
chem. Eng. J., 2008, vol. 39, no. 3, p. 566.
20. Vázquez, J.A. and Murado, M.A., Unstructured math-
ematical model for biomass, lactic acid and bacteriocin
production by lactic acid bacteria in batch fermenta-
tion, J. Chem. Technol. Biotechnol., 2008, vol. 83, no. 1,
p. 91.
21. Vázquez, J.A. and Murado, M.A., Mathematical tools
for objective comparison of microbial cultures: Appli-
cation to evaluation of 15 peptones for lactic acid bacte-
ria productions, Biochem. Eng. J., 2008, vol. 39, no. 2,
p. 276.
22. Roy, D., Leduy, A., and Goulet, J., Kinetics of growth
and lactic acid production from whey permeate by Lac-
tobacillus helveticus, Can. J. Chem. Eng., 1987, vol. 65,
no. 4, p. 597.
23. Luedeking, R. and Piret, E.L., A kinetic study of the
lactic acid fermentation. Batch process at controlled
pH, J. Biochem. Microbiol. Technol. Eng., 1959, vol. 1,
no. 4, p. 393.
24. Rogers, P.L., Bramall, L., and McDonald, I.J., Kinetic
analysis of batch and continuous culture of Streptococ-
cus cremoris HP, Can. J. Microbiol., 1978, vol. 24, no. 4,
p. 372.
25. Gonçalves, L.M.D., Xavier, A.M.R.B., Almeida, J.S.,
and Carrondo, M.J.T., Concomitant substrate and
product inhibition kinetics in lactic acid production,
Enzyme Microbiol. Technol., 1991, vol. 13, no. 4, p. 314.
26. Kumar Dutta, S., Mukherjee, A., and Chakraborty, P.,
Effect of product inhibition on lactic acid fermentation:
Simulation and modeling, Appl. Microbiol. Biotechnol.,
1996, vol. 46, no. 4, p. 410.
27. Monteagudo, J.M., Rodriguez, L., Rincón, J., and
Fuertes, J., Kinetics of lactic acid fermentation by Lac-
tobacillus delbrueckii grown on beet molasses, J. Chem.
Technol. Biotechnol., 1997, vol. 68, no. 3, p. 271.
28. Burgos-Rubio, C.N., Okos, M.R., and Wankat, P.C.,
Kinetic study of the conversion of different substrates to
lactic acid using Lactobacillus bulgaricus, Biotechnol.
Progr., 2000, vol. 16, no. 3, p. 305.
29. Åkerberg, C., Hofvendahl, K., Zacchi, G., and Hahn-
Hägerdal, B., Modeling the influence of pH, tempera-
ture, glucose and lactic acid concentrations on the
kinetics of lactic acid production by Lactococcus lactis
ssp. lactis ATCC 19435 in whole-wheat flour, Appl.
Microbiol. Biotechnol., 1998, vol. 49, no. 6, p. 682.
30. Sinclair, S.G., Microbial process kinetics, in Basic Bio-
technology, Bulock, J. and Kristiansen, B., Eds., Lon-
don: Academic, 1989, p. 102.
31. Gonzales, K., Tebbano, S., Lapes, F., Thorigne, A.,
Givry, S., Dumar, D., and Pareau, D., Modeling the
continuous lactic acid production process from wheat
THEORETICAL FOUNDATIONS OF CHEMICAL ENGINEERING
flour, Appl. Microbiol. Biotechnol., 2016, vol. 100, no. 1,
p. 147.
32. Ohara, H., Hiyama, K., and Yoshida, T., Kinetic study
on pH dependence of growth and death of Streptococcus
faecalis, Appl. Microbiol. Biotechnol., 1992, vol. 38,
no. 3, p. 403.
33. Ohara, H., Hiyama, K., and Yoshida, T., Non-compet-
itive product inhibition in lactic acid fermentation from
glucose, Appl. Microbiol. Biotechnol., 1992, vol. 36,
no. 6, p. 773.
34. Ohara, H., Hiyama, K., and Yoshida, T., Kinetics of
growth and lactic acid production in continuous and
batch culture, Appl. Microbiol. Biotechnol., 1992,
vol. 37, no. 5, p. 544.
35. Boonmee, M., Leksawasdi, N., Bridge, W., and Rog-
ers, P.L., Batch and continuous culture of Lactococcus
lactis NZ133: Experimental data and model develop-
ment, Biochem. Eng. J., 2003, vol. 14, no. 2, p. 127.
36. Nandasana, A.D. and Kumar, S., Kinetic modeling of
lactic acid production from molasses using Enterococ-
cus faecalis RKY1, Biochem. Eng. J., 2008, vol. 38,
no. 3, p. 277.
37. Gordeeva, Yu.L., Ivashkin. Yu.A., and Gordeev, L.S.,
Modeling of periodic process of microbiological syn-
thesis with nonlinear kinetics of microbial growth,
Vestn. Astrakhan. Gos. Tekh. Univ., Ser.: Upravl. Vychisl.
Tekh. Inf., 2011, no. 1, p. 37.
38. Gordeeva, Yu.L., Ivashkin. Yu.A., and Gordeev, L.S.,
Algorithms for calculating the process of microbiologi-
cal synthesis in periodic cultivation conditions, Vestn.
Astrakhan. Gos. Tekh. Univ., Ser.: Upravl. Vychisl. Tekh.
Inf., 2011, no. 2, p. 7.
39. Gordeeva, Yu.L., Ivashkin. Yu.A., and Gordeev, L.S.,
Stationary state stability of the process of lactic acid
biotechnological production, Vestn. Astrakhan. Gos.
Tekh. Univ., Ser.: Upravl. Vychisl. Tekh. Inf., 2012, no. 2,
p. 27.
40. Gordeeva, Yu.L., Komissarov, Yu.A., Gordeeva. E.L., and
Shcherbinin, M.Yu., Mathematical modeling of stabil-
ity of biotechnological process in its optimum condi-
tion, Vestn. Astrakhan. Gos. Tekh. Univ., Ser.: Upravl.
Vychisl. Tekh. Inf., 2015, no. 1, p. 105.
41. Feldbaum, A.A., Dudykin, A.D., Manovtsev, A.P., and
Mirolyubov, N.N., Teoreticheskie osnovy svyazi i uprav-
leniya (Theoretical Foundations of Communication
and Management), Moscow: FizMatLit, 1963.
42. Stepanov, V.V., Kurs differentsialnykh uravnenii
(Course of Differential Equations), Moscow: LKI,
2008.
43. Kulov, N.N. and Gordeev, L.S., Mathematical model-
ing in chemical engineering and biotechnology, Theor.
Found. Chem. Eng., 2014, vol. 48, no. 3, p. 225.
Translated by L. Brovko
Vol. 51 No. 3 2017