BIOTECHNOLOGY AND BIOENGINEERING,

VOL. XIV, PAGES 233-252 (1972)

Kinetic Studies of the Lactic Acid Fermentation

in Batch and Continuous Cultures*

T . P. HANSON, Shell Oil Company, Deer Park, Texas 77536, and

G. T . TSAO, Department of Chemical Engineering, Iowa State Univer-
sity, Antes, Iowa 50010

Summary
The fermentation kinetics of the homofermentative organism Lactobacillus
delbrueckii in a glucose-yeast extract medium is studied in both batch and con-
tinuous culture under conditions of controlled pH. From a graphical analysis of
the batch data, a mathematical model of the process is derived which relates
bacterial growth, glucose utilization, and lactic acid formation. The parameters
in the model represent the activity of the organism and are a function of pH,
having a maximum value a t about 5.90. In a continuous stirred tank fermentor
(CSTF), the effect of pH, feed concentration, and residence time is observed.
The feed medium is a constant ratio of two parts glucose t o one part yeast extract
plus added mineral salts. An approximate prediction of the steady-state be-
havior of the CSTF can be made using a method based on the kinetic model
derived for the batch case. In making step changes from one steady state t o
another, the transient response is observed. Using the kinetic model t o simulate
the transient period, the calculated behavior qualitatively predicts the observed
response.

INTRODUCTION
The application of kinetic models to ordinary chemical reactions
has long been regarded as being very important in the design and
process controI of chemical reactors.' However, the development and
application of fermentation kinetics has been somewhat slower as a
result of the complexity of the process. Nevertheless, a study of the
fermentation rates can be very useful for design of both continuous
and batch systems2 and during the last decade a number of models
have been proposed for bacterial g r o ~ t h . ~
* Presented a t 162nd National American Chemical Society meeting, Division
of Microbial Chemistry and Technology, Washington, D. C., September 14, 1971.
233
@ 1972 by John Wiley & Sons, Inc.
234 HANSON AND TSAO

I n this work, a study was made of the fermentation kinetics of

Lactobacillus delbrueckii on a glucose-yeast extract media in both
batch and continuous culture under conditions of controlled p H and
temperature. For the batch fermentation, a model was developed
that relates bacterial production, sugar utilization, and lactic acid
production. From this model the steady-state and transient be-
havior of a continuous, stirred tank fermentor (CSTF) was estimated.
The model parameters had a marked dependence on the p H of the
fermentation.

MODEL DEVELOPMENT
The fermentation was analyzed in terms of a general, irreversible
biological reaction,
V+S-+2V+P (1)
where the bacteria, V , consumes some substrate, S, to produce more
bacteria and a product, P . I n this study glucose was the primary
carbohydrate source for the fermentation and was considered to be
the limiting growth component, since the bacterial growth stopped
when the glucose concentration reached zero. Yeast extract pro-
vided the necessary amino acids and other nutrient requirements.
Trace minerals were supplied by adding inorganic salts. Previous
studies'-6 have shown that yeast extract has a stimulatory effect on
the L. delbrueckii fermentation and enhances the bacterial yield.
By using a fixed ratio of glucose to yeast extract in the culture
medium, changes in the growth kinetics due to yeast extract were
minimized. Lactic acid formation represented the fermentation
products. While some studies have indicated lactate inhibition of
this fermentati~n,'-~the concentration of the lactate produced in
this study was below the reported limiting concentration^.^
Six rate equations commonly used to represent the consumption or
formation of components in fermentations were fitted against batch
fermentation data.
dCi/dt = ai(dC,/dt) + pic, (2)
dCi/dt = ai(dC,/dt) (3)
dCi/dt = pic, (4)
dCi/dt = piC,C, (5)
BIOTECHNOLOGY AND BIOENGINEERING, VOL. XIV, ISSUE 2
 LACTIC ACID FERMENTATION KINETICS 235

dCi/dt = + PiCsC,
ai(dC,/dt) (6)
dCi/dt = ki[c’S/(K’ai + c’s)]cv (7)
C irefers t o the concentration of biomass, C,, glucose, C,, or lactate,
C,. ( Y ~and f i i are appropriate rate constants. I n eq. (7) K’,i and
C‘, are dimensionless quantities defined as, K‘,i = K,i/C,o and
C‘, = C8/Cso,where Ca0is the initial glucose concentration and K,i
is a constant..
Since eqs. (2), (3), and (6) contain a term of dC,/dt on the right-
hand side, they are not applicable to the case of bacterial growth.
Equations (2) to (7) could be compared to the batch data in their
differential form. However, due to the large error that can occur in
graphically differentiating experimental data, a n integral method
of analysis was used.’

EQUIPMENT AND PROCEDURES

The techniques used in this study were similar t o those employed
in previous investigations of fermentations of acid-producing bac-
teria.‘~’-~
Apparatus
A diagram of the apparatus used for the pH controlled fermenta-
tion is shown in Figure 1. A 7-liter New Brunswick Fermentor
(Model F-07) was used with a New Brunswick Fermentor Drive
Assembly (Series FS-300) which provides a temperature controlled
water bath and variable speed agitator drive. The p H of the medium
was measured with Leeds and Northrup heat-sterilizable electrodes
which were connected to a Heathkit pH Recording Electrometer
(Model EUW-301) which amplified the signal and drove a recorder
pen. When the pH fell below the set point, the recorder pen acti-
vated a microswitch which supplied power t o an Emdeco Micro Flow
Tubing Pump (Model No. 102-130) that accurately metered 4N
sodium hydroxide to the fermentor t o neutralize the acid formed.
The microswitch also activated a relay which sent a 1-V signal to
a strip chart recorder that was run continuously. Thus, when caustic
was being added to the fermentor, the trace on the chart showed a
hump indicating the time and duration the pump was on. From this
record, the rate of acid production was deduced. The set point was
adjusted by the zero setting of the recorder.
236 HANSON AND TSAO

Fig. 1. Diagram of fermentation system. A: Timer, B: Sample rotation motor,

C: Carousel test tube rack, D: Refrigerator (5"C), E: Photocell-light sensor,
F: Relay, G: Siphon breaker, H: Solenoid value (normally open), I: Waste reser-
voir, J: pH recorder-controller, K: Caustic addition recorder, L: Relay, M:
Caustic feed pump, N: Caustic reservoir, 0: Agitator shaft, P: Fermentor, Q : p H
electrodes, R: Quenching cooler, S: Media feed pump, T: Constant temperature
water bath, U: Feed reservoir. Inlet and outlet streams: 1. To aspirator,
2. To drain, 3. Tap water.

Agitation of the system was accomplished by two paddle stirrers

attached t o the drive shaft, one located 2 in. from the bottom and
the other 6 in. below the upper liquid surface. The shaft was run at
300 rpm for all batch and continuous fermentations.
When this fermentor was operated continuously, a Cole-Parmer
"Masterflex" Variable Speed Tubing Pump (Model No. 7020C)
metered the feed medium from a cotton plugged, 5-gal solution bottle
t o the fermentor, which was maintained at constant volume by an
overflow system. The overflow passed through a cold water heat
exchanger t o cool the culture down to 15"C, which quenched further
fermentation. At timed intervals a sample of the overflow was col-
lected in test tubes mounted on a rotary rack in a refrigerated box
maintained a t 5°C for further analysis.
BIOTECHNOLOGY AND BIOENGINEERING, VOL. XIV, ISSUE 2
 LACTIC ACID FERMENTATION KINETICS 237

Culture and Inoculum Preparation

The homofermentative organism Lactobacillus delbrueckii NRRL
B-445, obtained from the Northern Utilization Research Branch,
United States Department of Agriculture, Peoria, Ill., was used in
this study.
Stock cultures were maintained on stabs in a medium composed of
1% glucose, 1% yeast extract, 2y0 agar, and two marble chips. The
cultures were transferred frequently in order to maintain a satis-
factory and active inventory. After incubating the stabs for 48 hr
a t 45"C, a sizable colony was established. Then the stabs were
wrapped in polyethylene bags and stored in the refrigerator a t 5°C
for further use.
Nutrient Sources
The carbohydrate source was A.C.S. Standard Grade anhydrous
glucose. Yeast extract (Difco) was used to provide the organic
nutrilites. Inorganic salts were supplied in excess as described by
Gillies.Io
The feed medium for the continuous runs was made up in 5-gal
bottles and sterilized in a New Brunswick vertical autoclave for 2 hr
a t 15 psig steam pressure. The fermentor and medium were sterilized
for 1 hr.
The feed lines were sterilized prior to use by passing a 1% solution
of tetramethylammonium bromide through them, followed by a
thorough flushing with a 50% ethanol-water solution. In the process
of connecting and disconnecting feed lines from the fermentor, the
area around the input ports was thoroughly swabbed with the
ethanol solution.
The inoculum for the batch and continuous cultures was prepared
in 250-ml shaker flasks from the stab cultures and incubated for 18 hr
at 45°C. The medium consisted of 1% yeast extract, 1% glucose,
inorganic salts, and powdered calcium carbonate.

Analyses
The bacterial density of the culture was determined from turbidity
measurements. Since it was impractical to dilute each sample to the
range where the Beer-Lambert Law applies, a correction factor was
established that would compensate for such deviations. This factor
was determined by measuring the optical density at successive dilu-
238 HANSON AND TSAO

tions of a bacterial culture. For optical densities below 0.08, the

Beer-Lambert Law was observed to hold. At higher optical densi-
ties, the following function was found to apply:
ODC = (OD) exp [0.5745(0D - 0.08)] OD > 0.08 (8)
where ODC is the corrected optical density of the sample and is pro-
portional to the bacterial concentration. To find the bacterial
density of the undiluted culture, C,, the number of dilutions, n, is
multiplied by ODc and the optical density of the unfermented medium
ODB, is subtracted from the product.
C, = (n)(ODc) - ODe (9)
The units of C, are defined as units of optical density per milliliter
(UOD/ml).
The optical densities were measured on a Beckman DU Spectr2-
photometer (Model N. 2400) using light with a wavelength of 6100 A.
The samples were contained in standard silica absorption cells
(Beckman No. 75184). Before and after a measurement was made,
the instrument calibration was set to lOOyotransmission using dis-
tilled water for a reference.
Glucose was determined using the method of Shaffer and Somogyi,"
following the procedures outlined by Neish.I2 The lactic acid forma-
tion was determined from the record of the caustic addition rate.
Also as a check, a number of samples were analyzed for lactate ion
by the method of Friedmann and Graesser.13

RESULTS AND DISCUSSION

Batch Fermentations
Batch runs were made at 443°C and controlled pH levels of 4.95,
5.33, 5.85, and 6.35. The initial glucose concentration for these runs
was about 2% while the yeast extract concentration was about 1%.
The bacterial density, glucose concentration, and lactic acid concen-
tration were followed during the course of the fermentation. Samples
were withdrawn at approximately one-half hour intervals and immedi-
ately analyzed. Using a procedure similar to that employed by
Longsw-orth and M a ~ I n n e s , ' corrections
~ were applied to the ob-
served data in order to compensate for the effects of dilution of the
media by the neutralizing solution and removal of nutrients during
sampling.
BIOTECHNOLOGY A N D BIOENGINEERING, VOL. XIV. ISSUE 2
 LACTIC ACID FERMENTATION KINETICS 239

The overall lactic acid yield, HpIS, defined by,

- lactic acid produced (mg/ml)
YPIS =
glucose consumed (mg/ml)
is compared with the values obtained by Luedeking' and Finn15 in
Table I. The maximum yield occurred at a pH of about 5.8. While
the findings of Luedeking7 also show that the yield increases up to
pH 5.85, Finn's results indicated that the highest yield occurs a t
lower values of pH.
TABLE I
Lactic Acid and Bacterial Yields from Batch Fermentations
- - -
FPi9 YPlS YPIS Y v/s
PH This work Luedeking" Finn 8 This work

4.95 0.74 0.83 0.91 0.22

5.33 0.86 0.87 0.88 0.24
5.85 0.90 0.90 0.87 0.25
6.35 0.82 - - 0.17

Interpolated

The overall bacterial yield, Pv/sl as defined in eq. (11) and listed
in Table I, also had a maximum at pH 5.8.
bacteria produced (UOD/ml)
PVIS = (11)
glucose consumed (mg/ml)
In deriving a mathematical model of the batch fermentation, the
rate expressions given by eqs. ( 2 ) to (7) were integrated and the
terms rearranged so that a linear relationship existed between the
left- and right-hand sides:

, - C V O \Cn -
240 HANSON AND TSAO

ci - cio = @ J o t (C,C,)dt

A computer program compared the batch fermentation data with

the proposed rate expressions, using the trapizoidal rule for the inte-
gration of the experimental data. By using a least squares analysis
to compare these integrated expressions with the data, the best fitting
rate equations were determined.
The best fit of the bacterial growth data (see Fig. 2) was obtained
using eq. (7), which for bacterial growth is the Monod equation.
However, the calculated values of k , and K,, were negative which
makes the form of the equation different than the Monod equation.

-t
"
D

D
Y
I

Fig. 2. Comparison of integrated form of the rate eqs. (18)-(20) with batch
fermentation data.
BIOTECHNOLOGY AND BIOENGINEERING, VOL. XIV, ISSUE 2
 LACTIC ACID FERMENTATION KINETICS 24 1

It was convenient to alter eq. (7) for bacterial growth so that k , and
K,, are positive by changing the denominator from (KLw C,) +
t o (KL, - CL):
d_ - k,Cs C V
c, -
dt K,, - C ,
where
K,, = K6JAo.
The value of K,, was found to always be greater than the initial
glucose concentration, Cg0,so that the denominator of eq. (18) is
never zero or negative. The rate of glucose utilization and lactic
acid production were best fitted by forms of eq. (5):
dC,/dt = -pC,C, (19)
dC,/dt = p,C,C, (20)
A comparison of eqs. (18) to (20) in the integrated form with the batch
data is given in Figure 2 for a fermentation a t p H 5.33. From the
least squares analysis of the data at each p H level, the values of the
parameters k,, KL,, p8, and p, were found t o be strongly dependent
on p H as shown in Figures 3 and 4. The maximum values for the
rate constants occurred a t about pH 5.9 which confirmed the findings
of other^^.^.'^ who noted a maximum in the acid production and
bacterial growth activity of L. delbrueckii between pH 5.7 to 6.0.
Using the derived model, the actual batch fermentations were simu-
lated on a TR-48 analog computer. Figure 5 shows a comparison
of this simulation with the actual data a t p H 5.33.
For use in the graphical integration and differentiation of bacterial
growth curves, a modified form of the growth rate curve proposed by
Edwards and WilkelGwas applied.

(21)

Rearranging gave,

By arbitrarily choosing a value for $1, the left-hand side of eq. (22)
was calculated as a function of time from the experimental data.
242 HANSON AND TSAO

0.12

0.11

t a
0.10

m
0.09

0.08

0.07
5.0
PH - 6.0

0.14 1
0.13

0.12
m
0.11

0.10

0.09
I . I
5.0

Fig. 3. Variation of
PH
@S
- 1
6.0

and pP with pH.

Then, using multiple, linear regression analysis, the coefficients, ai,

were determined. It was necessary t o choose a value of M that was
greater than the final value of C , in order to keep the argument of the
natural logarithm from going negative. Choosing a value of M that
was 20 per cent higher than the largest value of C , was adequate for
this purpose. Figure 6 shows a fit of eq. (21) with the batch growth
data.
Continuous Fermentations
The variables investigated during the continuous fermentation
were pH, residence time, and feed concentration. The feed composi-
BIOTECHNOLOGY AND BIOENGINEERING, VOL. XIV, ISSUE 2
 LACTIC ACID FERMENTATION KINETICS 243

0
I
5.0
DH - I
6.0
I

Figure. 4. Variation of k v and K'sv with pH.

tion always was in the ratio of two parts glucose to one part yeast
extract, but the overall concentration was varied. The rate of acid
production, concentration of glucose and bacterial density were
measured at each steady state and during the transient periods.
The continuous fermentations were started in the same manner as
a batch fermentation. When the bacterial density reached the antici-
pated steady-state level for the continuous fermentation, the flow of
nutrients to the fermentor was commenced. The continuous fer-
mentor volume was 5.57 liters. The steady state was determined
244 HANSON AND TSAO

t
7

50 2
s
Y

>
U

2'
.d 1
I A d -
2 4 6 8
Time ( h r ) -o

Fig. 5. Mathematical simulation of batch fermentation at pH 5.33: (-) model;

( 0 , A, 0) data.

when the bacterial density, acid production rate, and glucose concen-
tration leveled out, which required from three to six residence times.
Samples were taken automatically a t 374 min intervals so that the
transient period between steady states could be observed. It was
assumed that the change from one set of operating conditions to the
next was instantaneous; however, it actually took about 5 min to
establish new conditions of feed rate, pH, and feed concentration.
The operating conditions for each run are listed in Table I1 with
the observed steady-state values of glucose concentration, bacterial
density, and lactate concentration. The lactate concentration was
determined analytically and also calculated from the rate of hydroxide
addition.
To systematically evaluate the effect of the three operating vari-
ables on the steady-state operation of the fermentor, the operating
conditions were chosen to follow a three-dimensional statistical de-
BIOTECHNOLOGY A N D BIOENGINEERING, VOL. XIV, ISSUE 2
 LACTIC ACID FERMENTATION KINETICS 245

I A - 0 -

M = 6.00
.a = 2.447
al = -9.731 x 10-1
a2 = -2.757 x '.01
/ a ? = 3.958 x lo-*-

I 0 / a; = -5.379 x '.01
a5 = 2.457 x
/

I I I I I
I I I

0 1 2 3 4 5 6 7
Time ( h r ) - - r

Fig. 6. Fit of eq. (21) to batch fermentation data at pH 5.85.

sign. The conditions for the center point of the design were a t p H
5.40, a dilution rate of 0.20 hr-l, and a glucose feed concentration
of 20.0 mg/ml. Because of experimental difficulties it was not
possible t o set the operating variables a t exactly the design criteria,
but they were reasonably close.
The reproducibility of the center point of the design was not good
as shown in Table I1 by steady states 2, 6, 10, 14, and 18. After
steady states were achieved at several different operating levels as
shown in Table 11, the pH, feed rate, and feed concentration were
returned to the center point conditions where the new steady-state
activity was observed to be significantly different than previous ones.
The reason for this change in bacterial activity could not be deter-
mined, although there was observed to be a relationship between the
pH of the culture a t previous operating levels and this change in
activity. When the culture was growing a t higher levels of pH, the
activity was lower than was previously observed when the base case
conditions were repeated, as illustrated by the sequence of runs 2 to 6
and 14 to 18. On the other hand, when the pH was lower during the
intervening runs, the activity increased as shown in the sequences
6 to 10 and 14 t o 18. This adaptation was not, completely reversible
246 HANSON AND TSAO

TABLE I1
Steady-State Operating Conditions

Steady
state pH
D,
hr-'
C~I, Cu, &, Rpa,
mg/rnl UOD/ml mg/ml mg/rnin mg/rnl
ePb, mg/rnl
Cp c,

1 5.34 0.201 9.8 1.84 3.75 80.0 4.29 3.98

2 5.40 0.201 19.3 2.78 10.00 123.0 6.59 5.86
3 6.20 0.205 20.0 3.95 0.00 332.0 17.44 15.46
4 5.90 0.122 26.4 3.32 6.10 160.0 14.12 15.58
5 5.95 0.252 26.6 3.44 9.10 285.0 12.19 9.86
6 5.50 0.215 18.3 2.15 10.50 108.0 5.41 5.20
7 5.38 0.291 20.0 1.56 11.90 94.0 3.48 3.24
8 5.00 0.246 24.4 1.20 19.80 69.0 3.02 3.56
9 4.85 0.245 11.4 1.26 9.40 64.0 2.82 3.85
10 5.45 0.208 21.7 4.38 8.00 185.0 9.60 9.82
11 4.60 0.197 19.6 1.46 18.10 60.0 3.28 3.66
12 5.00 0.132 13.9 2.77 3.41 100.0 8.13 7.92
13 6.10 0.134 14.8 3.73 0.00 135.0 10.85 13.58
14 5.40 0.198 20.5 4.58 2.36 240.0 13.11 14.34
15 5.38 0.201 30.8 5.47 7.63 300.0 16.09 20.78
16 4.85 0.122 26.6 3.24 8.98 141. 0 12.41 13.50
17 5.90 0.266 13.7 3.78 0.00 255.0 10.32 14.10
18 5.40 0.205 19.4 4.06 1.68 255.0 13.61 13.54
19 5.40 0.089 20.0 4.79 0.00 85.0 10.32 14.74

a Total acid production rate.

Determined from caustic addition rate.
Determined from chemical analysis.

as once the culture was exposed to lower pH levels it tended to main-

tain the higher level of activity. This phenomenom was also ob-
served in continuous fermentation runs of shorter duration.
The overall yield of lactic acid, FPIS,and bacetria, Y V / s ,(Table
111) based on sugar consumption did not show the pH dependence
observed in the batch fermentations. The average value of the
bacterial yield for the continuous fermentation was higher than the
average value for the batch fermentat,ion. However, the lactic acid
yield was lower. This indicates that the glucose may be utilized
more for cell production and less for lactic acid production during
the continuous process than in the batch process.
Samples from some of the steady-state continuous cultures were
used t o inoculate shaker flask cultures. After inoculation the bac-
terial density was followed for several hours to see if any lag phase
BIOTECHNOLOGY AND BIOENGINEERING, VOL. XIV, ISSUE 2
 LACTIC ACID FERMENTATION KINETICS 247

TABLE 111
Lactic Acid and Bacterial Yields from Continuous Fermentations
- -
Steady state YPIS y VIS
1 0.71 0.30
2 0.71 0.30
3 0.87 0.20
4 0.70 0.16
5 0.70 0.20
6 0.69 0.28
7 0.43 0.19
8 0.66 0.26
9 1.41 0.63
10 0.70 0.32
11 2.19 0.97
12 0.77 0.26
13 0.73 0.25
14 0.72 0.25
15 0.69 0.24
16 0.71 0.18
17 0.75 0.28
18 0.77 0.23
19 0.52 0.24

was observed. In all the cultures studied no discernible lag phase

was observed, even for the case of long residence times and zero
glucose concentrations in the medium. This implied that the bac-
teria are in an active state even under rather extreme conditions and
that the bacterial density determined from the turbidity measure-
ments was a valid representation of the active biomass. If there
had been a lag phase, this would have implied that a significant
number of cells were inactive or dead as the turbidity measurement
does not distinguish between dead and live cells.
Using the growth model developed for the batch fermentation case
given in eqs. (18) to (20) and a general material balance for a CSTF,
the unsteady-state equations for continuous fermentations were
derived :
dC,/’dt = [k,C,/iK,, - C,)]C, - DC, (23)
dC,/dt = (C8f - C,)D - pBC8C, (24)
dCp/dt = &,C,C, - DC, (2.5)
where L) is the dilution rate.
248 HANSON AND TSAO

By setting the left-hand side of the each of the above equations

equal to zero, the steady-state solution was determined:
cs = DK,,/(k, + D) (26)
C, = [(Cs, - C,)/PsC]D (27)
c, = P,C.C,/D (28)
K , , was found as a function of pH from Figure 4 and eq. (Ha). k,,
P., and p p were determined from Figures 3 and 4. From the operating
conditions (Table 11) the values of the steady-state concentrations of
each variable were calculated and are listed in Table IV.
Transient Behavior
Between each steady state, transient response data was obtained
for step changes in the operating conditions. These step changes

TABLE IV
Prediction of Steady-State Behavior from Batch Fermentation Data

From batch model

Steady C", c,, &I

state UOD/ml mg/ml mdml

1 1.74 5.40 3.56

2 2.16 9.30 8.00
3 1.97 8.40 9.90
4 2.44 5.96 16.30
5 2.05 11.20 12.40
6 2.50 7.70 8.10
7 1.92 12.20 6.40
8 1.08 17.60 5.40
9 1.02 8.40 2.40
10 1.28 13.00 6.80
11 0.70 15.10 3.50
12 1.24 7.62 5.00
13 2.30 4.00 8.90
14 2.16 9.80 8.60
15 1.98 15.70 12.30
16 1.19 14.50 9.70
17 2.31 5.50 6.60
18 2.06 9.60 7.80
19 2.92 4.10 12.10

BIOTECHNOLOGY AND BIOENGINEERING, VOL. XIV, ISSUE 2

 LACTIC ACID FERMENTATION KINETICS 249

involved changes in one, two, or three variables, depending on the

requirements of the experimental design. The resuks for one of these
transition periods is plotted in Figure 7, showing the response t o a
decrease in pH from 5.45 to 4.60. There did not appear to be any
oscillatory behavior in the cultures during the transient periods.
However, overshoot and undershoot of the final steady state was
observed in several of the runs.
Equations (23) t o (25) were solved on the analog computer to simu-
late the transient behavior of the system (Fig. 8). The value of K,,
was the same as that used t o predict the steady-state behavior. In
order that the computed transient run would have the same initial
and final values as those observed, the values of k,, PI, and P p were
redefined using eqs. (26) to (28).

k, = (DL/cJ - (29)
P S = [(CS, - cs)/c,c,lD (30)
Pp = Dcp/csc, (31)
where c,, c,, and c, were the steady-state values at the end of the
transient run.

D = 0.208 0 = 0.197 h r - ’
600 - pH = 5.45 pH = 4.60
C S f = 21.7 CSf = 19.6 mglml

0
I
10 - I

I I 1 I
0 0 4 8 12 16 20 24

Time ( h r ) +

Fig. 7. Transient behavior of the continuous fermentation following a step change

in the pH.
250 HANSON AND TSAO

I I I I I I 1

D = 0.197 hr-'
24 - pH = 5.45 pH = 4.60
CSf = 21.7 I Csf = 19.6 mglml
I -
u I -
8 - I
I
0 ;
I
-
t 16-

a. -

0 ;
I
I
4 -
1 1 -
2 - I -
I I I I I I I

Fig. 8. Mathematical simulation of the transient behavior of a continuous

fermentor following a step change in pH.

The correspondence between the model and the data was fairly
reasonable. The model predicted a much faster response in the
glucose concentration than was observed experimentally. The re-
sponse times for the other variables was approximately the same as
observed experimentally.

CONCLUSIONS
The batch and continuous fermentation of L. delbrueckii was
studied at controlled pH levels in a glucose-yeast extract medium.
The bacterial, glucose, and lactic acid concentrations were measured
as a function of time. Using a graphical analysis of the batch data,
a kinetic model was developed that closely fitted the observed batch
data. The parameters in the model, which reflected the activity of
the culture, were found t o be a function of pH and have maximum
values a t about pH 5.90.
Using a modified form of the growth curve equation proposed by
Edwards and WilkelGan adequate mathematical representation of the
experimental growth data was obtained. The modification allowed
the model t o be fitted by a simple linear multiple regression analysis.
BIOTECHNOLOGY A N D BIOENGINEERING. VOL. XIV, ISSUE 2
 LACTIC ACID FERMENTATION KINETICS 251

Both the overall bacterial and lactic acid yields based on sugar
consumption showed a maximum a t pH 5.9 for the batch fermenta-
tion. I n the continuous fermentation, no dependence of the yield on
p H was observed.
The effect of pH, feed concentration, and dilution rate on the
steady-state behavior of the CSTF was determined by carrying out
a n experimental program based on a three-dimensional statistical
design. I n most instances the observed values were approximately
predicted by the batch kinetic model.
I n moving from one steady state t o another, step changes were
made in the operating conditions, and the transient behavior of the
CSTF was observed in terms of acid production rate, glucose concen-
tration, and bacterial density. Using the batch kinetic model, this
transient behavior was simulated on the analog computer. The re-
sponse time for changes in the glucose concentration was much faster
for the simulated runs than was observed, but the response time of
the bacterial density was about the same in both cases.

Nomenclature
Ci concentration of component i, mg/ml or UOD/ml
CiO initial concentration of component i, mg/ml or UOD/ml
c* lactate concentration, mg/ml
CF. continucus fermentation steady-state lactate concentration, mg/ml
C, glucose concentration, mg/ml
c*0 initial glucose concentration in batch fermentation, mg/ml
C: dimensionless glucose concentration ( = C,/C,o)
C*/ glucose concentration in feed t o continuous fermentor, mg/ml
c. continuous fermentation steady-state glucose concentration, mg/ml
C" bacterial density, UOD/ml
C"0 initial bacterial density in batch fermentation, UOD/ml
C" continuous fermentation steady-state bacterial density, UOD/ml
D dilution rate, hr-1
K,i parameter in rate equation representing component i, mg/ml
K:i dimensionless parameter ( = K,i/C.o)
K." parameter in rate equation, mg/ml
K:" dimensionless parameter ( = K,,/C,o)
M parameter in growth curve equation, UOIl/ml
OD measured optical density, UOD/nil
ODB optical density of unfermented medium, UOl>/ml
ODc corrected optical density, UOD/ml
P fermentation product
S fermentation substrate
UOD units of optical density
V bacteria
2Fjz HANSON ANI) TSAO

lactic acid yield based on glucose consumed, dimensionless

bacterial yield based on glucose consumed, UOD/mg
regression parameter in growth curve equation
rate constant associated with component i, hr-1
bacterial growth constant, hr-1
number of dilutions prior to measuring optical density
time, hr
rate parameter associated with component i
rate parameter associated with component i
glucose rate constant
lactic acid rate constant

The research funds and equipment for this investigation were provided by the
Iowa State University Engineering Research Institute and the Iowa State Water
Resources Research Institute, whose support is greatly appreciated.

References
1. 0. Levenspiel, Chemical Reaction Engineering, John Wiley & Sons, New
York, 1962.
2. E. L. Gaden, A p p l . Microhiol., 8, 123 (1960).
3. H. M. Tsuchiya, A. G. Fredrickson, and R. Aris, Advan. Chem. Eng., 6,
125 (1966).
4. R. E . West, “The effect of agitation on the rate of acid formation by
Lactobacillus delbrzieckii,” Ph.l>. thesis, University of Michigan, Ann Arbor, 1957.
5. L. A. Rogers and E . 0. Whittier, J . Bacteriol., 16, 211 (1928).
6. A. E. Molini, “Proposed new process for the manufacture of lactic acid,”
Ph.D. thesis, University of Michigan, Ann Arbor, 195i.
7. R. Leudeking, “The lactic acid fermentation at controlled pH,” Ph.D.
thesis, University of Minnesota, Minneapolis, 1956.
8. L. L. Kempe, H. 0. Halvorson, and E. L. Piret, Znd. Eng. Chem., 42, 1852
(1950).
9. L. G. Longsworth and I). A. MacInnes, J. Racteriol., 29, 595 (1935).
10. R. A. Cillies, “A comparison of the effects of radiation and heat sterilization
of nutrilites on the rate and yield of lactic acid fermentation,” Ph.D. thesis,
Tiniversity of Michigan, Ann Arbor, 1955.
11. P. A. Shaffer and M. Somogyi, J . B id . Chem., 100, 695 (1933).
12. A. C. Neish, “Analytical methods for bacterial fermentations,” Nat. Ites.
Council of Canada Rept. No. 46-8-3 (Prairie Regional Lab., Saskatoon, Canada),
1952.
13. T. E. Friedmann and J. R. Graeser, J. Biol. Chem., 100, 291 (1933).
14. L. G. Longsworth and I). A. MacInnes, J . Hacteriol., 31, 287 (1936).
15. R . K. Finn, “The rate of formation of lactic acid by fermentation a t con-
trolled pfl,” Ph.1). thesis, University of Minnesota, Minneapolis, l94Y.
16. V. Edwards and C. It. Wilke, Biotechnml. Hioeng., 10, 205 (1968).

Accepted for Publication December 27, 1971

BIOTECIINOLOGY A N D BIOENGINEERING, VOL. XIV. ISSUE 2