Ethanol Production

Copyright in aPress
© 1999 by Humana CSTRInc. 535
All rights of any nature whatsoever reserved.
0273-2289/99/77–79/0535/$12.75

Ethanol Production
by a Flocculant Yeast Strain
in a CSTR Type Fermentor with Cell Recycling

OSSAMU HOJO,* CARLOS OSAMU HOKKA,

AND ANA MARIA SOUTO MAIOR
DEQ/UFSCar, Rua Francisco Degni, s/n,
Caixa Postal 335 CEP 14800-900, São Carlos, SP, Brazil

Abstract
Tests were performed in a continuous stirred tank reactor (CSTR), with
and without cell recycling, to produce ethanol. The reactor without cell recy-
cling produced the kinetic model of ethanol production, whereas the reactor
with cell recycling allowed for a study of process stability.
The Levenspiel kinetic model was adopted; however, in the case of fer-
mentation with cell recycling, the coefficient of cell death was added. It was
observed that cellular viability varied greatly throughout the fermenting
process and that microaeration is of fundamental importance in maintaining
the stability of the process.

Index Entries: Ethanol; flocculant yeast.

Introduction
The fermenting process of ethanol production has its roots dating back
to Brazil’s colonial period. This explains its choice, during the worldwide
oil crisis of the 1970s, as a motor fuel deriving from a renewable source.
Brazil today produces approx 12 million m3 of ethanol per year, with
the industry employing around 4 million workers per year. Most of the
ethanol produced is used as motor fuel or an additive in gasoline to improve
its octane level.
Several changes to the kinetic model proposed by Monod (1) have
been cited in the literature. In the specific case of alcohol fermentation, the
models of cell growth inhibition by the substrate or the product have been
suggested. Among these models, those that show more generic relations
for inhibition by the product are the ones proposed by Levenspiel (2) and
Luong (3).
*Author to whom all correspondence and reprint requests should be addressed.

Applied Biochemistry and Biotechnology 535 Vol. 77–79, 1999

536 Hojo, Hokka, and Maior
On the other hand, it seems that aeration exerts a beneficial effect
on cell growth and on ethanol productivity (4–6). Therefore, this study
was developed to propose an alternative process for ethanol production
using a flocculant yeast strain, in order to save electricity by substituting
the centrifuge for a settler and facilitating cell separation from the fer-
mented broth.
The study began with a search for a kinetic model that would describe
the continuous process of ethanol production, using a laboratory-produced
culture for improved control. Subsequently, the validity of this kinetic
model for ethanol production, using treated sugar-cane syrup diluted and
supplemented only with ammonia sulfate, was confirmed. Process stabil-
ity and improvements through microaeration were also studied.

Materials and Methods

Organism
The organism used was the naturally flocculent strain Saccharomyces
cerevisiae NRRL Y-265. The strain was kept on a Saboraud agar medium and
was subcultured monthly at 30°C for 5 d.

Medium
Four different media were used, two for the experiments without the
settler and two for the experiments with the settler. In the experiments
without the settler, the growth medium (medium A) contained per liter,
60 g sucrose, 6 g peptone, 5 g KH2PO4, 1 g MgSO4 · 7H2O, 5 g (NH4)2SO4,
5 g (NH 4 ) 2HPO 4 , 1 g yeast extract, and tap water at a final pH of 5.5.
The ethanol production medium (medium B) contained 100–200 g/L glucose,
5 g/L KH2PO4, 1 g/L MgSO4 · 7H2O, 10–20 g/L (NH4)2SO4, 10–20 g/L yeast
extract, and tap water at a final pH of 5.5.
The experiments with the settler first used growth medium A, so
a second growth medium (medium C) was also used containing per
liter: sugar cane syrup (about 50 g sucrose), 5 g KH 2PO 4, 12 g (NH 4) 2SO 4,
10 g (NH4)2HPO4, 0.5 g MgSO4 · 7H2O, 0.5 g CaCl2 · 2H2O, 1 g yeast extract,
and tap water at a final pH of 4.5. The production medium (medium D)
contained sugar cane syrup (150–260 g/L sucrose) and 1 g/L (NH4)2SO4 at
a final pH of 4.5. The pH of all media were corrected with 1.0 M NaOH or
1.0 M HCl.
Seed Culture (Inoculum)
The inoculum was prepared in two stages of successive cultures in
500 mL Erlenmeyers.
In the first stage (preinoculum), six loopsful of the stock test tube
were transferred for each 70 mL of growth medium A contained in three
500-mL Erlenmeyers and then maintained at 30°C for 48 h on a shaker
at 300 rpm.
Applied Biochemistry and Biotechnology Vol. 77–79, 1999
Ethanol Production in a CSTR 537

Fig. 1. Schematic diagram of the fermentation system.

In the second stage, actual inoculum, the previous product was trans-
ferred to the three 500-mL Erlenmeyers of the inoculation system, each jar
containing 70 mL of medium C, which was then also kept at a temperature
of 30°C for 48 h on a shaker at 300 rpm.
The inoculant obtained was then transferred to the sterilized 3-L vol-
ume fermenter to begin the growth phase in a fed-batch process.
Continuous Ethanol Production
Because the continuous production of ethanol was a sequential stage
to growth, the fermenter’s mixing and controls were already adjusted (tem-
perature 30°C, air injection 0.05 vvm, medium feed 140 mL/h, agitation
about 200 rpm). It was, therefore, necessary simply to wait for the fer-
menter to fill up and the medium to overflow into the settler when the cell
recycling pump was turned on and the flow adjusted. A diagram of the
system with cell recycling is shown in Fig. 1. The system consisted of a
stirred tank-type reactor (CSTR) with a total volume of 3 L and real vol-
ume maintained at 1.4 L, with stirring done by a magnetic stirrer. The
settler was jacketed and had a total volume of 500 mL, with a real volume
of 300 mL. It was 50 mm in diameter, with a conical bottom from which
the settled cells were removed for recycling to the fermenter through a
peristaltic pump. The settler’s temperature was kept at 10°C.

Biomass
Biomass was determined by filtering a 10-mL sample onto a 0.45-µm
Millipore membrane and drying it at 105°C until it reached a constant weight.

Viability of Cells
Ten milliliters of the sample were centrifuged and the precipitate was
suspended again for the cell viability analysis. The cells were deflocculated
with a 5% ethylenediamine tetra-acetic acid (EDTA) solution and treated
with a solution of methylene blue, as proposed by Lee et al. (7), for a viable
cell count under a microscope.
Applied Biochemistry and Biotechnology Vol. 77–79, 1999
538 Hojo, Hokka, and Maior
Glucose
The Somogyi method (8) was used to assay glucose. Media containing
sucrose were previously hydrolyzed by 0.9 N HCl. The hydrolysate was
neutralized with a solution of 2 N NaOH, using phenolphthalein as an
indicator.

Ethanol
Ethanol was assayed by gas chromatography under the following
conditions: The packing material used was Cromossorb W coated 15% Carbo-
wax 20 with a 2.2-m length and 1/8 in. inner diameter; and carrier gas,
nitrogen, and n-butanol as the internal standard. The temperature of the
column, detector, and vaporizer were 120, 170, and 150°C, respectively.
N-butanol (Merck, Rahway, NJ) was used in a 40 g/L concentration
weighed on an analytical scale. Synthetic air and hydrogen flows were
20 mL/4 s and 20 mL/28 s, respectively, both maintained at a line pres-
sure of 30 psi. The calculations were made by a processing integrator.
The samples taken from the process were diluted to a proportion of 1:10
and added to a quota of n-butanol solution in a 1:1 proportion before being
injected into the chromatograph. The volume injected was 1 µL.

Results and Discussion

Fermentation Without Cell Recycling
The continuous fermentations without cell recycling were made to
define the kinetic model and characterize alcohol fermentation using
flocculant yeast in a continuous stirred tank reactor. Six experiments were
carried out with different initial concentrations of glucose (Cso), 100, 110,
127, 146, 164, and 192 g/L. These experiments are illustrated in Figs. 2–4.
A considerable deviation from Monod’s kinetics can be observed when one
compares the residual concentrations of glucose at low dilution rates for
the different experiments. It can be seen that the concentration of residual
glucose is very low for 100 g/L Cso and that there is a gradual increase
when one increases the initial concentration of glucose. Since the concen-
tration of glucose is far below the inhibiting concentration, inhibition can
be attributed to the product, in this case to ethanol. The method employed
to make the necessary adjustment was the nonlinear regression. Of the sev-
eral models of inhibition by ethanol, the Monod model modified by
Levenspiel (2) proved to be the most suitable (9):

(1)

(2)

Applied Biochemistry and Biotechnology Vol. 77–79, 1999

Ethanol Production in a CSTR 539

Fig. 2. Variations of cell concentrations for different dilution rate and initial concentra-
tions of glucose (Cso). —䊏—, Cso 100 g/L; —䉱—, Cso 110 g/L; —䉬—, Cso 127 g/L;
— + —, Cso 146 g/L; —䊐—, Cso 164 g/L; —䊊—, Cso 192 g/L.

Fig. 3. Variations of ethanol for different dilution rates and initial concentrations
of glucose (Cso). —䊏—, Cso 100 g/L; —䉱—, Cso 110 g/L; —䉬—, Cso 127 g/L;
— + —, Cso 146 g/L; —䊐—, Cso 164 g/L; —䊊—, Cso 192 g/L.

Applied Biochemistry and Biotechnology Vol. 77–79, 1999

540 Hojo, Hokka, and Maior

Fig. 4. Variations of glucose for different dilution rates and initial concentrations
of glucose (Cso). —䊏—, Cso 100 g/L; —䉱—, Cso 110 g/L; —䉬—, Cso 127 g/L;
— + —, Cso 146 g/L; —䊐—, Cso 164 g/L; —䊊—, Cso 192 g/L.

(3)

where µmax = 0.6/h, C*p= 80 g/L, n = 1.8, and eKs = 0.57 g/L.
where 0.43 is the YP/S obtained from the angular coefficient of the graph of
specific rates of ethanol production (–σ) based on the specific rate of sub-
strate consumption (ν), as shown in Fig. 5.
From Fig. 6, one can see that cell yield varies with the initial concen-
tration of glucose, which makes its use necessary to calculate the medium’s
glucose concentration in the numerical integration of the differential equa-
tion to determine the cell concentration.

Fermentation with Cell Recycling

Stability Study Tests
Several experiments were carried out to study the stability of the pro-
cess. All the tests were made with high hydraulic residence times and high
concentrations of substrate to determine whether the inhibiting effect of
both ethanol and alcohol-related treatment were factors that would influ-
ence stability. Figure 7 groups the experiments carried out with hydraulic
residence times of 10–15 h with different cell residence times.
An analysis of the graphs of the experiments shown in Fig. 7 shows
that although the system is not unstable, it operates with a significant
fluctuation of cell, ethanol, and substrate concentrations. Tests with
Applied Biochemistry and Biotechnology Vol. 77–79, 1999
Ethanol Production in a CSTR 541

Fig. 5. Relationship between specific rate of ethanol production and specific rate of
glucose consumption for calculation coefficient YP/S. 䊏, Data; ——, linear fit of data 1_B.

Fig. 6. Cell yield in view of the initial concentration of glucose.

Applied Biochemistry and Biotechnology Vol. 77–79, 1999

542 Hojo, Hokka, and Maior

Fig. 7. Typical fermentation without microaeration and cell recycling. —䉱—, Cx

viable; —䉭—, Cx dry; —䊏—, Cp; —䊐—, Cs.

microaeration were therefore proposed as an attempt to stabilize the

concentration of ethanol produced, considering that it is produced for 7 mo/yr
in Brazil (Fig. 8). Figure 8 shows the result of tests with microaeration
(0.05 vvm or 70 mL of air/min), where the concentrations remain stable.
Of note is the influence of microaeration, particularly in the maintenance
of viable cell concentrations, which remain stable at approx 20 g/L despite
Applied Biochemistry and Biotechnology Vol. 77–79, 1999
Ethanol Production in a CSTR 543

Fig. 8. Typical fermentation in reactor type CSTR with cell recycling and without (A)
and with (B) microaeration with hydraulic residence time of 10 h. —䉱—, Cx viable;
—䉭—, Cx dry; —䊏—, Cp; —䊐—, Cs.

an increase in total cell concentration. Therefore, analyzing viable cell con-

centrations is extremely important in studying the process, since analyzing
the global cell concentration by means of the dry mass, assuming a
100% viability, does not correspond to the hypotheses established at the
time the problem was set forth.
Applied Biochemistry and Biotechnology Vol. 77–79, 1999
544 Hojo, Hokka, and Maior

Fig. 9. Schematic drawing for mass balance.

Analysis of the process with cell recycling was performed on the balance
of the system’s mass (considering the process as drawn in Fig. 9).
Assuming the system is homogeneous, perfectly stirred, isothermal,
and without any reaction in the settler and: Cxo, concentration of viable
cells in the feeding system; Cso, initial substrate concentration; Cpo, concen-
tration of the product in the feeding system; Cx, concentration of viable
cells in the reactor; Cs, concentration of substrate in the reactor; C p, con-
centration of the product in the reactor; Cxe, concentration of viable cells
in the exit of the settler; Cxx, concentration of purged viable cells; R, recy-
cling rate; Q, feed flow; and Qw, purge flow, the global balance of mass for
the cells is
(4)

(5)

assuming

(6)

(7)

when Cxe is low, so the substrate mass balance in steady state:

(8)

and remembering that and assuming

(9)

Applied Biochemistry and Biotechnology Vol. 77–79, 1999

Ethanol Production in a CSTR 545

so, plotting one can find Kd and Yg. The resulting values
were Kd = 0.0054/h eYg = 0.014. Thus, the kinetic model for the process with
cell recycling was established as (10):

(10)

Conclusions
1. Levenspiel’s kinetic model offers a good explanation of the process
of alcohol fermentation using NRRL Y 265 flocculant yeast.
2. The use of flocculant cells and their recycling may constitute a good
choice for modifying the classical process of alcohol fermentation.
3. The process of alcohol fermentation using flocculant yeast, with or
without cell recycling, proved to be quite stable.
4. Addition of the coefficient of cell death to the kinetic model in the
system with cell recycling indicates that cell death is a factor that
must be considered in continuous fermentation because of its long
periods of operation (7 mo) if the hydraulic residence time is high.
5. The culture medium based on sugar cane syrup and the form of
operation (microaeration) may alter the microorganism’s tolerance
to the inhibitor (ethanol), which can be observed by the modification
of the Cp* in the kinetic model.
6. The viable cell analysis was of fundamental importance in establish-
ing the kinetic model of the fermentation process with cell recycling.
7. Microaeration is extremely important to maintain the stability of the
process.

Acknowledgments
The authors express their gratitude for the financial support of CNPq,
CAPES, and CENPES/PETROBRÁS.

References
1. Monod, J. (1949), Ann. Ver. Microbiol. 3, 371–394.
2. Levenspiel, O. (1980), Biotechnol. Bioeng. 22, 1671–1687.
3. Luong, J. H. T. (1985), Biotechnol. Bioeng. 27, 280–285.
4. Cysewsky, G. B. and Wilke, C. R. (1978), Biotechnol. Bioeng. 20, 1421–1444.
5. Hoppe, G. K. and Hansford, G. S. (1982), Biotechnol. Lett. 4(1), 39–44.
6. Del Rosário, E. J., Lee, K. J., and Rogers, P. L. (1979), Biotechnol. Bioeng. 21, 1477–1482.
7. Lee, S. S., Robinson, F. M., and Wang, H. Y. (1981), Biotechnol. Bioeng. 11, 641–949.
8. Somogyi, M. (1952), J. Biol. Chem. 197, 19.
9. Souto Maior, A. M. (1986), Dissertação de Mestrado, Universidade Federal de São
Carlos, Brazil.
10. Hojo, O. (1990), Dissertação de mestrado, Universidade Federal de São Carlos, Brazil.

Applied Biochemistry and Biotechnology Vol. 77–79, 1999