Vous êtes sur la page 1sur 9

Bioenerg. Res.

(2008) 1:239–247
DOI 10.1007/s12155-008-9018-6

Optimization of Process Conditions Using Response Surface


Methodology (RSM) for Ethanol Production from Pretreated
Sugarcane Bagasse: Kinetics and Modeling
Ezhumalai Sasikumar & Thangavelu Viruthagiri

Published online: 19 September 2008


# Springer Science + Business Media, LLC. 2008

Abstract Based on a five level central composite design industrialization. Ethanol is a renewable energy source
(CCD) involving the variables substrate concentration (C), produced through fermentation of sugars unlike the fossil
pH (P), incubation temperature (T) and fermentation time fuels. Interest in the bioconversion of abundant and
(H), a response surface methodology (RSM) for the renewable cellulosic biomass into fuel ethanol as an
production of ethanol from pretreated sugarcane bagasse alternative to petroleum is rising around the world owing
by cellulase and yeast Kluyveromyces fragilis was stan- to the realization of diminishing natural oil and gas
dardized. The design contains a total of 31 experimental resources [20]. Lignocellulosic biomass, such as agricul-
trials in which the first 24 organized in a factorial design tural residues (corn stover, wheat straw, sugarcane bagasse),
and from 25 to 31 involving the replications of the central wood and energy crops, is an attractive material for
points. Data obtained from RSM on ethanol production bioethanol (biomass ethanol) fuel production since it is
were subjected to the analysis of variance (ANOVA) and the most abundant renewable resource on the earth [38].
analyzed using a second order polynomial equation. Lignocellulosic biomass could produce up to 442 billion
Maximum ethanol concentration (32.6 g/l) was obtained liters per year of bioethanol [5]. Lignocellulosic materials
from 180 g/l pretreated sugarcane bagasse at the optimized constitute an abundant and cheap feedstock, but the
process conditions (temperature 35°C, pH 5.5) in 72 h processing technique required for ethanol production are
aerobic batch fermentation. Various kinetic models such as presently extensive and costly. The cost of ethanol
logistic model, logistic incorporated leudeking piret model produced from lignocellulosic materials with currently
and logistic incorporated modified leudeking piret model available technology and under the present economic
have been evaluated and the constants were predicted. conditions is not competitive with the cost of gasoline
[11]. Comprehensive process development and optimiza-
Keywords Ethanol . Central composite design (CCD) . tion are still required to make the process economically
Response surface methodology (RSM) . Pretreated sugarcane viable. In reality, environmental considerations and energy
bagasse . Kluyveromyces fragilis and tax policies will determine the extent of fuel ethanol
utilization in the future. Approximately 80% of the ethanol
produced in the world is still obtained from fermentations;
Introduction the remainder comes largely by synthesis from the
petroleum product, ethylene [23]. Lignocellulosic materials
Over the last century, energy consumption has increased generally contain 75–80% of cellulose and hemicellulose,
progressively as the result of growing world population and which cannot be easily converted to simple monomeric
sugars due to their recalcitrant nature [29].
Furthermore, lignocellulose is a more complex substrate
E. Sasikumar (*) : T. Viruthagiri than starch. It is composed of a mixture of carbohydrate
Bioprocess Engineering Research Laboratory,
polymers (cellulose and hemicellulose) and lignin. One of
Department of Technology, Annamalai University,
Annamalai Nagar 608 002, India the largest cellulosic agro-industrial by-products is sugar-
e-mail: sashikumar_ess@yahoo.co.in cane bagasse (or, bagasse as it is generally called); a fibrous
240 Bioenerg. Res. (2008) 1:239–247

residue of cane stalks left over after the crushing and as high concentrations of glucose, so SSF is a good strategy
extraction of the juice from the sugar cane. Bagasse for increasing the overall rate of cellulose to ethanol
consists of approximately 450–500 mg/g cellulose and conversion [6]. Other advantages of this approach are a
250–300 mg/g hemicellulose and 150–200 mg/g lignin. shorter fermentation time and reduced risk of contamination
Chemically, bagasse contains about 500 mg/g cellulose, with external microflora, due to the high temperature of the
300 mg/g pentosans, and 24 mg/g ash. Because of its low process, the presence of ethanol in the reaction medium and
ash content, bagasse offers numerous advantages in the anaerobic conditions. One problem associated with SSF
comparison to other crop residues such as rice straw and is the different optimum temperatures of saccharification
wheat straw, which have 175 mg/g and 110 mg/g, and fermentation. The use of thermotolerant yeasts capable
respectively, ash contents, for usage in bioconversion of fermenting glucose to ethanol at temperatures above
processes using microbial cultures. Sugarcane bagasse is a 45°C, which are closer to the optima for the activity of the
suitable cellulosic substrate for a bioconversion process cellulolytic complex in the range of 35–45°C, is therefore
because of its relatively low lignin content and its advisable when employing coupled SSF processes [3, 7,
availability at central collection sites [12, 30]. 35].
The biological process for converting the lignocellulosic In the present study, optimization of process conditions
materials into fuel ethanol requires the following: deligni- (substrate concentration, temperature, pH and fermentation
fication to liberate cellulose and hemicellulose from their time) for the production of ethanol from pretreated
complex with lignin, depolymerization of the carbohydrate sugarcane bagasse by using cellulase and yeast Kluyver-
polymers to produce free sugars, and fermentation of mixed omyces fragilis NCIM 0557 has been carried out by
hexose and pentose sugars to produce ethanol. Lynd et al. response surface methodology and also various kinetic
[24] has summarized the desirable properties for an ideal models such as logistic model (growth kinetics), logistic
lignocellulose material after chemical pretreatment, i.e. it incorporated leudeking piret model (product formation) and
should (a) produce reactive fibers, (b) yield pentoses in logistic incorporated modified leudeking piret model
non-degraded form, (c) not release the compounds that (substrate utilization) has been evaluated.
significantly inhibit fermentation, (d) work in reactors of
reasonable size with moderate cost and (e) be effective at
low moisture contents. A number of pretreatment options Materials and Methods
are available; acid pretreatment, alkaline treatment, steam
explosion, wet oxidation, organic solvent pretreatment, and Materials
hot water [27]. Among all these methods, steam explosion
and acid pretreatment are still the methods of choice in Sugarcane bagasse sample was obtained from M.R.K.
several model processes [18, 21]. Sugar Mills Ltd. Sethiyathope, Tamilnadu. The bagasse
Enzymatic hydrolysis is the most promising approach to sample was made into 100 mesh (0.15 mm) fine powder by
get high product yields vital to economic success [31, 39]. use of laboratory blender at 3,000 rpm. Sample was
The high cost of cellulase enzyme production hinders the preserved in a sealed plastic bag at 4°C to prevent any
application of these enzymes to bioethanol production [1, possible degradation or spoilage. Pure cellulose powder
17]. The common end products of enzymatic cellulose was used in reference of cellulose estimation and fermen-
hydrolysis is cellobiose, a dimer of two glucose units joined tation tests. The control and pretreated bagasse samples
by a β-1,4 glycosidic linkage and glucose which were the were analyzed for cellulose content using anthrone reagent
major inhibitors for the hydrolysis process [4]. One way to at 630 nm in Bio spectrophotometer, Elico BL 198 [37].
overcome the inhibition problem is to ferment the glucose The estimated cellulose content of control and pretreated
into ethanol as soon as it appears in solution. Simultaneous sample was 420 mg/g and 420 mg/g respectively.
saccharification and fermentation (SSF) is the ideal method
of producing ethanol from sugarcane bagasse. In this Micro Organism and Culture Conditions
process, a cellulose hydrolyzing enzyme (cellulase) is
combined with an ethanol producing organism (yeast) to Commercially available cellulase enzyme was obtained
carry out simultaneous hydrolysis of cellulose to glucose from SISCO Laboratories, Mumbai. The activity of
and the conversion of glucose to ethanol in the same cellulase was found to be 15 FPU/ml and it was used
reactor. The result is improved hydrolysis rates and yields throughout the experimentation. The cellulase activity was
of ethanol when compared to those involving separate measured by standard Mandel’s method [25]. Yeast
hydrolysis and fermentation steps [32]. This method keeps Kluyveromyces fragilis-NCIM 0557 was obtained from
the concentration of glucose low and the accumulation of National Collection of Industrial Microorganism (NCIM),
ethanol in the fermentor does not inhibit cellulase as much National Chemical Laboratory, Pune, India. Culture was
Bioenerg. Res. (2008) 1:239–247 241

maintained on yeast extract–malt extract agar (YMA) eter (420–820). The sugarcane bagasse sample was ana-
medium. After 3 days incubation at 40°C the agar slants lyzed for sugar and Klason lignin content following the
were stored at 4°C.The liquid medium for the growth of procedures described in NREL Standard Procedure (No.
inoculum for yeast was yeast extract–malt extract–peptone– 002). Total reducing sugar concentration was determined by
glucose medium (YMP) composed of 3 g/l of yeast extract, the dinitrosalicylic acid (DNS) method [26]. Then it was
3 g/l of malt extract, 5 g/l of peptone and 10 g/l of glucose. confirmed by bio spectrophotometer (Elico BL 198) at
Inocula were grown aerobically in 250 ml Erlenmeyer 510 nm. Ethanol was estimated using NUCON 5765 gas
flasks containing the above mentioned medium at 40°C in chromatography (GC) with a flame ionization detector and
an Environmental Shaker (Remi Scientific) at 200 rpm for CHROMATO-PAK (10%C–20 M) column (3.0 m ×
24 h. Active cells were centrifuged in a clinical centrifuge 0.32 cm) using nitrogen as the carrier gas at 40 µl/min.
(1,200 rpm), washed with sterile water, and were used as The oven temperature was held at 80°C. The injector and
inoculum. Fermentations for ethanol production were detector temperature was maintained at 200°C.
conducted on a shaker at 200 rpm in 250 ml flasks with
100 ml medium. Experimental Design and Statistical Analysis

Pretreatment of Sugarcane Bagasse Optimization of process conditions is one of the most


critical stages in the development of an efficient and
The pretreatment of bagasse was generally done using economic bioprocess [34]. Classical and statistical method-
different quantities of sodium chlorite and acetic acid at ologies are available for optimizing process conditions such
100°C. Samples of 5 g of dry bagasse (with known as response surface methodology (RSM), Taguchi, SX, etc.
moisture) were transferred to 250 ml Erlenmeyer flasks RSM is a powerful mathematical model with a collection of
with 160 ml of distilled water, 0.5 ml of acetic acid and statistical techniques wherein, interactions between multiple
1.5 g of sodium chlorite. The samples were heated in a process variables can be identified with fewer experimental
water bath at 70–80°C with agitation at each 10 min. After trials [8, 9].
1 h reaction, 0.5 ml of acetic acid and 1.5 g of sodium The RSM used in the present study is a central
chlorite were added. The addition was repeated at each composite design (CCD) involving four different factors.
hour, until the final time of 4 h of reaction. The samples Experiments were conducted in a randomized fashion. The
were filtered in crucibles and washed with 1.6 l of hot CCD contains a total of 31 experimental trials with the first
distilled water under suction. Again the samples were 16 organized in a factorial design and from 25 to 31
washed with acetone and dried at room temperature. involving the replications of the central points (Table 1).
The dependent variables selected for this study were
Simultaneous Saccharification and Fermentation (SSF) ethanol yield (g/l). The independent variables chosen were
substrate composition, C (g/l); pH of the fermentation
The samples of pretreated sugarcane bagasse (PSB) were medium, P; incubation temperature, T (°C); fermentation
weighed (10 g/flask) and distributed into 250 ml Erlen- time, H (h). Once the experiments are performed, a second
meyer flasks with the addition of nutrient medium (YMP order polynomial Eq. (1) shown below was used to describe
without glucose) and 0.05 M Sodium phosphate buffer the effect of variables in terms of linear, quadratic and cross
(pH-5.0) followed by sterilization for 15 min, at 15 psi product terms [15].
(121°C) in an autoclave. Cellulase dosage of 15 FPU/g of
substrate was used for hydrolysis. Yeast cells at the amount X
k X
k k X
X k

of 2 g/l based on the dry weight were inoculated into the Y ¼ b0 þ bi Xi þ bij Xi2 þ bij Xi Xj þ e ð1Þ
i¼1 i¼1 ii <j j
production medium. Flasks were then covered with cotton
to allow CO2 produced during fermentation to escape. The
flasks were incubated in a rotary shaker (200 rpm) at 40°C Where, i, j are linear, quadratic coefficients, respectively,
for 96 h. Samples were withdrawn periodically, centrifuged while ‘b’ is regression coefficient, Y is the ethanol yield, k
in a laboratory desktop centrifuge at 1,200 rpm, and the the number of factors studied and optimized in the
supernatants were analyzed for glucose and ethanol experiment and ‘e’ is random error. The quality of fit of
concentrations. the second order equation was expressed by the coefficient
of determination R2, and its statistical significance was
Cell Growth and Chemical Analysis determined by F-test. The significance of each coefficient
was determined using Student’s t-test. The coefficients of
Cell growth (except during SSF) was monitored directly by the equation were determined by employing MINITAB 15
reading the optical density (OD) using systronics colorim- software. Analysis of variance (ANOVA) for the final
242 Bioenerg. Res. (2008) 1:239–247

predictive equation was done using the same software >F—0.005). ANOVA (F-test) for the model explained the
package. The response surface equation was optimized for response of the dependent variable Y. Table 3 also showed
maximum yield in the range of process variables using that the experimental yields fitted the second order
MATLAB 7.0.1 version. Three dimensional plots and their polynomial equation well as indicated by high R2 values
respective contour plots were obtained based on the effect (0.929). F-value several times greater than the tabulated
of the levels of two parameters (at five different levels each) F-value showed that the model predicted the experimental
and their interactions on the yield of ethanol by keeping the results well and the estimated factors effects were real.
other three parameters at their optimal concentrations. From The regression coefficients, along with the corresponding
these contour plots, the interaction of one parameter with P-values, for the model of ethanol production by K. fragilis,
another parameter was studied and also the model is are described in Table 3 and Table 4. It showed that the
developed to explain the quadratic interaction effects by regression coefficients of all the linear term and all quadratic
conducting the pairwise regression analysis of experimental coefficients of X1, X2, X3 and X4 were significant at <1%
data which is not fully described in the multiple regression level. ANOVA suggested the model to be significant at P<
analysis due to limited degrees of freedom. The optimum 0.01. The P-values are used as a tool to check the
concentration of each parameter was identified based on the significance of each of the coefficients, which in turn
hump in the three-dimensional plots. indicate the pattern of the interactions between the variables.
Smaller value of P then it was more significant to the
corresponding coefficient [19]. From the Figs. 1, 2, 3 and 4,
Results and Discussion it was evident that the interaction effects are significant but
the ANOVA and multiple regression analysis suggests that
Optimization of Ethanol Production from Pretreated the interaction effects are insignificant hence it is essential to
Sugarcane Bagasse analyze the experimental data for quadratic interactions by
pairwise regression analysis in an attempt to show the
Table 1 shows the four independent variables (substrate significance of interactions. Neglecting the unimportant
composition, pH, temperature, fermentation time) and their terms with the P-value greater than 0.01 the model is
concentrations at different coded and actual levels of the reduced and the stepwise regression analysis is conducted on
variables employed in the design matrix. The data obtained the experimental data to show the significance of quadratic
from the five level central composite design matrix were interactions. From Table 5, it is clear that the models
used to develop models in which each dependent variable developed by pairwise regression analysis of quadratic
(ethanol yield) was obtained as the sum of the contributions interactions P×T, C×T, P×H and T×H are significant with
of the independent variable through second order polyno- an R2 value of 0.581, 0.549, 0.395 and 0.439 respectively.
mial equation and interaction terms. The actual yields of The contour plots described by the regression model
ethanol obtained in the experiments and the yields predicted were drawn to illustrate the effects of the independent
by the model Eq. (1) are given in Table 2. variables, and effects of interactions of each independent
Based on the experimental response the quantity of variable, on the response variables. The shape of the
ethanol produced by K. fragilis ranged from 1.50 to 4.11 corresponding contour plots indicates whether the mutual
g/l. Runs # 1 and # 25 had the minimum and maximum interactions between the independent variables are signifi-
ethanol production respectively. The ANOVA result of cant or not. Interactions of variables can be better
quadratic regression model for ethanol yield is described in determined by the orientation of the principal axes of the
Table 3. ANOVA of the regression model for ethanol yield contour plots. From the contour plots, the optimal values of
demonstrated that the model was significant due to an the independent variables could be observed, and the
F-value of 14.96 and a very low probability value (P model interaction between each independent variable pair can de
described.
Table 1 Codes and actual levels of the independent variables for
The contour plots based on independent variable was
design of experiment obtained using the same software package (Figs. 1, 2, 3,
and 4) indicated that a local optimum exists in the area
Independent variables Symbols Coded levels
experimentally investigated. The orientation of the principal
−2 −1 0 1 2 axes of the contour plots between the variables pH and
temperature, substrate concentration and temperature, pH
Substrate conc. (g/l) C 10 20 30 40 50 and fermentation time and temperature and fermentation
pH P 4 5 6 7 8 time indicated that the mutual interactions between these set
Temperature (°C) T 25 30 35 40 45
of variables had a significant effect on the ethanol yield.
Fermentation time (h) H 24 48 72 96 120
The isoresponse contour plots of RSM as a function of two
Bioenerg. Res. (2008) 1:239–247 243

Table 2 Four level central composite design and the experimental responses of dependent variable (ethanol yield)

Run order Substrate conc. (g/l) pH Temp. (°C) Fermentation time (h) Ethanol conc. (g/l)

Experimental Predicted

1 −1(20) −1(5) −1(30) −1(48) 1.5 1.7


2 1(40) −1(5) −1(30) −1(48) 2.3 2.4
3 −1(20) 1(7) −1(30) −1(48) 2.1 2.0
4 1(40) 1(7) −1(30) −1(48) 1.9 2.1
5 −1(20) −1(5) 1(40) −1(48) 1.9 2.1
6 1(40) −1(5) 1(40) −1(48) 2.6 2.7
7 −1(20) 1(7) 1(40) −1(48) 3.3 3.5
8 1(40) 1(7) 1(40) −1(48) 2.2 2.3
9 −1(20) −1(5) −1(30) 1(96) 2.2 2.3
10 1(40) −1(5) −1(30) 1(96) 2.6 2.8
11 −1(20) 1(7) −1(30) 1(96) 1.9 2.1
12 1(40) 1(7) −1(30) 1(96) 2.9 3.1
13 −1(20) −1(5) 1(40) 1(96) 2.3 2.4
14 1(40) −1(5) 1(40) 1(96) 2.4 2.5
15 −1(20) 1(7) 1(40) 1(96) 2.5 2.7
16 1(40) 1(7) 1(40) 1(96) 2.8 2.9
17 −2(10) 0(6) 0(35) 0(72) 2.0 1.8
18 2(50) 0(6) 0(35) 0(72) 2.8 2.7
19 0(30) −2(4) 0(35) 0(72) 1.8 1.7
20 0(30) 2(8) 0(35) 0(72) 2.9 3.1
21 0(30) 0(6) −2(25) 0(72) 2.1 2.0
22 0(30) 0(6) 2(45) 0(72) 2.2 2.1
23 0(30) 0(6) 0(35) −2(24) 2.6 2.8
24 0(30) 0(6) 0(35) 2(120) 3.8 3.2
25 0(30) 0(6) 0(35) 0(72) 4.1 4.2
26 0(30) 0(6) 0(35) 0(72) 4.1 4.2
27 0(30) 0(6) 0(35) 0(72) 4.1 4.2
28 0(30) 0(6) 0(35) 0(72) 4.1 4.2
29 0(30) 0(6) 0(35) 0(72) 4.1 4.2
30 0(30) 0(6) 0(35) 0(72) 4.1 4.2
31 0(30) 0(6) 0(35) 0(72) 4.1 4.2

factors at a time, holding all other factors at fixed level, are The other pair of the independent variables pH and
helpful for understanding both the main and the interaction fermentation time showed similar effects while keeping the
effects of these two factors. The effect of varying levels of other independent variables, temperature and substrate
pH and temperature on the ethanol production, while other concentration as constant at 35°C and 30 g/l respectively
two variables (substrate concentration at 30 g/l and fer- (Fig. 3). Also the effect of temperature and fermentation
mentation time at 72 h) were fixed at central level, is shown time at a constant level of substrate concentration, (30 g/l)
in Fig. 1. When the other two variables kept constant, the and pH (6) was studied and was observed that the ethanol
interaction between the two variables (pH and temperature) yields were affected when the two independent variables
showed that the ethanol yield was sensitive even when pH (temperature and fermentation time) subject to small
and temperature were subject to small alterations (Fig. 1). alterations (Fig. 4). The results showed that as the values
Under certain conditions a maximal contour (ethanol of process variables increased, the yield also increased but
concentration of 4.10 g/l) could be determined, meaning only up to the midpoint of range of variables and thereafter
that further change in temperature and pH would not the yield decreased even though the values of variables
increase the ethanol yield any further. The contour plot for increased. The ethanol yield was significantly affected by
substrate concentration and temperature on the yield of substrate concentration, pH, temperature and fermentation
ethanol, where the two variables fermentation time and pH time, where pH produces greater effect.
were kept constant at 72 h and 6 h respectively, showed that Based on the model, the optimal working conditions
the ethanol yields were obtained in the middle level of the were obtained to attain high ethanol yield. Response
process variables (Fig. 2). analysis revealed the maximum ethanol yield by K. fragilis
244 Bioenerg. Res. (2008) 1:239–247

Table 3 Results of regression analysis and corresponding t and p- Table 5 Stepwise regression analysis for quadratic interactions
value of second order polynomial model for optimization of ethanol affecting ethanol yield of sugarcane bagasse
production
Model and independent Estimate Standard t-value P-
Term Regression Std t- P- variablesa error value
constant coefficient deviation statistic value
pH×temperature (P×T)
Intercept 4.1000 0.1179 34.834 <0.001 Intercept 3.627 0.180 20.151 < 0.001
Ca 0.1500 0.0637 2.354 0.032 P2 −0.319 0.115 −2.790 0.009
Pb 0.1667 0.0637 2.616 0.019 T2 −0.369 0.115 −3.231 0.003
Tc 0.1167 0.0637 1.831 0.086 P2 ×T2 −0.601 0.229 −2.630 0.014
Hd 0.1750 0.0637 2.746 0.014 Substrate conc.×temperature (C×T)
C×C −0.4588 0.0583 −7.859 <0.001 Intercept 3.625 0.183 19.850 < 0.001
P×P −0.4713 0.0583 −8.073 <0.001 C2 −0.306 0.116 −2.630 0.014
T×T −0.5213 0.0583 −8.929 <0.001 T2 −0.369 0.116 −3.170 0.004
H×H −0.2588 0.0583 −4.432 <0.001 C2 ×T2 −0.612 0.233 −2.630 0.014
C×P −0.1250 0.0780 −1.602 0.129 pH×fermentation time (P×H)
C×T −0.1250 0.0780 −1.602 0.129 Intercept 3.443 0.212 16.280 <0.001
C×H 0.1000 0.0780 −1.281 0.218 P2 −0.273 0.135 −2.030 0.053
P×T 0.0875 0.0780 1.121 0.279 H2 −0.061 0.135 −0.450 0.656
P×H −0.0375 0.0780 −0.480 0.637 P2 ×H2 −0.771 0.270 −2.860 0.008
T×H −0.1125 0.0780 −1.441 0.169 Temperature×fermentation time (T×H)
Intercept 0.3479 0.204 17.090 <0.001
R2 =0.929
a T2 −0.332 0.130 −2.560 0.016
Substrate concentration (g/l)
b
pH H2 −0.070 0.130 −0.540 0.595
c
Temperature (°C) T2 ×H2 −0.740 0.260 −2.850 0.008
d
Fermentation time (h) a
R2 =0.581 (P×T), 0.549 (C×T), 0.395 (P×H), 0.439 (T×H)

could be achieved at the conditions when substrate Kinetics and Modeling


concentration is 31 g/l, pH of growth media is 5.5;
temperature is 35°C and fermentation time is 72 h. The In general, formulating a useful kinetic model of a cell
matching quality, of the data obtained by the model population is an art which requires consideration of the
proposed in Eq. (1), was evaluated considering the intended end use of the kinetic model, careful thought and
correlation coefficient, R2, between the experimental and experiment to identify the key variables and parameters
modeled data. The mathematical adjust of those values which influence the processes of primary interest, and some
generated a R2 =0.962, revealing that the model could not conceptual and mathematical agility in translating this
explain only 3.74% of the overall effects, showing that it is qualitative view of the system into a workable mathematical
a robust statistical model. representation [2, 10]. Model predictions about the dynamic
behavior of the system offer a more stringent test of validity
than does comparison to steady state experiments [4, 16].
The process is simply that experimental evidence suggests
models which lead to testable predictions and to further
Table 4 Analysis of variance (ANOVA) for the fitted quadratic
experiments which lead to refinements in the models
polynomial model for ethanol production
resulting in new hypothesis and experiments etc. [13, 33].
Source Sum of Degrees of Mean F-value P-value The following desirable aspects are kept in mind for
squares freedom (df) square (MS) selecting the models (1) it should be capable of representing
the entire course of fermentation, (2) the equation has to be
Regression 20.4076 14 1.45768 14.960 <0.001
Linear 2.2683 4 0.56708 5.820 0.004 flexible to fit different types of data, (3) model parameters
Square 17.1317 4 4.28293 43.950 <0.001 should be such that they could be easily estimated and (4)
Interaction 1.0075 6 0.16792 1.720 0.180 the model should be easily usable once the parameters are
Residual 1.5592 16 0.09745 – – determined [9, 28]. Monod and logistic are the models for
error growth kinetics, modified leudeking piret and Logistic
Lack-of-fit 1.5592 10 0.15592 – – incorporated modified leudeking piret are the models for
Pure error 0.0000 6 0.00000 – –
product formation and substrate utilization kinetic models
Total 21.9667 30 – – –
respectively [14, 22].
Bioenerg. Res. (2008) 1:239–247 245

45 120

Fermentation time (h)


Temperature (°C)

40 96

35 4 72 4
3

2 3
30 48
2
1
25 24
4 5 6 7 8 4 5 6 7 8
pH pH
Hold values: Subst. conc.: 30.0 Time: 72.0 Hold values: Subst. conc.: 30.0 Temp: 35.0
Fig. 1 Contour plot for the effect of pH and temperature on ethanol Fig. 3 Contour plot for the effect of pH and fermentation time on
production ethanol production

the process simulation from the various models using the


experimental data are compared.
The objective of this modeling is to develop a
comprehensive mechanistic model based on the experimen- (a) Logistic model (growth)
tal observations. A comprehensive mechanistic kinetic Under optimal growth conditions and when the inhibi-
model has been derived based on the mechanism of ethanol tory effects of substrates and product play no role, the rate
production from lignocellulosic materials by using yeast of cell growth is given by Eq. (2)
strains. The validity of the proposed model under different
experimental conditions has been tested. The cellmass, dX
¼ m0 X ð2Þ
product formation and substrate utilization kinetics using dt
K. fragilis with different parameters are studied. Monod,
Logistic and Modified logistic models are attempted for where μ0 is a constant defined as the initial specific
representing the batch growth kinetics. Leudeking piret and growth rate. The logistic model equation implies that
Modified leudeking piret model, which are modified by the growth rate increases with increase in cellmass
incorporating the logistic models, are tried for product concentration and is independent of the substrate
formation and substrate utilization kinetics. The results of concentration. In reality the growth of cell is governed

45 2
Fermentation time (h)
Temperature (°C)

40 1

4
35 4 0 3
3

30 2 -1 2

1 1
25 -2
10 20 30 40 50 -2 -1 0 1 2
Substrate conc. (g/l) Temperature (°C)
Hold values: pH: 6.0 Time: 72.0 Hold values: Subst. conc.: 30.0 pH: 6.0

Fig. 2 Contour plot for the effect of substrate concentration and Fig. 4 Contour plot for the effect of temperature and fermentation
temperature on ethanol production time on ethanol production
246 Bioenerg. Res. (2008) 1:239–247

Table 6 Model parameters for ethanol production from pretreated The model parameter values are evaluated using MATLAB
sugarcane bagasse
program and are presented in Table 6. The simulation result
Model Models of the logistic incorporated leudeking piret model is in good
parameters agreement with the experimental data obtained from the
Logistic Logistic Logistic
pretreated sugarcane bagasse and the minimum average
incorporated incorporated
leudeking piret modified
error of 4.46%.
leudeking piret (c) Logistic incorporated modified leudeking piret model
(substrate utilization)
X0a 1.2 g/l – –
Xmb 5.5 g/l – – The substrate utilization kinetics is the modified form of
μ0c – 0.20 h−1 0.20 h−1 the leudeking piret model which can be used for substrate
μmd 0.086 h−1 –
utilization kinetics. Substrate consumption depends on the

Kse 0.52 g/l – –
magnitude of three sink terms, the instantaneous cellmass
αf 0.48 0.38 0.42 growth rate, the instantaneous product formation rate and a
βg 0.018 0.012 0.014 cellmass maintenance function. The assumed kinetic form
γh 5.80 8.80 8.20 is a linear combination of these terms.
ηi  
0.0542 0.016 0.008 Xo emt
Avg. error 6.47% 4.46% 3.02% St ¼ So  g  Xo
R2 0.942 0.966 0.972
1  ðXo =Xm Þð1  emt Þ
   
Xm Xo
a
Initial biomass concentration (g/l) h ln 1  ð1  emt Þ ð5Þ
b
Maximum biomass concentration (g/l) m Xm
c
Initial specific growth (h−1 )
d
Specific growth rate (h−1 ) The model parameter values shown in Table 6 are then used
e
Growth associated constant for substrate consumption (g substrate/g to simulate the experimental data of substrate concentration
biomass)
f
Non growth associated constant for substrate at any time during the entire fermentation. Better substrate
g
Consumption (g substrate/g biomass h) utilization kinetics is obtained using the logistic incorpo-
h
Growth associated constant (g product/g biomass) rated modified leudeking piret model Eq. (5) and is well
i
Non growth associated constant (g product/g biomass h) suited for ethanol production from pretreated sugarcane
by a hyperbolic relationship and the logistic Eq. (3) is bagasse with a minimum average error of 3.02%.
given by [36]
 
X ðt Þ ¼ X0 emo t =1  X0=Xm ð1  emo t Þ ð3Þ Conclusion

The logistic equation is utilized to describe the kinetics of Due to dwindling of fossil fuel microbial production of bio
several polysaccharides fermentation systems. The model fuel from biomass (mass of living matter) by products has
parameter values are evaluated using MATLAB 7.0.1 acquired significant fuel for the future. Bagasse, the solid
program and are shown in Table 6. A better prediction of residue after extraction of the sugarcane juice, is mostly
cellmass concentrations is obtained using the logistic utilized for producing steam and electricity required for the
model and it is most suited for ethanol production with cane processing plant. The price of bagasse ranges from Rs.
pretreated sugarcane bagasse as substrate. 100 to 400 (~$2 to 9) per tonne with an average of about Rs.
(b) Logistic incorporated leudeking piret model (product 100 ($2) per tonne. Because of its high carbohydrate content,
formation) relatively low lignin content and its availability as an
industrial waste product, bagasse is a particularly appropriate
substrate for bioconversion to ethanol. This study examines
Logistic incorporated leudeking piret model is developed the possibility of ethanol production from sugarcane bagasse
by rearranging the leudeking piret model as shown below using cellulase and K. fragilis. Conventional optimization
by equation (4). studies are time consuming and expensive. To overcome
2 3 these problems, a central composite design (CCD) was used
m0 t
Pt ¼ P0 þ aX0 4 
e
 5 þ bXm for the optimization of process conditions. From the
X
1  0=Xm ð1  e 0 Þm t m0 experimentation, it is evident that the use of statistical
  process condition optimization approach, response surface
X0 methodology has helped to locate the most significant
 ln 1  ð1  em0 t Þ ð4Þ
Xm conditions with minimum effort and time. In addition, it
Bioenerg. Res. (2008) 1:239–247 247

has also proved to be useful in increasing ethanol concentra- 16. Han K, Levenspiel O (1988) Extended Monod kinetics for substrate,
product and cell inhibition. Biotechnol Bioeng 32:430–437
tion. A maximum ethanol concentration of 32.6 g/l was
17. Himmel ME, Ruth MF, Wyman CE et al (1999) Production of
obtained from 180 g/l in of sugarcane bagasse at the optimized bioethanol from lignocellulosic materials. Curr Opin Biotechnol 10
process conditions of temperature 35°C, pH 5.5 and fermen- (4):358–364
tation time 72 h in aerobic batch fermentation. Kinetic models 18. Kaar WE, Gutierrez CV, Kinoshita CM et al (1998) Steam
explosion of sugarcane bagasse as a pretreatment for conversion
were evaluated and the model constants were predicted on
to ethanol. Biomass Bioenerg 14(3):277–287
ethanol production from pretreated sugarcane bagasse. 19. Kalil SJ, Maugeri F, Rodrigues MI et al (2000) Response surface
analysis and simulation as a tool for bioprocess design and
Acknowledgements The authors express their sincere thanks to optimization. Proc Biochem 35:539–550
UGC for the financial support through the Rajiv Gandhi National 20. Kim S, Dale BE (2004) Global potential bioethanol production from
Fellowship Scheme (RGNFS) for the fellowship to one of the authors wasted crops and crop residues. Biomass Bioenerg 26:361–375
(E. Sasikumar) and to the Department of Technology, Annamalai 21. Laser M, Schulman D, Allen SG, Lichwa J, Antal MJ, Lynd LR et
University for providing the necessary facilities for the successful al (2002) A comparison of liquid hot water and steam pretreat-
completion of this research work. ments of sugar cane bagasse for bioconversion to ethanol.
Bioresour Technol 81:33–44
22. Leudeking R, Piret EL (1959) A kinetic study of the lactic acid
References
fermentation. J Biochem Microbiol Technol Eng 1:393–400
23. Lin Y, Tanaka S (2006) Ethanol fermentation from biomass
1. Adsul MG, Ghule JE, Shaikh H, Singh R, Bastawde KB, Gokhale resources: current state and prospects. Appl Microbiol Biotechnol
DV, Varma AJ et al (2005) Enzymatic hydrolysis of delignified 69:627–642
bagasse polysaccharides. Carbohy Poly 62:6–10 24. Lynd LR, Elander RT, Wyman CE et al (1996) Evaluation of
2. Aiba S (1973) Biochemical engineering: comprehensive text on paper sludge for amenability to enzymatic hydrolysis and
fermentation treats batch kinetics. In: Aiba S (ed) Biochemical conversion to ethanol. Appl Biochem Biotechnol 57/58:741–761
engineering. 2nd edn. Academic, New York 25. Mandel M, Andreotti R, Roche C et al (1976) Measurement of
3. Anderson PJ, Neil Mc K, Watson C et al (1986) High efficiency saccharifying cellulase. Biotechnol Bioeng Symp 6:21–33
carbohydrate fermentation to ethanol at temperatures above 40°C 26. Miller GL (1959) Use of dinitrosalicylic acid reagent for
by Kluyveromyces marxianus var. marxianus isolated form sugar determination of total reducing sugar. Anal Chem 31:420–426
mills. Appl Env Microbio 51:1314–1320 27. Morjanoff PJ, Gray PP (1987) Optimization of steam explosion as
4. Bailey JE (1986) Kinetics of substrate utilization, product formation a method for increasing susceptibility of sugarcane bagasse to
and biomass production in cell cultures. In: Bailey JE (ed) enzymatic saccharification. Biotechnol Bioeng 24:733–741
Biochemical engineering fundamentals. 3rd edn. McGraw Hill, 28. Okpokwasili GC, Nweke CO (2005) Microbial growth and
New York substrate utilization. African J Biotechnol 5(4):305–317
5. Balat M, Balat H, Cahide Z et al (2007) Progress in bioethanol 29. Pala H, Mota M, Miguel Gama F et al (2007) Enzymatic
processing. Prog Energ Combus Sci 12:137–145 depolymerization of cellulose. Carbohy Poly 68:101–108
6. Ballesteros M, Oliva JM, Ballesteros I et al (1993) Optimization of 30. Pandey A, Soccol CR, Nigam P, Soccol VT et al (2000)
the simultaneous saccharification and fermentation process using Biotechnological potential of agro-industrial residues. I: Sugar-
thermotolerant yeasts. Appl Biochem Biotechnol 40:201–211 cane bagasse. Bioresour Technol 74:69–81
7. Ballesteros M, Oliva JM, Negro MJ, Manzanares P, Ballesteros I 31. Philiphidis GP, Smith TK, Wyman CE et al (1993) Study of the
et al (2004) Ethanol from lignocellulosic materials by a enzymatic hydrolysis of cellulose for production of fuel ethanol
simultaneous saccharification and fermentation process with by the simultaneous Saccharification and fermentation process.
Kluyveromyces marxianus. Proc Biochem 391:1843–1848 Biotechnol Bioeng 41:846–853
8. Balusu R, Paduru RR, Kuravi SK, Seenaya G, Reddy G et al 32. Chandrakant P, Bisaria VS (1998) Simultaneous bioconversion of
(2005) Optimization of critical medium components using cellulose and hemicellulose to ethanol. Critic Rev Biotechnol 18
response surface methodology for ethanol production from (4):295–331
cellulosic biomass by Clostridium thermocellum SS19. Proc 33. Roels JA (1983) An excellent integrated treatment of macroscopic
Biochem 40:3025–3030 thermodynamics and kinetic models for microbial growth and
9. Bas D, Ismail H, Boyaci J et al (2007) Modeling and optimization I: product formation. In: Roels JA (ed) Energetics and kinetics in
usability of response surface methodology. J Food Eng 78:836–845 biotechnology. 2nd edn. Elsevier, Amsterdam
10. Blakebrough N (1967) A useful reference for several aspects of 34. Sanjay KR, Kumaresan N, Manohar B, Umeshkumar S, Vijaya-
biological reactor design and fermentation process kinetics. In: lakshmi G et al (2007) Optimization of carotenoid production by
Blakebrough N (ed) Biochemical and biological engineering Aspergillus carbonarius in submerged fermentation using RSM.
science. 1st edn. Academic, New York Inter J Food Eng 3(5):19–27
11. Cardona CA, Sanchez OJ (2007) Fuel ethanol production: process 35. Szczodrak J, Targonski Z (1988) Selection of thermotolerant yeast
design trends and integration opportunities. Bioresour Technol strains for simultaneous saccharification and fermentation of
98:2415–2457 cellulose. Biotechnol Bioeng 31:300–303
12. Dawson L, Boopathy R (2007) Use of post harvest sugarcane 36. Tsao GT, Hanson TP (1975) Extended Monod equation for batch
residue for ethanol production. Bioresour Technol 98:1695–1699 cultures with multiple exponential phases. Biotechnol Bioeng
13. Deindoerfer FH (1960) Fermentation kinetics and model process- 17:1591–1598
es. Adv Appl Microbiol 2:321–327 37. Updegroff DM (1969) Estimation of cellulose by anthrone
14. Fredrickson AG, Megee RD, Tsuchiya HM et al (1970) reagent. Anal Chem 32:420–423
Mathematical models for fermentation processes. Adv Appl 38. Wyman CE (1994) Ethanol from lignocellulosic biomass: tech-
Microbiol 23:419–425 nology economics, and opportunities. Bioresour Technol 50:3–15
15. Giovanni M (1983) Response surface methodology and product 39. Sun Y, Cheng J (2002) Hydrolysis of lignocellulosic materials for
optimization. J Food Technol 37:41–45 ethanol production: a review. Bioresour Technol 83:1–11

Vous aimerez peut-être aussi