1394 Biotechnol. Prog.

2007, 23, 1394−1403

Process Design and Optimization of Novel Wheat-Based Continuous Bioethanol

Production System
Najmul Arifeen,† Ruohang Wang,† Ioannis K. Kookos,‡ Colin Webb,† and Apostolis A. Koutinas*,†
Satake Centre for Grain Process Engineering, School of Chemical Engineering and Analytical Science, The University of
Manchester, P.O. Box 88, Manchester M60 1QD, United Kingdom, and Department of Chemical Engineering, University of
Patras, 26500 Rio-Patras, Greece

A novel design of a wheat-based biorefinery for bioethanol production, including wheat milling,
gluten extraction as byproduct, fungal submerged fermentation for enzyme production, starch
hydrolysis, fungal biomass autolysis for nutrient regeneration, yeast fermentation with recycling
integrated with a pervaporation membrane for ethanol concentration, and fuel-grade ethanol
purification by pressure swing distillation (PSD), was optimized in continuous mode using the
equation-based software General Algebraic Modelling System (GAMS). The novel wheat
biorefining strategy could result in a production cost within the range of $0.96-0.50 gal-1 ethanol
($0.25-0.13 L-1 ethanol) when the production capacity of the plant is within the range of 10-
33.5 million gal y-1 (37.85-126.8 million L y-1). The production of value-added byproducts
(e.g., bran-rich pearlings, gluten, pure yeast cells) was identified as a crucial factor for improving
the economics of fuel ethanol production from wheat. Integration of yeast fermentation with
pervaporation membrane could result in the concentration of ethanol in the fermentation outlet
stream (up to 40 mol %). The application of a PSD system that consisted of a low-pressure and
a high-pressure column and employing heat integration between the high- and low-pressure
columns resulted in reduced operating cost (up to 44%) for fuel-grade ethanol production.

Introduction tion. In the conventional ethanol industry, hydrolytic enzymes

are supplied by another bioindustry (enzyme producer) that
Commercial simulators such as Aspen plus (1-3) and employs a similar fungal or bacterial fermentation, uses whole
SuperPro Designer (4) have been used to model and predict
cereal flour or pure cereal components supplemented with
the performance of various fuel ethanol production technologies.
synthetic chemicals or protein hydrolysates, and purifies or at
The rigorous process design studies by simulators provide the
opportunities to perceive the effects of variations in process least concentrates them before being transferred to the fuel
parameters, prices of raw material and cost of utilities on process ethanol production plant. Furthermore, enzyme production is
economics. However, most studies do not focus on simultaneous optimized to suit the needs of the enzyme producer. The
process design and parametric and structural optimization of integration of such a fungal fermentation in the proposed wheat-
the proposed flowsheet. This study is focused on the theoretical based biorefinery would optimize this process to provide
optimization of a novel wheat-based biorefinery for fuel ethanol optimum amount of enzymes and minimize the use of wheat
production. The technologies used for upstream processing, yeast for this purpose, reduce non-renewable energy requirements,
fermentation, and downstream purification of fuel-grade ethanol and utilize only whole wheat flour as medium for enzyme
were selected on the basis of experimental and literature-cited production. In addition, the use of fungal autolysis would lead
results. to the natural and low-cost regeneration of a significant amount
Upstream processing is based on a novel wheat-based of nutrients consumed during fungal fermentation, minimizing
biorefining concept for the production of a nutrient-complete waste production and the amount of wheat lost throughout this
fermentation feedstock that could replace the current wheat dry process.
milling process employed in industry for bioethanol production The use of crude enzymatic filtrates and novel processing
(5-10). This process could lead to the production of value- could lead to the production of a low-cost and nutrient-complete
added coproducts (pearlings, gluten) with many food and non- microbial feedstock (9, 11). Arifeen et al. (9) used computer
food applications that could significantly improve process optimization to evaluate a continuous process for the production
economics. An on-site fungal fermentation is used in the of fermentation feedstock in two stages, including fungal
proposed process to produce amylolytic enzymes for starch fermentation for enzyme production (first stage) and integrated
hydrolysis, while the fungal biomass produced during fermenta- starch hydrolysis and fungal autolysis (second stage) using
tion is autolysed to provide nutrient supplements for yeast experimental results and operating parameters from batch
fermentation. experiments (9, 10). A process producing 120 m3/h nutrient-
Integrating a fungal fermentation in the proposed biorefinery complete feedstock for bioethanol production containing 250
would reduce the environmental impact of fuel ethanol produc- g/L glucose and 0.85 g/L free amino nitrogen would result in a
production cost of $0.126/kg glucose that is less than the current
production cost of glucose solution from starch wet milling in
* To whom correspondence should be addressed. Tel: +44 (0)161 306
4418. E-mail: apostolis.koutinas@manchester.ac.uk.. the USA (9).
† The University of Manchester. Several bioreactor configurations have been tested for fuel
‡ University of Patras. ethanol production. Recycling of yeast cells is one of the
10.1021/bp0701517 CCC: $37.00 © 2007 American Chemical Society and American Institute of Chemical Engineers
Published on Web 10/10/2007
Biotechnol. Prog., 2007, Vol. 23, No. 6 1395

Figure 1. Optimized flowsheet of bioethanol production for a production capacity of 10 million gal y-1.

techniques that have been used to increase the productivity. The It should be stressed that the main target of this optimization
fermentation technology used in this study was a hybrid system study was to present the potential improvement of bioethanol
involving continuous fermentation with recycling and ethanol production economics from wheat that could be achieved
concentration by pervaporation membrane for selective removal through continuous processing, production of added-value
of ethanol from the fermentation broth. Pervaporation was byproducts, and employing novel technologies throughout the
selected because it is one of the most promising technologies process. In addition, this optimization study could be used for
for improving ethanol production economics and reducing non- evaluating promising technologies in upstream, bioconversion,
renewable energy requirements (12). Although this technology and downstream stages for advancing fuel ethanol production.
is not well-developed for industrial implementation of fuel-grade
ethanol production, its potential application would lead to Materials and Methods
ethanol concentration with low energy requirements and produc- The modeling equations and parameters used in this study to
tion of ethanol-free, live yeast as a high-value byproduct. optimize the continuous upstream process have been presented
Many downstream separation systems have been proposed in previous publications (9, 10). Those equations were used
for ethanol purification as substitutes for conventional extractive along with models developed in this study for yeast fermentation
distillation (ED), including PSD, hybrid membrane/distillation and PSD for optimizing the complete flowsheet.
(HMD) systems, and hybrid separation systems involving Modeling of Yeast Bioconversion. A continuous bioreactor
distillation and molecular sieve adsorption (HSDMS). HMD (CSTR) with cell recycling was connected to a pervaporation
systems have not been industrially applied because of the lack membrane to separate ethanol selectively from the fermentation
of detailed process know-how and membrane replacement cost broth in continuous mode (Figure 1). At an ethanol concentration
(13, 14). Similar cost reduction as compared to extractive in the fermentation outlet of 9 mol %, the separation factor given
distillation has been reported for PSD and hybrid separation by Hoover and Hwang (18) for the silicon-rubber membrane
systems (15-17). Jaksland et al. (15) reported 30% reduction employed in this study is maximum with a value of 9. Therefore,
in the operating cost of PSD as compared to ED for ethanol it could be assumed that almost all of the ethanol present in the
separation without heat integration between the high- and low- broth is transferred through the membrane because the separation
pressure columns. In this study, PSD including a heat integration is independent of vapor/liquid equilibrium. The product stream
scheme has been used to evaluate potential reduction in contained ethanol and water, whereas the yeast along with other
processing cost. byproducts were retained by the membrane in the ethanol-free
1396 Biotechnol. Prog., 2007, Vol. 23, No. 6

retentate stream, which was divided into the recycle (FR) and
bleed (FB) stream. The recycle stream was mixed with the feed
stream to yeast fermentation.
Modeling equations of the pervaporation system and biore-
actor-cell-recycle system were combined to design an inte-
grated bioreactor-pervaporator-recycle continuous system as
shown below:

[ ( ) (
µ ) µmax
s
Ksx + s
-
KxpP
Px,max - P )] (1)

[ ( ) (
DP ) ηmax
s
Ksp + s
-
KppP
Pp,max - P
X )] (2)

(1 - γ)
µ)D (3)
(1 + R - γ)
βF0 ) FB (4)
RF0 ) FR (5)
γF0 ) FP (6)
F 0 - FB - FP ) 0 (7)
F0 - FR - Fs ) 0 (8)
F s - FR - FB - F P ) 0 (9)
F0(1 + R) ) D × V (10)
(1 - γ)X P
(Si - S) - - )0 (11) Figure 2. Superstructure of a single distillation column.
(1 + R - γ)Yx/s Yp/s
R
Xi ) X1 (12) structure and the specification of the process parameters and
1+R operating conditions. In comparison to conventional algorithms
1+R
X1 ) X (13) based on nonlinear programming (NLP), feed tray locations,
1+R-γ the number of overall stages, the number of columns, and the
Si(1 + R) ) (R S1 + S0) (14) column interconnections can be optimized by mixed integer
S1(1 + R - γ) ) (1 + R)S (15) nonlinear programming (MINLP).
The MINLP model developed by Viswanathan and Gross-
β+γ)1 (16) mann (19) for a single distillation column was used to develop
Fs × P - Fm,P ) 0 (17) an MINLP model for the PSD system in order to solve the
distillation design problem for the nonideal ethanol-water
V × D - Fs ) 0 (18) mixture. The model determines the minimum cost of a distil-
lation column by varying the location of reflux and reboil in
Q′jAmpsat
j xpj the column with N potential equilibrium stages including a total
Fpj ) (19)
δm condenser and a kettle-type reboiler. The feed consists of a
mixture of ethanol and water entering at bubble point. The stages
Hoover and Hwang (18) have proposed the use of the
are numbered bottom upward. The reboiler is at stage (tray)
continuous membrane column concept for a pervaporation
no. 1, and the condenser is the last tray (i.e., N) as shown in
system and developed the following empirical correlations of
Figure 2. In this work two columns at different pressures are
permeability Q′j for ethanol-water separation:
interconnected in order to break the azeotrope between ethanol-
water mixtures.
Q′E ) 49.09 - 86.67xE + 65.80x2E (20)
In order to present the mathematical model used to optimize
the PSD system, the relevant sets and indices should be defined:
Q′w ) 13.95 - 9.49xw + 48.15x2w (21)
C ) {LP, HP} denotes the set of columns connected in the
system
Modeling of Pressure Swing Distillation System. Distilla-
tion design problems involve both continuous (i.e., flows, J ) {ethanol, water} denotes set of components involved
concentrations, pressure, and temperature) and integer variables Ic ) {1, 2, ..., ntc} denotes all possible trays in the column c
(i.e., number of stages, feed tray location, locations of reflux The set Ic contains the following subsets (k and m are the
and reboil trays). Therefore, the discrete nature of a distillation indexes of subsets refluxc and reboilc, respectively):
design problem cannot be modeled and optimized by a Rebc ) {1} denotes the reboiler tray in column c
conventional algorithm that deals only with continuous variables. Conc ) {ntc} denotes the condenser tray in column c
Mixed integer nonlinear programming allows the optimization Colc ) {2, 3,..., ntc-1} denotes stages in the column c
of a system of nonlinear equations, which contain both continu- excluding reboiler and condenser
ous and integer variables. Hence, the mathematical description Flocc ) {ifeed} denotes feed tray location in column c
of a distillation process as a mixed integer nonlinear program- refluxc ) {i : ifeed < i < nt} denotes possible locations of
ming problem considers the optimization of both the process reflux stream in column c
Biotechnol. Prog., 2007, Vol. 23, No. 6 1397

reboilc ) {i :1 < i < ifeed} denotes possible locations of reboil l

li,chi,c V
+ Vi,chi,c - li+1,chi+1,c
l V
- Vi-1,chi-1,c V
- bui,ch1,c )
stream in column c
0, i ∈ reboilc (39)
The model equations for the PSD system are
Phase equilibrium Definitions (for c ∈ C
fLi,j,c ) fVi,j,c i ∈ I, j ∈ J, c ∈ C (22) ∑
k∈refluxc
refk,c ) rc P1c, i ∈ con (40)
In this study the vapour phase was assumed to be ideal, whereas
the Van Laar activity model was adopted for the nonideal liquid li,c - P2c ) 0, i ∈ reb (41)
phase.
Phase equilibrium error Balance on mixer
f ) P1HP + f1
∑ yi,j,c - ∑ xi,j,c ) 0, i ∈ I, c ∈ C
(42)
(23)
j∈J j∈J
f xfj ) P1HPxnt,j,HP + f1 xf1j, j ∈ J (43)
Total material balance (for c ∈C)
Logical constraints (for c ∈ C)
∑
k∈refluxc
refk,c + P1c + li,c - Vi-1,c ) 0, i ∈ con (24) zi,c - z1-1,c e 0, i ∈ refluxc (44)

zi,c - z1+1,c e 0, i ∈ reboilc (45)

li,c + Vi,c - li+1,c - Vi-1,c - refi,c ) 0, i ∈ refluxc (25)
refi,c - fmax(zi,c - zi+1,c) e 0, i ∈ refluxc (46)
li,LP + Vi,LP - li+1,LP - Vi-1,LP - f ) 0, i ∈ flocLP (26)
bui,c - fmax (zi,c - zi-1,c) e 0, i ∈ reboilc (47)
li,HP + Vi,HP - li+1,HP - Vi+1,HP - P11 ) 0, i ∈ flocHP (27)
Other constraints (for c ∈ C and i ∈ I)
li,c + Vi,c - li+1,c - Vi-1,c - bui,c ) 0, i ∈ reboilc (28)
li,c ) 0, i ∈ con (48)
li,c + Vi,c + ∑
m ∈ reboilc
bum,c - li+1,c ) 0, i ∈ reb (29) Vi,c ) 0, i ∈ reb (49)

z1,c ) 0 (50)
Components material balance (for j ∈ J and c ∈ C)
znt,c ) 0 (51)
( ∑ refk,c + P1c + li,c)xi,j,c - Vi-1,cyi-1,c ) 0, i ∈ con
k ∈ reflux
c A heat integration scheme was applied in the PSD system
(30)
by minimizing the difference between the reboiler duty in the
li,cxi,j,c + Vi,cyi,j,c - li+1,cxi+1,j,c - Vi-1,cyi-1,j,c - refi,cxnt,j,c ) low-pressure column (Qreb,LP) and the heat duty in the condenser
of the high-pressure column (Qcon,HP). The optimal solution for
0, i ∈ refluxc (31)
the economic objectives was achieved at zero value of this
difference.
li,LPxi,j,1 + Vi,1yi,j,LP - li+1,LPxi+1,j,LP - Vi-1,LPyi-1,j,LP - f xf ) The MINLP problem was solved by using the OA/ER/AP
0, i ∈ refluxLP (32) algorithm, which was developed in the software DICOPT++
by Viswanathan and Grossmann (19), integrated in GAMS
li,HPxi,j,HP + Vi,HPyi,j,HP - li+1,HPxi+1,j,HP - Vi-1,HPyi-1,j,HP - (www.gams.com). For the NLP program, CONOPT was used,
P1xnt,j,LP ) 0, i ∈ flocHP (33) which is based on a generalized reduced gradient (GRG)
algorithm, since it was found to be the most effective for solving
li,cxi,j,c + Vi,cyi,j,c - li+1,cxi+1,j,c - Vi-1,cyi-1,j,c - bui,cy1,j,c ) relaxed MINLP problems. For solving the mixed integer
0, i ∈ reboilc (34) programming (MIP) problem, the optimization subroutine library
(OSL) of IBM was utilized.

∑
Process Design and Optimization of Bioethanol Produc-
li,cxi,j,c + Vi,cyi,j,c + bum,cym,j,c - li+1,cxi+1,j,c ) tion System. The detailed process design and optimization of
m∈reboilc
the continuous upstream process were carried out using process
0, i ∈ reb (35) parameters (9) obtained by modeling studies of experimental
Energy balance (for c ∈ C) results (10). The simulation studies were carried out by
integrating the continuous feedstock production system with the
V V yeast bioconversion system that included a yeast bioreactor with
l
li,chi,c + Vi,chi,c - li+1,chi+1,c
l
- Vi-1,chi-1,c - refi,chlnt,c )
recycling and a pervaporation membrane (Figure 1). Model
0, i ∈ reflux (36) equations describing yeast fermentation carried out on the
V V generic feedstock have been developed by Arifeen et al. (11).
l
li,LPhi,LP + Vi,LPhi,LP - li+1,LPhi+1,LP
l
- Vi-1,LPhi-1,LP - The continuous fermentation was optimized using parameters
f hflLP ) 0, i ∈ flocHP (37) estimated from experimental results in this study. In this
integrated flowsheet, the bleed stream after yeast separation was
V V
l
li,HPhi,HP + Vi,HPhi,HP - li+1,HPhli+1,HP - Vi-1,HPhi-1,HP - recycled to the upstream process as it contained a glucose
concentration of around 35 g L-1. In this way, the cost of
P11 hflHP ) 0, i ∈ flocHP (38)
producing glucose solution was further reduced by around 6%.
1398 Biotechnol. Prog., 2007, Vol. 23, No. 6

Table 1. Data for PSD Model

low-pressure high-pressure
data type column column
system ethanol-water
thermodynamic model nonideal liquid
phase-van Laar model;
ideal vapor phase
source of thermody- Reid et al. (1987)
namic data
condenser type total
max no. of trays 70 70
operating pressure 1 bar 10 bar
overall feed rate 217 kmol h-1
ethanol concn in feed 40 mol %
purity constraints xnt,E g ynt,E - 0.005 x1,E g 0.99
recovery constraints V1 g 0.995 f xfE
objective function total annualized cost
direction minimization
Figure 3. Experimental and simulated data of yeast, glucose, and
ethanol concentration during the growth phase of yeast fermentation
The steam used to sterilize the fermentation feedstock was used using a wheat-derived feedstock with a glucose concentration of 114 g
for heating the flour suspension to 68 °C. L-1: (b) ethanol, (() yeast, and (2) glucose.
The fermentation outlet was concentrated with a pervaporation
membrane to 40% ethanol (mole basis). This was used as the Table 2. Kinetic Parameters for Yeast Fermentation
feed in the PSD system. Optimization of the PSD system parameter value
involved thermal integration between the two columns. The µmax 0.154
design data for the PSD system are shown in Table 1. The hot ηmax 1.2
product stream from the high-pressure column at 10 bar (150 Ksx 1.24
°C) was used to sterilize the flour suspension used in fungal Ksp 80.8
Kxp 0.03
fermentation. The permeate stream from the pervaporation Px,max 70.16
membrane was fed to the PSD system where the following Kpp 0.01
equations were used: Pp,max 45.42
Yx/s 0.42
f1 ) ∑j pm j
(52)
Yp/s 0.75

using Guthrie’s capital cost correlations (26). Equipment cost

f ) P1HP + f1 (53) was updated to current prices using the Marshall and Swift
index, which was 1310 in 2006 (27).
The cost of wheat was taken as $0.16/kg, which was the UK
f xfj ) P1HPxnt,j,HP + f1xpj, j ∈ J (54)
price for the year 2003 (www.faostat.fao.org). Utility cost
(process water, cooling water, and electricity) and labor cost
f hf 1LP ) P1HPhnt,HP + f1hp(xpj, tp), j ∈ J (55) were taken from Coulson et al. (28). Steam generation cost was
adopted from the values given by Blanch and Clark (20) and
Upstream and bioconversion processes were optimized for Turton et al. (21). Operating labor requirements was estimated
fixed structure only by nonlinear programming. In the upstream according to the procedure given by Reisman (29). To calculate
process, the minimum volume of reactors for fungal fermenta- the gluten revenue, the market price of gluten was taken from
tion and flour hydrolysis and minimum power requirements were Pallos et al. (30), and the market price of yeast was taken from
considered for process optimization in order to minimize cost. Koutinas et al. (31).
The bioconversion process was optimized on the basis of
minimum cost of the bioreactor and pervaporation membrane. Results and Discussion
In the PSD system, the optimization parameters were the Bioconversion Process. Arifeen et al. (11) presented experi-
numbers of trays and reflux ratios. The discrete variable was mental and modeling studies on bioethanol production by
used to find out the number of trays and location of reflux and Saccharomyces cereVisiae using the wheat-derived feedstock.
reboil streams in high-pressure and low-pressure columns. The The parameters (Table 2) used in this optimization study for
objective function used for minimizing the cost of fuel ethanol continuous yeast fermentation were obtained by fitting the
production was modeling equations developed by Arifeen et al. (11) on
experimental measurements taken only during the yeast growth
(cost of ethanol per kg)min ) phase from a batch fermentation (Figure 3) carried out on wheat-
TOTAL PRODUCTION COST
(56) derived media with a glucose concentration of 114 g L-1. This
P2HPx1,E,HPMWE procedure was followed because of the absence of experimental
data from continuous fermentations on the wheat-derived media.
Cost Estimation. Capital and manufacturing costs were The parameters obtained (Table 2) were within the range of
calculated by the procedures given by Blanch and Clark (20) those reported in the literature for yeast fermentation for fuel
and Turton et al. (21), respectively. The cost of the equipment ethanol production (32-36).
involved in upstream and bioconversion processes was estimated The performance of three bioconversion systems, including
by equations given by Blanch and Clark (20). In the case of a simple bioreactor, a bioreactor with yeast cell recycling, and
wheat storage tank, hammer mill, kneader, boiler, and pumps, a bioreactor with cell recycling integrated with a pervaporation
the costs were calculated from estimating charts (22-25). The membrane, was compared by optimizing each one of those by
cost of equipment involved in the PSD system was estimated nonlinear programming for a fermentation carried out at an inlet
Biotechnol. Prog., 2007, Vol. 23, No. 6 1399

glucose concentration of 114 g L-1. Cell recycling was chosen Table 3. Pressure Swing Distillation System Design Variables
in order to increase productivity. In the case of continuous low-pressure high-pressure
fermentation with cell recycling, the outlet from the bioreactor variables column column
was split equally into two streams, a bleed and a recycle stream. reflux ratio 4.75 8.88
The productivity of this system was two times higher than a top product rate, 227.9 142.9
bioreactor without cell recycling, which most probably occurred kmol h-1 (89.75% ethanol) (84.95% ethanol)
as a result of the high cell concentration in the bioreactor. bottom product rate, 132 84.97
kmol h-1 (1.09% ethanol) (99% ethanol)
Although the bioreactor-cell recycle system is very efficient condenser temp, K 351.35 422.31
in terms of productivity, there are certain disadvantages. Ethanol reboil rate, kmol h-1 1298.8 1413.8
is not completely removed from the recycle stream because reboiler temp, K 369.63 422.82
diameter, m 2.9 1.85
conventional centrifugation or filtration cannot achieve complete height, m 33.18 28.15
separation of ethanol from the yeast slurry. Ethanol concentra- no. of stages 69 59
tion in the inlet of the PSD system is the same as in the case feed tray location 6 40
that the bioreactor without cell recycle is used. In addition, the operating pressure (bar) 1 10
yeast obtained from the bleed stream is not of high value because
dead yeast cells are only obtained as stillage at the bottom of shown in Table 1 was used for the first column due to the
the first column. similarity of liquid and vapor compositions in the condensate.
In the case that the conventional cell separation device is The PSD system was designed to produce 10 million gal ethanol
replaced by the pervaporation membrane module, the concentra- y-1. It was found that a feed flow rate of 217 kmol h-1 with
tion of ethanol in the inlet of the PSD system could be increased 40% (mol %) ethanol was required for this production capacity.
and ethanol-free, active yeast cells could be obtained. In The feed stream was obtained from the combined bioreactor-
addition, the high ethanol concentration at the feed of the PSD cell recycle-pervaporation system as mentioned above. In
system would result in reduced energy requirements. Therefore, contrast to conventional downstream processes used currently
simulation studies of the whole biorefinery were carried out in the fuel ethanol industry, the feed to distillation system was
using the combined bioreactor-cell recycle-pervaporation system. more concentrated and free of solids. Therefore, reduction in
The continuous membrane fermentation (CMF) system operating and capital costs could be achieved.
proposed by Cho and Hwang (37) and its modification (38) is The overall recovery of ethanol in the PSD system was 98%.
an attractive integrated process for ethanol production. The CMF The optimal number of trays in the low pressure column was
system used in this study was based on the pervaporation 69, whereas 59 trays were required in the optimal configuration
membrane proposed by Hoover and Hwang (18). The pervapo- of the high pressure column. The optimum flow rates and sizes
ration membrane was made of silicon rubber. The separation obtained from the optimization program are given in Table 3.
factor for this membrane was reported in the range of 8-9 for When the heat integration scheme was applied, the hot vapor
an inlet ethanol concentration of 5-10 mol % (18). Therefore, stream from the top of the high-pressure column was condensed
for the ethanol concentration (9 mol %) obtained in this in the reboiler of the low-pressure column in order to reboil
simulation study at the membrane inlet, a product concentration the bottom stream of the low-pressure column. A fraction of
up to 40 mol % could be achieved. This system was further
the condensate stream from the reboiler of the low-pressure
modified in this study by introducing liquid recycle and bleed
column was used as reflux to the high-pressure column (Figure
streams in order to make the process more economical. In the
1). The remaining part of the condensate stream was mixed with
proposed system (Figure 1), the bioreactor outlet was circulated
the feed stream to the low-pressure column. Around 44%
through a continuous pervaporation system and the ethanol free
reduction in annualized cost of PSD system was achieved by
retentate stream was split into a bleed and a recycle stream. In
implementing thermal integration between the two columns.
this way, the ethanol concentration can be increased in the PSD
inlet and the yeast concentration can be increased in the recycle Flowsheet Optimization. The aim of this study was to utilize
stream. The recycle stream could be used to dilute the modeling, process design and optimization for the evaluation
fermentation feedstock produced in the upstream processing of promising technologies that could improve the economics
stage from a glucose concentration of 252 to 114 g L-1. and efficiency of ethanol production from wheat. Fuel ethanol
A portion of the retentate stream was taken as bleed in order production by the proposed biorefinery was optimized for a
to avoid accumulation of unwanted byproducts produced during production capacity range of 10-33.5 million gal y-1. The
fermentation in the bioreactor. In addition, pure yeast cells could optimum design variables including flow rates, concentrations
be recovered as coproduct from the ethanol-free bleed stream. and sizes of major equipments involved in a plant capacity of
Since no solids were present in the medium, the pure yeast 10 million gal y-1 are shown in Figure 1. Tables 4 and 5 present
produced from the generic feedstock could be sold as a value- the procedure followed to estimate the total capital cost and
added coproduct. total manufacturing cost, respectively.
The productivity of the bioreactor-cell recycle-pervapora- The optimum dilution rate for a continuous fungal fermenta-
tion system was two times higher than that obtained in the tion with an inlet flour concentration of 85 kg m-3 has been
bioreactor-cell recycle system. The minimum flow rates and estimated by Arifeen et al. (9) at 0.0498 h-1. The estimation of
equipment sizes required to operate the system at optimum the volume for each bioreactor was based on equations proposed
conditions are given in Figure 1. The bleed stream after cell by Blanch and Clark (20) where the maximum size for each
separation can be recycled to the upstream process in order to bioreactor unit is 250 m3. For this reason, whenever in a
recover any remaining glucose or other nutrients such as FAN, simulation the volume needed became higher than 250 m3,
phosphorus and vitamins. In this way the production cost of another bioreactor was added. The upstream process was
fermentation feedstock can be further reduced. designed to produce 39.2 m3 h-1 fermentation feedstock
Pressure Swing Distillation. The optimum design variables containing 252 kg m-3 glucose and 0.85 kg m-3 FAN
for the PSD system are given in Table 3. The purity constraint concentrations.
1400 Biotechnol. Prog., 2007, Vol. 23, No. 6

Table 4. Total Capital Cost for a Plant Producing 10 Million gal y-1 Fuel Ethanol
equipment specifications unit total cost (M$)
wheat storage tank 96,076 t, silo 1 0.019
hammer mill 10.7 t /h 1 0.049
kneader 6.26 m3/h, stationary upright double-arm 1 1.4
kneader, 304 stainless steel
centrifuge 57,758 L/h, stainless-steel solid bowl 1 0.51
rotary drier 48.13 m2 peripheral area, hot air heat, 1 0.38
carbon steel
continuous sterilizer 12.42 m3/h and 39.2 m3/h, insulated 316 2 0.23
stainless steel pipes plus heat exchangers
suspension tank 12.42 m3, 316 stainless steel 1 0.125
fungal bioreactors 250 m3, skid-mounted units, 316 stainless steel 1 4.8
rotary vacuum filters 6.2 m2 and 19.2 m2 filter area, carbon steel 2 0.015
gelatinizer 15 m2, 304 stainless steel 1 0.036
hydrolyzis tanks 120 m3, 304 stainless steel 1 2.8
yeast bioreactors 250 m3, 47 m3 skid-mounted units, 316 stainless steel 2 6.26
membrane unit including 493.6 m2 Pervaporation, hollow fiber, 1 µm thickness; 1 0.126
condenser feed pump and 108.5 m3/h feed flow, 7.36 m3/h permeate flow,
vacuum pump 2,631.65 kW condenser load
yeast centrifuge 31,800 L/h, stainless-steel solid bowl 1 0.33
yeast drier 106 m2 peripheral area, hot air heat, carbon steel 1 0.71
low-pressure column, including 56 stages (sieve trays), 3.14 diameter, 27.65 height, 1 2.46
bare module, condenser, reboiler, carbon steel, 2,178.5 m2 condenser area,
reflux drum and trays 519 m2 reboiler area
high-pressure column, including 53 stages (sieve trays), 2.03 diameter, 26.14 height, 1 1.07
bare module, reboiler, reflux drum carbon steel, 518.87 m2 reboiler area
and trays
boiler 33,948 kg/h steam, oil- and gas-fired 1 0.68
air compressor 151.5 hp, 30 m3/h air 1 0.61
cooling tower 28.22 m3/min cooling water 1 0.76
pumps 2.1 m3/h, 32 m head, centrifugal, 316 stainless steel 12 0.19
bare module costs (BMC), sum of above costs 23.63
contingency and fees (40% of BMC) 9.45
total purchased equipment cost (TEC ) 1.4 ×BMC) 33.08
buildings costs (BC ) 15% of total equipment) 4.96
land and site development (LSD ) 7% of total equipment) 2.32
startup costs (STC ) 10% of total equipment) 3.31
fixed capital investment (FCI ) TEC + BC + LSD + STC) 43.66
working capital (3 months raw material and supplies, 1 month labor), WC 3.96
total capital investment (FCI + WC) 47.62

Table 5. Total Production Cost for a Plant Producing 10 Million gal y-1 Fuel Ethanol
item details cost (million $/y)
1. Direct Manufacturing Costs
raw material cost 96,076 t/y, $0.16/kg 15.36
utilities
i) process water 7.36 m3/h, $1.14/m3 0.066
ii) process steam 33,948 kg/h, $0.015/kg 4.15
iii) cooling water 28.22m3/min, $0.028/m3 0.38
iv) electricity cost $0.0228/MJ & $0.06/kWh 0.53
operating labor cost (OLC) 9 workers, 3 shifts/day, $19/h 1.35
direct supervisory and clerical labor 0.18 × OLC 0.243
maintenance and repair 0.06 × FCI 2.62
operating supplies 0.009 × FCI 0.39
laboratory charges 0.15 × OLC 0.20
patent and royalties 0.03 × TPC 1.23
2. Fixed Manufacturing Costs
depreciation 0.1 × FCI 4.37
local taxes and insurance 0.032 × FCI 1.4
plant overhead costs 0.708 × OLC + 0.036 × FCI 2.53
3. General Manufacturing Expenses
administration costs 0.177 × OLC + 0.009 × FCI 0.63
distribution and selling costs 0.11 × TPC 4.52
research and development 0.05 × TPC 2.06
total production cost (TPC) sum of items 1 to 3 41.1
total production cost per gal ($/gal) 4.0
credit from gluten revenue 11,676.5 t/y, $1.32/kg 15.4
total production cost per gal ($/gal), including gluten credit 2.5
credit from yeast revenue 12,994.2 t/y, $1.22/kg 15.85
total production cost per gal ($/gal), including gluten and yeast credit 0.963

The dilution rate for yeast fermentation was optimized at kg kg-1, respectively. The ethanol production yield achieved
0.369 h-1, whereas the productivity was 13 kg m-3 h-1. The was 82% of the maximum. It should be stressed that wheat (or
membrane permeabilities for ethanol and water were estimated glucose) was also utilized for enzyme production in fungal
at 0.68 and 0.28 g m m-2 h-1 bar-1, respectively, from the fermentation and yeast growth. Some ethanol (around 2% of
correlations given by Hoover and Hwang (18). the total production) was lost from the bottom stream of the
The raw material (wheat) required for the designed production low-pressure column, which could not be avoided.
capacity would be 96,075 t y-1, whereas the amount of gluten For a production capacity of 10 million gal y-1, the ethanol
extracted from wheat and yeast recovered from the bleed stream production cost was $2.5 gal-1 by considering gluten as the
would be 11,676 and 12,994 t y-1, respectively. The yields of only byproduct. In the case that gluten and pure yeast cells were
ethanol and yeast production from wheat were 0.312 and 0.12 considered as byproducts, the fuel ethanol production cost was
Biotechnol. Prog., 2007, Vol. 23, No. 6 1401

be used for the production of biodegradable plastics (30). In

addition, gluten hydrolysates could be used as fermentation
feedstock for the production of recombinant proteins via
recombinant Escherichia coli cultivations (current research in
the Satake Centre for Grain Process Engineering).
The removal of pearlings (outer layer of wheat kernel) could
generate additional revenue by identifying novel value-added
applications. Koutinas et al. (7) reported that pearlings could
be used for the production of functional foods, monosaccharides,
ferulic acid, arabinoxylan, and germ-rich fractions. In addition,
enzymatic hydrolysis of the hemicellulose fraction of bran-rich
pearlings would result in the release of xylose, fermentative
conversion of which could lead to the production of xylitol,
which has been listed by the U.S. Department of Energy among
the top 12 value-added platform molecules for the production
Figure 4. Effect of bioethanol plant capacity on production cost with of a range of chemicals, including xylaric acid, propylene glycol,
or without byproduct credits. ethylene glycol, mixture of hydroxyl-furans, and polyesters (44).
It should be stressed that the revenue from pearlings was not
$0.96 gal-1. The production cost of fuel ethanol could be further taken into consideration in this study which means that further
reduced to $0.50 gal-1 (Figure 4) by increasing the production process improvements should be anticipated.
capacity up to 33.5 million gal y-1. Figure 4 shows that the
revenue from byproducts influences significantly fuel ethanol Conclusions
production cost. A novel wheat-based biorefining strategy has been evaluated
Wooley et al. (1) reported that bioethanol could be produced for the production of fuel ethanol. The processing schemes that
from lignocellulosic biomass within the range of $1.16-1.44 were chosen for upstream, bioconversion, and downstream
gal-1 depending on the technology and the availability of low stages were selected from experimental or literature-cited
cost feedstocks for conversion to ethanol. Kwaitkowski et al. information. A nutrient-complete fermentation feedstock was
(4) used the software Superpro Designer to estimate the cost of produced based on a continuous process including fungal
bioethanol production by a corn dry milling process at $1.04 fermentation to produce enzymes and fungal cells integrated
gal-1 for 40 million gal y-1 production capacity. Similarly with feedstock production by starch hydrolysis and fungal
Krishnan et al. (2) estimated the production cost of bioethanol autolysis. Gluten was extracted from wheat as byproduct. The
at $1.11 and $1.08 gal-1, using the simulator Aspen plus, for a proposed bioconversion system involved continuous yeast
15 million gal y-1 corn dry milling plant (base case) and a corn fermentation with cell recycling and ethanol concentration by
dry milling plant using fluidized-bed bioreactors, respectively. pervaporation. Pure yeast cells were separated as byproduct.
McAloon et al. (39) reported a fuel ethanol production cost of The implementation of heat integration between the low- and
$0.88 gal y-1 for a plant capacity of 25 million gal y-1 using high-pressure columns involved in the proposed PSD system
a corn dry milling process. The proposed process evaluated in for fuel-grade ethanol production resulted in 44% reduction in
this study could lead to fuel ethanol production from wheat at the operating cost of the downstream process. Theoretical
a comparable cost to the U.S. corn-based bioethanol production optimization of the proposed wheat-based biorefinery resulted
process. in a fuel ethanol production cost that is lower than the current
production cost of petrol in the UK for 10 millions gal y-1
The cost of bioethanol from wheat dry milling was reported
production capacity.
to be (Can)$2 gal-1 for 2.6 million gal y-1 production capacity
(40) and (Can)$1.75 gal-1 for 26.4 million gal y-1 plant capacity Acknowledgment
(41). The cost of bioethanol production based on UK wheat
was reported to be $2.4 gal-1 for 7.93 million gal y-1 plant The authors gratefully acknowledge the generous contribution
of the Satake Corporation of Japan in providing financial support
capacity (42, 43). It is obvious that the cost of bioethanol from
for much of the research carried out in the Satake Centre for
wheat is higher than the corn-based ethanol due to the high
Grain Process Engineering (SCGPE, University of Manchester,
market price of wheat and the low value of wheat-derived UK).
byproducts. However, the cost of bioethanol could be further
reduced by using set-aside land (43). The price of wheat on Notation
set-aside land could be reduced more than half of the actual
price. The proposed wheat-based process evaluated in this study For bioconversion system
could lead to fuel ethanol production at a lower cost as compared Am membrane area, m2
to current wheat-based processes. D dilution rate, h-1
The current production cost of petrol in the UK was reported Fs volumetric feed flowrate of fermenter at steady state,
to be $1.55 gal-1 (42). Therefore, the process evaluated in this m3 h-1
study could lead to bioethanol production at a lower cost than FB bleed flow rate, m3 h-1
the UK petrol production cost for a plant capacity of 10 millions FR recycle flow rate, m3 h-1
gal y-1. FP product flow rate, m3 h-1
The production cost of bioethanol depends on the selling price F1 retentate flow rate, m3 h-1
of the byproducts (i.e., gluten and yeast). In addition, increased Fp,j flow of permeating components from membrane unit,
ethanol production would increase the production of such kmol h-1
byproducts, reducing their market value for conventional Fm,p mass flow rate of product from reactor, kg h-1
applications. Therefore, it is imperative to identify non-food F0 volumetric feed flowrate to proposed bioconversion
applications for these byproducts. For example, gluten could system m3 h-1
1402 Biotechnol. Prog., 2007, Vol. 23, No. 6

J set of no. of components involved in the system, i.e., E xfj mole fraction of jth component in liquid feed to first
(for ethanol) and W (for water) in this case column
Ksx substrate saturation constant for cell growth, kg m-3 xf1j mole fraction of jth component in overall feed to
Ksp substrate saturation constant for product formation, kg distillation system
m-3 yijc mole fractions of jth component in vapour on ith tray
Kxp inhibition constant for cell growth, h-1 of column c
Kpp inhibition constant for product formation, kg kg-1 h-1 zic binary variable associated with tray i in column c; value
P ethanol concentration, kg m-3 of zic is one when tray i exists in column c and 0
otherwise
Px,max maximum ethanol concentration above which cells do
not grow, kg m-3 Abbreviations
Pp,max maximum ethanol concentration above which cells do CMF continuous membrane fermentation
not produce ethanol, kg m-3 ED extractive distillation
psat
j
saturated vapor pressure of component j, bar FAN free amino acid nitrogen
Q′j permeabilities of permeating component j, mol m m-2 GAMS general algebraic modelling
s-1 kpa-1 LP low pressure column
S final glucose concentration, kg m-3 HP high pressure column
S1 glucose concentration in retentate stream, kg m-3 HMD hybrid membrane distillation
S0 glucose concentration in feed stream, kg m-3 HSDMS hybrid separation involving distillation and molecular
Si glucose concentration in reactor inlet kg m-3 sieve
V volume of fermenter, m3 NLP nonlinear programming
X biomass concentration, kg m-3 MINLP mixed integer nonlinear programming
X1 yeast concentration in retentate stream, kg m-3 PSD pressure swing distillation
Xi yeast concentration in reactor inlet, kg m-3
xp,j molar fractions of components j in product stream
Yx/s yield coefficient (kg of cell mass per kg of glucose) References and Notes
Yp/s yield coefficient (kg of product per kg of glucose)
(1) Wooley, R.; Ruth, M.; Glassner, D.; Sheehan, J. Process design
R recycle flow fraction
and costing of bioethanol technology: A tool for determining the
β bleed flow fraction status and direction of research and development. Biotechnol. Prog.
γ product flow fraction 1999, 15, 794-803.
ηmax maximum specific production rate, kg kg-1 h-1 (2) Krishnan, M. S.; Taylor, F.; Davison, B. H.; Nghiem, N. P.
µ specific growth rate, h-1 Economic analysis of fuel ethanol production from corn starch using
µmax maximum specific growth rate, h-1 fluidized-bed bioreactors. Bioresour. Technol. 2000, 75, 99-105.
δm membrane thickness, m (3) Brien, D. J. O.; Craig, J. C. Ethanol production in a continuous
fermentation/membrane pervaporation system. Appl. Microbiol.
For PSD system Biotechnol. 1996, V44, 699-704.
buic amount of reboil entering tray i of column c, kmol h-1 (4) Kwiatkowski, J. R.; McAloon, A. J.; Taylor, F.; Johnston, D. B.
Modeling the process and costs of fuel ethanol production by the
f molar flow rate of feed to first column, kmol h-1 corn dry-grind process. Ind. Crops Prod. 2006, 23, 288-296.
f1 molar flow rate of fresh feed to PSD system, kmol h-1 (5) Koutinas, A. A.; Wang, R.-H.; Webb, C. Restructuring upstream
L fagacity coefficient of jth component in liquid on ith
fi,j,c bioprocessing: technological and economical aspects for production
tray of column c of a generic microbial feedstock from wheat. Biotechnol. Bioeng.
H fagacity coefficient of jth component in vapour on ith
fi,j,c 2004, 85, 524-538.
tray of column c (6) Webb, C.; Koutinas, A. A.; Wang, R.-H. Developing a sustainable
fmax maximum flow rate, kmol h-1 bioprocessing strategy based on a generic feedstock. AdV. Biochem.
hlic enthalpy of liquid stream entering in tray i of column Eng./Biotechnol. 2004, 87, 195-268.
c, kJ kmol-1 (7) Koutinas, A. A.; Wang, R.-H.; Campbell, G. M.; Webb, C. A whole
hvic enthalpy of vapour stream entering in tray i of column crop biorefinery system: A closed system for the manufacture of
c, kJ kmol-1 non-food products from cereals. In Biorefineries-Industrial Pro-
hlfc enthalpy of feed stream entering in tray i of column c, cesses and Products; Kamm, B., Gruber, P. R., Kamm, M., Eds.;
kJ kmol-1 Wiley-VCH: Weinheim, 2006; Vol. 1, Chapter 8, pp 165-191.
hp enthalpy of overall feed to PSD system, kJ kmol-1 (8) Koutinas, A. A.; Wang, R.-H.; Webb, C. Development of a process
lic molar flow rate of liquid leaving tray i of column c, for the production of nutrient supplements for fermentations based
kmol h-1 on fungal autolysis. Enzyme Microb. Technol. 2005, 36, 629-638.
MWE molecular weight of ethanol (9) Arifeen, N.; Wang, R.-H.; Kookos, I. K.; Webb, C.; Koutinas, A.
P1c top product rate from column c, kmol h-1 A. Optimisation and cost estimation of novel wheat biorefining for
continuous production of fermentation feedstock. Biotechnol. Prog.
P2c bottom product rate from column c, kmol h-1
2007, 23, 872-880.
pic pressure of tray i of column c, bar
(10) Koutinas, A. A.; Arifeen, N.; Wang, R.-H.; Webb, C. Cereal-
refic amount of reflux entering tray i of column c, kmol h-1 based biorefinery development: Integrated enzyme production for
rc reflux ratio in column c cereal flour hydrolysis. Biotechnol. Bioeng. 2007, 97, 61-72.
tic temperature of tray i of column c, K (11) Arifeen, N.; Koutinas, A. A.; Wang, R.-H.; Kookos, I. K.; Webb,
tp temperature of overall feed to PSD system, K C. Experimental and modelling studies of bioethanol production via
Vic molar flow rate of vapour leaving tray i of column c, novel wheat biorefining. Biotechnol. Prog. Submitted for publication.
kmol h-1 (12) Vane, L. M. A review of pervaporation for product recovery from
xijc mole fractions of jth component in liquid on ith tray of biomass fermentation processes. J. Chem. Technol. Biotechnol. 2005,
column c 80, 603-629.
Biotechnol. Prog., 2007, Vol. 23, No. 6 1403

(13) Kreis, P.; Gorak, A. Process analysis of hybrid separation (32) Ghose, T. K.; Tyagi, R. D. Rapid ethanol fermentation of cellulose
processes: Combination of distillation and pervaporation. Chem. Eng. hydrolysate. II. Product and substrate inhibition and optimization
Res. Des. 2006, 84 (A7), 595-600. of fermentor design. Biotechnol. Bioeng. 1979, 21, 1401-1420.
(14) Bausa, J.; Marquardt, W. Shortcut design methods for hybrid (33) Warren, R. K.; Macdonald, D. G.; Hill, G. A. Improved bioreaction
membrane/distillation processes for the separation of nonideal kinetics for the simulation of continuous ethanol fermentation by
multicomponent mixtures. Ind. Eng. Chem. Res. 2000, 39, 1658- Saccharomyces cereVisiae. Biotechnol. Prog. 1990, 6, 319-325.
1672. (34) Ge, X. M.; Bai, F. W. Intrinsic kinetics of continuous growth
(15) Jaksland, C.; Hytoft, G.; Gani, R. Computer-aided process design and ethanol production of a flocculating fusant yeast strain SPSC01.
and optimization with novel separation units. Appl. Thermal Eng. J. Biotechnol. 2006, 124, 363-372.
1997, 17, 973-980. (35) Aiba, S.; Shoda, M.; Nagatani, M. Kinetics of product inhibition
(16) Pressly, T. G.; Ng, K. M. A break-even analysis of distillation- in alcohol fermentation. Biotechnol. Bioeng. 2000, 67, 671-690.
membrane hybrids. AIChE J. 1998, 44, 93-105. (36) Phisalaphong, M.; Srirattana, N.; Tanthapanichakoon, W. Math-
(17) Lipnizki, F.; Field, R. W.; Ten, P.-K. Pervaporation-based hybrid ematical modeling to investigate temperature effect on kinetic
process: a review of process design, applications and economics. parameters of ethanol fermentation. Biochem. Eng. J. 2006, 8, 36-
J. Membr. Sci. 1999, 153, 183-210. 43.
(18) Hoover, K. C.; Hwang, S.-T. Pervaporation by a continuous
(37) Cho, C.-W.; S.-Hwang, T. Continuous membrane fermentor
membrane column. J. Membr. Sci. 1982, 10, 253-271.
separator for ethanol fermentation. J. Membr. Sci 1991, 57, 21-42.
(19) Viswanathan, J.; Grossmann, I. E. A combined penalty function
and outer-approximation method for MINLP optimization. Comp. (38) Kargupta, K.; Datta, S.; Sanyal, S. K. Analysis of the performance
Chem. Eng. 1990, 14, 769-782. of a continuous membrane bioreactor with cell recycling during
(20) Blanch, H. W.; Clark, D. S. Biochemical Engineering; Marcel ethanol fermentation. Biochem. Eng. J. 1998, 1, 31-37.
Dekker: New York, 1997; pp 643-682. (39) McAloon, A.; Taylor, F.; Yee, W.; Ibsen, K.; Wooley, R.
(21) Turton, R.; Bailie, R. C.; Whiting, W. B.; Shaeiwitz J. A. Analysis, Determining the cost of producing ethanol from corn starch and
Synthesis, and Design of Chemical Processes; Prentice Hall: New lignocellulosic feedstocks. A Joint Study sponsored by U.S. Depart-
Jersey, 2003; pp 187-197. ment of Agriculture and U.S. Department of Energy. National
(22) Garrett, D. E. Chemical Engineering Economics; Van Nostrand Renewable Energy Laboratory, http://www.nrel.gov/docs/fy01osti/
Reinhold: New York, 1989; pp 256-307. 28893.pdf; 2000.
(23) Perry, R. H.; Green, D. W.; Maloney, J. O. Perry’s Chemical (40) Lang, X.; MacDonald, D. G.; Hill, G. A. Recycle bioreactor for
Engineers’ Handbook; McGraw-Hill: New York, 1984. bioethanol production from wheat starch II. Fermentation and
(24) Peters, M. S.; Timmerhaus, K. D. Plant Design and Economics economics. Energy Sources 2001, 23, 427-436.
for Chemical Engineers; McGraw-Hill: New York, 1991; pp 515- (41) Warren, R. K.; Macdonald, D. G.; Hill, G. A. The design and
534. costing of a continuous ethanol process using wheat and cell recycle
(25) Ulrich, G. D. A Guide to Chemical Engineering Process Design fermentation. Bioresour. Technol. 1994, 47, 121-129.
and Economics; John Wiley & Sons: New York, 1984. (42) Anonymous. Proposed UK policy. http://www.esru.strath.ac.uk/
(26) Guthrie, K. M. Capital cost estimating. Chem. Eng. 1969, 76, 114. EandE/Web_sites/02-03/biofuels/proposed_uk.htm; 2007.
(27) Anonymous. Economic indicators. Chem. Eng. 2006, 113 (4), 104. (43) Batchelor, S. E.; Cook, P.; Booth, E. J.; Walker, K. C. Economics
(28) Coulson, J. M.; Richardson, J. F.; Sinnott, R. K. Chemical of bioethanol production from wheat in the UK. Renewable Energy
Engineering-Design; Butterworth and Heinemann: Oxford, 1999; 1994, 5, 807-809.
pp 242-279. (44) Werpy, T.; Petersen, G. Top value added chemicals from biomass.
(29) Reisman, H. B. Economic Analysis of Fermentation Processes; Volume I. Results of screening for potential candidates from sugars
CRC Press: Boca Raton, FL, 1988. and synthesis gas. Produced by staff at the Pacific Northwest National
(30) Pallos, F. M.; Robertson, G. H.; Pavlath, A. E.; Orts, W. J. Laboratory (PNNL) and the National Renewable Energy Laboratory
Thermoformed wheat gluten biopolymers. J. Agric. Food Chem. (NREL). http://www1.eere.energy.gov/biomass/pdfs/35523.pdf; 2004.
2006, 54, 349-352.
(31) Koutinas, A. A.; Athanasiadis, I.; Bekatorou, A.; Iconomopoulou, Received May 15, 2007. Accepted July 30, 2007.
M.; Blekas, G. Kefir yeast technology: Scale-up in SCP production
using milk whey. Biotechnol. Bioeng. 2005, 89, 788-796. BP0701517