Bioresource Technology 144 (2013) 311320Contents lists available at SciVerse ScienceDirect

Bioresource Technology

journal homepage: www.elsevier .com/locate /bior techScale-up and kinetic modeling for bioethanol production0960-8524/$ - see front matter 2013 Elsevier Ltd. All rights reserved.http://dx.doi.org/10.1016/j.biortech.2013.06.118

Corresponding author. Tel./fax: +90 2323884955.E-mail address: esraimamoglu@yahoo.com (E. Imamoglu).Esra Imamoglu , Fazilet Vardar SukanDepartment of Bioengineering, Faculty of Engineering, University of Ege, 35100 Bornova, Izmir, Turkey

h i g h l i g h t s

Methodologies of scale up procedure for bioethanol production. Kinetic modeling for the behaviors of Escherichia coli KO11. Importance of the rheological behavior of fermentation broth. Hydrodynamic parameters of bioreactors were evaluated.a r t i c l e i n f o

Article history:Received 4 May 2013Received in revised form 27 June 2013Accepted 29 June 2013Available online 5 July 2013

Keywords:Bioethanol productionScale-upShaken flaskStirred-tank bioreactorsKinetic modelinga b s t r a c t

Bioethanol was produced from acidic hydrolysate of rice hulls using recombinant Escherichia coli KO11.Two different issues (scale-up and kinetic modeling) were evaluated simultaneously and concomitantlyfor bioethanol production. During the step-wise scale-up process from 100 mL shaken flask to 10 L stir-red-tank bioreactor, the constant Reynolds number and the constant impeller tip speed were evaluated asscale-up methodologies under laboratory conditions. It was determined that the volumetric bioethanolproductivity was 88% higher in 10 L bioreactor in comparison to the value of 0.21 g L1 h1 in shakenflask. The modified Monod and Luedeking-Piret models provided an accurate approach for the modelingof the experimental data. Ethanol concentration reached the maximum level of 29.03 g/L, which was5% higher than the value of model prediction in 10 L bioreactor. The findings of this research could con-tribute to the industrial scale productions especially from lignocellulosic raw materials.

2013 Elsevier Ltd. All rights reserved.1. Introduction

Bioethanol is considered as one of the most promising next-generation automotive fuels, as it is carbon neutral and can be pro-duced from renewable resources, like lignocellulosic materials(Tamburini et al., 2011). The demand for bioethanol in the EU is ex-pected to rise to 28.5 billion liters by 2020 (Wang et al., 2012). Sta-tistics indicate that the world ethanol production is 60% from sugarcrops, 33% from other crops and 7% from chemical synthesis (Dex-ter and Fu, 2009) and only two countries U.S. (corn-derived) andBrazil (sugar cane-derived) monopolize more than 90% of theworlds bioethanol production (Serrano-Ruiz et al., 2012). Eventhough bioethanol is the predominant biofuel today, it has impor-tant compatibility, energydensity and water-absorption issuesthat limit its further implementation in the current fuel infrastruc-ture (Serrano-Ruiz and Dumesic, 2011; Serrano-Ruiz et al., 2012).

The biorefinery industry is looking for cheaper, abundant andmore accessible feedstock for producing bioethanol (Wang et al.,2012). In rice industry, biomass residues that are commonly uti-lized for energy generation are rice straw and rice hull, which arecrop residue and agro-industrial residue, respectively. Rice hull isthe outer layer of a rice seed, and removed from the rice seed asa by-product during the milling process. It consists of 28.6% cellu-lose, 28.6% hemicelluloses, 24.4% lignin and 18.4% extractive mat-ter. Its annual production is approximately 137.05 million tonswhich is distributed in Africa (4.9 million tons), America (7.62 mil-lion tons), Asia (123.65 million tons), Europe (0.82 million tons),and Oceania (0.06 million tons) (Lim et al., 2012). The rice hullsin these quantities could be converted to 38.4 billion liters bioeth-anol per year, approximately.

With increasing interest in the industrial application of batchalcoholic fermentation, various kinetic models have been exam-ined for microbial growth, product formation and substrate con-sumption. Kinetic modeling may be regarded as an importantstep in developing a fermentation process, since models help inprocess control, reducing process costs and increasing productquality (Dodic et al., 2012). Microbial processes are inherentlycomplex, and it is of critical importance in practical applicationsto develop models that provide an accurate description of the pro-cess without unnecessary complexity (Luong et al., 1988). Unstruc-tured kinetic models are frequently used because of theirsimplicity and adequacy for technical purposes (Garcia-Ochoaet al., 1995; Richard and Margaritis, 2004). In general, these models

http://crossmark.dyndns.org/dialog/?doi=10.1016/j.biortech.2013.06.118&domain=pdfhttp://dx.doi.org/10.1016/j.biortech.2013.06.118mailto:esraimamoglu@yahoo.comhttp://dx.doi.org/10.1016/j.biortech.2013.06.118http://www.sciencedirect.com/science/journal/09608524http://www.elsevier.com/locate/biortech

Nomenclature

Ac cross-sectional area or area available for oxygen ex-change, m2

EtOH ethanolDi impeller diameter, mDint internal diameter of vessel, mdf largest inner diameter of shaken flask, mdo shaking diameter, mDT doubling time, hFr Froude number, dimensionlessg gravitational acceleration, 9.81 m s2

HAc acetic acidHFor formic acidHLac lactic acidHSuc succinic acidi the number of experimental data points, dimensionlessj degree of product inhibition, dimensionlessk constant with magnitude dependent on the geometry of

the impeller, dimensionless,Ks saturation constant, g/LKI inhibition constant, g/LN impeller rotation speed, rpmNp power number for stirred tank, dimensionlessN0p modified power number for unbaffled shaken flask,

dimensionlessn shaking frequency, rpmP product concentration, g/LPm maximum product concentration, g/LPh phase number, dimensionlessPo unaerated power consumption for stirred tank, W

Pf power consumption for shaken flask, WPo/VL power consumption per unit volume of liquid, W m3

qA specific reaction rate of A, h1

Rei impeller Reynolds number, dimensionlessRef Reynolds number for shaken flask, dimensionlessS substrate concentration, g/LVL working volume, m3

ttip impeller tip speed, m s1

X biomass concentration, g/LYAB yield of A from B, g g1

Z minimizing objective function of the discrepancy be-tween results, dimensionless

Greek symbolsl Specific growth rate, h1

lmax Maximum specific growth rate, h1

a Inhibition constant for nonspecific inhibitors in thehydrolysate, dimensionless

e The rate of energy dissipation per unit mass of fluid,W kg1

k Kolmogorovs eddy size, lmc Shear rate, s1

s Shear stress, N m2

q Fluid density, kg m3

g Dynamic viscosity of fluid, kg m1 s1

t Kinematic viscosity of fluid, m2s1

312 E. Imamoglu, F.V. Sukan / Bioresource Technology 144 (2013) 311320use a form of the logistical equation to describe biomass growthand a form of the Luedeking-Piret model to describe biopolymerformation as a function of biomass. In order to fit their experimen-tal data, several researchers have also used Monod type equationsto better describe biomass growth and product formation in termsof the limiting substrate (Moraine and Rogovin, 1966; Garcia-Ochoa et al., 1995) and others have proposed substrate dependentequations which are not of theMonod type (Quinlan, 1986; Richardand Margaritis, 2004).

The two main objectives of the present study were: (i) to eval-uate the use of both the constant Reynolds number and the con-stant impeller tip speed as scale-up methodologies underlaboratory conditions for the step-wise scale-up process from100 mL shaken flask to 10 L stirred-tank bioreactor, consideringwhether an increase in the bioethanol yield can be achieved, and(ii) to validate the kinetic models applied for better describingthe behaviors of Escherichia coli KO11 during bioethanol fermenta-tion from rice hulls hydrolysate.2. Methods

2.1. The preparation and dilute acid pre-treatment of rice hulls

The rice hulls were obtained from Kahya rice factory, Izmir, Tur-key. They were milled in a hammer mill (Brook Crompton Series2000, UK) to pass through a 1 mm screen. The milled rice hullswere dried in an oven (Memmert GmbH, Germany) at 70 C forone night.

The milled rice hulls at a solid loading of 30% (w/w) were mixedwith 0.40 M H2SO4 and pretreated in an autoclave at 121 C underthe pressure of 0.10 MPa for 60 min. The detoxification processwas carried out using 340 mM Ca(OH)2 at 60 C for 30 min at400 rpm. The hydrolysate was adjusted to initial pH 6 with 6 MKOH and then separated (10,000 rpm, 30 min) using centrifugeseparator (Westfalia Seperator Mineral oil Systems GmbH D-59302, Germany) to remove any precipitate formed before usingit as substrate. The dry LB ingredients were added to the hydroly-sate and were not sterilized for fermentation (Moniruzzaman andIngram, 1998).

2.2. Strain maintenance and preparation of Inocula

E. coli KO11 was provided by courtesy of Professor L.O. Ingram(University of Florida, USA). The recombinant E. coli KO11 is thederivative of E. coli B and contains the chloramphenicol acyl trans-ferase gene (cat) and the Z. mobilis genes encoding alcohol dehy-drogenase (adhB) and pyruvate decarboxylase (pdc) for ethanolproduction. Stock cultures were stored in 40% glycerol at 80 C.

The cells from a single well-isolated colony were inoculatedinto 250 ml cotton-plugged-conical flasks containing 50 ml ofmodified LuriaBertani (LB) medium (Moniruzamman et al.,1997) with 50 g/L glucose. The cultures were incubated under sta-tic condition for 16 h at 30 C in the absence of antibiotic. Then, thecells were harvested by centrifugation (6000g, 5 min and 5 C) andused as inocula for the experiments at the initial dry weight of0.72 g cell/L.

2.3. Bioethanol production conditions

For the small-scale production, the cells were incubated on anorbital shaker (IKA KS 4000ic Thermoshake, Werke GmbH & Co.KG, Germany) with a 20 mm shaking diameter at a shaking fre-quency of 228 rpm in 100 mL flask (62 mm diameter) containing80 mL of rice hulls hydrolysate with dry LB ingredients at the tem-perature of 30 C under semi-anaerobic condition during 72 h of

E. Imamoglu, F.V. Sukan / Bioresource Technology 144 (2013) 311320 313the fermentation period. The semi-anaerobic condition of fermen-tation was allowed to reach anaerobic conditions through con-sumption of dissolved oxygen (DO) during the early stage offermentation.

For the stirred-tank bioreactor productions, bioethanol fermen-tations were carried out in 2 L (Sartorius Biostat B-plus, Ger-many), 5 L (Sartorius Biostat A-plus, Germany) and 10 L(Sartorius Biostat B-plus, Germany) stirred-tank bioreactors atthe stirrer rates of 312 rpm, 254 rpm and 220 rpm with the work-ing volumes of 1.5, 4 and 8 L, respectively at 30 C under semi-anaerobic conditions during 72 h of the fermentation period. Allbioreactors were equipped with 6-blade Rushton turbine impellersand the probes of pH and DO. The pH was maintained at 6.0 byautomatic addition of 2 N KOH to prevent acidification. The baffleinsert consisted of a holder with four vertical baffles. For 10 L bio-reactor, one baffle was attached to the gassing pipe and the otherwas mounted into a 6 mm lid port. Details of bioreactor dimen-sions and configurations are given in Table 1.

2.4. Scale-up strategies

The bioethanol production by E.coli KO11 was step-wise scaled upfrom 100mL shaken flask to 10 L stirred-tank bioreactor under labo-ratory conditions as shown in Fig. 1. The scale-up was conductedfrom shaken flask to 2 L bioreactor using the constant Reynolds num-ber. For the stirred-tank bioreactors, the scale-up was performedusing the constant impeller tip speed as a scale-up methodology.Mathematical equations of scale-up process for bioethanol produc-tion are presented in Table 2. For stirred-tank bioreactors, the powernumbers were obtained from a Power number curve as a function ofthe impeller Reynolds number and the impeller type (Geankoplis,1983). The characteristic empirical constant (k) for a standardRushton turbine impeller was taken as 10 (Doran, 1995).Table 1Studied bioreactor dimensions AND configurations.

Dimensions andConfigurations

2 LBioreactor

5 LBioreactor

10 LBioreactor

VL (L) 1.50 4.00 8.00Number of impellers 1.00 2.00 3.00Number of blades per impeller 6.00 6.00 6.00Number of baffles 4.00 4.00 2.00Dt (cm) 16.0 20.0 22.0Dint (cm) 13.0 16.0 19.0Di (cm) 5.30 6.50 7.50Db (cm) 12.5 15.4 NaDs (cm) 4.00 6.00 8.00Ht (cm) 24.0 34.5 47.0Hl (cm) 11.7 19.8 28.8Hi (cm) 4.00 5.50 8.00Hii (cm) Na 14.4 12.5Hsi (cm) 2.50 3.50 5.50Hs (cm) 1.50 2.00 2.50Hb (cm) 13.0 20.4 28.0Wb (cm) 1.00 1.50 2.00Li (cm) 1.40 1.60 2.00Wi (cm) 1.00 1.50 1.50Dt/Di 3.00 3.08 2.93Ds/Di 0.76 0.92 1.06Hl/Di 2.21 3.05 3.84Hi/Di 0.76 0.85 1.06Ht/Dt 1.50 1.73 2.14Wb/Dt 0.06 0.08 0.09Wi/Di 0.19 0.23 0.20Li/Di 0.26 0.25 0.27

VL, working volume; Dt, diameter of vessel; Dint, internal diameter of vessel; Di,diameter of impeller; Db, diameter of baffle; Ds, diameter of ring sparger; Ht, heightof vessel; Hl, height of liquid; Hi, distance of the low impeller to the bottom ofvessel; Hi-i, distance of two impellers; Hs-i, distance of the sparger to the impeller;Hs, distance of the sparger to the bottom of vessel; Hb, height of baffle; Wb, width ofbaffle; Li, length of impeller; Wi, width of impeller.2.5. Description of kinetic modeling

The kinetic modeling was developed based on the followingassumptions:

1. The mutual inhibition effects of glucose and xylose consumptionswere determined by experimental analysis for bioethanol produc-tion process from rice hulls hydrolysate using E. coli KO11. Theexperimental data showed that in the rice hulls hydrolysate, glu-cosewaspresentat about twoandahalf timeshigherconcentrationthan xylose at the beginning of the fermentation period. The xyloseconsumption rate was slow compared with the glucose consump-tion rate.When the concentrationsof glucose andxylosewere closeto each other, the xylose consumption rate increased in themiddleof fermentationperiod. Fromaprocesspointof view, themodelwascreated based on the diauxic phenomenon: First the glucose andsmall amounts of xylose (neglected) were fermented simulta-neously during the first 36 h of the fermentation period, and thenthe remaining parts of xylose were fermented until the end ofthe fermentation period.l; P; S lglu; Pglu; Sglu 0 t 36lxyl; Pxyl; Sxyl 36 t 72

" #172. It is assumed that the specific growth rate, l, could be ade-quately described by the modification of the well-knownMonodmodel (Monod, 1949), which includes uncompetitive substrateinhibition and the product inhibition by the generalized nonlin-ear equation suggested by Levenspiel (1980):l either lglu or lxyl 18lglu lmax

gluSglu

KSglu Sglu S2gluKIglu

SgluSxylKlxyl

1 PPm

jglu19lxyl lmax

xylSxyl

KSxyl Sxyl S2xylKIxyl

SxylSgluKlglu

1 PPm

jxyl20The above model equations are applicable for both glucose (sub-script glu) and xylose (subscript xyl) as substrates.

3. Assuming cell death was negligible, the biomass formation ratewas shown in Eq. (21):dXdt

lX 21Integrating Eq. (21) with the boundary condition produced:Xt either X0glu elglut or X0xyl e

lxylt 224. A large number of inhibitors (such as furfural and hydroxymeth-ylfurfural) appeared in the hydrolysate after the dilute acid pre-treatment of rice hulls. aglu and axyl are the inhibition constantsfor the product formation from glucose and xylose, respectively.

5. Ethanol was produced as a major fermentation product andorganic acids (acetic, formic, lactic and succinic acids) weresimultaneously formed as by-products during the fermentationby E. coli KO11. The formation rates of ethanol and organic acidscould be described using the Luedeking-Piret model (Luedekingand Piret, 1959) for the growth-associated type of products as afunction of biomass. The maintenance coefficient for productformation was neglected. The product formation rate was givenby Eq. (23):dPdt

aYpXlX 23

Fig. 1. Flow diagram of scale-up bioethanol production.

Table 2Mathematical equations of scale-up process for bioethanol production.

Scale-up Model Equations

Rei NiqD2i

g(1) e NpN3i D

2i

(9)

Ref nd2ft

(2)k t3e

1=4 (10)t gq (3) c kNi (11)#t pNiDi (4) s gc (12)P0 NpqN3i D

5i

(5) sf et 1=2g (13)

Pf N0pqn3d4f V

1=3L

(6) Fri N2i Dig

(14)

N0p 70Re1f 25Re

0:6f 1:5Re

0:2f

(7) Frf 2pn2do

2g > 0:4(15)

Ac pD2int4

(8)Ph d0df 1 3log10

q2png

d2f4 1

ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi1 4p

V0:33Ldf

h i2r !22435

8 0.05) difference in bioethanol concentrations, whereasbiomass concentrations (p < 0.05) were found significantly differ-ent. The ethanol concentrations in shaken flask, 2 and 5 L bioreac-tors were 48%, 28% and 16% lower than in 10 L bioreactor,respectively. On the other hand, ethanol concentrations did notreach inhibitory levels in the experimental studies. As seen fromthe plots, the modifiedMonod and Luedeking-Piretmodels providedan accurate approach for the modeling of the experimental data(Fig. 2). The ethanol production rate in the first 36 h of the fermen-tation period was faster than in the second stage of the periodwhere the remaining parts of the xylose were fermented (Figs. 2and 3). Additionally, it was recorded that the ethanol productionFig. 3. Experimental data and model predictions of glucose and xylose concentrations focoli KO11 in different scales of production: (a) shaken flask, (b) 2 L bioreactor, (c) 5 L brepresent experimental data. (j) Glucose, (N) Xylose, () model.rate decreased as the lactic acid formation rate increased in thesecond stage of the fermentation period.

According to literature, the highest level of ethanol (68 g/L) wasachieved by E. coli KO11 after 108 h in Luria broth containing140 g lactose/L (Guimaraes et al., 1992). In another research, max-imum ethanol concentration and degree of ethanol inhibition werefound to be 49.20 g/L and 0.52, respectively for E. coli KO11 (Olssonand Hahn-Hagerdal, 1995). For Zymomonas mobilis ZM4, the max-imum inhibitory ethanol concentration for growth on glucose(57.20 g/L) was similar to that for growth on xylose (56.30 g/L)(Leksawasdi et al., 2001). In this study, during the fitting procedurethe model parameter values indicated that maximum ethanol con-centration was 54 g/L for each sugar and the degrees of ethanolinhibition on growth for glucose and xylose were found 0.22 and0.06, respectively (Table 5). When the experimental data is takeninto consideration, it was concluded that the inhibitory effect ofethanol on growth for E. coli KO11 is above a threshold level of5458 g/L, except for some extreme cases. This result is in agree-ment with the literature values. The determined values of KS, KI,and a are listed in Table 5.

During the fermentation period, the cells predominantly usedsugar resources to build up and produce primary metabolites.r bioethanol production from rice hulls hydrolysate using recombinant Escherichiaioreactor and (d) 10 L bioreactor. Lines represent model predictions and symbols

318 E. Imamoglu, F.V. Sukan / Bioresource Technology 144 (2013) 311320The glucose and xylose were not consumed simultaneously for thebioethanol production, possibly because the same amounts of glu-cose and xylose were not present in the hydrolysate at the begin-ning of the fermentation. Little decline in xylose concentration wasobserved with a sharp decrease in glucose concentration, espe-cially in the first 36 h of the fermentation period. In short, glucosewas consumedmuch more rapidly than xylose. As shown in Fig. 3c,the glucose concentration reduced gradually from 37.77 g/L to11.58 g/L, whereas xylose concentration decreased by only 23%(p > 0.05), which resulted in an ethanol concentration of 18.54 g/L within 36 h in 5 L bioreactor, since sugar content is directly asso-ciated to ethanol yield. But the stronger mutual inhibition effects ofglucose and xylose on each other appeared after the 42 h of fer-mentation period. This may cause the unpredictable variations inmetabolic pathway of E. coli KO11 for the production of ethanol.The experimental glucose concentration decreased 27 times incomparison to the initial glucose concentration (38.20 g/L) whenthe value of model prediction was 0.88 g/L in 2 L bioreactor atthe end of the fermentation period (Fig. 3b). Parallel changes onthe sugar consumptions were monitored for all different scales ofethanol productions. The model accurately predicts the glucoseand xylose concentration profiles.

In the mixed-substrate fermentation, the presence of glucoseaffects the uptake of xylose and vice versa because transport ofboth sugars uses the same facilitated diffusion system (Kilian andVan Uden, 1988; Tochampa et al., 2005). In this study, the inhibi-tion constant of glucose and xylose were found 14.50 and 0.27,respectively. The (un)competitive substrate inhibition dependson the inhibition constants, while the mutual inhibition of sugarconsumption rates depends only on the type of microorganism.The phenomenon of mutual sugar inhibition has been suggestedpreviously for the fermentation of E. coli by Lawford and Rousseau(1991) and Olsson and Hahn-Hagerdal (1995).Fig. 4. Experimental data and model predictions of organic acid concentrations for bioethin different scales of production: (a) shaken flask, (b) 2 L bioreactor, (c) 5 L bioreactorexperimental data. (e) HAc, () HFor, () HLac, () HSuc, () model.The major by-products of bioethanol productions were aceticand lactic acids (p < 0.05). Lactic acid was produced in the late offermentation stage, especially in 5 and 10 L bioreactors when thesecond main carbon source was consumed (Fig. 4c and d). Ingeneral, whether the second carbon source is glucose or xylosedid not make a difference on lactic acid formation. On the otherhand, in shaken flask, glucose was consumed, xylose was inhib-ited and also lactic acid was not produced during the 72 h of fer-mentation period, as shown in Figs. 3a and 4a. This might bedue to the high stirrer rate. It is noteworthy to mention thatmorphological state affects the metabolic change and vice versa.The model predictions were close to the experimental datathroughout fermentation; only some minor deviations of organicacid concentration profile were monitored for all different scalesof ethanol productions (Fig. 4).

The probable cause on the increase of the formation of lacticacid in bioethanol production is the saturation of the ethanologenicpathway, and therefore additional NAD+ can be generated by lac-tate production (Olsson and Hahn-Hagerdal, 1995). As also re-ported by Song et al. (2008), lactic acid and succinic acidformations share one common physiological meaning of NADHregeneration during the anaerobic fermentation; NADH producedduring glycolysis needs to be regenerated to continuefermentation.

The predicted values for the parameters were obtained by min-imizing the least-square discrepancies between experimental dataand model prediction results. The predicted values of the variablekinetic model parameters are given in Table 6.

Kinetic models contribute to interpretation of cell behavior,essential for successful scale-up procedure. It is important tounderstand why and when this expression is meaningful. If thepredictions provide accurate approaches, the reliabilities of boththe expression and the applied model will not be judged. Oneanol production from rice hulls hydrolysate using recombinant Escherichia coli KO11and (d) 10 L bioreactor. Lines represent model predictions and symbols represent

Table 6Predicted values of the variable kinetic model parameters for the bioethanol production.

Parameter Shaken flask 2 L Bioreactor 5 L Bioreactor 10 L Bioreactor

Glucose t1 Xylose t2 Glucose t1 Xylose t2 Glucose t1 Xylose t2 Glucose t1 Xylose t2

lmax (h1) 0.57 0.10 1.18 0.12 1.34 0.07 3.65 0.15YEtOHX (g g1) 8.89 11.82 10.52 8.81 8.79 14.29 7.70 9.98YHAcX (g g1) 0.45 0.86 1.12 0.02 0.05 0.66 0.10 0.28YHForX (g g1) 0.37 0.24 0.42 0.51 0.13 0.24 0.03 0.25YHLacX (g g1) 0.00 0.00 0.80 1.34 0.00 5.55 0.00 3.56YHSucX (g g1) 0.00 0.81 0.13 0.18 0.19 1.04 0.00 0.79YXGlu (g g1) 0.11 0.01 0.05 0.09 0.08 0.04 0.21 0.02YEtOHGlu (g g1) 0.98 0.09 0.51 0.68 0.76 0.22 1.64 0.16YHAcGlu (g g1) 0.05 0.01 0.05 0.02 0.01 0.01 0.02 0.003YHForGlu (g g1) 0.04 0.003 0.02 0.05 0.01 0.004 0.01 0.003YHLacGlu (g g1) 0.00 0.00 0.04 0.11 0.00 0.22 0.00 0.14YHSucGlu (g g1) 0.00 0.01 0.01 0.02 0.02 0.02 0.00 0.02YXXyl (g g1) 1.35 0.07 0.30 0.16 0.63 0.04 0.62 0.06YEtOHXyl (g g1) 12.31 0.86 3.08 1.27 5.77 0.575 4.87 0.55YHAcXyl (g g1) 0.66 0.07 0.30 0.03 0.06 0.024 0.07 0.01YHForXyl (g g1) 0.49 0.02 0.13 0.09 0.09 0.011 0.03 0.01YHLacXyl (g g1) 0.00 0.00 0.256 0.208 0.00 0.566 0 0.466YHSucXyl (g g1) 0.31 0.07 0.041 0.029 0.16 0.045 0 0.051

t1 = 0 < t 6 36; t2 = 36 < t 6 72.

E. Imamoglu, F.V. Sukan / Bioresource Technology 144 (2013) 311320 319important point to note here is that complicated models offeropportunities for further research requiring numerical analysis inorder to better describe biomass growth and product formation.4. Conclusion

The bioethanol production from rice hulls hydrolysate using re-combinant E. coli KO11 was successfully step-wise scaled-up from100 mL shaken flask to 10 L stirred-tank bioreactor under labora-tory conditions. The experiments on scale-up have given valuableinformation on ways to improve process control. The results ofthe model predictions showed very good agreement with theexperimental data. The findings of this research could contributeto the industrial scale productions especially from lignocellulosicraw materials. It will also provide a useful guide for further re-search aiming to elucidate the interaction between scale-up andkinetic modeling for bioethanol production.Acknowledgements

This study was a part of Cost action FP0602 and the authorswish to thank The Scientific and Technical Research Council of Tur-key (Tubitak) for the financial support.References

Azbar, N., Tutuk, F., Keskin, T., 2009. Effect of organic loading rate on theperformance of an up-flow anaerobic sludge blanket reactor treating olivemill effluent. Biotechnol. Bioprocess Eng. 14, 99104.

Bchs, J., Maier, U., Milbradt, C., Zoels, B., 2000. Power consumption in shakingflasks on rotary shaking machines: II: nondimensional description of specificpower consumption and flow regimes in unbaffled flasks at elevated liquidviscosity. Biotechnol. Bioeng. 68 (6), 594601.

Charles, M., Wilson, J., 1999. Fermenter design. In: Flickinger, M.C., Drew, S.W.(Eds.), Bioprocess Technology: Fermentation, Biocatalysis, and Bioseparation.John Wiley and Sons, New York, pp. 11571189.

Dexter, J., Fu, P., 2009. Metabolic engineering of cyanobacteria for ethanolproduction. Energy Environ. Sci. 2, 857864.

Dodic, J.M., Vucurovic, D.G., Dodic, S.N., Grahovac, J.A., Popov, S.D., Nedeljkovic,N.M., 2012. Kinetic modelling of batch ethanol production from sugar beet rawjuice. Appl. Energy 99, 192197.

Doran, P.M., 1995. Fluid flow and mixing. In: Bioprocess Engineering Principles.Elsevier Science & Technology Books, pp. 129164.

Garcia-Ochoa, F., Gomez, E., 2009. Bioreactor scale-up and oxygen transfer rate inmicrobial processes: an overview. Biotechnol. Adv. 27, 153176.

Garcia-Ochoa, F., Santos, V., Alcon, A., 1995. Xanthan gum production: anunstructured kinetic model. Enzyme Microb. Technol. 17, 206217.Geankoplis, C.J., 1983. Transport principles of momentum transfer and applications.In: Processes and Unit Operations. Allyn and Bacon, Inc., Massachusetts, pp.125195.

Guimaraes, W.V., Dudey, G.L., Ingram, L.O., 1992. Fermentation of sweet whey byethanologenic Escherichia coli. Biotechnol. Bioeng. 40, 4145.

Gungormusler, M., Gonen, C., Azbar, N., 2011. Continuous production of 1,3-propanediol using raw glycerol with immobilized Clostridium beijerinckii NRRLB-593 in comparison to suspended culture. Bioprocess Biosyst. Eng. 34 (6), 727733.

Hiruta, O., Futamura, T., Takebe, H., Satoh, A., Kamisaka, Y., Yokochi, T., 1996.Optimization and scale-up of c-linolenic acid production by Mortierellaramanniana MM15-1, a high c-linolenic acid producing mutant. J. Ferment.Bioeng. 82, 366370.

Jem, K.J., 1989. Scale-down techniques for fermentation. Biopharm Int. 3, 3039.Junker, B.H., 2004. Scale-up methodologies for Escherichia coli and yeast

fermentation processes. J. Biosci. Bioeng. 97 (6), 347364.Kilian, S.G., van Uden, N., 1988. Transport of xylose and glucose in the xylose-

fermenting yeast Pichia stipitis. Appl. Microbiol. Biotechnol. 27, 545548.Lawford, H.G., Rousseau, J.D., 1991. Ethanol production by recombinant Escherichia

coli carrying genes from Zymomoms mobilis. Appl. Biochem. Biotechnol. 28 (29),221236.

Leksawasdi, N., Joachimsthal, E.L., Rogers, P.L., 2001. Mathematical modelling ofethanol production from glucose/xylose mixtures by recombinant Zymomonasmobilis. Biotechnol. Lett. 23, 10871093.

Levenspiel, O., 1980. The monod equation: a revisit and a generalization to productinhibit situations. Biotechnol. Bioeng. 22, 671687.

Lim, J.S., Manan, Z.A., Wan Alwi, S.R., Hashim, H., 2012. A review on utilisation ofbiomass from rice industry as a source of renewable energy. Renew. Sust. Energ.Rev. 16, 30843094.

Luedeking, R., Piret, E.L., 1959. Kinetic study of the lactic acid fermentation: batchprocess at controlled ph. J. Biol. Microbiol. Technol. Eng. 1, 393424.

Luong, J., Mulchandani, A., LeDuy, A., 1988. Kinetics of biopolymer synthesis: arevisit. Enzyme Microb. Technol. 10, 326332.

Margaritis, A., Zajic, J.E., 1978. Biotechnology review: mixing, mass transfer andscale-up of polysaccharide fermentations. Biotechnol. Bioeng. 20, 9391001.

Marques, M., Joaquim, P.C., Cabral, M.S., Fernandes, P., 2010. Bioprocess scale-up:quest for the parameters to be used as criterion to move from microreactors tolab-scale. J. Chem. Technol. Biotechnol. 85, 11841198.

Moniruzzaman, M., Ingram, L.O., 1998. Ethanol production from dilute acidhydrolysate of rice hulls using genetically engineered Escherichia coli.Biotechnol. Lett. 20 (10), 943947.

Moniruzamman, M., Lai, X., York, S.W., Ingram, L.O., 1997. Isolation and molecularcharacterization of high-performance cellobiose-fermenting spontaneousmutants of ethanologenic Escherichia coli KO11 containing the Klebsiellaoxytoca casAB operon. Appl. Environ. Microbiol. 63 (12), 46334637.

Monod, J., 1949. The growth of bacterial cultures. Annu. Rev. Microbiol. 3, 371394.Moraine, R., Rogovin, P., 1966. Kinetics of polysaccharide B-1459 fermentation.

Biotechnol. Bioeng. 8, 511524.Olsson, L., Hahn-Hagerdal, B., 1995. Kinetics of ethanol production by recombinant

Escherichia coli KO11. Biotechnol. Bioeng. 45, 356365.Quinlan, A., 1986. Kinetics of secondary metabolite synthesis in batch culture when

two different substrates limit cell growth and metabolite production: xanthansynthesis by Xanthomonas campestris. Biochem. Eng. 469, 259269.

Richard, A., Margaritis, A., 2004. Empirical modeling of batch fermentation kineticsfor poly(glutamic acid) production and other microbial biopolymers.Biotechnol. Bioeng. 87 (4), 501515.

http://refhub.elsevier.com/S0960-8524(13)01047-X/h0005http://refhub.elsevier.com/S0960-8524(13)01047-X/h0005http://refhub.elsevier.com/S0960-8524(13)01047-X/h0005http://refhub.elsevier.com/S0960-8524(13)01047-X/h0010http://refhub.elsevier.com/S0960-8524(13)01047-X/h0010http://refhub.elsevier.com/S0960-8524(13)01047-X/h0010http://refhub.elsevier.com/S0960-8524(13)01047-X/h0010http://refhub.elsevier.com/S0960-8524(13)01047-X/h0015http://refhub.elsevier.com/S0960-8524(13)01047-X/h0015http://refhub.elsevier.com/S0960-8524(13)01047-X/h0015http://refhub.elsevier.com/S0960-8524(13)01047-X/h0020http://refhub.elsevier.com/S0960-8524(13)01047-X/h0020http://refhub.elsevier.com/S0960-8524(13)01047-X/h0025http://refhub.elsevier.com/S0960-8524(13)01047-X/h0025http://refhub.elsevier.com/S0960-8524(13)01047-X/h0025http://refhub.elsevier.com/S0960-8524(13)01047-X/h0030http://refhub.elsevier.com/S0960-8524(13)01047-X/h0030http://refhub.elsevier.com/S0960-8524(13)01047-X/h0035http://refhub.elsevier.com/S0960-8524(13)01047-X/h0035http://refhub.elsevier.com/S0960-8524(13)01047-X/h0040http://refhub.elsevier.com/S0960-8524(13)01047-X/h0040http://refhub.elsevier.com/S0960-8524(13)01047-X/h0045http://refhub.elsevier.com/S0960-8524(13)01047-X/h0045http://refhub.elsevier.com/S0960-8524(13)01047-X/h0045http://refhub.elsevier.com/S0960-8524(13)01047-X/h0050http://refhub.elsevier.com/S0960-8524(13)01047-X/h0050http://refhub.elsevier.com/S0960-8524(13)01047-X/h0055http://refhub.elsevier.com/S0960-8524(13)01047-X/h0055http://refhub.elsevier.com/S0960-8524(13)01047-X/h0055http://refhub.elsevier.com/S0960-8524(13)01047-X/h0055http://refhub.elsevier.com/S0960-8524(13)01047-X/h0060http://refhub.elsevier.com/S0960-8524(13)01047-X/h0060http://refhub.elsevier.com/S0960-8524(13)01047-X/h0060http://refhub.elsevier.com/S0960-8524(13)01047-X/h0060http://refhub.elsevier.com/S0960-8524(13)01047-X/h0065http://refhub.elsevier.com/S0960-8524(13)01047-X/h0070http://refhub.elsevier.com/S0960-8524(13)01047-X/h0070http://refhub.elsevier.com/S0960-8524(13)01047-X/h0075http://refhub.elsevier.com/S0960-8524(13)01047-X/h0075http://refhub.elsevier.com/S0960-8524(13)01047-X/h0080http://refhub.elsevier.com/S0960-8524(13)01047-X/h0080http://refhub.elsevier.com/S0960-8524(13)01047-X/h0080http://refhub.elsevier.com/S0960-8524(13)01047-X/h0085http://refhub.elsevier.com/S0960-8524(13)01047-X/h0085http://refhub.elsevier.com/S0960-8524(13)01047-X/h0085http://refhub.elsevier.com/S0960-8524(13)01047-X/h0090http://refhub.elsevier.com/S0960-8524(13)01047-X/h0090http://refhub.elsevier.com/S0960-8524(13)01047-X/h0095http://refhub.elsevier.com/S0960-8524(13)01047-X/h0095http://refhub.elsevier.com/S0960-8524(13)01047-X/h0095http://refhub.elsevier.com/S0960-8524(13)01047-X/h0100http://refhub.elsevier.com/S0960-8524(13)01047-X/h0100http://refhub.elsevier.com/S0960-8524(13)01047-X/h0105http://refhub.elsevier.com/S0960-8524(13)01047-X/h0105http://refhub.elsevier.com/S0960-8524(13)01047-X/h0110http://refhub.elsevier.com/S0960-8524(13)01047-X/h0110http://refhub.elsevier.com/S0960-8524(13)01047-X/h0115http://refhub.elsevier.com/S0960-8524(13)01047-X/h0115http://refhub.elsevier.com/S0960-8524(13)01047-X/h0115http://refhub.elsevier.com/S0960-8524(13)01047-X/h0120http://refhub.elsevier.com/S0960-8524(13)01047-X/h0120http://refhub.elsevier.com/S0960-8524(13)01047-X/h0120http://refhub.elsevier.com/S0960-8524(13)01047-X/h0125http://refhub.elsevier.com/S0960-8524(13)01047-X/h0125http://refhub.elsevier.com/S0960-8524(13)01047-X/h0125http://refhub.elsevier.com/S0960-8524(13)01047-X/h0125http://refhub.elsevier.com/S0960-8524(13)01047-X/h0130http://refhub.elsevier.com/S0960-8524(13)01047-X/h0135http://refhub.elsevier.com/S0960-8524(13)01047-X/h0135http://refhub.elsevier.com/S0960-8524(13)01047-X/h0140http://refhub.elsevier.com/S0960-8524(13)01047-X/h0140http://refhub.elsevier.com/S0960-8524(13)01047-X/h0145http://refhub.elsevier.com/S0960-8524(13)01047-X/h0145http://refhub.elsevier.com/S0960-8524(13)01047-X/h0145http://refhub.elsevier.com/S0960-8524(13)01047-X/h0150http://refhub.elsevier.com/S0960-8524(13)01047-X/h0150http://refhub.elsevier.com/S0960-8524(13)01047-X/h0150

320 E. Imamoglu, F.V. Sukan / Bioresource Technology 144 (2013) 311320Rocha-Valadez, J.A., Estrada, M., Galindo, E., Serrano-Carren, L., 2006. From shakeflasks to stirred fermentors: Scale-up of an extractive fermentation process for6-pentyl-a-pyrone production by Trichoderma harzianum using volumetricpower input. Process Biochem. 41, 13471352.

Serrano-Ruiz, J.C., Ramos-Fernandez, E.V., Sepulveda-Escribano, A., 2012. Frombiodiesel and bioethanol to liquid hydrocarbon fuels: new hydrotreating andadvanced microbial technologies. Energy Environ. Sci. 5, 56385652.

Serrano-Ruiz, J.C., Dumesic, J.A., 2011. Catalytic routes for the conversion of biomassinto liquid hydrocarbon transportation fuels. Energy Environ. Sci. 4, 8399.

Song, H., Jang, S.H., Park, J.M., Lee, S.Y., 2008. Modeling of batch fermentationkinetics for succinic acid production by Mannheimia succiniciproducens.Biochem. Eng. J. 40, 107115.Tamburini, E., Bernardi, T., Castaldelli, G., Tumiatti, G., Ferro, S., 2011. Greenelectrochemical approach for delignification of wheat straw in second-generation bioethanol production. Energy Environ. Sci. 4, 551557.

Tochampa, W., Sirisansaneeyakul, S., Vanichsriratana, W., Srinophakun, P., Bakker,H.H.C., Chisti, Y., 2005. A model of xylitol production by the yeast Candida mogii.Bioprocess Biosyst. Eng. 28, 175183.

Wang, L., Sharifzadeh, M., Templer, R., Murphy, R.J., 2012. Technology performanceand economic feasibility of bioethanol production from various waste papers.Energy Environ. Sci. 5, 57175730.

http://refhub.elsevier.com/S0960-8524(13)01047-X/h0155http://refhub.elsevier.com/S0960-8524(13)01047-X/h0155http://refhub.elsevier.com/S0960-8524(13)01047-X/h0155http://refhub.elsevier.com/S0960-8524(13)01047-X/h0155http://refhub.elsevier.com/S0960-8524(13)01047-X/h0160http://refhub.elsevier.com/S0960-8524(13)01047-X/h0160http://refhub.elsevier.com/S0960-8524(13)01047-X/h0160http://refhub.elsevier.com/S0960-8524(13)01047-X/h0165http://refhub.elsevier.com/S0960-8524(13)01047-X/h0165http://refhub.elsevier.com/S0960-8524(13)01047-X/h0170http://refhub.elsevier.com/S0960-8524(13)01047-X/h0170http://refhub.elsevier.com/S0960-8524(13)01047-X/h0170http://refhub.elsevier.com/S0960-8524(13)01047-X/h0175http://refhub.elsevier.com/S0960-8524(13)01047-X/h0175http://refhub.elsevier.com/S0960-8524(13)01047-X/h0175http://refhub.elsevier.com/S0960-8524(13)01047-X/h0180http://refhub.elsevier.com/S0960-8524(13)01047-X/h0180http://refhub.elsevier.com/S0960-8524(13)01047-X/h0180http://refhub.elsevier.com/S0960-8524(13)01047-X/h0185http://refhub.elsevier.com/S0960-8524(13)01047-X/h0185http://refhub.elsevier.com/S0960-8524(13)01047-X/h0185

Scale-up and kinetic modeling for bioethanol production1 Introduction2 Methods2.1 The preparation and dilute acid pre-treatment of rice hulls2.2 Strain maintenance and preparation of Inocula2.3 Bioethanol production conditions2.4 Scale-up strategies2.5 Description of kinetic modeling2.5.1 Model parameters prediction

2.6 Experimental analysis

3 Results and discussion3.1 Scale-up for bioethanol production3.2 Kinetic modeling for bioethanol production

4 ConclusionAcknowledgementsReferences