Gn

Ma

5b

a

ARR1A

KFFGBP

1

tbtCdwddgoi

1d

Process Biochemistry 45 (2010) 1023–1029

Contents lists available at ScienceDirect

Process Biochemistry

journa l homepage: www.e lsev ier .com/ locate /procbio

as–liquid dispersion in a fibrous fixed bed biofilm reactor at growth andon-growth conditions

artin Martinova,∗, Dimiter Hadjieva, Serafim Vlaevb

Laboratoire de Biotechnologie et Chimie Marine, Université Européenne de Bretagne-Université de Bretagne Sud (UEB-UBS), Centre de Recherche, Rue Saint Maude,6321 Lorient, Bretagne, FranceInstitute of Chemical Engineering, Bulgarian Academy of Sciences, Sofia 1113, Bulgaria

r t i c l e i n f o

rticle history:eceived 8 June 2009eceived in revised form9 November 2009ccepted 10 March 2010

eywords:ibrous fixed bed bioreactorluid dynamicsas hold-upiofilm reactionseudomonas putida

a b s t r a c t

There is limited data on gas dispersion characteristics of fixed bed biofilm reactors under growth andnon-growth conditions. In this paper, the gas–liquid dispersion of a bubble bed packed with a fibrousstructured packing for biofilm application is studied. The reactor is operated with Pseudomonas putidaaimed at aniline degradation in wastewater. Gas hold-up and bubble size distribution are determined.Running gas–liquid reaction conditions as well as non-reactive flow gas hold-up and bubble size distribu-tion in the presence of surface-active and viscous components were measured. The properties of the gasdispersion proved to be stabilized by the fibrous bed presence and showed improvement of the dispersionparameter by the packing. Gas hold-up was found to increase monotonously with the rise of gas superfi-cial velocity and viscosity and with surface tension fall. Liquid superficial velocity showed marginal effect.Apart from showing high gas hold-up and low bubble size due to surface-active and viscous dissolvedelements, the biochemical reaction did not pose any significant additional effect. In agreement with theexpected lack of bubble coalescence and break-up in the highly ionic solution practiced, the population

size distribution and average bubble size were found to vary with the major operation factors opposite totheir gas hold-up contribution. Gas hold-up was correlated with the specific bubble-to-channel size ratioand further with the variables considered. An empirical equation is proposed that relates gas hold-upwith all studied variables. Assuming geometric similarity of the prototype and the real vessels, the equa-tion as well as its corresponding range of fluid velocities can be used for bioreactor design and scale-up.The results concerning the gas hold-up are shown to be comparable with previous studies of mesh wire packing.

. Introduction

Referring to the last decennial, biological fixed bed bioreac-ors are increasingly used for wastewater treatment processesecause of their flexibility and compactness [1–3]. Recent prac-ical examples are: treatment of �-galactosidase entrapped ina-alginate-K-carrageenan gels for lactose hydrolysis [4], biodegra-ation and methane production from glycerol-containing syntheticastes [5], and azo-dye decolorization in continuous mode by Pseu-

omonas luteola [6]. In parallel to these, aerobic (thus, gas–liquid)

egradation has been experienced recently in some fixed bedas–liquid systems, e.g. phenol degradation [7] and degradation ofrganics [8]. Such treatments involve gas–liquid equipment includ-ng various structured and unstructured packings.

Abbreviations: PEVA, polyethylenevinylacetate; BSD, bubble size distribution.∗ Corresponding author. Tel.: +33 297874594; fax: +33 297874500.

E-mail address: martinov@univ-ubs.fr (M. Martinov).

359-5113/$ – see front matter © 2010 Elsevier Ltd. All rights reserved.oi:10.1016/j.procbio.2010.03.008

© 2010 Elsevier Ltd. All rights reserved.

A packing itself is attractive for biomass immobilizationforming biofilms for large-scale biological treatment. Focusingon structured packing known to enhance transversal flow inviscous batches, it was indicated as prospective for complex non-Newtonian bio-fluid flow stabilization [9]. At high fluid velocity, itcould serve as replacement of highly energy consuming mechanicalmixers in mixing bioreactors. Its applications are multi-functionaland require additional studies to uncover the character of func-tional variation in the case of the particular internal body. In thecase of this study, the use of a fibrous plastic matter has been consid-ered for gas–liquid dispersion analysis. Compared to conventionalgranular packing, such as clay balls, ceramic pieces, volcanic rocks,the fibrous packing presents a novelty due to its extended spe-cific surface and plastic material (PEVA) reported to enhance the

adhesion of biofilm cells [10].

Alternatively, gas–liquid bubble columns with structured pack-ing have large potential of applications in biochemical processing.An overview of the literature [11] shows just several studies tohave been devoted to low velocity bubble column operation, the

1024 M. Martinov et al. / Process Biochemistry 45 (2010) 1023–1029

Fwga

hp[sabMargeoosptc

fitba

2

2

iwpcom

gtgparw

mgv

Table 1Characteristics of the reactor and packing used.

Parameter Value

Column diameter (m) 0.1Clear liquid height (m) 0.6Height of a pack module (m) 0.13Packing material Polyethylenevinylacetate (PEVA)Packing specific area average (m2/m3) 1800Packed bed porosity (%) 70Gas distribution plate diameter (m) 0.06

Table 2Physicochemical properties of the solutions used.

Concentration Surface tension,� (mN/m)

Viscosity, � (mPa s) Density, �(kg m−3)

0 (water) 71.40 1 99820% saccharose 73.00 1.97 110440% saccharose 74.10 6.22 11761 g/l methanol 61.02 1 9982 g/l methanol 54.11 1 997

ig. 1. Schematics of experimental vessel and PEVA packing: (1) packing sectionith fiber modules, (2) fluid distribution section, (3) gas disengagement section, (4)

as flow controller, (5) gas outlet, (6) liquid feed pump, (7) liquid circulation pump,nd (8) liquid outlet.

ydrodynamic performance analysis of reactors with structuredacking being revealed by just a few recent papers. Birrer and Böhm9] have reported an overview analysis on the matter. Before them,tructured packing parameters have been studied by Urseanu etl. [12]. Recently, packing hydrodynamics has been studied alsoy Moustiri et al. [13], Bhatia et al. [14], Nikakhtari and Hill [15],aldonado et al. [16], and Monsalvo and Böhm [17]. The studies

re restricted to specific designs. Most of the work concerned non-eactive model systems. Apart from column design, an importantroup of studies have examined the effect of liquid physical prop-rties [18–21]. From the point of view of diversity of bioreactorbjectives, the present state of knowledge points at the importancef studying the performance effects of the various classes of systemseparately. These systems include both reactor geometry and fluidhysical properties, e.g. pure liquids and their solutions of addi-ives, i.e. low-viscous and highly viscous liquids and liquid phasesontaining surface-active matter.

In this paper, gas–liquid dispersion in bubble bed packed withbrous structured mesh material prospective for biofilm applica-ion is studied. Gas hold-up and bubble size distribution (BSD) atoth growth and non-growth conditions of batches with surface-ctive and viscous components are measured and analyzed.

. Materials and methods

.1. Bioreactor setup and packing

All experiments were carried out in a bioreactor column containing the pack-ng. The column and packing pattern are shown schematically in Fig. 1. The packing

as made of polymer fiber material 1.2–1.5 mm (Norten Ltd., France) produced asolyethylene (PE) and vinylacetate (VA) co-polymer, termed PEVA. The geometryomprised a pattern of thermally soldered fibers winded and folded to form a packf concentric layers filling the circular column cross-section. The details are sum-arized in Table 1. Cases of no packing, one and two stage packing were examined.

A porous plate of diameter 0.06 m mounted on the bottom lid carried out theas distribution. Gas flow controller and two liquid pumps were used to operatehe fluids in up-flow mode with counter-current liquid recycle. Air was used as theas phase. The bioreactor was equipped with a pH electrode, pH titration unit, O2

robes, and thermometer probe (not shown). During the fermentation process inletnd outlet dissolved oxygen concentration, temperature of the medium inside the

eactor, air and liquid flow rate were measured and controlled on-line, while pHas measured and controlled off-line.

The experiments were carried out in both continuous and semi-batch arrange-ent at continuous gas flow and at zero or very low liquid feed flow rate. Superficial

as velocity was varied between 0 and 22 mm/s and superficial liquid velocity wasaried in the range 0–12.6 mm/s involving a liquid recycle 25–280% of liquid feed.

5 g/l methanol 42.74 1 99610 g/l methanol 34.87 1 9932 g/l biomass 57.7 0.98 1020

2.2. Reactor start-up and operation

Growth and non-growth conditions were studied.At growth conditions, a biofilm containing Pseudomonas putida was cultivated

in aqueous saccharose solutions to remove aniline from the up-flowing water. Theaerobic strain Ps. putida ATCC 21812 was cultivated. The inoculum was prepared inadvance in flasks for 24 h starting with a sample culture in 10 cm3 LB medium. Arotary shaker for 24 h at 37 ◦C and 180 rpm was employed.

The bioreactor with volume 6 dm3 was inoculated by 0.4 dm3 initial culture.Liquid batches of nutrients were prepared in tap water by adding 10 g/dm3 saccha-rose, 0.5 g/dm3 aniline and salts, as follows: KH2PO4 0.5 g/dm3, K2HPO4 1 g/dm3,(NH4)SO4 0.5 g/dm3, Na2SO4 0.5 g/dm3, NaCl 0.5 g/dm3, MgSO4 0.5 g/dm3, CaCl20.02 g/dm3, and FeSO4 0.02 g/dm3. The reactor temperature was maintained at22 ± 2 ◦C. Air was fed in the bottom section for the experiments. The liquid feedwas maintained constant. Gas flow rate (at 22 ◦C and atmospheric pressure) andliquid circulation were varied. The pH value of the culture was maintained at 7–7.5.

Prior to continuous operation, the reactor was operated batch wise for 24 h withculture medium (5.6 dm3) and 0.4 dm3 culture; then the liquid feed was started(at 0.04 dm3/h) and the reactor was operated continuously for 48 h to develop thebiofilm.

At non-growth conditions, analytical grade methanol and saccharose were usedto arrange the physicochemical properties of the model solutions. Pure water wasused as reference. Table 2 contains the physical properties of the systems employed.

2.3. Measurement techniques

The surface tension was measured using a TD1 Lauda tensiometer. The viscositywas measured using viscometer Micro – Ubbelohde (Schott-Geräte GmbH).

Determining gas hold-up both at growth and non-growth conditions, the gasdisplacement technique was used. Recent experiments in structured packing byMonsalvo and Böhm [17] showed good reproducibility of the method when checkedagainst pressure tap technique, especially so in the range of gas hold-up 0–30%. Thepercent volumetric gas fraction was determined as

εG = HD − H0

HD(1)

where H0 stands for initial liquid level corresponding to equilibrium level in water,and HD indicates additional head in gas presence.

Bubble size distribution was determined by photographic imaging technique, asfollows: using electronic digital camera Nikon GSP 25, the dispersion field of the flowoutlet off the fibrous bed at 5 mm over the bed top surface was photographed andthe images were processed by scanning the images electronically on a PC screen.Number size distributions and average bubble size were determined. Sauter meanbubble diameter dS was calculated, as follows:

∑

dS = i

nid3bi∑

inid2

bi

(2)

where ni is the number of bubbles of size dbi .

M. Martinov et al. / Process Biochemistry 45 (2010) 1023–1029 1025

Fig. 2. Gas hold-up (a) and Sauter mean bubble dia

3

3

aruUU

ifFpt

Fig. 3. The effect of superficial gas velocity at growth conditions.

. Results

.1. The effect of packing height

Gas hold-up was measured first at zero condition of no packingnd the data obtained in this case were compared with previousesults in conventional bubble columns [19]. The values of gas hold-p obtained at the low gas velocity considered, namely, εG ∼ 10% atG ∼ 0.02 m/s coincided with the reference results, e.g. 4–10% atG ∼ 0.01–0.02 m/s.

Further, the case of one (0.13 m) and two (0.26 m) packing units

n the whole range of gas flow rates was examined. The resultsor gas hold-up and Sauter mean bubble diameter are shown inig. 2a and b, respectively. Gas hold-up is rising almost linearly withacking height (Fig. 2a). The Sauter mean bubble diameter shows arend towards decrease (Fig. 2b). The effect of the packing is seen as

Fig. 4. The effect of superficial liquid velocity (UL) on the Sauter mean bubble diam

meter (b) in modular packing 13 and 26 cm.

higher dispersion capacity represented by smaller bubbles at risingpacking height.

Considering the monotonous gas hold-up effect of packingheight, it was inferred that the properties of a single packing unitare representative enough to proceed in the further analysis.

3.2. Gas–liquid dispersion at growth conditions

It is extremely important to evaluate the estimates of the majorparameters characterizing a gas–liquid dispersion at condition ofgrowth. This is because the biological process is strongly dependenton dissolved oxygen and gas hold-up is a major variable componentof the mass transfer interfacial area. The other component vari-able is bubble size. Both parameters were determined in this study.Figs. 3 and 4 contain the results.

Gas superficial velocity is the major factor affecting the gas dis-persion; its characteristic impact on the reactive gas–liquid flow isillustrated in Fig. 3. A 3-fold increase of UG was found to cause a 5-fold gas hold-up increase. Liquid superficial velocity being assumeda parameter in this plot, it did not affect gas hold-up considerably: a3-fold increase in UL lead to a just 15% increase of gas hold-up whatwas within the range of the experimental error. Consequently, themajor control on the quality of gas dispersion was related to gasflow rate and the effect of liquid flow on the gas–liquid interfacewas considered marginal.

Fig. 4 shows (a) the relationship of bubble size distribution(BSD), and (b) the Sauter mean bubble diameter dS vs. UL. Forthe aim of comparison, Fig. 4b shows also the dS profile in water.

As expected, in both cases liquid velocity caused only a minorchange on the Sauter mean bubble diameter dS. For example, thebell-shaped distribution at the end points of the UL range stud-ied in Fig. 4a (being represented by mean bubble size 3.6 and4.3 mm) showed a difference of less than 1 mm both in real flow

eter (dS) at UG = 6.3 mm/s and conditions of growth: BSD (a) and dS vs. UL (b).

1026 M. Martinov et al. / Process Biochemistry 45 (2010) 1023–1029

anteabcopwmgTa

3

3w

FoTtoqtep

3

eI

Fig. 7. The effect of surface tension upon εG .

Fig. 5. The effect of superficial gas velocity UG on gas hold-up in water.

nd water. However, comparing the plots of growth (ca. 4 mm) andon-growth (ca. 5 mm) bubble diameters in Fig. 4b, one could inferhat the smaller bubbles of the reactive flow represent a largerxtent of limited coalescence. Referring to the packing functions a contact unit, one would expect the packing impact to causereak-up and collision. Regarding the highly ionic non-coalescingharacter of the solution of the bioreactor at conditions of growth,ne might expect bubble size to be frozen. In order to clarify thisoint, additional studies were accomplished using model solutionshere the solutes were changed according to pre-selected experi-ental program. Interest has been invoked in further comparison of

as hold-up at controlled physical properties of the liquid medium.o obtain the effect of the liquid properties, water with controlledmount of additives – alcohols or saccharose – was used.

.3. Gas–liquid dispersion at non-growth conditions

.3.1. The effect of UG for the fibrous packing operating in pureater

Firstly, the effect of superficial gas velocity was examined.ig. 5 contains the data. Comparable with the one at conditionsf growth, a 3-fold increase of UG yielded a 5-fold gas hold-up rise.he corresponding bubble size distribution in Fig. 6 shows trendowards average bubble size rise (up to 15%) in the velocity rangef UG ∼ 6–21 mm/s; gas velocity rising, it looks likely that the fre-uency of bubble collisions in the bubble swarm increased leadingo increased coalescence. In this sense, the lower the bubble diam-ter rise, the higher the packing efficiency to disperse the gas inacking presence.

.3.2. The effect of liquid physical propertiesThe effect of surface tension is shown in Figs. 7 and 9, and the

ffect of viscosity in Figs. 8 and 10, all data valid at UL = 8.3 mm/s.t is seen that the presence of surface-active additives causes the

Fig. 6. The effect of superficial gas velocity UG on b

Fig. 8. The effect of viscosity upon gas holdup.

gas hold-up to increase up to 30% and dS to decrease significantly.Evidently, the dispersion capacity of the unit was extended.

As seen, viscosity affects the parameter similarly by decreas-ing dS and increasing gas hold-up. Qualitatively, the latter effectis similar to the one reported by Zahradnik et al. [19] and Ruzickaet al. [21] for conventional bubble columns (without packing). Thisis however the opposite of what was expected from most of thereference studies (see Kantarci et al., 2009 in Ref. [11]). In general,liquid viscosity is known to affect gas dispersion in two routes:effect on bubble rise velocity, i.e. increasing friction decreases bub-ble rise velocity thus increasing hold-up, and effect on bubblegeneration, e.g. producing various quantities of small bubbles andthus increasing or decreasing hold-up. While increasing liquid vis-

cosity, these two component effects compete and bring about anextremal behavior, especially in the viscosity range � < 10 mPa s.Similar behavior has been reported as early as 1981 by Schügerl[22]. While not excluding a gas hold-up fall at higher or wider rangeof liquid viscosity [11], we believe that the effect observed in this

ubble size in water: BSD (a) and dS vs. UG (b).

M. Martinov et al. / Process Biochemistry 45 (2010) 1023–1029 1027

Fig. 9. The effect of surface tension upon BSD.

sfig

on

4

4

inmrtr3

Fig. 10. The effect of viscosity upon BSD.

tudy is due to the viscosity induced hindered bubble rise in theber section that also hindered bubble coalescence and promotedeneration of small-size bubbles.

The reliability of the packing-induced behavior of εG vs. �bserved is confirmed by the size analysis in Fig. 10 showing defi-ite diminishing trend of the Sauter mean bubble size.

. Discussion

.1. The packing performance in real flow

A summary of reactor dispersion potential, depicted as dS plots,s shown in Fig. 11. Comparing bubble diameter at growth andon-growth conditions, namely, the reactive (real) nutrient ionic

edium (line 1) on one hand, and water with additives – saccha-

ose and methanol, lines 2 and 3, respectively – on the other hand,he packing dispersion performance could be inferred. In Fig. 11, theelationship dS vs. viscosity (line 2) and dS vs. surface tension (line) are plotted together and their intersection with dS = const (the

Fig. 12. Photograph of gas–liquid dispersion at gr

Fig. 11. Comparison of bubble size in model and real reactive flow at UG = 6.3 mm/s.

dotted horizontal line) are shown as points A and B, respectively;both model and reactive flows had equal velocity UL. By intersectionA, one could infer that the real media exhibit as high a dispersionpotential (e.g. estimated as dS) as water with added up to 20% sac-charose. The intersection at B shows that water with added 2 g/dm3

methanol produced bubble size as large as the real medium did.Because methanol and saccharose are known to suppress coales-cence [18–20], the data in Fig. 11 prove that the real flow createsnon-coalescing conditions.

Additional information on the packing operation is shown inFig. 12. It represents an image of bubbles (type 1) captured at thepacking channel outlets, and the bubbles (type 2) rising up freelyin the bulk. By the difference of bubble size in both cases, one cansee clearly that the packing exhibits both a dispersion function in away that it ensures small bubble size, and a stabilizing function—ina way that the bubbles remain uniform compared to those per-forming out of the packing. Consequently, packing performance iseffective.

4.2. Comparison with reference data of similar systems

In order to verify the dispersion property of the packing, theresults of the study were compared with reference data reportedby Moustiri et al. [13], Bhatia et al. [14] and Maldonado et al. [16]in Fig. 13.

The data sets of the various studies are comparable in a waythat they all study gas displacement in packed beds of diameters0.1–0.15 m at similar superficial liquid velocity, e.g. 12 mm/s, bysimilar methods and differ within a range of values εG ∼ 1–2%. Such

differences can be accepted in view of the differences of the packingdesigns. However, if one considers the data as linear relationships,one could register the deviation of the slope of the curves reportedby Moustiri et al. [13] and Maldonado et al. [16]. To our opinion,the reason for such inconsistency could be attributed to large vol-

owth conditions in real flow: views 1 and 2.

1028 M. Martinov et al. / Process Biochem

Fig. 13. Comparison with reference data.

F

utpc

arMlhtbtsb

4

ibcaa

d

wi

nori

ig. 14. Comparison between experimental and predicted gas hold-up values.

metric differences of reported packing porosity. In other words,he major reason for this deviation might be due to the larger bedorosity of the packing studied in Refs. [13,16] than the one in thease of this study.

Among the reference data, only Bhatia [14] and Maldonado etl. [16] have reported bubble size data. The Sauter bubble diametereported by Bhatia et al. [14] is 4.5 mm and the one reported byaldonado et al. in Ref. [16] is 3–4.6. Both studies reported 1–2 mm

ower bubble size compared to the one measured in this study. Weave accepted both the values and the techniques employed byhese authors as precise enough to believe that the differences inoth cases lie entirely within the range of experimental capacity ofhe techniques employed. Consequently, referring to variation oflopes, we do not believe that the discrepancy in Fig. 12 is causedy variation in bubble size.

.3. The effect of packing channel size

The gas hold-up data can be discussed in terms of bubble flown the packing. The packing hydraulic diameter dH and the specificubble-to-channel size ratio dS/dH were determined. Based on theoncept of a capillary model of bed channels regarding the packings a system of capillaries, the hydraulic diameter was determineds

H = 23

�dP

(1 − �)(3)

here dP is the solid (soldered fiber) diameter (ca. 2–3 mm) and �s bed porosity (Table 1).

As gas flow in the packing is ruled by the size of the porous chan-els formed by the mesh material void spaces represented by dH,ne could infer the regime of gas–liquid flow in the packing. Refer-ing to the image of particle size and packing dimensions shownn Fig. 12, one could see the small size of the bubbles coming out

istry 45 (2010) 1023–1029

of the packing, as well as the enlarged bubbles following bubblecollisions in the packing-free column section. As pointed by Molgaand Westerterp [23], comparing the Sauter bubble size relative tothe channel hydraulic diameter, one could define the gas–liquidflow in the bed voids. Two limiting flow patterns have been consid-ered: ‘bubble flow’ and ‘continuous streams of gas flowing throughwetted packing’. In this study the value of dH being determined tobe about 3 mm and observing dS/dH ∼ 1, bubble flow regime witha trend to continuous streams for larger bubbles was assumed asflow model for PEVA packing.

4.4. Correlation to operation variables

In order to use the results of this study for extrapolation, anempirical equation was determined, as follows:

εG = 1.07U0.61G U0.21

L

(�

�w

)0.1( �

�w

)−0.16( HB

HBo

)0.3(4)

Data processing over 70 data sets including 22 sets at growthconditions showed an approximation representative enough by itsdeviation of 15%, as evaluated in the parity plot in Fig. 14.

5. Conclusions

The gas–liquid dispersion characteristics of bubble columnspacked with fibrous structured material PEVA for biofilm appli-cation have been studied. Gas hold-up of both reactive andnon-reactive batches with surface-active and viscous componentswas measured. Bubble size distribution was determined. Gas hold-up was found to increase significantly with gas superficial velocityand viscosity and to decrease with surface tension. Liquid super-ficial velocity showed marginal effect. Gas–liquid flow regime inthe packing, as inferred from the specific bubble-to-channel sizeratio was classified as bubbly flow. An empirical correlation isproposed that relates gas hold-up with all studied variables. Gasdispersion properties proved to stabilize by the fibrous packingand showed improvement of dispersion parameter by the pack-ing. Assuming geometric similarity of the prototype and the realvessels, the equation at the relevant range of fluid velocities can beused for bioreactor design and scale-up. The results concerning thegas hold-up are shown to be comparable with previous studies ofmesh wire packing.

Acknowledgement

Authors MM and SDV acknowledge grants of PRIR BGF – RegionBretagne in support of this study.

Appendix A. Symbols

db bubble diameter [mm]dS Sauter mean bubble diameter [mm]dH packing hydraulic diameter [mm]dP solid (fiber) diameter [mm]dS/dH specific bubble-to-channel size ratioH height [m]U superficial velocity [m/s]V liquid volume [m3]

Greek symbols

ε gas hold-up� bed porosity� density [kg m−3]� surface tension [N m−1]� dynamic viscosity [Pa s]

iochem

SLGBDW0

R

[

[

[

[

[

[

[

[

[

[

[

[

M. Martinov et al. / Process B

ubscriptsliquidgasbeddeterminedwaterinitial

eferences

[1] Iliuta I. Performance of fixed bed reactors with two-phase upflow and down-flow. J Chem Technol Biotechnol 1997;68(1):47–56.

[2] Deront M, Samb F, Adler N, Péringer P. Biomass growth monitoring usingpressure drop in a cocurrent biofilter. Biotechnol Bioeng 1998;60(1):97–104.

[3] Benthack C, Srinivasan B, Bonvin D. An optimal operating strategy forfixed-bed bioreactors used in wastewater treatment. Biotechnol Bioeng2001;72(1):34–40.

[4] Mammarella E, Rubiolo A. Effect of biocatalyst swelling on the oper-ation of packed-bed immobilized enzyme bioreactor. Process Biochem2009;44(2):183–90.

[5] Yang Y, Tsukahara K, Sawayama S. Biodegradation and methane produc-tion from glycerol-containing synthetic wastes with fixed-bed bioreactorunder mesophilic and thermophilic anaerobic conditions. Process Biochem2008;43(4):362–7.

[6] Chen B, Chen S, Chang J. Immobilized cell fixed-bed bioreactor for wastewaterdecolorization. Process Biochem 2005;40(11):3434–40.

[7] Bajaj M, Gallert C, Winter J. Biodegradation of high phenol containingsynthetic waste-water by an aerobic fixed bed reactor. Bioresour Technol

2008;99(17):8376–81.

[8] Borghei S, Sharbatmaleki M, Pourrezaie P, Borghei G. Kinetics of organicremoval in fixed-bed aerobic biological reactor. Bioresour Technol2008;99(5):1118–24.

[9] Birrer FR, Böhm U. Gas–liquid dispersions in structured packing with high-viscosity liquids. Chem Eng Sci 2004;59:4385–92.

[

[

istry 45 (2010) 1023–1029 1029

10] Hadjiev D, Dimitrov D, Martinov M, Sire O. Enhancement of the biofilm forma-tion on polymeric supports by surface conditioning. Enzyme Microb Technol2007;40(4):840–8.

11] Kantarci N, Borak F, Ulgen K. Bubble column reactors. Process Biochem2005;40:2263–83.

12] Urseanu MI, Ellenberger J, Krishna R. A structured catalytic bubble columnreactor: hydrodynamics and mixing studies. Catal Today 2001;69:105–13.

13] Moustiri S, Herbrard G, Roustan M. Effect of a new high porosity pack-ing on hydrodynamics of bubble columns. Chem Eng Process 2002;41:419–26.

14] Bhatia B, Nigam KDP, Auban D, Hebrard G. Effect of new high porosity packingon hydrodynamics and mass transfer in bubble columns. Chem Eng Process2004;43:1371–80.

15] Nikakhtari H, Hill GA. Hydrodynamic and oxygen mass transfer in an exter-nal loop airlift bioreactor with a packed bed. Biochem Eng J 2005;27:138–45.

16] Maldonado JG, Bastoul D, Baig S, Roustan M, Hebrard G. Effect of solidcharacteristics on hydrodynamics and mass transfer in a fixed bed reactoroperating in co-current gas–liquid up flow. Chem Eng Process 2008;47:1190–200.

17] Monsalvo M, Böhm U. Influence of structured packing on gas hold-up in a three-phase bubble column. Chem Eng Sci 2007;62:6595–603.

18] Kastanek F, Zahradnik J, Kratochvil J, Cermak J. Chemical reactors for gas–liquidsystems. Praha: Ellis Horwood Ltd./Academia; 1993. p. 260.

19] Zahradnik J, Fialova M, Ruzicka M, Drahos J, Kastanek F, Thomas NH. Dual-ity of the gas–liquid flow regimes in bubble column reactors. Chem Eng Sci1997;52:3811–26.

20] Zahradnik J, Kuncova G, Fialova M. The effect of surface active additiveson bubble coalescence and gas hold-up in viscous batches. Chem Eng Sci1999;54:2401–8.

21] Ruzicka MC, Drahos J, Mena PC, Teixeira JA. Effect of viscosity onhomogeneous–heterogeneous flow regime transition in bubble columns. Chem

Eng J 2003;96:15–22.

22] Schügerl K. Oxygen transfer into highly viscous media. Adv Biochem Eng1981;19:71–174.

23] Molga EJ, Westerterp KR. Gas–liquid interfacial area and hold-up in a co-currentup-flow packed bed bubble column reactor at elevated pressures. Ind Eng ChemRes 1997;36:622–31.