Design and characterization of a scalable airlift flat panelphotobioreactor for microalgae cultivation

Jian Li & Marisa Stamato & Eirini Velliou &

Clayton Jeffryes & Spiros N. Agathos

Received: 22 February 2014 /Revised and accepted: 8 May 2014# Springer Science+Business Media Dordrecht 2014

Abstract A novel flat panel photobioreactor prototype withbulk liquid flow driven by an external airlift was designed,modeled, and experimentally characterized for the purpose ofdeveloping scalable industrial photobioreactors. Baffles werebuilt inside the flat panel part of the reactor, directing theliquid bulk flow in a serpentine way, and the external airliftdrove the liquid flow and facilitated gas mass transfer. The gasholdup, liquid flow velocity, and oxygen mass transfer of thisprototype were experimentally determined and mathematical-ly modeled, and the performance of the reactor was tested bycultivating two species of microalgae, Scenedesmus obliquusand Chlorella sorokiniana. The model-predicted trends corre-lated well with experimental data, indicating that the reactormight be scaled up using these models. A high cell concen-tration of C. sorokiniana was achieved under controlled in-door cultivation conditions although serious biofouling oc-curred in the case of S. obliquus cultivation. The results favorthe possibility of scaling up the reactor to industrial scales,based on the models employed, and the potential advantagesand disadvantages of the reactor are discussed regarding thisindustry-oriented photobioreactor configuration in compari-son with current industrial photobioreactors.

Keywords Microalgae . Photobioreactor .Chlorellasorokiniana . Scenedesmus obliquus . Flat panel . Airlift .

Biofouling

Introduction

The potential of microalgae to produce biofuels or otherproducts has been under intensive exploration in recentyears (Haag 2007; Chisti 2007; Stephens et al. 2010;Wijffels et al. 2013). One of the technical challenges isto cut down the capital and operational costs ofmicroalgae production systems (Wijffels and Barbosa2010; Stephens et al. 2010). Although many cultivationsystems have been proposed and studied, mostly inlaboratory settings, for microalgae biomass production,only raceway ponds, tubular photobioreactors, and flatpanel photobioreactors have the potential to be scaledup to industrial scales without forbidding costs (Norskeret al. 2011; Ugwu et al. 2008). However, the capital andoperational cost of such systems are still too high toproduce microalgae biomass as feedstock for biofuels orother low-value products with currently available tech-nologies (Li et al. 2011; Petkov et al. 2012), and thus,it is still desirable to develop cost-effective productionsystems to release the full potential of microalgal bio-technology (Acién et al. 2012).

Although raceway ponds are believed to be the mostinexpensive systems for the industrial autotrophic culti-vation of microalgae and might even be the only choicefor microalgal biodiesel production, industrial tubularand flat panel photobioreactor systems are still expectedto play a vital role in the future microalgae industry.For the production of low-value products such as bio-diesel or animal feed at large scales, the utilization ofphotobioreactors might be necessary to provide inocu-lum culture for raceway ponds to achieve sustainable

Electronic supplementary material The online version of this article(doi:10.1007/s10811-014-0335-1) contains supplementary material,which is available to authorized users.

J. Li :M. Stamato : E. Velliou : C. Jeffryes : S. N. Agathos (*)Bioengineering Laboratory, Earth and Life Institute, UniversitéCatholique de Louvain, Place Croix du Sud 2, Box L07.05.19,1348 Louvain-la-Neuve, Belgiume-mail: Spiros.Agathos@uclouvain.be

Present Address:E. VelliouCentre for Process Systems Engineering (CPSE) and BiologicalSystems Engineering Laboratory (BSEL), Department of ChemicalEngineering, Imperial College London, London SW7 2AZ, UK

J Appl PhycolDOI 10.1007/s10811-014-0335-1

production (Delrue et al. 2012; Lundquist et al. 2010).For high-value products such as nutraceuticals or phar-maceuticals, the production system might need to ex-clude the utilization of raceway ponds for hygienicreasons (Ugwu et al. 2008). Thus, advancements ofphotobioreactor research could have a crucial impactfor the growth of the microalgae industry (Chen et al.2011; Lehr and Posten 2009).

Both tubular and flat panel photobioreactors havebeen scaled up to large scales, yet neither of them hasbeen shown to be clearly superior to the other one.Tubular photobioreactors can be constructed using inex-pensive PVC film tubes with an external airlift made ofplastic material, and the tubular part of the bioreactorcan be laid on the ground without complicatedsupporting infrastructure. The oxygen removal andCO2 addition can be both achieved through airlift mod-ules, which are inexpensive, reliable, and efficient(Molina et al. 2000; Acién et al. 2001). Due to thiskind of design configuration, the material and operation-al costs of tubular reactors can be moderate. The indus-trial flat panel photobioreactors can also be constructedusing low-cost disposable PVC film bags, but theseneed to be supported by scaffolds, which can be expen-sive (Sierra et al. 2008). In addition, the sparging of thereactors requires larger volumes of air, which mightmean higher energy consumption and CO2 lost to theatmosphere. Comparing with industr ial tubularphotobioreactors, both the material and operational costsof the flat panel reactor can be higher per unit area, butthe areal production can also be higher due to shorterlight path length and vertical alignment of flat panels(Norsker et al. 2011; Janssen et al. 2003; Slegers et al.2011).

In this research, we cultivated microalgae in a flatpanel photobioreactor with internal bulk liquid flow andan external airlift, which we designed and characterized.Although a similar concept was proposed and tested in1992, it has never been fully exploited and developedfor industrial photobioreactor design (Ratchford andFallowfield 1992). In currently employed industrial flatpanel photobioreactors, the air is bubbled directly fromthe bottom of the flat panels and there is no liquid bulkflow in the reactor. In this way, the efficiency of mixingis low, and thus, the energy efficiency of the reactor isalso low due to high consumption of compressed air(Leupold et al. 2013; Sierra et al. 2008). Another majorproblem of this kind of reactor is biofouling, which isalso partially caused by inefficient mixing (AciénFernández et al. 2013). Although some teams havereported occasionally flat panel photobioreactor designswith internal airlift for laboratory studies, most of themhave never been scaled up for industrial applications

(Reyna-Velarde et al. 2010; Degen et al. 2001) or haveonly been of the order of 30 L and used in industry inthe form of multiple units (Münkel et al. 2013).External airlift pumps are widely used in tubularphotobioreactor systems; they are inexpensive to buildand can be easily scaled up. In order to improve theenergy efficiency and enhance the mixing of the flatpanel photobioreactor systems, we modified the currentflat panel photobioreactor design configuration by sup-plying an external airlift pump and adding internal baf-fles to drive and direct the liquid bulk flow in thereactor. With our design, the flat panel part of thephotobioreactor was configured in a way that liquidmedium can display bulk f low as in tubularphotobioreactors, and the oxygen removal and CO2 ad-dition is realized through an external airlift module. Atthe same time, the alignment of multiple flat panelreactors and the light path length of individual flatpanels can be maintained similarly to current industrialflat panel photobioreactors if the new prototype were tobe scaled up for industrial applications. In this way, thenew bioreactor might be able to have high areal pro-ductivity, low energy consumption, and less seriousbiofouling, and simultaneously, the construction and op-eration costs can be kept low. Therefore, this airlift flatpanel reactor might overcome some disadvantages ofboth industrial flat panel and tubular photobioreactorsand lead to a novel class of industrial photobioreactordesign which is superior to both of them.

Materials and methods

The airlift flat panel photobioreactor

The prototype airlift flat panel photobioreactor wascomprised of a flat panel and an external airlift pump,as shown schematically in Fig. 1. The airlift had a riserand a downcomer, both of which were attached to agas–liquid separator on their top ends. Both the riserand downcomer were attached to a T-shaped sparger attheir bottom ends, and a perforated plate was used forair distribution in the sparger. The joint flow of air viathe perforated plate sparger and of the liquid from theoutlet of the flat panel started at the bottom of the riserthen flowed to the gas–liquid separator where the airand liquid were separated. The air flowed out, and theliquid flowed through the downcomer to the inlet of theflat panel. In the flat panel, the liquid flowed frombottom to top in a serpentine manner.

The flat panel was made of transparent polycarbonate(PC) plates, with a surface area of 1 m2 (1 m×1 m) andan internal thickness of 3.6 cm, and was split into nine

J Appl Phycol

lanes of a continuous flow chamber. The airlift modulewas made of plexiglass tubes and plates, with a tubulardiameter of 3.2 cm and a height of 2.35 m. The probesand sensors were installed at the top of the gas–liquidseparator, which linked riser and downcomer at its bot-tom. The flat panel and the airlift modules were linkedby soft translucent PVC tubes, and the total workingvolume of the whole reactor was about 48 L.

The photobioreactor was equipped with a NI cRIO-9074CompactRIO data acquisition and control chassis and twomodules from National Instruments Corporation (USA). Thedissolved oxygen (DO) level and the pH of the cell culturecould be constantly monitored by a ProfiLine Oxi 340i oxy-gen meter and ProfiLine pH 340i pH meter, respectively, bothof which were from WTW (Germany). Temperature wasmonitored continuously with an additional function of thepH meter. The data were transferred between the meters andthe CompactRIO controller through a NI 9870 serial portmodule (National Instruments). The temperature could becontrolled through the automatic turning on and off of tworadiating heaters facing the flat panel, and the pH could also beconstantly controlled by on–off injection of CO2 through thesparger at the bottom end of the downcomer. The controlfunctions were realized using a NI 9481 relay module(National Instruments).

Organisms and cultivation conditions

The microalgal strain Chlorella sorokiniana SAG 211-32 wasfrom the SAG Culture Collection of Algae (Göttingen,Germany), and Scenedesmus obliquus CCAP 276-2 was fromthe Culture Collection of Algae and Protozoa (Oban, UK).The average surface light intensity of the flat panel was about150 μmol photons m−2 s−1 (continuous illumination withfluorescent light bulbs) which was determined by a PARquantum sensor from Skye Instruments (UK). The pH of the

photobioreactor was maintained constant at 7.5 by controlledinjection of CO2. The aeration rate was kept at 4 L min−1 forboth species, and the temperature was maintained at 22 °C.The reactor was first filled with demineralized water andsterilized by ozone for 2 h, and the stock solutions of BBMmedium with threefold sodium nitrate (Stein 1979) werepumped into the reactor through a 1.0-μm liquid filter fromSartorius AG (Germany). The reactor was then inoculatedwith axenic seed cell culture.

Cell density measurements

The cell culture density of samples was quantified by spec-trophotometry and gravimetry. For the spectrophotometricmethod, the optical density (OD) of the cell culture wasmeasured in a 1-cm light path cuvette at wavelengths of530 and 680 nm using a DU 640 Beckman (USA) spectro-photometer (Kliphuis et al. 2010). For the gravimetric meth-od, duplicate cell culture samples of 40 mL were centrifugedin Eppendorf centrifuge tubes at 2,000×g for 10 min, thesupernatant was carefully discarded, and the pellets ofmicroalgae were resuspended in 2 mL distilled water andtransferred to pre-weighed 2-mL Eppendorf tubes. The sam-ples were then centrifuged at 2,000×g for another 10 min, andthe supernatants were again carefully discarded. Finally, thepellets were dried at 70 °C for 24 h, and the weight of thebiomass was measured.

Liquid flow and gas holdup measurements

The liquid flow in the photobioreactor was measured with aconductivity meter, model ProfiLine340i from WTW(Germany). The probe of the meter was placed on the top ofthe gas–liquid separator, and sodium chloride was used as atracer. The reactor was first filled with demineralized water,and the aeration was turned on to drive the liquid flow. Then, ashot of saturated sodium chloride solution, about 50 mL involume, was introduced in the reactor at the gas–liquid sepa-rator region, and the conductivity of the liquid was constantlymonitored and recorded. With the liquid circulating throughthe reactor, a series of peaks of conductivity values wereobserved, and the time interval between peaks was estimatedbased on the data and regarded as the time duration for theliquid to travel across one cycle. This time duration wasdefined as the circulation time. The superficial liquid flowvelocity at the riser region,UL, could then be calculated basedon the following equation (Rubio et al. 1999):

UL ¼ 4V T

πtcd2r

ð1Þ

where VT is the total volume of the reactor, tc is the circulationtime, and dr is the diameter of the riser.Fig. 1 Representation of the airlift flat panel photobioreactor

J Appl Phycol

The gas holdup in the riser was measured by the differencesin liquid level between the airlift and an external tube whichwas linked to an opening at the bottom level of the riser. Atcertain aeration conditions, the liquid levels in the externaltube were compared with those in the gas–liquid separator onthe top of the riser, and the height differences divided by thetotal height of the riser were considered the values of gasholdup in the riser (Rubio et al. 1999).

Oxygen transfer measurement

The oxygen transfer was measured dynamically by a DOmeter, model ProfiLine Oxi 340i from WTW (Germany).The photobioreactor was bubbled with pure nitrogen gasthrough the airlift module at a flow rate equal to the aerationrate until the DO level dropped close to zero. Then, thenitrogen gas was switched off and air switched on, after whicha leap of DO level was observed and recorded. After a timeinterval of circular flow, another leap of DO level was ob-served and recorded. Equation 2 can be easily derived toestimate the oxygen transfer coefficient kLa in the riser(Acién et al. 2001; Rubio et al. 1999):

kLa ¼ UL

hrln

Cin−C*

Cout−C*

� �ð2Þ

where UL is the superficial liquid flow velocity; hr is the totalheight of the riser; C∗ is the saturated DO level; Cin is the DOlevel at the inlet of the riser, which is the value before the leap;and Cout is the oxygen level at the outlet of the riser which isthe value after the leap.

Modeling fluid dynamics and mass transfer

The equations for modeling the photobioreactor processeswere adapted from literature data on conventional tubularphotobioreactor systems and cell culture fluid dynamics andapplied to the operation scenarios of the flat panelphotobioreactor. All the software codes were programmed inMATLAB software from MathWorks (USA).

Since the liquid flow in the flat panel was driven by theairlift module, we used the previously established mathemat-ical relationships between liquid flow velocity and gas holdupbased on energy balance (Chisti 1989; Chisti et al. 1988) anddeveloped (Molina et al. 2001) to be

UL ¼ gεrhrd1:25t

0:3164 μL

.ρL

� �0:25Leq

264

375

4=7

ð3Þ

where UL is the superficial liquid velocity in the tube, g is thegravitational acceleration constant, εr is the gas holdup in the

riser, hr is the height of the riser, dt is the diameter ofthe liquid flow channel, μL is the viscosity of theculture broth, ρL is the density of the culture broth,and Leq is the equivalent length of the liquid flowchannel. The relationship between gas holdup in theriser εr and superficial liquid flow velocity in the riserUL had also been established (Zuber and Findlay 1965)semi-experimentally based on mass balance:

εr ¼ UG

λ UG þ ULð Þ þ Ubð4Þ

where UG is the superficial gas velocity in the riser, λ is aparameter related to the fluid dynamics in the riser, and Ub isthe bubble drift velocity in the riser. In our experimentalsettings where the liquid flow can be characterized as a one-dimensional vertically upward homogeneous bubbly flow, λand Ub can be estimated by the following equations (Liaoet al. 1985; Wallis 1969):

U b ¼ 1:53 1−εrð Þ2 σg ρL−ρGð ÞρL2

� �1=4

; λ ¼ 1:0 ð5Þ

where σ is the surface tension of the liquid and ρG is thedensity of the gas phase.

The oxygen transfer in the airlift was estimated accordingto the same literature reports as liquid flow (Chisti et al. 1988;Chisti 1989; Molina et al. 2001; Chisti 1999), which haveestablished the guiding principles for airlift tubularphotobioreactor design. The following equation, which wasdeveloped based on fundamental geometrical concepts, wasfound to be valid under various aeration conditions for airliftdevices (Chisti and Moo-Young 1987; Chisti 1989; Cerriet al. 2010):

kLdB

¼ kLa 1−εrð Þ6εr

ð6Þ

where kL is the true convective mass transfer coefficient, dB isthe average air bubble diameter, a is the volumetric interfacialarea, and kLa is the volumetric mass transfer coefficient. Forthe air–water system, the following equation (Chisti 1989) canbe used to estimate kL/dB:

kLdB

¼ 5:63� 10−5gDLρ2Lσ

μ3L

� �1=2

ð7Þ

where DL is the diffusivity of oxygen in water and μL is theviscosity of the culture medium.

Microalgae growth modeling and parameter determination

To simplify analysis, we assume in our experimentalconditions that the growth of microalgae is only light

J Appl Phycol

limited, light inhibition and oxygen inhibition are neg-ligible, and light transmission in the cell culture obeysthe Beer–Lambert law. Thus, the following equationscan be modified from literature to describe the cellgrowth in the flat panel photobioreactor (Li et al.2003; Su et al. 2003):

dX

dt¼ μX ð8Þ

μ ¼ μmaxI av

K I þ I av

� �−r ð9Þ

Iav ¼ IoL

Z L

0e−KaXldl ¼ Io

1−e−KaXL

LKaXð10Þ

where X is the biomass concentration in the reactor, t is time, μis the specific growth rate, μmax is the maximum specificgrowth rate at the prevailing temperature, Iav is the averagelight intensity in the reactor chamber, r is the specific respira-tion rate at the given temperature, KI is the light saturationconstant, L is the thickness of the reactor chamber, Io is theincident light intensity, and Ka is the mass-specific light ab-sorption coefficient.

We estimated μmax, r, and KI, which are intrinsic propertiesof a given microalgal strain at a defined temperature, for thestrain C. sorokiniana SAG 211-32 that we used in this re-search, using an OxyLab Plus System from HanatechInstruments Inc. (UK) and under the same temperature andnutritional conditions as described above. The oxygen levelsof the microalgae cell culture in the chamber under differentlevels of light illumination were monitored, and the rates ofoxygen production were thus derived, based on the rate ofoxygen level changes (Huang et al. 1998). The specific oxy-gen production rates were transformed into biomass growthrates, according to the chemical stoichiometry Eq. 11 listedbelow (Munoz and Guieysse 2006):

5:02CO2 þ 3:82H2Oþ 0:75NO−3 þ 1:89Hþ þ 0:047PO3−

4

¼ 5:02CH1:7O0:4N0:15P0:009 þ 7:15O2 ð11Þ

The growth rates were then fitted with light intensity levelsto get the values of μmax, r, and KI, based on Eq. 9. The valueof Ka is estimated from the light absorption levels with differ-ent cell concentrations of Chlorella culture, based on the

Beer–Lambert law. The values of the above parameters aretabulated in Table 1.

Results and discussion

Characterization of fluid dynamics in the reactor

The fluid dynamics in photobioreactors is of primary impor-tance as in any other bioreactor type because the turbulenceaffects cell physiology, keeps the cells in suspension, mixesnutrients, and facilitates gas and heat transfer. In fact, ad-vanced engineering tools such as computational fluid dynam-ics (CFD) have already been applied in photobioreactor de-sign and development (Bitog et al. 2011). However, to esti-mate the most basic parameters of fluid dynamics in airliftphotobioreactors, such as liquid flow velocity and gas holdup,some semi-empirical equations might be both sufficient andless onerous (Znad et al. 2004).

In this research, we employed established semi-empiricalrelationships, i.e., Eqs. 3, 4, and 5, to estimate liquid flowvelocity and gas holdup and we compared experimentallymeasured values with the mathematically estimated trends.The experimental data fitted the model-predicted trends rea-sonably well, and we believe these model equations could beappropriate for the purpose of scaling up the airlift flat panelphotobioreactor. The liquid flow pattern in the riser of theairlift could be categorized as bubbly flow regime by visualobservation, based on the near homogeneity of the bubblesand their small size compared to the tube diameter (Shaikhand Al-Dahhan 2007). The observation could hold true for airflow rates up to 5 L min−1 (equivalent to superficial velocityup to about 0.1 m s−1), above which slug flow would takeplace. Thus, we used fluid dynamic equations that characterizebubbly flow regime tomodel the liquid flow and gas holdup inthe photobioreactor, as described in the “Materials andmethods.” Figure 2 shows model-predicted and experimental-ly measured data of the gas holdup in the riser of the airlift,and Fig. 3 shows the superficial liquid flow velocity data inthe riser together with model predictions. Considering the factthat in both figures the experimental data were fitted with only

Table 1 Fitted and measured microalgae growth model parameters

Items Description Values Units

μmax Maximum specific growth rate 0.041 h−1

KI Light saturation constant 35.4 μmol photonsm−2 s−1

r Specific respiration rate 0.01 h−1

Ka Mass-specific light absorption coefficient 297 kg−1 m2

J Appl Phycol

one fitting variable, Leq, the equivalent length of the pipe,reasonably good fittings were obtained. Although the fittingof the experimental data showed some deviation from themodel-predicted gas holdup at lower aeration conditions(Fig. 2), the fitting of the experimentally obtained data ofliquid flow velocity corresponded well with the model curvefrom low to high aeration conditions (Fig. 3).

The fitted value of Leq was 15.0m equivalent length of tubewith an inner diameter of 32mm, and this number was close tothe value of the actual experimental settings. The riser anddowncomer were 2.35 m high (together=4.7 m in totallength), and the PVC pipe connecting the flat panel and airliftmodules was, in total, about 5.0 m in length. The two tees atthe bottom of the riser and the downcomer were estimated tototal 3.2 m in equivalent length (Green and Perry 2007), andthus, the equivalent length of the panel was only about 2.1 m.If we consider that the actual conduit length of all the panellanes was about 9 m and there were eight return bends, theestimated equivalent length of the panel seemed shorter thanexpected. This can be explained by the fact that the hydraulicdiameter of the panel lane was 1.7 times more than the tube.

Estimation of oxygen transfer in the riser

The estimation of oxygen mass transfer is crucial in designingphotobioreactor systems, especially for large-scale industrialapplications. Oxygen buildup by photosynthesis harmsmicroalgal growth in photobioreactors in at least two ways,though the significance of the harm differs among differentspecies and strains. First, a high level of DO inhibits photo-synthesis (Sánchez Mirón et al. 1999). Secondly, when com-bining with high levels of irradiance in outdoor conditions,

oxygen induces photooxidation, which leads to serious bio-mass losses (Carvalho et al. 2011). Usually, oxygen removal isnot reported to be a problem because gas–liquid transfer canbe sufficient for small-scale devices used for scientific re-search and published in the literature. However, the accumu-lation of oxygen might be the primary drawback havingcontributed to failures of large-scale outdoor photobioreactorsystems (Sánchez Mirón et al. 1999). Thus, engineering solu-tions for oxygen removal are important for photobioreactordesign and development and should even be customized forspecific species or strains of microalgae.

Numerous factors can affect the oxygen transfer in three-phase flow airlift devices, and accurate estimation of theoxygen volumetric mass transfer coefficient kLa using a vari-ety of available equations can be difficult for new prototypeslike the airlift used in this research. Here, we employedequations that had been used by other researchers in situationssimilar to ours (Molina et al. 2001). Both the model-estimatedvalues and experimentally measured data for kLa are shown inFig. 4. It appears that, at very low aeration rates, there is acertain deviation of the model estimation from the experimen-tal data; however, at middle to high aeration rates, the predic-tions are reasonably accurate. Considering the fact that inmostcases high oxygen transfer and mixing is desirable, we believethat Eqs. 6 and 7 may be precise enough to estimate oxygentransfer for the analysis and scale-up of this photobioreactortype. It is worth mentioning here that kLa was measured in anair–water system in the absence of any solid. The actual kLafor a culture broth with microalgal cells and ingredients suchas antifoams or other macromolecules might differ somewhatfrom our measurements and predictions. However, again, wedo not believe that considering those details is as crucial forthe purpose of photobioreactor design, especially given the

Fig. 2 Gas holdup measurement and model prediction. The dots repre-sent experimental data, the continuous line represents the model-predict-ed results, and the capped vertical lines show the standard errors of themeasurements

Fig. 3 Liquid flow velocitymeasurement andmodel prediction. The dotsrepresent experimental data, the continuous line represents the model-predicted results, and the capped vertical lines show the standard errors ofthe measurements

J Appl Phycol

generally dilute nature of algal cultures and the simple mineralmedia used. Thus, the approach employed here would bereasonably suitable for the purpose, considering the fact thatmany photobioreactors have been devised quite empiricallyinstead of being designed on the basis of engineeringprinciples.

The cultivation of microalgae in the reactor

S. obliquus and C. sorokiniana were cultivated in the airliftflat panel photobioreactor to test the reactor performance. Thecultivation of Scenedesmuswas hampered by a very slow pacein the development of a suspension cell culture accompaniedby serious biofouling, as shown in Fig. 5a. The biofoulingcould be primarily explained by the low turbulence in the flatpanel part of the reactor. The highest turbulent flow at the flatpanel region could only reach a Reynolds number of about5,500 at the maximum aeration rate within the bubble flowregime in the riser region. This hypothesis was supported bythe fact that biofouling was much less serious in the areawhere high turbulence would occur, as shown in Fig. 5a. Toobtain further support for the hypothesis, we cultivated theexact same Scenedesmus strain in an airlift tubularphotobioreactor that was operated at high turbulent flow con-ditions, and we observed that the suspension cell culture couldbe developed to high concentrations without biofouling on thereactor’s inner wall surface (data not shown).

Although it was possible to raise the liquid flow velocityand thus the Reynolds number by either increasing the heightof the riser or the diameter of the tubing system, we decided tocultivate another species, C. sorokiniana, in place ofS. obliquus in this reactor. Increasing the height of the riserwould increase the liquid pressure at least at the bottom of the

riser and thus increase the inlet air pressure, and increasing thetubing sizes would increase the light–dark zone of the reactorvolume and capital cost. We expected that Chlorella cellsmight create less or no fouling because these cells weresmaller in size and more spherical than those ofScenedesmus, and at the same time Chlorella did not havespines, which we thought also to be a factor contributing to theserious biofouling of Scenedesmus.

The cultivation of Chlorella, under the conditions definedin the “Materials and methods” section, was successful in thereactor, and an axenic homogeneous cell culture could bedeveloped to high cell concentrations, as shown in Fig. 5b.During the whole cultivation process, the DO level at the inletof the flat panel could be kept well below 120 % of airsaturation, which justified our assumption of no oxygen inhi-bition in the light-limited growth model (“Microalgae growthmodeling and parameter determination” section). The averagelight intensity on the surface of the reactor was about150 μmol photons m−2 s−1, and the biomass yield on lightenergy ranged from 0.2 to 0.6 g mol−1 photons, which wascomparable with data reported for the samemicroalgal speciesin a flat panel reactor (Kliphuis et al. 2010). The OD680/OD530

ratio was constant at about 1.2 for all the samples taken duringthe whole growth period, which indicated that pigment levelswere marginally, if at all, affected by the average light inten-sity in the reactor (Kliphuis et al. 2010). A typical growthcurve and the model-predicted growth are shown in Fig. 6.The close match of the model-predicted growth and measuredbiomass and OD data indicated that microalgal growth in thereactor could be described by the light-limited growth modelspresented in the “Materials and methods” section. The light-limited growth of microalgae indicated the mass transfer of thereactor was adequate. In conclusion, the airlift flat panelphotobioreactor enabled to develop axenicC. sorokiniana cellculture to adequate biomass concentrations with controlledpH, light input, DO, and temperature conditions.

The scalability of the airlift flat panel photobioreactor

The scale-up of photobioreactors constitutes a major challenge(Molina et al. 2000; Janssen et al. 2003). Just like bioreactorsfor heterotrophic microorganisms, photobioreactors need tobe scaled up to cut down capital and operation costs.However, the scale-up of photobioreactors is confronted withtwo major constraints. First, upon increase in thephotobioreactor volume, the light penetration becomes anissue due to the mutual shading of the cells. The light pathin the cell culture should not exceed about 10 cm to prevent alarge portion of the culture from falling into the dark zone ofthe reactor, where only respiration instead of photosynthesisoccurs. Secondly, for closed transparent vessels with a shortlight path, such as long tubes, the accumulation of oxygenproduced by microalgae photosynthesis limits the length of

Fig. 4 Estimated and experimentally measured values of kLa in the riser.The dots represent experimental data, the continuous line represents themodel-predicted results, and the capped vertical lines show the standarderrors of the measurements

J Appl Phycol

the vessel for a single photobioreactor device (Molina et al.2001).

The present airlift flat panel photobioreactor should bescalable to large scales for industrial applications. The scale-up of the reactor volume can be realized by enlarging thesurface area and specifically by the elongation of the individ-ual flowing lanes inside of the flat panel portion withoutchanging the light path. In this way, the light path is main-tained, and thus, the light penetration can be kept constant forthe scaled-up reactors. The light-dependent growth would notbe changed, and similar light utilization efficiency would beexpected, comparing with small-scale reactors of this type or,generally, with conventional flat panel reactors. The airliftportion can also be scaled up for O2 removal or CO2 additionsimilarly to the scale-up of airlift tubular photobioreactors(Molina et al. 2001). If we imagine the flowing lanes of theflat panel module as transparent tubes, the airlift flat panelreactor would share a similar liquid flow pattern as airlifttubular photobioreactors. The O2 accumulated by photosyn-thesis while the cell culture is circulating through the flowinglanes would be stripped out at the riser of the airlift, and the

CO2 consumed would be replaced by injecting CO2 at thebottom of the downcomer (Rubio et al. 1999).

The scale-up and engineering of the airlift flat panelphotobioreactor can be based on the equations tested in thisresearch. Similarly to airlift tubular photobioreactors, the sizeof the reactor will also be limited by the oxygen removalcapacity of the external airlift module (Molina et al. 2001).Since the oxygen removal capacity can be estimated by Eqs. 6and 7, the size limits of the flat panel can be estimated, giventhe lighting conditions and provided that the relevant physio-logical data of the microalgae species concerned are known.The liquid flow velocities and turbulence levels can also beestimated, using Eqs. 3, 4, and 5. Putting all the predictedparameters together, the geometry of the scaled-up airlift flatpanel photobioreactor and the productivity of the microalgaespecies cultivated in the reactor can be estimated, providedthat the environmental conditions and growth characteristicsof the microalgae species are known (Slegers et al. 2011;Quinn et al. 2012).

As we discussed above, the size of the reactor is limited bythe oxygen removal capacity of the airlift module. To scale upthe reactor, we can first scale up the airlift module, and then,we can define the volume and configuration of the panel onthe principle of avoiding prohibitive oxygen levels. For con-venience of operation and referring to the literature (Aciénet al. 2001), we scale up the airlift to a configuration of 2.0 mhigh and 0.1 m in diameter. Assuming the total aquatic equiv-alent length of the reactor to be roughly 80 m long and 0.1 min diameter and according to Eqs. 3, 4, 5, 6, and 7, the gasholdup, superficial liquid flow velocity, and kLa can be pre-dicted versus the aeration rates (shown in Figs. 1, 2, and 3 insupplementary materials) with the highest aeration rate at50 L min−1, which is still in the bubbly flow regime (testedby experiments, data not shown). On the one hand, if theupper limit of DO concentration of the incoming flow cellculture to the airlift is decided (which can be a design param-eter related to certain microalgae species), the maximal oxy-gen removal capacity of the airlift can be further estimated,

Fig. 5 Cultivation of Scenedesmus obliquus (a) and of Chlorellasorokiniana (b) in the airlift flat panel photobioreactor. Serious biofoulingoccurred while cultivating Scenedesmus obliquus due to low turbulence

in the flat panel part of the reactor (a). On the contrary, the reactor enabledthe cultivation of Chlorella sorokiniana to adequate biomass concentra-tions under controlled conditions (b)

Fig. 6 Cell growth of Chlorella sorokiniana in the airlift flat panelphotobioreactor. The diamonds are OD measurement data, the trianglesare dry weight data, and the standard errors are shown by capped verticallines. The continuous line represents the model-predicted growth data

J Appl Phycol

according to Eq. 2. For example, if the incoming DO level ofthis scaled-up airlift is 20 mg L−1, the outflow DO could reachas low as 16.1mg L−1 when the aeration is at the upper limit ofbubby flow regime (shown in Fig. S4 of supplementary ma-terial). Considering the superficial liquid flow velocity anddiameter of the airlift tube, the maximal oxygen removalcapacity is about 10.5 mg s−1 by the airlift module (shownin Fig. S5 of supplementary material). On the other hand, if wecan assume that the highest production of the flat panel reactoroccurs at a surface light intensity of 1,000 μmolphotons m−2 s−1 (considering the outdoor light conditionsand potential light inhibition effects, Sierra et al. 2008), themaximal productivity would occur at a biomass concentrationof about 1.8 g L−1 forC. sorokiniana used in this research, andthe maximal productivity would be about 0.033 g L−1 h−1 for a10-cm-thick flat panel reactor (shown in Fig. S6 of the sup-plementary material), according to Eqs. 8, 9, and 10 and usingparameters in Table 1. Thus, based on the stoichiometricrelationship of O2 and biomass established in Eq. 11, the airliftmodule would be sufficient for a total reactor volume of about520 L. The airlift module and connecting tubingmight occupyas much as 40 L, and a 480-L panel can be attached to theairlift, i.e., 0.1 m in width, 1.0 m in height, and 4.8 m in length.The flat panel can be further divided into ten (or nine) lanes tomake bulk flow possible in that panel, the liquid flow velocityand Reynolds number are reduced to the levels appropriate formicroalgae growth, and at the same time, homogeneousmixing of the cell culture is guaranteed. The total equivalentlength of the reactor can also be roughly 80 m in length and0.1 m in diameter.

The maximal power consumption of the above design canbe easily predicted or derived, i.e., 16.3 W for 520 L or31.3 W m−3. This number is slightly lower than that reportedfor some flat panel reactors (Sierra et al. 2008; Qiang et al.1998), which might be explained to some degree by the highturbulence caused by bubbles in the airlift and/or the spargerdesign. It is difficult to predict the productivity of the reactordue to lack of outdoor cultivation data, but it should bereasonable to expect that it would yield a value similar to thatof traditional flat panel reactors of similar sizes. The reactorcan be further scaled up, either by enlarging the diameter orincreasing the height of the airlift tube, to enhance the oxygen-stripping capacity and thus the reactor volume. For example, ifthe diameter of the airlift module increases twofold and thesuperficial liquid and gas velocity are kept the same as in theabove design example, the oxygen removal capacity couldrise asmuch as fourfold. The water head provided by the airliftcan be almost proportionally increased by increasing theheight of the airlift. The dividers inside the panel can be alsoadjusted appropriately in accordance with the cell culture’srequirements in terms of flow resistance and Reynolds num-ber. Thus, there should not be any major limitations in reactorsizing except for the characteristics of the materials of

construction, especially with respect to the resistance to thepressure caused by the water in the airlift.

The novelty of the airlift flat panel photobioreactorand the justification of the design

Although numerous photobioreactor designs have been pro-posed especially for research purposes, most of them cannotbe scaled up and are not appropriate for large-scale applica-tions. The large-scale industrial photobioreactor requires largeworking volume, inexpensive infrastructure, and easy opera-tion to minimize the unit production cost. If the volume of thereactor is small, not only is the construction cost of reactor perunit of volume high due to the sensors and valves installed andmaterials used, but also operations such as inoculation andharvest can be troublesome. This concept is similar to con-ventional bioreactors.

The only photobioreactors that have been scaled up forindustrial applications are tubular and flat panelphotobioreactors due to their inexpensive setup. Both typesutilize inexpensive transparent materials such as glass,plexiglass, or PVC membrane with low-cost structure.Usually, natural sunlight instead of artificial illuminationis employed for lighting of reactors to cut down the energyconsumption cost. The volume of both reactor types can beas large as several hundred liters and even up to severalthousand liters. Recently, flat panel reactors have drawnmore attention than tubular reactors due to their high sun-light utilization efficiency, and several new flat panel reac-tors have been developed for industrial applications. Somereactors utilize lower-cost PVC membranes replacing rela-tively more expensive plexiglass and use water to supportthe reactor and to compensate for environmental tempera-ture changes (Morweiser et al. 2010; Wijffels and Barbosa2010). Another example is to utilize static mixers for bettersunlight exposure, in order to enhance the light utilizationefficiency (Bergmann et al. 2013; Degen et al. 2001;Münkel et al. 2013).

Our design consisted in adding an external airlift andbaffles to the flat panel reactor in order to reduce the energyconsumption and enhance the mixing. In conventional flatpanel photobioreactors, the compressed air is bubbled directlyfrom the bottom of the flat panels and the cell culture does nothave bulk flow (Sierra et al. 2008). Consequently, a largevolume of compressed air is consumed to strip the oxygenand to mix the cell culture, and serious biofouling can occurpartially due to the low level of turbulence. The added externalairlift has dual functions—one is to strip oxygen and the otheris to drive the liquid flow. Airlift pumps have advantages overmechanical pumps because not only are they inexpensive andlong lasting but they also impose much less mechanical stressto microalgal cells. The airlift pump can be easily scaled upaccording to engineering equations and correlations such as

J Appl Phycol

those employed in this report and can even be standardized tofurther decrease the cost. Thus, adding an external airlift doesnot substantially raise the cost and improves the cultivatingconditions of photobioreactors. Added baffles in the flat panelpart of the reactor help direct the liquid flow in a serpentinemanner, which enhances the degree of mixing and turbulence.The bulk flow of the cell culture would be expected to makealgal cells experience the same light regime and uniformnutrient supply, enhance the mixing and turbulence, and re-duce energy consumption (Zhang et al. 2013). Although itmight be challenging to scale up a flat panel with baffles in thesense of maintaining the same dynamic environment at thehigher scale, the recent trends in developing plasticmembrane-based flat panel reactors indicate that the cost ofscale might still be affordable (Degen et al. 2001).

The advantages and disadvantages of the airlift flat panelreactor

The present airlift flat panel photobioreactor appears toshare some advantages of both airlift tubular and flat panelphotobioreactors. First, it is expected that this new reactorwould have high areal production and biomass concentra-tion levels like traditional flat panel photobioreactors be-cause both of them share the same light penetration anddistribution pattern due to similar geometries. Hence, thescalability of the flat panel section can be assured as itwould not affect light utilization efficiency. Secondly, thereactor could be scaled up just like airlift tubularphotobioreactors because of the similar pattern of liquidflow. The equations regarding cell growth, light penetra-tion, mixing, and oxygen removal that have been devel-oped for flat panel and airlift tubular reactors can bereliably applied for the scale up of this kind ofphotobioreactor. Thirdly, the industrial photobioreactorbased on this concept would be stable and inexpensiveto be operated like any airlift tubular photobioreactor.Comparing compressed air consumption between the airliftflat panel and a traditional flat panel reactor (Sierra et al.2008), it can be concluded that the former might need lesscompressed air for the same level of oxygen removalcapacity and has a higher energy efficiency than the latter.Lastly, comparing with traditional flat panel reactors, theairlift flat panel photobioreactor should have a muchhigher CO2 utilization efficiency. By using the externalairlift, the CO2 addition and O2 removal are uncoupled.The CO2 can be injected at the downcomer of the airliftas was done in this research, and bubbles of nondissolvedCO2 would circulate with the fluid to the flat panel regionfor further absorption before being lost to the atmosphere(Rubio et al. 1999). In this way, the residence time ofCO2 bubbles could be increased to tens of seconds, andCO2 utilization efficiency could be substantially improved.

The airlift flat panel photobioreactor might also havesome limitations. First, it might be more challenging tomake flat panels with internal flow lanes in an inexpen-sive way at a large scale. Using thin film plastic materialsmight be a solution to scale up the flat panel reactor, butstill, the supporting structure would definitely increase thecost (Dillschneider and Posten 2013). Secondly, the work-ing volume of the photobioreactor would also be restrictedby the oxygen removal ability of the airlift as in the caseof airlift tubular photobioreactors. Thirdly, if the liquidflow is only driven by the airlift, the turbulence of theflow might not be high enough for some microalgalspecies which need high turbulence to be kept insuspension.

Conclusion

The newly designed prototype air l i f t f la t panelphotobioreactor is equipped with an external airlift pumpand enables bulk liquid flow in the flat panel part. This reactorcan be modeled using simple equations and should be scalablefor industrial applications. Since it shares a similar liquid flowpattern with the airlift tubular photobioreactor and the lightpenetration can be kept constant upon scale-up as in conven-tional flat panel photobioreactors, this industrial design shouldhave both higher light utilization efficiency and lower con-struction and operation costs. The scale-up and engineering ofindustrial airlift flat panel photobioreactors can be based onthe mathematical equations employed in this research.

Acknowledgments This work is supported by the BEMA project (Bio-Energy from Micro-Algae, contract no. 6161, TWEED Cluster) of theWalloon Region, Belgium, and The Belgian National Fund for ScientificResearch (Fonds de la Recherche Scientifique—FNRS) projectPHOTOFUNDS. Philippe Demarche, Rakesh Nair, Emna Bouhajja,and Charles Junghanns are acknowledged for useful discussions andtechnical assistance.

References

Acién Fernández FG, Fernández Sevilla JM, Molina Grima E (2013)Photobioreactors for the production of microalgae. Rev Environ SciBiotechnol 12:131–151

Acién FG, Fernández JM, Sánchez JA, Molina Grima E, Chisti Y (2001)Airlift-driven external-loop tubular photobioreactors for outdoorproduction of microalgae: assessment of design and performance.Chem Eng Sci 56:2721–2732

Acién FG, Fernández JM, Magán JJ, Molina E (2012) Production cost ofa real microalgae production plant and strategies to reduce it.Biotechnol Adv 30:1344–1353

Bergmann P, Ripplinger P, Beyer L, Trösch W (2013) Disposable flatpanel airlift photobioreactors. Chem Ing Tech 85:202–205

J Appl Phycol

Bitog J, Lee IB, Lee CG, Kim KS, Hwang HS, Hong SW, Seo IH,Kwon KS, Mostafa E (2011) Application of computationalfluid dynamics for modeling and designing photobioreactorsfor microalgae production: a review. Comput Electron Agric76:131–147

Carvalho A, Silva S, Baptista J, Malcata F (2011) Light require-ments in microalgal photobioreactors: an overview ofbiophotonic aspects. Appl Microbiol Biotechnol 89:1275–1288

Cerri MO, Baldacin JC, Cruz AJG, Hokka CO, Badino AC (2010)Prediction of mean bubble size in pneumatic reactors. BiochemEng J 53:12–17

Chen CY, Yeh KL, Aisyah R, Lee DJ, Chang JS (2011) Cultivation,photobioreactor design and harvesting of microalgae for biodieselproduction: a critical review. Bioresour Technol 102:71–81

Chisti Y (1989) Airlift bioreactors. Elsevier Applied Science, New YorkChisti MY (1999) Mass transfer. In: Flickinger MC, Drew SW

(eds) Encyclopedia of bioprocess technology: fermentation,biocatalysis and bioseperation, vol 3. Wiley, New York, pp1607–1640

Chisti Y (2007) Biodiesel from microalgae. Biotechnol Adv 25:294–306Chisti MY, Moo-Young M (1987) Airlift reactors: characteristics, ap-

plications, and design considerations. Chem Eng Commun 60:195–242

Chisti MY, Halard B, Moo-Young M (1988) Liquid circulation in airliftreactors. Chem Eng Sci 43:451–457

Degen J, Uebele A, Retze A, Schmid-Staiger U, Trösch W (2001)A novel airlift photobioreactor with baffles for improved lightutilization through the flashing light effect. J Biotechnol 92:89–94

Delrue F, Setier PA, Sahut C, Cournac L, Roubaud A, Peltier G, FromentAK (2012) An economic, sustainability, and energetic model ofbiodiesel production from microalgae. Bioresour Technol 111:191–200

Dillschneider R, Posten C (2013) Closed bioreactors as tools formicroalgae production. In: Lee JW (ed) Advanced biofuels andbioproducts. Springer, New York, pp 629–649

Green D, Perry R (2007) Perry’s chemical engineers’ handbook, 8th edn.McGraw-Hill, New York

Haag A (2007) Algae bloom again. Nature 447:520–521Huang YM, Maliakal S, Cheney DP, Rorrer GL (1998) Comparison of

development and photosynthetic growth for filament clumps andregenerated microplantlet cultures of Agardhiella subulata(Rhodophyta, Gigartinales). J Phycol 34:893–901

Janssen M, Tramper J, Mur LR, Wijffels RH (2003) Enclosedoutdoor photobioreactors: light regime, photosynthetic effi-ciency, scale-up, and future prospects. Biotechnol Bioeng 81:193–210

Kliphuis AMJ, deWinter L, Vejrazka C,Martens DE, JanssenM,WijffelsRH (2010) Photosynthetic efficiency of Chlorella sorokiniana in aturbulently mixed short light-path photobioreactor. Biotechnol Prog26:687–696

Lehr F, Posten C (2009) Closed photo-bioreactors as tools for biofuelproduction. Curr Opin Biotechnol 20:280–285

Leupold M, Hindersin S, Kerner M, Hanelt D (2013) The effect ofdiscontinuous airlift mixing in outdoor flat panel photobioreactorson growth of Scenedesmus obliquus. Bioprocess Biosyst Eng 36:1653–1663

Li J, Xu NS, SuWW (2003) Online estimation of stirred-tank microalgalphotobioreactor cultures based on dissolved oxygen measurement.Biochem Eng J 14:51–65

Li J, Zhu D, Niu J, Wang G (2011) An economic assessment ofastaxanthin production by large scale cul t ivation ofHaematococcus pluvialis. Biotechnol Adv 29:568–574

Liao LH, Parlos A, Griffith P (1985) Heat transfer, carryover, and fallback in PWR steam generators during transients. Division of

Accident Evaluation, Office of Nuclear Regulatory Research, U.S.Nuclear Regulatory Commission

Lundquist T, Woertz I, Quinn N, Benemann J (2010) A realistic technol-ogy and engineering assessment of algae biofuel production. EnergyBiosciences Institute, University of California Berkeley, California.Final Report. (http://www.energybiosciencesinstitute.org/sites/default/files/media/AlgaeReportFINAL.pdf)

Molina E, Acién Fernández F, García Camacho F, Camacho Rubio F,Chisti Y (2000) Scale-up of tubular photobioreactors. J Appl Phycol12:355–368

Molina E, Fernández J, Acién F, Chisti Y (2001) Tubular photobioreactordesign for algal cultures. J Biotechnol 92:113–131

Morweiser M, Kruse O, Hankamer B, Posten C (2010) Developmentsand perspectives of photobioreactors for biofuel production. ApplMicrobiol Biotechnol 87:1291–1301

Münkel R, Schmid-Staiger U, Werner A, Hirth T (2013)Optimization of outdoor cultivation in flat panel airlift reactorsfor lipid production by Chlorella vulgaris. Biotechnol Bioeng110:2882–2893

Munoz R, Guieysse B (2006) Algal–bacterial processes for the treatmentof hazardous contaminants: a review. Water Res 40:2799–2815

Norsker N-H, Barbosa MJ, Vermuë MH, Wijffels RH (2011) Microalgalproduction—a close look at the economics. Biotechnol Adv 29:24–27

Petkov G, Ivanova A, Iliev I, Vaseva I (2012) A critical look at themicroalgae biodiesel. Eur J Lipid Sci Technol 114:103–111

Qiang H, Zarmi Y, Richmond A (1998) Combined effects of lightintensity, light-path and culture density on output rate of Spirulinaplatensis (Cyanobacteria). Eur J Phycol 33:165–171

Quinn JC, Turner CW, Bradley TH (2012) Scale-up of flat platephotobioreactors considering diffuse and direct light characteristics.Biotechnol Bioeng 109:363–370

Ratchford IAJ, Fallowfield HJ (1992) Performance of a flat plate, air-liftreactor for the growth of high biomass algal cultures. J Appl Phycol4:1–9

Reyna-Velarde R, Cristiani-Urbina E, Hernández-Melchor DJ, ThalassoF, Cañizares-Villanueva RO (2010) Hydrodynamic and mass trans-fer characterization of a flat-panel airlift photobioreactor with highlight path. Chem Eng Process: Process Intensif 49:97–103

Rubio FC, Fernández FGA, Pérez JAS, Camacho FG, Grima EM (1999)Prediction of dissolved oxygen and carbon dioxide concentrationprofiles in tubular photobioreactors for microalgal culture.Biotechnol Bioeng 62:71–86

Sánchez Mirón A, Contreras Gómez A, Garcı́a Camacho F, MolinaGrima E, Chisti Y (1999) Comparative evaluation of compactphotobioreactors for large-scale monoculture of microalgae. JBiotechnol 70:249–270

Shaikh A, Al-Dahhan MH (2007) A review on flow regime transition inbubble columns. International J Chem React Eng 5 (1): art. no. R1

Sierra E, Acien F, Fernandez J, Garcia J, Gonzalez C, Molina E (2008)Characterization of a flat plate photobioreactor for the production ofmicroalgae. Chem Eng J 138:136–147

Slegers PM, Wijffels RH, van Straten G, van Boxtel AJB (2011) Designscenarios for flat panel photobioreactors. Appl Energy 88:3342–3353

Stein JR (ed) (1979) Handbook of phycological methods: culturemethods and growth measurements. Cambridge University Press,Cambridge

Stephens E, Ross IL, King Z, Mussgnug JH, Kruse O, Posten C,Borowitzka MA, Hankamer B (2010) An economic and tech-nical evaluation of microalgal biofuels. Nat Biotechnol 28:126–128

Su WW, Li J, Xu N-S (2003) State and parameter estimation ofmicroalgal photobioreactor cultures based on local irradiance mea-surement. J Biotechnol 105:165–178

Ugwu C, Aoyagi H, Uchiyama H (2008) Photobioreactors for masscultivation of algae. Bioresour Technol 99:4021–4028

J Appl Phycol

Wallis GB (1969) One-dimensional two-phase flow. McGraw-Hill,New York

Wijffels RH, Barbosa MJ (2010) An outlook on microalgal biofuels.Science 329:796–799

Wijffels RH, Kruse O, Hellingwerf KJ (2013) Potential of industrialbiotechnology with cyanobacteria and eukaryotic microalgae. CurrOpin Biotechnol 24:405–413

Zhang QH, Wu X, Xue SZ, Wang ZH, Yan CH, Cong W (2013)Hydrodynamic characteristics and microalgae cultivation in a novelflat-plate photobioreactor. Biotechnol Prog 29:127–134

Znad H, Bales V, Kawase Y (2004) Modeling and scale up of airliftbioreactor. Comput Chem Eng 28:2765–2777

Zuber N, Findlay JA (1965) Average volumetric concentration in two-phase flow systems. J Heat Transf 87:453–468

J Appl Phycol