Biochemical Engineering Journal 87 (2014) 25–32

Contents lists available at ScienceDirect

Biochemical Engineering Journal

journal homepage: www.elsevier.com/locate/bej

Regular Article

Improvement of mixing time, mass transfer, and power consumption

in an external loop airlift photobioreactor for microalgae cultures
Ali Pirouzi a , Mohsen Nosrati a,∗ , Seyed Abbas Shojaosadati a , Saeed Shakhesi b
a
Biotechnology Group, Faculty of Chemical Engineering, Tarbiat Modares University, P.O. Box 14115-143, Tehran, Iran
b
Engineering Research Institute, P.O. Box 13445-754, Tehran, Iran

a r t i c l e i n f o a b s t r a c t

Article history: Longer mixing times and higher power consumption are common problems in the design of photobiore-
Received 16 November 2013 actors. In this study, a vertical triangular external airlift loop photobioreactor was designed, constructed
Received in revised form 16 March 2014 and operated for microalgae production studies. Gas feeding was performed by two spargers: one at
Accepted 22 March 2014
the bottom of the hypotenuse (downcomer) and another at the bottom of the vertical side (riser).
Available online 31 March 2014
This configuration provided more effective countercurrent liquid–gas flow in the hypotenuse. The mass
transfer coefficient, gas hold-up, mixing time, circulation time, dimensionless mixing time, bubble size,
Keywords:
and volumetric power consumption were measured and optimized using response surface methodol-
Airlift bioreactors
Mixing
ogy. Investigations were carried out on the performance of the riser (the vertical side), downcomer
Gas hold-up (the hypotenuse), and separator. The countercurrent flow in the hypotenuse provided sufficient contact
Mass transfer between gas and liquid phases, and increased mixing and mass transfer rates, in contrast to the results
Power consumption of previous studies. The promising results of this geometry were shorter mixing time and a significant
Counter-current flow decrease in volumetric power consumption in comparison with other configurations for photobioreac-
tors.
© 2014 Elsevier B.V. All rights reserved.

1. Introduction Microalgae are sensitive to stress [18–21] and harsh mechanical

Hydrodynamics and transport phenomena are crucial and
complicated aspects of sparged photobioreactor cultivation of
microalgae [1–3]. The level of complexity increases strongly in
response to interactions between the fluid phases, biomass, and
nutrients generally found in photobioreactors [4]. The growth
rate of most microalgae is rapid [5–11], thus, photosynthesis in
microalgae demands high CO2 absorption and high O2 release.
Consequently, the rate of CO2 absorption and O2 desorption are
usually limiting factors inhibiting the overall growth rate of the
biomass [2,12–14]. Photosynthesis depends on sufficient mass
transfer between the three phases involved; liquid culture, sus-
pended solid biomass, and sparged gas, which is the main carbon
source required for growth [13,15–17]. Superficial fluid velocity,
gas hold-up, photobioreactor geometry, and mixing time influence
overall mass transfer [2,10], which in turn affects biomass function
and process productivity [10,16].
∗ Corresponding author. Tel.: +98 21 82884372; fax: +98 21 82883381.
E-mail address: mnosrati20@modares.ac.ir (M. Nosrati).
stirring may discourage growth [16,22–26]. Photobioreactors have
been designed and tested to address this limitation, the majority of
which are gas-lift photobioreactors [3,13,17,27]. Both the intensity
and history of illumination influence the growth rate of microalgae
[2,3]. Loop photobioreactors may yield better performance in this
area over flat-panel or bubble column air-lift photobioreactors.
External loop airlift bioreactors show good performance at dif-
ferent gas velocities, good compatibility with sensitive organisms,
are easy to maintain, low cost, and have low energy consump-
tion [5,28]. They can accommodate multiphase and heterogeneous
gas-liquid systems, allow once to scale-up, and provide good
sterilization, making them promising candidates for microalgae
cultivation [13,16,17,29].
The present study designed, built, and operated a specially
shaped external loop airlift photobioreactor. The mass transfer
coefficient, gas hold-up, mixing time, and circulation time were
studied and correlated for the riser (vertical side), downcomer
(hypotenuse), and separator. The main variables in this configu-
ration were superficial gas velocities sparged from the bottom of
the vertical side (Vgs2 ) and the hypotenuse (Vgs1 ). The results of
the experimental study were compared with other configurations
for bioreactors. Better results were found in all aspects, especially

http://dx.doi.org/10.1016/j.bej.2014.03.012
1369-703X/© 2014 Elsevier B.V. All rights reserved.
26 A. Pirouzi et al. / Biochemical Engineering Journal 87 (2014) 25–32

Nomenclature

1 gas feeding from bottom of the hypotenuse

2 gas feeding from bottom of the riser
A cross-sectional area, m2
C concentration of dissolved oxygen, mg l−1
d bubble diameter size, mm
D downcomer
DO dissolved oxygen, mg l−1
e volumetric power consumption, W m−3
g global gravity acceleration, m s−2
G gas
H height
kL a volumetric oxygen mass transfer coefficient, h−1
L liquid
P/V volumetric power consumption, W m−3
R riser
RSM response surface method
S separator
t time, arbitrary time, h
t0 zero time state, s
tc circulation time, s
tm mixing time, s
T total
Vgs superficial gas velocity, m/s
X1 , X2 RSM parameters
Y expected response value predicted from RSM,
dimensionless

Greek letters
˛i , ˛j , ˛ij parameters estimated from the RSM regression
ε gas hold-up, dimensionless
m dimensionless mixing time
 liquid density, kg m−3
* saturation state
◦ degree

mixing time (tm ), mass transfer coefficient (kL a), and volumetric
power consumption (P/V).

2. Materials and methods

2.1. Photobioreactor

The 63 l Plexiglas (with PVC joints and separator) photobioreac-

tor is a hybrid of an external loop and a vertical isosceles triangle
configuration (45◦ ) in which gassed liquid moves up the vertical
side and comes down the hypotenuse. To provide better mixing
(low tm ), a countercurrent flow was introduced into the hypotenuse
by gas feeding at the bottom of the photobioreactor. Achieving a
high mixing rate in the hypotenuse was essential to the geometry.
A schematic diagram of the photobioreactor is illustrated in Fig. 1.
The volume of the separator (No. 1) was 18 l and the combined vol-
ume of the vertical (No. 2) and horizontal (No. 3) sides was 8.9 l. The Fig. 1. Schematic of external airlift loop photobioreactor. 1: Separator; 2: Vertical
diameter of the hypotenuse (No. 4) was 14 cm and the diameter of side, riser; 3: Horizontal side; 4: Diameter of hypotenuse; 5: Hypotenuse, down-
other two sides was 7 cm. The hypotenuse had a length of 234 cm comer; 6: Bubble; 7: Bubble sparger; 8: Light source; 9: Digital gas flow meter; 10:
Compressor; 11: Gas sterilizer, filter; 12: PC; 13: Thermometer; 14: Heater; 15: pH
and the lengths of other two sides, including joints, were 170 cm
probe and controller; 16: NaOH solution; 17: HCl solution; 18: Dozing pump; 19:
each. The diameter of separator was 20 cm. Dissolved oxygen (DO) sensor; 20: Heat exchanger.
In earlier photobioreactors, the downcomers (No. 5) have
demonstrated poor mass transfer and gas hold-up, which lim-
its the residence time of air bubbles (No. 6) in the downcomer.
In the present study, the creation of a high countercurrent flow
and vortices in the downcomer increased the performance of the
mixing process and decreased tm , eliminating this limitation. The
 A. Pirouzi et al. / Biochemical Engineering Journal 87 (2014) 25–32 27

Table 1 The pH was measured by 4 similar online probes (No. 15) (Lutron
Photobioreactor geometry and operational parameters.
TR–pHT1A4) having an accuracy of 0.1; 3 were installed on the
Description Unit Value downcomer and one at the top of the vertical side near the sep-
Photobioreactor volume m3 0.063 arator. NaOH (No. 16) and HCl (No. 17) solutions, both 1.00 normal
Separator diameter m 0.2 concentrations, were injected to control pH as directed by the com-
Separator height m 0.44 puter (No. 12) using two dosing pumps (No. 18) (Vigor magnet
Separator volume m3 0.018 pump) and a control circuit with adjustable work and pause times.
Riser diameter m 0.06
Three dissolved oxygen (DO) sensors (No. 19) (Lutron TR–DOT1A4)
Riser height m 1.46
Riser volume m3 0.0041 with a response time of 4 s for 63% saturation were placed (one
Horizontal side diameter m 0.06 each for downcomer, riser, and separator). A heat exchanger (No.
Horizontal side length m 1.70 20) was used in the middle of the riser for temperature control.
Horizontal side volume m3 0.0048
Hypotenuse diameter m 0.14
Hypotenuse length m 2.34 2.2. Measurements
Hypotenuse volume m3 0.036
◦
Hypotenuse angle 45 2.2.1. Mixing time, liquid circulation time and linear liquid
◦
Operating temperature C 25
velocity
Operating pH # 8
Air filter size ␮ 2 Mixing time was defined as the time needed to reach 95% com-
Acid and base concentration N 1 plete mixing [30–32], which is commonly determined by sensing
Vgs1 m/s 0.0005, 0.0014, 0.0037, the acid traces using pH meters [31,33,34]. The circulation time
0.0060, 0.0069 (tc ) and the mean linear liquid velocity were determined using the
Vgs2 m/s 0.0040, 0.0069, 0.0139,
0.0209, 0.0238
same technique [30,35]. The circulation time in the photobioreac-
tor was estimated from the time period between the two adjacent
peaks of the response curves [36]. Each experiment was carried out
in triplicate by calculating the mean of three sensors.
downcomer (hypotenuse) was itself sparged from the bottom by a
moderate stream of very fine bubbles from the sparger (No. 7). The
2.2.2. Gas hold-up and mass transfer coefficient (kL a)
sparged downcomer, however, acted like a mixer with low power
The gas hold-up (ε) was evaluated by volumetric expansion as
consumption. The hypotenuse was exposed to a light source (No.
proposed by Chisti [32] and is shown in Eq. (1); where HG and HL ,
8) and the other two regions were dark. The level of liquid in the
respectively, are the gassed and un-gassed heights of the fluid in
top section was kept at 190 cm in the absence of gas.
each part of the photobioreactor. Total gas hold-up (εT ), is described
The gas flow rates were measured and controlled by two digi-
by Eq. (2). The εT relates the gas hold-up of the riser, downcomer,
tally calibrated gas flow meters (No. 9). A clean and dry pressured
and separator zones with the respective cross-sectional area.
air stream was available from a central compressor (No. 10) and
a gas sterilizer filter (2 ␮m) (No. 11) located after the compres- HG − HL
ε= (1)
sor. Also, a 40 l high purity (99.99% vol.) pressurized (150 bar) N2 HG
cylinder was used to measure kL a throughout the dynamic gassing-
AR · εR + AD · εD + AS · εS
out method. Geometric details of the photobioreactor are shown in εT = (2)
AR + AD + AS
Table 1.
All experiments were carried out at 25 ◦ C (±0.1). This was con- A is the cross-sectional area and subscripts R, D, and S represent the
trolled using a computer (No. 12) and a temperature loop controller riser, downcomer, and separator, respectively. The volumetric oxy-
(TLC) with 3 thermometers (No. 13) at 3 equidistant points on the gen mass transfer coefficient (kL a) was determined by the dynamic
hypotenuse. In the two-phase flow, critical gas and liquid zones method [37]; and the slope of the graph was calculated by Eq. (3).
formed in the downcomer, and most mass transfer was expected  C∗ − C 
0
to occur there; thus, all 3 thermometers were installed on the ln = kL a · (t − t0 ) (3)
C∗ − C
hypotenuse for accurate recording. The average temperature was
fed to the TLC, which controlled the overall temperature by means where C* , C0 , and C are, respectively, the DO saturation concentra-
of an on/off electrical heater (No. 14). tion, DO concentration at zero time (t0 ), and DO concentration at an

Table 2
Performance of hydrodynamic and mass transfer properties for Vgs1 and Vgs2 in RSM.

Run Superficial gas velocities Mass transfers Gas hold-up Bubble diameter Dimensionless Power Input
mixing time

(m/s) (1/h) (−) (mm) (−) (W/m3 )

X1 ; Vgs1 X2 ; Vgs2 kL a R kL a D kL a S εR εD εS dR dD m P/V

1 0.0014 0.0069 12.62 11.61 18.90 0.0680 0.0730 0.0850 3.9300 5.9700 1.1813 22.1200
2 0.0037 0.0040 18.00 14.00 19.00 0.0990 0.0895 0.1022 4.2150 5.0056 0.5660 37.1722
3 0.0037 0.0238 26.00 17.00 22.00 0.1649 0.1541 0.1976 3.8750 6.1682 0.7685 67.6229
4 0.0037 0.0139 14.00 16.00 22.00 0.1293 0.1254 0.1580 4.1910 5.6021 0.8570 52.3980
5 0.0060 0.0209 23.72 20.26 23.77 0.1700 0.1600 0.2200 4.3200 6.9701 0.2559 82.3200
6 0.0037 0.0139 14.90 15.10 21.00 0.1290 0.1270 0.1579 4.1905 5.6010 0.8571 52.3971
7 0.0037 0.0139 14.00 16.00 22.00 0.1292 0.1273 0.1581 4.1909 5.6005 0.8573 52.3979
8 0.0005 0.0139 9.00 13.00 18.00 0.0823 0.0885 0.1014 3.4000 6.7250 1.3592 25.4597
9 0.0069 0.0139 20.00 22.00 26.00 0.1654 0.1567 0.1983 4.5100 8.0800 0.1500 79.3354
10 0.0060 0.0069 18.39 19.23 22.76 0.1400 0.1200 0.1500 4.4800 6.9900 0.6337 60.4800
11 0.0037 0.0139 14.00 16.00 22.00 0.1290 0.1278 0.1583 4.1909 5.6009 0.8569 52.3980
12 0.0014 0.0209 17.82 12.64 19.56 0.1300 0.1200 0.1500 3.8010 5.9800 2.5579 44.0500
13 0.0037 0.0139 14.00 16.00 22.00 0.1297 0.1275 0.1583 4.1911 5.6010 0.8571 52.3975
28 A. Pirouzi et al. / Biochemical Engineering Journal 87 (2014) 25–32

Vgs2=0.0209 m/s Vgs2=0.0040 m/s

25 0.4
160 Vgs2(Exp)=0.0238 m/s Vgs2=0.0238 m/s
Vgs2(Exp)=0.0040 m/s Vgs2=0.0209 m/s
Vgs2(Exp)=0.0209 m/s Vgs2=0.0139 m/s
Vgs2=0.0069 m/s 0.35
Vgs2=0.0069 m/s
140 Vgs2(exp)=0.0069 m/s
Vgs2(Exp)=0.0139 m/s Vgs2=0.0040 m/s
20 Vgs2(Exp)=0.0069 m/s
Vgs2=0.0139 m/s
0.3
120 Vgs2(Exp)=0.0139 m/s Vgs2(Exp)=0.00397 m/s
Vgs2=0.0238 m/s
Vgs2=0.0139 m/s Vgs2=0.0209 m/s

100 Vgs2(Exp)=0.0209 m/s 0.25

Vgs2=0.0069 m/s 15

kLa D (1/h)
Vgs2=0.0238 m/s
tm (s)

εD (-)
Vgs2( Exp)=0.0238 m/s 0.2
80
Vgs2=0.0238 m/s
Vgs2=0.0209 m/s
10
60 Vgs2=0.0139 m/s
0.15
Vgs2=0.0040 m/s Vgs2=0.0069 m/s

Vgs2=0.0040 m/s
40 0.1
5
20 0.05

0
0 0.002 0.004 0.006 0.008 0 0
0 0.002 0.004 0.006 0.008 0.01
Vgs1 (m/s) Vgs1 (m/s)

Fig. 2. Mixing time vs. superficial gas velocities at T = 25 ◦ C (±0.1). Fig. 3. kL a D and εD vs. superficial gas velocities at T = 25 ◦ C (±0.1).

different operating conditions. A stepwise regression analysis was

arbitrary time (t). The total mass transfer coefficient (kL a)T , in Eq. (4)
was calculated by relating the kL a of the vertical side, hypotenuse
and separator zones with their respective cross-sectional areas
[32,38].
AR · (kL a)R + AD · (kL a)D + AS · (kL a)S
(kL a)T =
AR + AD + AS
P/V was calculated as a function of liquid density, superficial gas
velocity, and the ratio of the cross-sectional area of the vertical side
to the hypotenuse [32] as:
L · g · (AD · Vgs1 + AR · Vgs2
P/V = eT =
AD + AR
where L and g are the liquid density and global gravity accelera-
tion, respectively.
2.3. Experimental design to examine the effects of gas feeding
An on-face central composite design was used to investigate
the interactive effects of Vgs1 and Vgs2 on hydrodynamic and mass
transfer properties. Thirteen experiments were designed and con-
ducted using response surface methodology (RSM) with a central
composite design. The range of Vgs1 and Vgs2 , as key operating
parameters, were 0.0005–0.0069 m/s (central value = 0.0037 m/s)
and 0.0040–0.0238 m/s (central value = 0.0139 m/s), respectively.
The experimental design for X1 (Vgs1 ) and X2 (Vgs2 ) is shown in
Table 2. This table also lists the results of all responses under the
conducted by generating a second-order polynomial equation:
Y = ˛0 + ˛1 · X1 + ˛2 · X2 + ˛11 · X12 + ˛22 · X22 + ˛12 · X1 · X2 (6)
in which Y is the expected response value predicted by RSM; ˛i , ˛j
and ˛ij are the parameters estimated from the regression results
(4)
(i = 1,2; and j = 1,2). Ten performance indices were used to assess
the influence of these variants on hydrodynamic and mass transfer
properties.
3. Results and discussion
(5) The index of tm was used to analyze the quality of mixing in
the new photobioreactor. As shown in Fig. 2, this parameter was
strictly influenced by Vgs1 and Vgs2 . Fig. 2 shows that tm always
decreased linearly as Vgs1 increased, therefore, at a low Vgs1 value,
if Vgs2 increased, more time was required for complete mixing in the
photobioreactor. The reason for this is that the high liquid circula-
tion velocity causes the liquid to rapidly pass from the hypotenuse.
Despite the presence of a countercurrent flow in the hypotenuse,
less mixing occurred in this part of the photobiorector (which is
the main zone of mass transfer). Consequently, longer mixing times
were observed in the downcomer.
For low Vgs1 values, the downcomer could not produce making
strong vortices or retain liquid. As Vgs1 increased, small eddies were
generated and the intensity of the countercurrent flow improved,
resulting in shorter mixing times. As expected, and as Fig. 3 shows,
kL a increased as Vgs1 increased in the downcomer. Although Vgs1
is the main factor affecting kL a in the downcomer, it increased

Table 3
Final equation for actual factors in RSM analysis using Vgs1 and Vgs2 as key operating parameters.

Response Final equation in terms of actual factors R-squared

kL a R = + 18.4618 + 1.4809·Vgs1 − 7.1123·Vgs2 + 1.2367·Vgs2 2 0.9894

kL a D = + 10.31111 + 0.47315·Vgs1 + 0.4457·Vgs2 + 0.1398·Vgs1 2 0.9700
kL a S = + 13.3481 + 1.0930·Vgs1 + 2.0412·Vgs2 − 0.2247·Vgs2 2 0.9650
εR = + 6.7389E-003 + 0.0237·Vgs1 + 0.0183·Vgs2 − 1.9692E-003·Vgs1 ·Vgs2 − 5.7977E-004·Vgs1 2 + 3.1531E-004·Vgs2 2 0.9995
εD = + 0.0230 + 0.0145·Vgs1 + 0.0206·Vgs2 − 6.01228E-004·Vgs1 2 − 1.1425E-003·Vgs2 2 0.9929
εS = + 0.0147 + 0.0192·Vgs1 + 0.0265·Vgs2 + 3.0769E-004·Vgs1 ·Vgs2 − 7.3284E-004·Vgs1 2 − 1.2246E-003·Vgs2 2 0.9998
tm = −18.3899 + 2.2299·Vgs1 + 78.5629·Vgs2 − 5.9982·Vgs1 ·Vgs2 − 7.8103·Vgs2 2 0.9409
tc = −11.5318 + 2.1954·Vgs1 + 66.9539·Vgs2 − 10.1784·Vgs2 2 0.9864
 A. Pirouzi et al. / Biochemical Engineering Journal 87 (2014) 25–32 29

40 0.45
Vgs1(Exp)=0.0069 m/s 120
35 Vgs1(Exp)=0.0060 m/s
0.4
Vgs1(Exp)=0.0037 m/s
Vgs1=0.0069 m/s
30 Vgs1(Exp)=0.0014 m/s Vgs1=0.0060 m/s
0.35 100
Vgs1(Exp)=0.0005 m/s Vgs1=0.0037 m/s
25
Vgs1=0.0014 m/s
Vgs1=0.0005 m/s 0.3
20
kLa R (1/h)

80

εR (-)
15 0.25

tC (s)
10 0.2
Vgs1=0.0069 m/s

5
Vgs1=0.0060 m/s
60
Vgs1=0.0037 m/s
Vgs1=0.0014 m/s 0.15
Vgs1=0.0005 m/s
0
0.1 Vgs1=0.0005 m/s Vgs1(Exp)=0.0005 m/s
-5 40 Vgs1=0.0014 m/s Vgs1(Exp)=0.0014 m/s
0.05 Vgs1=0.0040 m/s Vgs1(Exp)=0.0040 m/s
-10 Vgs1=0.0060 m/s Vgs1(Exp)=0.0060 m/s
Vgs1=0.0069 m/s Vgs1(Exp)=0.0069 m/s
-15 0 20
0 0.005 0.01 0.015 0.02 0.025 0.03 0.035
0 0.005 0.01 0.015 0.02 0.025
V s2 (m/s)
g Vgs2 (m/s)

Fig. 4. kL a R and εR vs. superficial gas velocities at T = 25 ◦ C (±0.1).

Fig. 5. Circulation time vs. superficial gas velocities at T = 25 ◦ C (±0.1).

slightly when Vgs2 increased. Fig. 3 illustrates that gas hold-up in the
downcomer increased when Vgs1 increased, causing a significant Low liquid circulation caused by weak gas feeds was the main
improvement in kL a in the downcomer. reason for the decrease in kL a in the riser. When Vgs1 exceeded
Unlike Fig. 3, in which downcomer kL a increased slightly when the critical value of 0.013 m/s, kL a in the riser increased as Vgs1
Vgs2 increased, gas hold-up in the downcomer increased as Vgs2 increased. The kL a in the riser was more sensitive to Vgs2 than to
increased significantly. The correlation coefficients for superfi- Vgs1 . This can be seen in Fig. 4 and was confirmed by the correlation
cial gas velocities shown in Table 3 affirm these findings. Where coefficients in Table 3.
Vgs1 < 0.004 m/s (critical superficial gas velocity), similar gas hold- Although circulation and mixing times are vital variables for
up values were observed in the downcomer; however, where photobioreactors, they are not usually used separately in fluid
Vgs1 > 0.004 m/s, the gas hold-up in the downcomer increased hydrodynamics because there is a strong effect from the geometry
smoothly. of the photobioreactor. Dimensionless mixing time ( m = tm /tc ) is
Fig. 4 illustrates the variations of gas hold-up and kL a in the
riser. Since the gas feeds from bottom to top in both the riser and
downcomer, and gas bubbles were encountered at the top of the
riser, both Vgs1 and Vgs2 should have had a significant effect on gas
hold-up in the riser. The amount of gas hold-up in the riser at low
Vgs2 depended on the amount of gas feeding into the downcomer;
however, by increasing aeration in the riser, the effect of Vgs1 on
gas hold-up in the riser decreased.
The results show that, in most experiments, gas hold-up in the
riser was slightly more than in the downcomer (Table 2). The over-
all gas hold-up (riser, downcomer, and separator) is expressed by
Eq. (2). As expected, the influences of Vgs1 and Vgs2 on the over-
all gas hold-up were almost equal, as confirmed by the correlation
coefficients in Table 3.
The tc in loop photobioreactors that have a reverse interrela-
tion with liquid circulation velocity is equal to the time needed for
a single complete pass of the liquid through the entire length of
the loop photobioreactor. Fig. 5 shows changes in tc as a function
of superficial gas velocity. The figure indicates that tc was more
sensitive to Vgs2 than to Vgs1 . When Vgs2 ≤ 0.013 m/s, tc increased
linearly as Vgs1 increased; when Vgs2 > 0.013 m/s, tc decreased. This
range of superficial gas velocity can be considered the critical range
of velocity for tc .
The critical velocity and the parabolic form of the tc curve is
a result of the domination of the weight of the downcoming liq-
uid and the force of Vgs2 over the up-flow force resulting from
Vgs1 . When the downcoming liquid overcame Vgs1 , it decreased the Fig. 6. (a):  m and (b): kL a S vs. superficial gas velocities at T = 25 ◦ C (±0.1). Calcu-
lated data at: Vgs2 = 0.0040 m/s (------), Vgs2 = 0.0069 m/s (– – – –), Vgs2 = 0.0139 m/s
retention time required for liquid in the downcomer to mix, thus, (- - -), Vgs2 = 0.0209 m/s (-·-), Vgs2 = 0.0238 m/s (— —); Experimental data at:
decreasing tc . This is evident in Fig. 4, where Vgs2 ≤ 0.013 m/s, kL a in Vgs2 = 0.0040 m/s (), Vgs2 = 0.0069 m/s (), Vgs2 = 0.0139 m/s (䊉), Vgs2 = 0.0209 m/s
the riser decreased; it began to increase as soon as Vgs2 > 0.013 m/s. (), Vgs2 = 0.0238 m/s ().
30 A. Pirouzi et al. / Biochemical Engineering Journal 87 (2014) 25–32
Fig. 7. Mean bubble diameter of (a): downcomer and (b): riser vs. superfi-
cial gas velocities at T = 25 ◦ C (±0.1). (a): Calculated data at: Vgs2 = 0.0040 m/s
(-·-·-·-), Vgs2 = 0.0069 m/s (—), Vgs2 = 0.0139 m/s (—··), Vgs2 = 0.0209 m/s (- - -··),
Vgs2 = 0.0238 m/s (-·-); Experimental data at: Vgs2 = 0.0040 m/s (䊉), Vgs2 = 0.0069 m/s
(), Vgs2 = 0.0139 m/s (), Vgs2 = 0.0209 m/s (), Vgs2 = 0.0238 m/s ( ). (b): Calcu-
lated data at: Vgs1 = 0.0005 m/s (-·-), Vgs1 = 0.0014 m/s (- - - - -), Vgs1 = 0.0037 m/s
(-··), Vgs1 = 0.0060 m/s (- - -), Vgs1 = 0.0070 m/s (—·-); Experimental data at:
Vgs1 = 0.0005 m/s (), Vgs1 = 0.0014 m/s ( ), Vgs1 = 0.0037 m/s ( ), Vgs1 = 0.0060 m/s
(♦), Vgs1 = 0.0070 m/s ().
an appropriate parameter for studying fluid hydrodynamics in air-
lift photobioreactors [29]. In this study,  m is introduced in Fig. 6a
as a dependent variable to Vgs1 . Like Vgs2 in Fig. 4, which is the criti-
cal velocity for kL a in the riser, Vgs1 reaches 0.004 m/s in Fig. 6a and
is a critical velocity for  m . In Fig. 6a, when  m < 1 (Vgs1 > 0.004 m/s),
the effects of mixing surpass that of liquid circulation.
Fig. 6b shows that kL a in the separator is more sensitive to Vgs1 .
This occurs because the mixing process prevails over liquid circula-
tion in the separator. This, however, shows that kL a in the separator
is sustained by the downcomer rather than the riser. As seen in
Fig. 6b, kL a values for the separator and downcomer are similar
and do not have an extremum. The higher mass transfer coefficient
in the separator is the effect of increased mass transfer coefficient
in the riser (Figs. 3 and 6b).
Fig. 7a shows that the main factor affecting the size of rising
bubbles in the downcomer (dD ) is Vgs1 . Fig. 7a shows that varia-
tions in Vgs2 have a weak effect on bubble size in the downcomer.
Where Vgs1 < 0.004 m/s, bubble size in the downcomer decreased
as linear velocity increased. When Vgs1 > 0.004 m/s, it overcame the
downcoming liquid force and the up-flow stream in the vertical
Table 4
Actual and predicted values for Vgs1 and Vgs2 and the responses.
Variables Responses
Vgs1 Vgs2 kL a R kL a D
(m/s) (m/s) (1/h) (1/h)
Predicted 0.0050 0.0209 22.4507 18.1851
Actual 0.0050 0.0209 22.9194 19.1467
Fig. 8. kL a and tm vs. P/V in present and previous studies. Read left axis: This work
(), Loubière et al. [39]: VF1 ( ), Loubière et al. [39]: VF4 (), Yazdian et al. [40]:
VL = 0.12 m/s ( ); Read right axis: This work ( ), Reyna-Velard et al. [31] ( ),
Fadavi et al. [41]: VL = 0.013 m/s ( ), Fadavi et al. [41]: VL = 0.025 m/s ( ).
side caused by Vgs2 . This decreased  m , indicating that the mixing
process surpassed liquid circulation (Fig. 6a).
Increased Vgs1 in the downcomer created conflicting counter-
current gas–liquid streams, an increase in gas retention time and
gas hold-up in the downcomer (Fig. 3). The dynamic behavior
between the countercurrent fluids decreased  m in the hypotenuse.
The countercurrent flow created an agglomeration of bubbles
of increased size as a result of bubble movement on the surface
of the downcomer. As bubble size increased and their surface-
to-volume ratio (specific surface) decreased, kL a decreased in the
downcomer, as expected; however, the increase in gas hold-up in
the downcomer was so significant that it halted deterioration of
kL a in the downcomer (Fig. 3). The kL a in the downcomer increased
(Fig. 3). The retention time increased as Vgs2 increased, which lin-
early increased bubble size in the downcomer.
As stated,  m represents the importance of liquid mixing over
liquid circulation (Fig. 6a). The variation of  m for Vgs1 values of
about 0.004 m/s converts velocity to an inflection point on the
curves in Figs. 3 and 7a. When Vgs1 > 0.004 m/s, tm is greater than tc ;
when Vgs1 < 0.004 m/s, tc is greater than tm . When Vgs1 < 0.004 m/s,
liquid circulation surpasses liquid mixing; when Vgs1 > 0.004 m/s,
the opposite is true.
The existence of co-current liquid and gas flow in the riser
increased Vgs2 (Fig. 7b) and bubble size decreased linearly. In this
part of the photobioreactor, Vgs1 influenced bubble size, but its
effect was less than that of Vgs2 . This increased gas retention time
and the agglomeration of bubbles. Table 4 demonstrates that there
is good agreement between the optimized RSM parameters, which
is verified by the experimental results.
The results of present study can be compared with the results
of other investigators. Table 5 shows the different bioreactor con-
figurations for comparison. 7 bioreactors (2 external loop airlifts,
kL a S tm tc P/V
(1/h) (s) (s) (W/m3 )
23.1329 37.1967 69.3017 72.0943
24.0097 36.1084 68.2556 71.1840
 A. Pirouzi et al. / Biochemical Engineering Journal 87 (2014) 25–32 31

Table 5
Configurations of bioreactors studied by other investigators for comparison with present study.

Reactor type Case of comparison Remark Ref.

External-loop airlift kL a vs. P/V With annular light chambers in which a swirling motion was induced. [39]
External-loop airlift kL a vs. P/V, P/V vs. Vgs Vertical tubular loop bioreactor with forced-liquid flow [40]
Internal-loop draft-tube tm vs. P/V, P/V vs. Vgs , tm vs. Vgs With forced circulation [41]
Flat-panel airlift tm vs. P/V, P/V vs. Vgs , kL a vs. tm – [31]
Bubble column tm vs. Vgs – [29]
Internal-loop draft-tube tm vs. Vgs – [29]
Split-cylinder vessel tm vs. Vgs – [29]

250 2 internal loop draft tubes, 1 flat panel airlift, 1 bubble column,
Reyna-Velarde et. al. 2010 and 1 split cylinder) are presented in Table 5 [29,31,39–41]. The
corresponding results are given in Figs. 8–11.
Fadavi et. al. 2007: VL=0.013 m/s
Fig. 8 compares kL a and power consumption of the photobiore-
200 Yazdian et. al. 2010: VL=0.12 m/s actor with the results of the 3 external loop airlift bioreactors. At
This work similar levels of power consumption, the geometry of the new
photobioreactor shows better mass transfer than the other config-
urations. This is a result of better mixing (decreased mixing time)
P/V (W/m3)

150 as a result of countercurrent flow in the downcomer when power

consumption slightly increased. Fig. 8 shows the rapid decrease in
mixing time (good mixing) of the new photobioreactor compared
to that of the other bioreactors.
100 Power consumption in this study was influenced by the two
gas feeders from the bottom of the riser and downcomer (Fig. 9).
Extrema shown in Fig. 9 resulted from the influence of Vgs1 , which
was greater than that of Vgs2 . In all cases, the power consumption of
50 the new photobioreactor was less than that for the other reactors.
It appears that this is a result of the specific shape of the new pho-
tobioreactor (triangle geometry) and the lack of pumps. This shape
increased kL a over that of previous reactors at similar mixing times.
0 Fig. 10 compares the kL a of the present study with those of Reyna-
Velarde [31]. In all mixings, the kL a was significantly greater than
0.005 0.015 0.025 0.035 0.045
that from Reyna-Velarde. This was true even for low mixing (high
Vgs (m/s) mixing time) and less power consumption.

Fig. 9. P/V vs. Vgs in present and previous studies.

160
30

140

25
120

20 100
kLa (1/h)

tm (s)

80
15

60
10
40 Fadavi et.al. 2007
Sanchez Miron et. al. 2004 (Bubble column)
5 Reyna-Velarde et. al. 2010
Sanchez Miron et. al. 2004 (Interna loop)
20
This work Sanchez Miron et.al. 2004 (Split cylinder vesel)
This work
0 0
0 50 100 150 0.005 0.015 0.025 0.035 0.045 0.055
tm (s) Vgs (m/s)

Fig. 10. kL a vs. tm from the present study and Reyna-Velard et al. [31].
Fig. 11. tm vs. Vgs in the present and previous studies.
32 A. Pirouzi et al. / Biochemical Engineering Journal 87 (2014) 25–32
For all the bioreactors, when gas feeding increased, mixing
improved (mixing time decreased). Fig. 11 demonstrates that mix-
ing time is very sensitive to gas feeding.
4. Conclusion
The countercurrent flow in the downcomer of the new photo-
bioreactor provided adequate contact time between gas and liquid,
and increased the mixing and mass transfer rates. This was not
true for existing external loop photobioreactors. Downcomers in
conventional bioreactors achieved weak mass transfer rates and
gas hold-up because of limited bubble resident time. In the present
study, the high countercurrent flow and vortices in the downcomer
increased the performance of mixing process and decreased the
mixing time, eliminating this limitation.
The geometry of the new reactor significantly decreased vol-
umetric power consumption, which decreased input energy over
that of other configurations. Overall energy balance is the main
challenge to indirect production of bioenergy by photobioreactors;
the new geometry can be a good selection for a full-scale system.
The countercurrent flow in the hypotenuse increased contact time
for liquid and gas mass transfer and provided adequate duration
for the microalgae to absorb light. With sufficient access to light
and nutrients, the growth of the microalgae also increased. When
microalgae are cultivated in the new photobioreactor, maximum
biomass productivity can be achieved by adjusting Vgs1 and Vgs2 .
References
[1] E. Suali, R. Sarbatly, Conversion of microalgae to biofuel, Renew. Sustain. Energy
Rev. 16 (2012) 4316–4342.
[2] J.C. Merchuk, Y. Rosenblat, I. Berzin, Fluid flow and mass transfer in a counter-
current gas–liquid inclined tubes photo-bioreactor, Chem. Eng. Sci. 62 (2007)
7414–7425.
[3] G. Vunjak-Novakovic, Y. Kim, X. Wu, I. Berzin, J.C. Merchuk, Air-lift bioreactors
for algal growth on flue gas: mathematical modeling and pilot-plant studies,
Ind. Eng. Chem. Res. 44 (2005) 6154–6163.
[4] J. Merchuk, Shear effects on suspended cells, in: Bioreactor Systems and Effects,
Springer, Berlin/Heidelberg, 1991, pp. 65–95.
[5] A. Bahadar, M. Bilal Khan, Progress in energy from microalgae: a review, Renew.
Sustain. Energy Rev. 27 (2013) 128–148.
[6] R. Balgobin, Bubble-free oxygen and carbon dioxide mass transfer in bioreac-
tors using microporous membranes, in: Chemical and Biochemical Engineering,
University of Western Ontario, 2012, pp. 438.
[7] M.K. Lam, K.T. Lee, A.R. Mohamed, Current status and challenges on
microalgae-based carbon capture, Int. J. Greenhouse Gas Control 10 (2012)
456–469.
[8] M.R. Tredici, Photobiology of microalgae mass cultures: understanding the
tools for the next green revolution, Biofuels 1 (2009) 143–162.
[9] Y. Chisti, Biodiesel from microalgae, Biotechnol. Adv. 25 (2007) 294–306.
[10] H. Weiwei, Characterization of hydrodynamic forces and interfacial
phenomena in cell culture processes, in: Chemical Engineering, Ohio
State University, 2007, pp. 176.
[11] M.E. de Swaaf, T.C. de Rijk, G. Eggink, L. Sijtsma, Optimisation of docosahe-
xaenoic acid production in batch cultivations by Crypthecodinium cohnii, in:
R. Osinga, J. Tramper, J.G. Burgess, R.H. Wijffels (Eds.), Progress in Industrial
Microbiology, Elsevier, 1999, pp. 185–192.
[12] M. Scarsella, G. Torzillo, A. Cicci, G. Belotti, P. De Filippis, M. Bravi, Mechanical
stress tolerance of two microalgae, Process Biochem. 47 (2012) 1603–1611.
[13] N.M. Langley, S.T.L. Harrison, R.P. van Hille, A critical evaluation of CO2 supple-
mentation to algal systems by direct injection, Biochem. Eng. J. 68 (2012) 70–75.
[14] C.U. Ugwu, H. Aoyagi, H. Uchiyama, Photobioreactors for mass cultivation of
algae, Bioresour. Technol. 99 (2008) 4021–4028.
[15] T.M. Sobczuk, F.G. Camacho, E.M. Grima, Y. Chisti, Effects of agitation on the
microalgae Phaeodactylum tricornutum and Porphyridium cruentum, Bioprocess
Biosyst. Eng. 28 (2006) 243–250.
[16] S. Powtongsook, K. Kaewpintong, A. Shotipruk, P. Pavasant, Effect of superficial
gas velocity on growth of the green microalga Haematococcus pluvialis in airlift
photobioreactor, Stud. Surf. Sci. Catal. 159 (2006) 481–484.
[17] S. Divya, Biokinetic behaviour of Chlorella vulgaris in a continuous stirred biore-
actor and a novel circulating loop photobioreactor, in: Chemical Engineering,
University of Saskatchewan, 2009.
[18] H. Hadiyanto, S. Elmore, T. Van Gerven, A. Stankiewicz, Hydrodynamic eval-
uations in high rate algae pond (HRAP) design, Chem. Eng. J. 217 (2013)
231–239.
[19] R. Rosello Sastre, Z. Csögör, I. Perner-Nochta, P. Fleck-Schneider, C. Posten,
Scale-down of microalgae cultivations in tubular photo-bioreactors—a concep-
tual approach, J. Biotechnol. 132 (2007) 127–133.
[20] M.J. Barbosa, Hadiyanto, R.H. Wijffels, Overcoming shear stress of microalgae
cultures in sparged photobioreactors, Biotechnol. Bioeng. 85 (2004) 78–85.
[21] X. Wu, J.C. Merchuk, Simulation of algae growth in a bench-scale bubble column
reactor, Biotechnol. Bioeng. 80 (2002) 156–168.
[22] B. Wang, C.Q. Lan, M. Horsman, Closed photobioreactors for production of
microalgal biomasses, Biotechnol. Adv. 30 (2012) 904–912.
[23] L. Xu, P.J. Weathers, X.-R. Xiong, C.-Z. Liu, Microalgal bioreactors: challenges
and opportunities, Eng. Life Sci. 9 (2009) 178–189.
[24] F.G. García, E.M. Grima, A.S. Mirón, V.G. Pascual, Y. Chisti, Carboxymethyl cellu-
lose protects algal cells against hydrodynamic stress, Enzyme Microb. Technol.
29 (2001) 602–610.
[25] Y. Chisti, M.C. Flickinger, Shear sensitivity, in: Encyclopedia of Industrial
Biotechnology, John Wiley & Sons, Inc., 2009.
[26] C.R. Thomas, Problems of shear in biotechnology, in: M.A. Winkler (Ed.), Chem-
ical Engineering Problems in Biotechnology, Elsevier, New York, 1990, pp.
23–93.
[27] H. Nikakhtari, G.A. Hill, Enhanced oxygen mass transfer in an external loop
airlift bioreactor using a packed bed, Ind. Eng. Chem. Res. 44 (2005) 1067–1072.
[28] M.J. Barbosa, M. Janssen, N. Ham, J. Tramper, Microalgae cultivation in air-
lift reactors: modeling biomass yield and growth rate as a function of mixing
frequency, Biotechnol. Bioeng. 82 (2003) 170–179.
[29] A. Sánchez Mirón, M.C. Cerón García, F. García Camacho, E. Molina Grima, Y.
Chisti, Mixing in bubble column and airlift reactors, Chem. Eng. Res. Des. 82
(2004) 1367–1374.
[30] J.L. Mendoza, M.R. Granados, I. de Godos, F.G. Acién, E. Molina, C. Banks,
S. Heaven, Fluid-dynamic characterization of real-scale raceway reactors for
microalgae production, Biomass Bioenergy 54 (2013) 267–275.
[31] R. Reyna-Velarde, E. Cristiani-Urbina, D.J. Hernández-Melchor, F. Thalasso, R.O.
Cañizares-Villanueva, Hydrodynamic and mass transfer characterization of a
flat-panel airlift photobioreactor with high light path, Chem. Eng. Process. Pro-
cess Intensif. 49 (2010) 97–103.
[32] Y. Chisti, Airlift Bioreactors, Elsevier Applied Science, 1989.
[33] R. Babcock, J. Malda, J. Radway, Hydrodynamics and mass transfer in a tubular
airlift photobioreactor, J. Appl. Phycol. 14 (2002) 169–184.
[34] A. Contreras, F. García, E. Molina, J.C. Merchuk, Interaction between CO2 -mass
transfer, light availability, and hydrodynamic stress in the growth of Phaeo-
dactylum tricornutum in a concentric tube airlift photobioreactor, Biotechnol.
Bioeng. 60 (1998) 317–325.
[35] F. Yazdian, S.A. Shojaosadati, M. Nosrati, M.P. Hajiabbas, E. Vasheghani-
Farahani, Investigation of gas properties, design, and operational parameters
on hydrodynamic characteristics, mass transfer, and biomass production from
natural gas in an external airlift loop bioreactor, Chem. Eng. Sci. 64 (2009)
2455–2465.
[36] F. Yazdian, S.A. Shojaosadati, M. Nosrati, M.R. Mehrnia, E. Vasheghani-Farahani,
Study of geometry and operational conditions on mixing time, gas hold up, mass
transfer, flow regime and biomass production from natural gas in a horizontal
tubular loop bioreactor, Chem. Eng. Sci. 64 (2009) 540–547.
[37] F. Kargi, M. Moo-Young, Transport phenomena in bioprocesses, in: M. Moo-
Young (Ed.), Comprehensive Biotechnology, Pergamon Press, Oxford, 1985, pp.
5–55.
[38] A. Sánchez Mirón, F. García Camacho, A. Contreras Gómez, E.M. Grima, Y. Chisti,
Bubble-column and airlift photobioreactors for algal culture, AIChE J. 46 (2000)
1872–1887.
[39] K. Loubière, E. Olivo, G. Bougaran, J. Pruvost, R. Robert, J. Legrand, A new
photobioreactor for continuous microalgal production in hatcheries based
on external-loop airlift and swirling flow, Biotechnol. Bioeng. 102 (2009)
132–147.
[40] F. Yazdian, M.P. Hajiabbas, S.A. Shojaosadati, M. Nosrati, E. Vasheghani-
Farahani, M.R. Mehrnia, Study of hydrodynamics, mass transfer, energy
consumption, and biomass production from natural gas in a forced-liquid ver-
tical tubular loop bioreactor, Biochem. Eng. J. 49 (2010) 192–200.
[41] A. Fadavi, Y. Chisti, Gas holdup and mixing characteristics of a novel forced
circulation loop reactor, Chem. Eng. J. 131 (2007) 105–111.