Bioresource Technology 137 (2013) 188–195

Contents lists available at SciVerse ScienceDirect

Bioresource Technology
journal homepage: www.elsevier.com/locate/biortech

Oxygen transfer and evolution in microalgal culture in open raceways

J.L. Mendoza a, M.R. Granados b, I. de Godos c, F.G. Acién b,⇑, E. Molina b, S. Heaven a, C.J. Banks a
a
Faculty of Engineering and the Environment, University of Southampton, Southampton, UK
b
Department of Chemical Engineering, University of Almería, 04120 Almería, Spain
c
Aqualia Gestión Integral del Agua S.A., C/Ulises, 18, 28043 Madrid, Spain

h i g h l i g h t s

 Oxygen mass transfer in raceway reactors mainly occurs place in paddlewheel and sump.
 Mass transfer capacity of raceway reactors is a limiting factor for productivity.
 Productivity increases at high mass transfer outweigh the higher power consumption.

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

Article history: The mass transfer characteristics of all sections of a 100 m2 raceway were evaluated. The efficiency of dif-
Received 25 January 2013 ferent diffusers was determined dynamically and the most effective was used for steady state system
Received in revised form 12 March 2013 characterisation at water depth 0.2 m and velocity 0.22 m s1. Mass transfer coefficients at a gas flow rate
Accepted 14 March 2013
of 6 m3 h1 were 164.50, 63.66, 0.87 and 0.94 h1 for the paddlewheel, sump, straight and curved channel
Available online 26 March 2013
sections, with associated oxygen transfer rates of 106, 172, 27 and 39 g h1. Oxygen supersaturation dur-
ing algal cultivation led to a reduction in biomass productivity, which was more severe with pure CO2
Keywords:
than flue gas. Simulations showed the energy required to increase mass transfer and reduce oxygen con-
Raceway
Microalgae
centrations was more than compensated for by increased biomass and potential energy yields. Oxygen
Mass transfer removal is likely to be a critical criterion, and maintenance of mass transfer by sparging may be necessary
Oxygen accumulation even when CO2 is not required.
Ó 2013 Elsevier Ltd. All rights reserved.

1. Introduction Raceway reactors have some serious drawbacks, however, that

Microalgae offer the potential for production of high-value
compounds (Pulz and Gross, 2004), wastewater treatment (Muñoz
et al., 2009), CO2 mitigation (Brune et al., 2009) and bioenergy pro-
duction (Chisti, 2007). A variety of different photobioreactor types
can be used for this purpose, from tubular to flat panels and race-
ways (Posten, 2009). At a commercial scale, however, raceway
reactors are the most common and have been used for the large-
scale production of Spirulina and Dunaliella, as well as at smaller
scale for other strains (Moheimani and Borowitzka, 2007; Moreno
et al., 2003; Jiménez et al., 2003; Richmond and Cheng-Wu, 2001;
Radmann et al., 2007). The popularity of raceway reactors is due to
their low construction costs and the low energy requirement for
mixing, which may be on the order of 4 W m3 (Jorquera et al.,
2010; Mendoza et al., 2013). For these reasons raceways have been
proposed as the most feasible system for microalgal energy pro-
duction (Jorquera et al., 2010).
⇑ Corresponding author. Tel./fax: +34 950 015484.
E-mail address: facien@ual.es (F.G. Acién).
may limit their performance: these include the high risk of culture
contamination, low final biomass concentrations incurring high
harvesting costs, the lack of temperature control, and the poor
gas/liquid mass transfer (Posten, 2009; Richmond, 2004; Carvalho
et al., 2006). A typical design for raceway reactors was suggested
more than 40 years ago by Oswald (Oswald and Golueke, 1968),
and engineering aspects of the design were evaluated by Weiss-
man (Weissman et al., 1988). The current interest in energy pro-
duction from micro-algae has led to re-examination of these
cultivation systems in order to improve their performance in terms
of energy and materials inputs (Mendoza et al., 2013; Sompech
et al., 2012; Chiaramonti et al., 2013). One area receiving particular
attention has been the supply of carbon, due to the limitations on
mass transfer and the potential for losses when CO2 is bubbled into
a shallow channel. Devices such as sumps and mixing columns
have been proposed as a means of increasing the gas/liquid contact
time and the efficiency of CO2 absorption (Weissman et al., 1988;
Azov and Shelef, 1982; Weissmann and Goebel, 1987; Park et al.,
2011; Putt et al., 2011). When sumps are used, the provision of a
baffle has also been suggested to allow counter-current injection
of CO2 as a means of increasing the mass transfer (Weissman

0960-8524/$ - see front matter Ó 2013 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.biortech.2013.03.127
 J.L. Mendoza et al. / Bioresource Technology 137 (2013) 188–195
et al., 1988). The use of a baffle may considerably increase the
power requirement for mixing, however, and careful consideration
must be given to its design (Mendoza et al., 2013).
In addition to the transfer of CO2 to the culture, the removal of
photosynthetically generated oxygen is necessary. The adverse ef-
fects of dissolved oxygen are widely reported as its accumulation is
one of major problems in the design and operation of closed pho-
tobioreactors (Camacho Rubio et al., 1999). It is generally assumed
that in raceway reactors the oxygen produced will be released to
the atmosphere along the channels, with no major accumulation
in dissolved oxygen: this is considered to be one of the advantages
of this type of reactor. In practice, however, several studies have re-
ported dissolved oxygen concentrations in raceways as high as
500% saturation, causing inhibition of photosynthesis and growth,
and eventually leading to culture death (Marquez et al., 1995;
Singh et al., 1995; Vonshak et al., 1997). It is therefore necessary
to provide mechanisms for the removal of dissolved oxygen, and
to consider these when designing raceway reactors and establish-
ing operational protocols. To develop an oxygen control strategy, a
knowledge of the mass transfer capacity of each section of the
raceway reactor is needed (Camacho Rubio et al., 1999).
The present work used a raceway reactor with a 100 m channel
length and a gas transfer sump through which gas injection could
be controlled by varying the flow rate and diffuser type. The mass
transfer capacity for oxygen was determined for the whole reactor
at different liquid velocities, and also for specific sections of the
reactor, which included straight and curved channel sections, the
paddlewheel section, and sump section. The data were used to de-
velop a model describing the overall behaviour of the system, and
the equations obtained were used to calculate mass transfer
requirements and associated power consumption for the design.
Oxygen accumulation as a result of algal cultivation was deter-
mined in continuous growth trials using an inoculum of Scenedes-
mus spp. in two identical raceways, one of which was supplied
with pure CO2 and the other with flue gas.
2. Methods
2.1. Raceway reactors
Two identical raceway reactors were used, located at the Esta-
ción Experimental Las Palmerillas of Fundación Cajamar in El Ejido,
Spain. Each raceway consisted of two channels 50 m long and 1 m
wide, connected by 180° bends at each end, with a 0.59 m3 sump
(0.65 m length  0.90 m width  1 m depth) located part of the
way down one channel (Fig. 1). The entire reactor, including the
sump, was made of white 3 mm thick fibreglass. The liquid in the
raceway was circulated by a marine plywood paddlewheel 1.2 m
Fig. 1. Schematic of the raceway reactor showing dimensions and position of DO probes used during experiments. Raceway reactor consisted of two channels 50 m long and
1 m wide, connected by 180° bends, with a 0.59 m3 sump (0.65 m length  0.90 m width  1 m depth) located part of the way down one channel. Numbers indicates the
position where the probes were located.
189
in diameter, which was driven by an electric motor (Ebarba, Barce-
lona, Spain) with gear reduction and speed control using a fre-
quency inverter (Ibérica, S.A., Barcelona, Spain). The fluid
dynamic characteristics of the reactor had previously been deter-
mined to define the optimal operating conditions (Mendoza
et al., 2013). The raceway was equipped with four pH-T probes
(5083T, Crison, Barcelona, Spain) and four dissolved oxygen (DO)
probes (5120, Crison, Barcelona, Spain) connected to transmitters
(MM44, Crison, Barcelona, Spain) and data acquisition software
(Daqfactory, Azeotech, Arizona, USA). The gas flow rate entering
the reactor was measured by a mass flow meter (PFM 725S-F01-
F, SMC, Tokyo, Japan).
2.2. Mass-transfer characterisation
2.2.1. Sump and diffuser
Mass transfer was determined using a dynamic method in
which the sump was filled with water, and nitrogen gas was bub-
bled through a diffuser at the bottom of the sump until the dis-
solved oxygen concentration reached zero. At this point the gas
flow was switched to air which was bubbled until the water be-
came saturated with dissolved oxygen in equilibrium with the
air. Three different types of diffuser commonly used in wastewater
treatment were tested at different pressures. These were: type 1 –
a membrane tubing diffuser giving a small bubble size but with a
high pressure drop; type 2 – a plate diffuser giving a small bubble
size with a low pressure drop; and type 3 – a porous tubing dif-
fuser giving medium-size bubbles with a low pressure drop. All
three diffusers had the same surface area of 0.18 m2. Dissolved
oxygen concentrations were measured at the liquid surface, and
it was assumed that the sump was well mixed and that readings
taken at this point were representative of the sump contents. The
variation in dissolved oxygen with time was assumed to be a func-
tion of the mass transfer coefficient (Klal) and the driving force
([O2⁄]  [O2]) (Eq. (1)), where Klal is the mass transfer coefficient
(s1), [O2] is the DO concentration (mg l1), [O2⁄] is the equilibrium
DO concentration (mg l1) and t is time (s).
d½O2    
¼ K l al  O2  ½O2  ð1Þ
dt
By integrating this equation between time zero and time t the
mass transfer coefficient can be obtained (Eq. (2)); in particular,
by defining the fractional approach to equilibrium (E) (Eq. (3))
the mass transfer coefficient can be obtained as a function of E
and time (Eq. (4)).
!
 
O2  ½O2 t¼0
Ln   ¼ K l al  t ð2Þ
O2  ½O2 
190 J.L. Mendoza et al. / Bioresource Technology 137 (2013) 188–195
!
½O   ½O2 t¼0
E¼  2 
O2  ½O2 t¼0
Lnð1  EÞ
K l al ¼
t til a biomass concentration of 0.7 g total solids (TS) l1 was
The power consumption for gas transfer was calculated as the
work required to compress air from atmospheric pressure to over-
pressure under isothermal conditions at 20 °C, using Eq. (5) where
P is the power consumption (J s1), n is the molar flow (mol s1), R
is the ideal gas constant (J K1 mol1), T is the temperature (K), p1
is the atmospheric pressure (atm) and p2 is the pressure in the dif-
fuser (atm).
 
p
P ¼ nRT Ln 2
p1
This was determined by passing different gas flows through the
diffusers and measuring the pressure in the diffuser using a stan-
dard glycerine manometer.
2.2.2. Raceway
This used a steady state method in which the reactor was filled
with water to the test depth of 20 cm, and circulation at different
velocities was achieved by adjusting the rotational speed of the
paddlewheel. The water flow rate under these conditions was
determined using pulsed tracer measurements (Mendoza et al.,
2013). The tests were carried out by bubbling nitrogen into the
sump at a gas flow rate of 6 m3 h1, using the diffuser identified
as having the best performance in terms of bubble size and pres-
sure drop. For the purposes of the analysis the reactor was consid-
ered to have four zones: the paddlewheel, the sump, the first
channel, and the second channel including the two bends. Dis-
solved oxygen measurements were taken at the beginning and
end of each zone from DO probes located at the inlet to the paddle-
wheel, inlet of the sump, outlet of the sump, and end of the first
channel (Fig. 1). From these measurements the mass flow of oxy-
gen transferred in each section was calculated from the difference
in oxygen concentration at the inlet and outlet to the section and
the flow of water in the channel. This was expressed in terms of
the oxygen transfer coefficient, driving force and volume in each
zone (Eq. (6)), where mO2, zone is mass flow of O2 into the zone
(mg s1), Q is the water flow in the channel (l s1), [O2]in(out), zone
is the DO concentration entering (leaving) the zone (mg l1), Kl
alzone is the O2 transfer coefficient for the zone (s1),
([O2]  [O⁄2]) zone is the average driving force in the zone (mg l1),
and Vzone is the volume of the zone (l). The mass of oxygen trans-
ferred in each section was confirmed by the overall steady state
mass balance (Eq. (7)).
mO2; zone ¼ Q ð½O2 in  ½O2 out Þzone ¼ K l alzone ð½O2   ½O2 Þzone V zone
mO2; sump þ mO2; paddlewheel þ mO2; channel 1 þ mO2; bendsþ channel 2
¼0
2.3. Algal growth conditions, inoculum and raceway operation
2.3.1. Algal culture medium
This was prepared in fresh water using commercial fertilisers
and supplements to give the following additional ionic concentra-
tions (mmol l1) to those naturally present in the fresh water
source: NO þ  +
3 (9.49); NH4 (0.59); H2 PO4 (1.00); K (1.60); Ca
2+
2+ ¼  +
(5.00); Mg (3.25); SO4 (1.20); HCO3 (2.80); Na (8.20); Cl
(11.10) Fe3+ (2.00); Mn2+ (0.84); Zn2+ (0.56); B3+ (0.49) and Cu2+
(0.08).
2.3.2. Inoculum
ð3Þ The raceway was inoculated with 3 m3 of Scenedesmus sp.
grown in the culture medium under aseptic conditions in a tubular
photobioreactor at a dilution rate of 0.3 day1, temperature 25–
30 °C and pH 8.0. The raceway was operated as a batch reactor un-
ð4Þ
reached. At this point half of the volume of the first raceway was
transferred into the second raceway reactor as inoculum. Culture
medium was then added to both reactors to reach the working vol-
ume and they were again operated in batch mode until a biomass
concentration of 0.7 g TS l1 was achieved.
2.3.3. Continuous operation
Reactor operation was switched from batch to semi-continuous
ð5Þ by additions of culture medium to give a dilution rate of 0.25 d1.
The raceway pH was controlled between set points of 7.8 and 8.0
by bubbling pure CO2 at 0.6 m3 h1 in the first reactor, and flue
gas from a diesel boiler containing 10% CO2 (and also 6% O2,
150 ppm NOx, 0% SOx) at 6.0 m3 h1 in the second raceway. The li-
quid velocity in the reactors was 0.22 m s1 and the operational
depth was 0.2 m; temperature was dependent on ambient air tem-
perature and incident solar radiation.
2.4. Productivity versus consumption simulation
To predict energy productivity in different conditions a set of
simulations were carried out using empirical data and the follow-
ing assumptions. The culture was assumed to have a biomass con-
centration Cb of 1 g TS l1 and a biomass extinction coefficient Ka of
0.18 m2 g1 biomass. Hourly values for solar radiation I0 (lE m2
s1) were available for the raceway location, and from these data
the average irradiance inside the culture Iav (lE m2 s1) was cal-
culated for a constant culture depth d of 0.2 m using Eq. (8).
I0
Iav ¼ ð1  expðK a d C b ÞÞ ð8Þ
K a d Cb
The rate of photosynthetic oxygen production RO2phot is a func-
tion of Iav and specific maximum photosynthesis rate RO2max
(mg l1 h1), and was assumed to fit a hyperbolic model with
n = 2 and Ik = 60 lE m2 s1 (Camacho Rubio et al., 1999) (Eq. (9)).
RO2 max Inav
RO2phot ¼ ð9Þ
Ink þ Inav
RO2max was calculated from Eq. (10), using the experimentally-
obtained mass transfer coefficient values and assuming a maxi-
mum acceptable DO concentration in the raceway of 250%
saturation.
d½O2 
¼ RO2phot  K l al ð½O2   ½O2 Þ ð10Þ
ð6Þ dt
The overall biomass productivity in the raceway was then cal-
culated based on the daily total oxygen production, using a conver-
sion factor of 1.33 g O2 g1 biomass. The energy potential of the
ð7Þ
biomass was calculated based on a typical calorific value of
20 kJ g1 biomass. The biomass potential energy yields were then
compared with the additional energy consumption required to
achieve this mass transfer capacity in the sump using the empiri-
cally-derived Eq. (12) below.
3. Results
3.1. Sump characterisation
Fig. 2 shows the differential pressure (overpressure) required to
obtain different gas flow rates for each type of diffuser, including
 J.L. Mendoza et al. / Bioresource Technology 137 (2013) 188–195
Fig. 2. Variation of overpressure, compression work and mass transfer coefficient
with the gas flow rate and diffuser type.
the 0.10 atm of hydrostatic pressure at the bottom of the sump.
The type 1 diffuser required an overpressure of 3.0 atm to reach
a gas flow rate of 3.6 m3 h1, whereas the type 2 and 3 diffusers at-
tained the same gas flow rate with only 0.7 and 1.0 atm of over-
pressure, respectively. A power consumption of up to 150 W was
required when using membrane tube diffuser, which decreased
to 70 and 56 W when using the porous tube and plate diffuser,
respectively. Irrespective of the type of diffuser the gas flow rate
showed a linear relationship to the overpressure with slopes of
0.84, 0.28 and 0.21 m3 h1 atm1 for types 1, 2 and 3, respectively.
The results showed that the small bubble size diffusers gave a
higher gas transfer coefficient than the one with a larger size bub-
ble at the same gas flow rate, clearly demonstrating that this coef-
ficient is only a function of bubble size and gas flow rate and is
independent of pressure drop. Mass transfer coefficients of 32.4
and 29.2 h1 were obtained for type 2 and 3 diffusers, compared
to 10.8 h1 for the type 1 diffuser when operated at the highest
gas flow rate of 3.6 m3 h1. The plate diffuser (type 2) was there-
fore used in the steady state characterisation of the raceway. The
191
mass transfer coefficient obtained was compared to the calculated
power consumption (Eq. (5)) for each of the three diffusers
used and gave the relationships shown in Eqs. (11)–(13), where
P/V = power consumption per unit volume (W m3) taking into
account the 0.1 atm sump pressure.
0:4635
K l al ¼ 0:00063ðP=VÞ for small bubble high pressure diffuserðtype 1Þ
ð11Þ
0:5707
K l al ¼ 0:00057ðP=VÞ for small bubble low pressure diffuserðtype 2Þ
ð12Þ
K l al ¼ 0:00016ðP=VÞ0:6209 for medium bubble low pressure diffuser ðtype3Þ
ð13Þ
3.1.1. Reactor characterisation
Results from the steady state reactor characterisation are
shown in Table 1. The total oxygen desorption in the sump was
171.9 g h1, and oxygen absorption occurred mainly in the pad-
dlewheel section at 105.8 g h1. Oxygen transfer into the channel
1 straight section was 27.2 g h1, and in the channel 2 straight sec-
tion + bends was 38.9 g h1. This indicates that the oxygen transfer
capability of the two bends was equivalent to 0.43 of the capability
of the straight channel 1 despite being only 0.06 of the length. The
sections of the reactor where gas transfer is most effective are thus
the sump and the paddlewheel, with both the straight section and
bends in the channel contributing 19% and 4% respectively. The
mass transfer coefficient determined for the sump using the steady
state method was 63.7 h1 which is higher than the 32.4 h1 mea-
sured previously using the dynamic method. This difference is due
to the lower gas flow rate of 3.6 m3 h1 used in the dynamic meth-
od; the value obtained using the steady state method at a flow rate
of 6 m3 h1 also fitted the empirical correlation obtained for this
diffuser (Eq. (12)). The highest mass transfer coefficient
(164.5 h1) was determined for the paddlewheel section and the
lowest for the channel (0.9 h1).
To determine the influence of liquid velocity on the mass trans-
fer capacity of the reactor, experiments were performed using the
steady state method at different velocities. Results showed the
mass transfer coefficients in all sections of the reactor increased
with increasing liquid velocity (Fig. 3). The percentage change
was similar in each section: in the paddlewheel section it increased
from 132 to 189 h1 (30%) and in the sump from 63 to 95 h1 (34%)
as the liquid velocity increased from 0.17 to 0.39 m s1 (Fig. 3a).
Values in the channels and bends were much lower (Fig. 3b) and
ranged from 0.7 to 2.1 h1. Overall the mass transfer coefficient
for the entire reactor increased from 3.7 to 6.1 h1 with the in-
crease in liquid velocity. In an experiment where no gas was bub-
bled into the sump, the mass transfer coefficient of the reactor was
reduced to 1.8 h1.
3.1.2. Microalgal growth experiments
The experiments aimed to determine if the mass transfer capac-
ity of the reactor was sufficient to allow unrestricted growth and
productivity of the microalgal culture. The results for the reactor
Table 1
Steady state values from experimental characterisation of the raceway, operated at a
liquid depth of 0.2 m and velocity of 0.22 m s1 with a nitrogen flow rate of 6 m3 h1.
Paddlewheel Sump Channel 1 Bends + channel 2
Volume, m3 0.20 0.59 8.00 11.26
DO, %Sat. 63.98 50.42 52.56 55.63
1
mO2, g h 105.8 171.9 27.2 38.9
Driving force, g m3 3.22 4.58 3.88 3.67
Kla, h1 164.5 63.7 0.9 0.9
192 J.L. Mendoza et al. / Bioresource Technology 137 (2013) 188–195
Fig. 3. Variation of mass transfer coefficient in each section of the raceway with
liquid velocity.
supplied with pure CO2 are shown in Fig. 4and those for the reactor
supplied with flue gas at 10% CO2 concentration in Fig. 5. Apart
from the mode in which CO2 was supplied, the reactors were ex-
posed to the same conditions of solar radiation, temperature and
dilution rate. Dissolved oxygen concentrations in the reactor with
pure CO2 reached 25–30 mg l1 around noon, but were lower in
the reactor using flue gas at 20–25 mg l1.
Using the mass transfer coefficients it was possible to calculate
the oxygen production by photosynthesis based on the rates of
oxygen desorption and accumulation (Eq. (10)). When pure CO2
was used the results showed the desorption capacity of the reactor
was low, with a maximum value of 37 mg l1 h1 despite the high
dissolved oxygen concentrations: this was due to the low mass
transfer coefficient (1.83 h1). The low mass transfer can also be
attributed to the low gas flow rate of 0.6 m3 h1 in the sump when
using pure CO2. When flue gas was used the mass transfer coeffi-
cient increased to 4.0 h1 because of the greater volume of gas in-
jected, giving a desorption capacity of 70 mg l1 h1. If the
photosynthetic oxygen productivities based on oxygen accumula-
tion and desorption are considered (Eq. (10)), a much higher value
of 65 mg l1 h1 is obtained with flue gas (Fig. 5b) compared to the
35 mg l1 h1 when pure CO2 was used (Fig. 4b). These results sup-
port the view that an accumulation of dissolved oxygen has a neg-
ative effect on photosynthesis. To verify this, the overall yield of
the cultures grown in the two reactors was analysed over 5 days
and daily values are shown in Table 2. The average biomass yield
of 0.33 g TS l1 day1 in the reactor using flue gas was higher than
the 0.24 g TS l1 day1 using pure CO2. Using flue gas the average
oxygen-to-biomass ratio was 1.46 g O2 g1 biomass while using
pure CO2 the ratio was 0.99 g O2 g1 biomass. The photosynthetic
yield was 61.0 E mol1 O2 when using pure CO2 and 31.1 E mol1 -
O2 when flue gas was used.
3.1.3. Productivity versus consumption simulation
The power consumption required to increase the mass transfer
coefficient in the sump from 0 to 150 h1 increased exponentially,
with a power requirement of 1800 W m3 in the sump to achieve
the maximum value. As the volume of the sump is small in com-
parison with the entire raceway, however, the impact on the over-
all power consumption is relatively small. Thus, the overall power
consumption increases from 4 W m3 when no gas is supplied into
the sump (only liquid circulation), to 13 W m3 when gas is sup-
plied into the sump at maximum flow rate of 200 m3 h1 to achieve
the maximum mass transfer coefficient into the sump of 150 h1.
The results of the simulation (Fig. 6) indicated that the increased
mass transfer should allow an increase in biomass productivity
from 0.17 g l1 day1 without gas supply into the sump, up to
0.50 g l1 day1 with a mass transfer coefficient of 150 h1. On
the basis of an energy value of 20 kJ g1 biomass, there is a sub-
stantial increase in potential net energy production from 3400 to
9700 kJ m3 day1 (Fig. 6b).
4. Discussion
High dissolved oxygen concentrations have previously been
found in raceway reactors of the scale used in the current experi-
ments. Values from 25 to 40 mg DO l1 were reported in 100 m2
raceway reactors used for the production of Chlorella sp. (Weiss-
man et al., 1988). In 450 m2 raceway reactors producing Spirulina
platensis the dissolved oxygen concentration ranged from
10 mg l1 in winter (115% of O2 saturation) to 30 mg l1 in summer
(375% saturation); a decrease in biomass concentration was found
when the dissolved oxygen concentration was higher than
25 mg l1 (Jiménez et al., 2003). Dissolved oxygen concentrations
as high as 500% saturation have also been reported in large ponds
with limited mixing (Marquez et al., 1995; Singh et al., 1995; Von-
shak, 1997). In the current experiments the raceway to which pure
CO2 was added showed reduced biomass productivity and higher
dissolved oxygen concentrations compared to the raceway receiv-
ing flue gas. The lower productivity was probably a result of pho-
torespiration, which has been well documented (Eckardt, 2005).
USEPA (1985) quotes values for the oxygen yield from algal bio-
mass production from 1.24 to 1.8 g O2 g1 biomass from various
studies. In the current work the average value for the flue gas reac-
tor was 1.46 g O2 g1 biomass which was considerably higher than
the 0.99 g O2 g1 biomass for pure CO2. Both values suggest that
some of the CO2 is being diverted from assimilative growth, and
the photorespiration mechanism could account for this difference.
Similarly in ideal conditions the theoretical photosynthetic conver-
sion efficiency can be as low as 10 E mol1 O2: in the current work
the efficiency of the culture using pure CO2 was only 16% of this
theoretical value, compared to 32% for the flue gas.
The results reported here showed that the mass transfer coeffi-
cient was high where the mixing between gas and liquid was effec-
tive, and this occurred mainly in the sump and paddlewheel
sections, with values of up to 90 and 160 h1. This compares to
0.7 h1 in the channel where the mixing between liquid and gas
phases is poor: this result agrees well with (Weissman et al.,
1988).who found a value of 0.5 h1 for CO2 released from the sur-
face of a 100 m2 raceway pond. In general, where gas is bubbled
through the water the overall value of the mass transfer coefficient
Klal is higher than for diffusion alone, although a wide range of Klal
values between 0.4 and 350 h1 has been reported for aerated sys-
tems of different designs (Carvalho et al., 2006). In tubular photo-
bioreactors a typical value for the mass transfer coefficient in the
closed loop is 1.1 h1, but can be as high as 90 h1 in the degassing
section where air is bubbled (Camacho Rubio et al., 1999). These
values are similar to the values found in this study for the raceway,
 J.L. Mendoza et al. / Bioresource Technology 137 (2013) 188–195 193

Fig. 4. Variation in dissolved oxygen concentration in a continuous culture of Scenedesmus sp. with pH controlled by on-demand injection of pure CO2. Values of
photosynthesis production, desorption and accumulation of oxygen calculated by mass balance.

Fig. 5. Variation in dissolved oxygen concentration in a continuous culture of Scenedesmus sp. with pH controlled by on-demand injection of flue gas with 10% CO2. Values of
photosynthesis production, desorption and accumulation of oxygen calculated by mass balance.
194 J.L. Mendoza et al. / Bioresource Technology 137 (2013) 188–195

Table 2
Influence of solar radiation on cultures grown using pure CO2 and flue gases for pH control. Values are average for daylight period based on experimental values recorded at one-
minute intervals.

Day Solar radiation, lE m2 s1 Photosynthesis rate, Biomass productivity, Oxygen to biomass, g O2 g1 Photosynthetic efficiency,
mg l1 h1 g l1 day1 biomass E mol1
Pure CO2 Flue gas Pure CO2 Flue gas Pure CO2 Flue gas Pure CO2 Flue gas
1 543.92 9.3 16.8 0.26 0.41 0.85 0.98 67.08 37.20
2 548.95 9.4 17.6 0.24 0.37 0.95 1.13 67.57 35.95
3 527.63 10.3 19.7 0.28 0.35 0.87 1.33 58.90 30.84
4 474.21 9.8 20.4 0.26 0.31 0.91 1.57 55.79 26.80
5 382.20 7.9 17.9 0.14 0.19 1.36 2.26 55.48 24.59

maximum dissolved oxygen concentration of 250% saturation to

Fig. 6. Variation of biomass productivity, power consumption (A) and energy
balance (B) with mass transfer into the sump. Values obtained from simulations
performed considering a hyperbolic photosynthesis model (mO2max = 190 mg O2 -
l1 h1, n = 2, Ik = 60 lE m2 s1) for a raceway of 0.2 m depth operated at
0.22 m s1, biomass concentration of 1.0 g TS l1 and biomass extinction coefficient
of 0.18 m2 g1.
taking the loop as equivalent to the channel sections and the
degassing column equivalent to the sump; the major difference lies
in the volume ratios of these coupled sections in each type of reac-
tor, and hence their overall ability to release oxygen. In the race-
way only 4%, of the total liquid volume is aerated at any one
time compared to a minimum volume of 15% in tubular
photobioreactors.
In the current study the mass transfer capacity for removing
oxygen from the raceway appeared to be directly related to volume
of gas injected into the raceway in order to satisfy the carbon
requirements, as all other components in the system were the
same. Even when flue gas was used, however, the volume of gas
bubbled was not sufficient to reduce the oxygen to low concentra-
tions. Using the data on the contribution made to the mass transfer
in each of the sections, the maximum photosynthetic rate and
hence the productivity were modelled, on the basis of allowing a
accumulate. Where mass transfer occurred at the culture surface
only, the maximum photosynthesis rate ranged from 8.4 to
27.6 mg O2 l1 h1, equivalent to biomass productivities of 6–
20 g m2 day1, similar to the values usually found in raceway
reactors. This result is further supported by the work of Boussiba
et al. (1988) who found that under the same culture conditions
the productivity of Isochrysis galbana cultures was 30 g m2 day1
when using a 2.5 m2 raceway reactor, whereas in a 100 m2 raceway
reactor, in which the mass transfer contribution of paddlewheel is
reduced, productivity decreased to 22 g m2 day1.
To improve productivity, measures to enhance mass transfer are
needed, which requires improved mixing between gas and liquid
phases. This could be by increasing the volume of the sump,
improving the performance of the sparger, or increasing the gas
flow to the sump. An alternative is to increase the gas transfer at
the water surface, or to improve the mass transfer capacity of the
paddlewheel section, both of which are likely to have a significant
effect on the raceway fluid dynamics and energy consumption. Ma-
jor improvements therefore appear to be best focussed on the
sump, and as a first measure the choice of a small bubble size dif-
fuser is essential, preferably one with a low pressure drop. The
oxygen desorbed will also depend on the gas flow rate, and the
capacity for oxygen removal may be more rate-determining than
the supply of CO2: the experimental results showed that, even
when using flue gas sparging with 10% CO2 coupled to carbon de-
mand through pH control, the volume of gas used was insufficient
to reduce DO concentrations to <20 mg l1. It may therefore be
necessary to continue to sparge the culture even when there is
no demand for carbon, and this might be done by switching the
gas stream to air rather than flue gas when there is sufficient dis-
solved carbon in the system. Sparging of larger volumes of gas
leads to greater power consumption, which will depend on the
type of diffuser used, sump depth and other characteristics. The
benefits in biomass productivity achieved by increasing the mass
transfer rate appear, however, to give a net gain in energy
production.
5. Conclusions
The dynamic method successfully determined diffuser effi-
ciency. Mass transfer occurred primarily in the paddlewheel and
sump which accounted for 75% of the total: mass transfer coeffi-
cients in all raceway sections increased with liquid velocity. Algal
cultivation led to oxygen supersaturation and reduced productiv-
ity, especially when using pure CO2 rather than flue gas Simula-
tions showed that reducing DO concentrations by improving
mass transfer rates increased the net potential energy yield. Oxy-
gen removal is likely to be a more critical design criterion than car-
bon supply, and it may be necessary to maintain mass transfer by
sparging even when no carbon is required.
 J.L. Mendoza et al. / Bioresource Technology 137 (2013) 188–195
Acknowledgements
This study was supported by a University of Southampton post-
graduate scholarship and by the company Aqualia as part of the EU
FP7 ALL-GAS project ‘Industrial-scale demonstration of microalgae
biofuels’ (ENERGY.2010.3.4-1, n°268208). The practical assistance
of the staff of the Estación Experimental Las Palmerillas from Fun-
dación Cajamar is gratefully acknowledged.
References
Azov, Y., Shelef, G., 1982. Operation of high-rate oxidation ponds: theory and
experiments. Water Res. 16, 1153–1160.
Boussiba, S., Sandbank, E., Shelef, G., Cohen, Z., Vonshak, A., Ben-Amotz, A., Arad, S.,
Richmond, A., 1988. Outdoor cultivation of the marine microalga Isochrysis
galbana in open reactors. Aquaculture 72, 247–253.
Brune, D.E., Lundquist, T.J., Benemann, J.R., 2009. Microalgal biomass for greenhouse
gas reductions: potential for replacement of fossil fuels and animal feeds. J.
Environ. Eng. 135, 1136–1144.
Camacho Rubio, F., Acién Fernández, F., Sánchez Pérez, J.A., García Camacho, F.,
Molina Grima, E., 1999. Prediction of dissolved oxygen and carbon dioxide
concentration profiles in tubular photobioreactors for microalgal culture.
Biotechnol. Bioeng. 62, 71–86.
Carvalho, A.P., Meireles, L.A., Malcata, F.X., 2006. Microalgal reactors: a review of
enclosed system designs and performances. Biotechnol. Prog. 22, 1490–1506.
Chiaramonti, D., Prussi, M., Casini, D., Tredici, M.R., Rodolfi, L., Bassi, N., Zittelli, G.C.,
Bondioli, P., 2013. Review of energy balance in raceway ponds for microalgae
cultivation: re-thinking a traditional system is possible. Appl. Energy 102, 101–
111.
Chisti, Y., 2007. Biodiesel from microalgae. Biotechnol. Adv. 25, 294–306.
Eckardt, N.A., 2005. Photorespiration revisited. Plant Cell 17, 2139–2141.
Jiménez, C., Cossío, B.R., Niell, F.X., 2003. Relationship between physicochemical
variables and productivity in open ponds for the production of Spirulina: a
predictive model of algal yield. Aquaculture 221, 331–345.
Jorquera, O., Kiperstok, A., Sales, E.A., Embiruçu, M., Ghirardi, M.L., 2010.
Comparative energy life-cycle analyses of microalgal biomass production in
open ponds and photobioreactors. Bioresour. Technol. 101, 1406–1413.
Marquez, F.J., Sasaki, K., Nishio, N., Nagai, S., 1995. Inhibitory effect of oxygen
accumulation on the growth of Spirulina platensis. Biotechnol. Lett. 17, 225–228.
195
Mendoza, J.L., Granados, M.R., Godos, I., Acién, F.G., Molina, E., Banks, C., Heaven, S.,
2013. Fluid-dynamic characterization of real-scale raceway reactors for
microalgae production, Biomass Bioenergy.
Moheimani, N.R., Borowitzka, M.A., 2007. Limits to productivity of the alga
Pleurochrysis carterae (haptophyta) grown in outdoor raceway ponds.
Biotechnol. Bioeng. 96, 27–36.
Moreno, J., Vargas, M.A., Rodríghez, H., Rivas, J., Guerrero, M.G., 2003. Outdoor
cultivation of a nitrogen-fixing marine cyanobacterium, Anabaena sp. ATCC
33047. Biomol. Eng. 20, 191–197.
Muñoz, R., Köllner, C., Guieysse, B., 2009. Biofilm photobioreactors for the treatment
of industrial wastewaters. J. Hazard. Mater. 161, 29–34.
Oswald, W.J., Golueke, C.G., 1968. Large scale production of microalgae. In:
Mateless, R.I., Tannenbaum, S.R. (Eds.), Single Cell Protein. MIT Press,
Cambridge, MA, pp. 271–305.
Park, J.B.K., Craggs, R.J., Shilton, A.N., 2011. Wastewater treatment high rate algal
ponds for biofuel production. Bioresour. Technol. 102, 35–42.
Posten, C., 2009. Design principles of photo-bioreactors for cultivation of
microalgae. Eng. Life Sci. 9, 165–177.
Pulz, O., Gross, W., 2004. Valuable products from biotechnology of microalgae. Appl.
Microbiol. Biotechnol. 65, 635–648.
Putt, R., Singh, M., Chinnasamy, S., Das, K.C., 2011. An efficient system for
carbonation of high-rate algae pond water to enhance CO2 mass transfer.
Bioresour. Technol. 102, 3240–3245.
Radmann, E.M., Reinehr, C.O., Costa, J.A.V., 2007. Optimization of the repeated batch
cultivation of microalga Spirulina platensis in open raceway ponds. Aquaculture
265, 118–126.
Richmond, A., 2004. Principles for attaining maximal microalgal productivity in
photobioreactors: An overview. Hydrobiologia 512, 33–37.
Richmond, A., Cheng-Wu, Z., 2001. Optimization of a flat plate glass reactor for mass
production of Nannochloropsis sp. outdoors. J. Biotechnol. 85, 259–269.
Singh, D.P., Singh, N., Verma, K., 1995. Photooxidative damage to the
cyanobacterium Spirulina platensis mediated by singlet oxygen. Curr.
Microbiol. 31, 44–48.
Sompech, K., Chisti, Y., Srinophakun, T., 2012. Design of raceway ponds for
producing microalgae. Biofuels 3, 387–397.
Vonshak, A., 1997. Spirulina: growth, physiology and biochemistry. In: Vonhask, A.
(Ed.), Spirulina platensis (Arthrospira): Physiology, Cell-Biology and
Biotechnology. Taylor and Francis, London, pp. 43–65.
Weissman, J.C., Goebel, R.P., Benemann, J.R., 1988. Photobioreactor design: mixing,
carbon utilization, and oxygen accumulation. Biotechnol. Bioeng. 31, 336–344.
Weissmann, J.C., Goebel, R.P., 1987. Design and analysis of microalgal open pond
systems for the purpose of producing fuels: A subcontract report. SERI/STR-231-
2840, United States.