Algal Research 25 (2017) 381–390

Contents lists available at ScienceDirect

Algal Research
journal homepage: www.elsevier.com/locate/algal

Open thin-layer cascade reactors for saline microalgae production evaluated MARK
in a physically simulated Mediterranean summer climate
A.C. Apela,c, C.E. Pfaffingera,c, N. Basedahla,c, N. Mittwollena,c, J. Göbela,c, J. Sautera,c,
T. Brückb,c, D. Weuster-Botza,c,⁎
a
Institute of Biochemical Engineering, Technical University of Munich, Garching, Germany
b
Professorship for Industrial Biocatalysis, Technical University of Munich, Garching, Germany
c
TUM AlgaeTec Center, Technical University of Munich, Ottobrunn, Germany

A R T I C L E I N F O A B S T R A C T

Keywords: While microalgae hold the promise for conversion of sunlight and CO2 to a wide variety of products, the eco-
Open thin-layer cascade photobioreactor nomics of algae processes are still debatable. We have designed an open thin-layer cascade photobioreactor for
LED-based climate simulation high-cell density cultivation of saline microalgae to advance economic microalgae mass production. Pilot-scale
Nannochloropsis salina reactors with a surface area of up to 8 m2 (cultivation volume 50–140 L) were constructed and evaluated using a
Pilot-scale batch cultivation
dynamic climate simulation technology (light, air temperature and humidity) integrating natural sunlight and
Economic bioprocess metrics
multi-color LED arrays for a highly realistic reproduction of the sunlight spectrum. Batch processes with
Nannochloropsis salina were performed in these reactors in the physically simulated Mediterranean summer
climate of Almería, Spain – an ideal location for outdoor microalgae cultivation. Two reactor variants were
examined: one with a smooth but expensive rigid channel made of polyethylene sheets, and one with a more
uneven but significantly less expensive channel made of pond liner. Maximal intra-day growth rates of 1.9 d−1
were observed at a cell density of 1–3 g L−1. The maximal cell density of 50 g L−1 was obtained within 25 days.
These high growth rates and cell densities markedly exceed literature data. No difference in growth between the
channel variants was observed. This suggests that cost-efficient large-scale thin-layer cascade reactors with in-
expensive pond liner channels are feasible. The high cell density allows a reduction of harvesting cost. Optimal
process conditions were identified by analyzing the batch and daily economic bioprocess metrics: At a cell
density of 17 g L−1, an areal biomass productivity of 25 g m−2 d−1 (volumetric productivity 4 g L−1 d−1) and a
photosynthetic conversion efficiency of 4.6% were observed. The reactor design is discussed in detail to en-
courage further advancement of thin-layer algal cultivation technology.

1. Introduction
Outdoor mass production of microalgae is poised to impact re-
newable bio-production of food, chemicals and energy [1]. Algae cul-
tivation is sustainable as it does not compete with agricultural activities
and can utilize salt- or waste water, thereby circumventing depletion of
valuable fresh water resources [2]. At present, only few high-value
industrial products such as pigments are generated by outdoor micro-
algae processes. For low-value bulk chemicals or biofuels, production
cost are currently too high to provide economic viability [3].
The choice of a cultivation system has a major effect on production
cost. Open cultivation systems are suitable to generate low-value pro-
ducts [4–6]. The most widely used open cultivation system is the ra-
ceway pond. However, with its depth of 15–30 cm, low cell densities of
only 1–1.5 g L−1 are achieved [7–9]. Hence, energy costs for circulation
and harvesting of the algae are high and the cultures are susceptible to
contamination [10,11]. A promising alternative is the thin-layer culti-
vation concept pioneered in Třeboň, Czech Republic [12–14]. This
bioreactor concept provides for a microalgae suspension to flow down a
sloped channel in a layer of < 1 cm thickness. At the end of the
channel, the suspension is pumped up again to the starting point. High
cell densities of 30–50 g L−1 after 2–3 weeks have been reported in

Abbreviations: CDW, cell dry weight; LED, light-emitting diode; PAR, photosynthetically active radiation; PPFD, photosynthetic photon flux density; TUM, Technical University of
Munich
⁎
Corresponding author at: Institute of Biochemical Engineering, Boltzmannstr. 15, 85748 Garching, Germany.
E-mail addresses: a.apel@lrz.tum.de (A.C. Apel), c.pfaffinger@lrz.tum.de (C.E. Pfaffinger), natascha.basedahl@gmail.com (N. Basedahl),
natascha.mittwollen@hotmail.de (N. Mittwollen), johannagoebel.de@gmail.com (J. Göbel), tschuja@gmail.com (J. Sauter), brueck@tum.de (T. Brück),
d.weuster-botz@lrz.tum.de (D. Weuster-Botz).

http://dx.doi.org/10.1016/j.algal.2017.06.004
Received 3 March 2017; Received in revised form 9 May 2017; Accepted 8 June 2017
2211-9264/ © 2017 Elsevier B.V. All rights reserved.
A.C. Apel et al. Algal Research 25 (2017) 381–390

outdoor cultivation of Chlorella and Scenedesmus freshwater strains in

the Czech Republic and Greece [15–17]. Moreover, the microalgae
production cost in thin-layer systems were estimated to be only 20% of
the cost in raceway ponds [11].
Unfortunately, technical information on thin-layer cascade reactors
is scarce [11,12,18] and a detailed functional evaluation of all reactor
parts is not available, hampering the advancement of thin-layer culti-
vation. Furthermore, a comparison of different reactor setups is difficult
because precise intra-day growth rates have not been determined yet.
Moreover, no thin-layer cascades for saltwater have been developed
yet, although saline cultivation is desirable because of reduced fresh-
water use and contamination risk as well as improved CO2 containment
in the aqueous phase due to higher pH. Finally, the channels of pub-
lished thin-layer cascade reactors were made from expensive rigid
materials like steel, glass, concrete or fiberglass [19,20], resulting in a
high investment cost compared to other open cultivation systems. Thus,
the development of thin-layer cascade reactors addressing these issues
appears to be a valuable addition to the growing field of microalgae
research.
Testing bioreactors and processes on a larger scale usually requires
outdoor experiments. However, achieving a thorough understanding of
an outdoor process is complicated by randomly changing environ-
mental conditions. A solution to this is physical and dynamic day and
night climate simulation – the realistic indoor reproduction of outdoor
environmental conditions [8,21]. With the recent advent of light-
emitting diode (LED) technology in microalgae research [22], faithful Fig. 1. The open thin-layer cascade photobioreactor developed and used in this study (3D
CAD drawing and photo). 1 inlet module, 2 upper channel, 3 flow reversal module, 4
sunlight simulation has become possible. Experiments under controlled
lower channel, 5 retention tank. The photo shows a small volume of foam in the retention
environmental conditions allow informed decisions on worthwhile tank. No anti-foaming agent was required because of the mechanical foam destruction
improvements. The development process is accelerated since the ex- inherent in the reactor design: the accumulating foam was continuously destroyed by the
periments are independent of outdoor weather conditions. As a result, algae suspension falling from the lower channel into the retention tank.
the investment for a large-scale outdoor microalgae production facility
will more likely be profitable if the process has been thoroughly tested weeks without manual intervention, and detailed bioprocess data is
and optimized indoors under the environmental conditions of the en- automatically acquired), easily cleanable, and corrosion-resistant. The
visaged outdoor site. reactor modules (inlet module, flow reversal module, retention tank)
This paper reports on the design, construction, operation, and eva- were made of white high-density polyethylene (manufactured by
luation of open thin-layer cascade photobioreactors at TUM AlgaeTec Rauch, Feldkirchen, Germany). Two channel types were examined. An
Center (Technical University of Munich, Germany), a microalgae re- initial design involved a very smooth but expensive rigid channel made
search facility enabling physical and dynamic day and night climate of white high-density polyethylene sheets (10 mm thickness, Rauch,
simulation on a pilot scale. The LED-supported climate simulation Feldkirchen, Germany). A significantly less expensive design en-
system is presented and the thin-layer cascade reactor design is dis- compassed a slightly more uneven pond liner channel made of white
cussed. The saline microalga Nannochloropsis salina was chosen as a woven coated polyethylene fabric (areal weight 320 g m−2, Daedler,
model to evaluate batch process performance in open thin-layer cascade Trittau, Germany) placed on galvanized welded steel wire mesh (wire
photobioreactors with a surface area of up to 8 m2. In these processes, diameter 3 mm, 25 mm square spacing, Driller, Freiburg, Germany).
the climate simulation reproduced the environmental conditions of a The reactor parts were mounted on aluminum construction profiles
Mediterranean summer in Almería, Spain, a suitable location for a (Alváris, Suhl, Germany) and arranged in such a way that there was no
large-scale outdoor microalgae cultivation site. contact between the reactor parts, i.e. the algae suspension fell freely
from one part to the next. The retention tank was connected to the inlet
2. Materials and methods module by a food safe PVC hose (Rauspiraflex Liquitec, Rehau, Rehau,
Germany) via a magnetically coupled centrifugal pump (MKPG,
2.1. Open thin-layer cascade photobioreactor Ventaix, Monschau, Germany) operated by a frequency inverter
(Movitrac, SEW Eurodrive, Bruchsal, Germany). The volume flow rate
The thin-layer cascade reactor design developed and used in this of the algae suspension was measured using a magnetic-inductive
study (Fig. 1) consisted of five modular reactor parts: inlet module, sensor (MIK, Kobold, Hofheim, Germany) between pump and inlet
upper channel, flow reversal module, lower channel, and retention tank module. Temperature and pH of the algae suspension were measured
(parts given in order of the flow of the algae suspension). A detailed using a combination electrode (tecLine 201020/51-18-04-18-120,
technical description of the reactor parts is given in the Supplementary Jumo, Fulda, Germany) connected to a transmitter (ecoTrans pH 03,
materials. Jumo, Fulda, Germany). Evaporated water was automatically replaced
An off-the-shelf centrifugal pump was used to lift the algae sus- with tap water added via a solenoid valve (Type 52, Gemü, Ingelfingen,
pension from the retention tank to the inlet module, reducing invest- Germany) when the level of the algae suspension in the retention tank
ment and operational cost compared to raceway ponds where custom- fell below a level switch (LFFS, Baumer, Friedberg, Germany). CO2 was
engineered paddlewheels are required. In the pilot-scale implementa- added in the retention tank via a perforated hose (Solvocarb, Linde,
tions of this reactor design, the channel width was fixed to 1 m and the Pullach, Germany) connected to a CO2 mass flow rate controller (red-y
channel length was varied. In total, two reactors of 4 m2 each and six smart, Voegtlin, Aesch, Switzerland). The frequency inverter, volume
reactors of 8 m2 each were constructed. flow rate sensor, pH/temperature transmitter, level switch, solenoid
The reactor was designed to be modular (the geometry is easily valve, and CO2 mass flow rate controller were connected to a data
modifiable), automated (after inoculation the bioprocess can run for

382
A.C. Apel et al.
acquisition unit (Labjack U6, Labjack, Lakewood, Colorado, USA). The
data acquisition unit was connected to a computer running a process
control system written in Labview (Labview 2015, National
Instruments, Austin, Texas, USA). The pH of the algae suspension was
controlled by the CO2 mass flow rate using Labview's PID algorithm
(Kc = − 5, Ti = 1, Td = 0.1). Process data were recorded in an SQL
database. The temperature, pH, volume flow rate of the algae suspen-
sion, and CO2 mass flow rate were recorded every 10 s. Make-up water
addition events were recorded when they occurred.
2.2. Climate simulation
The climate simulation technology at TUM AlgaeTec Center was
designed to realistically reproduce the light and air conditions of any
location in the world suitable for microalgae mass production. This
allows the microalgae to experience the same environmental conditions
indoors with respect to temperature, air humidity and irradiance as
they would experience outdoors at the simulated location. Accordingly,
the microalgae can grow indoors as they would grow outdoors, making
efficient development of bioreactors and bioprocesses possible. As the
environmental conditions in an entire hall at TUM AlgaeTec Center are
controlled, any type of open cultivation system (e.g. raceway pond,
thin-layer cascade, pivot pond) can be used without modification of the
climate simulation equipment. The facility consists of three separate
glass halls with a pitched roof (angle 20°, average hall height 7.5 m), an
air conditioning system, and LED lamps (Fig. 2). To save energy and
assure the most natural sunlight spectrum, photons are supplied to the
microalgae by both solar and LED radiation, with the LEDs supplying
only those photons that are not supplied by the solar radiation.
The air conditioning system automatically selects the best operation
mode (natural or mechanical ventilation) to achieve the desired air
conditions (temperature, humidity) in the respective hall. The light
conditions of the target climate are simulated on the reactor surface via
real sunlight and artificial LED-based sunlight. The roof consists of la-
minated safety glass with high light transmission in the wavelength
ranges of 400–700 nm (average transmission 70%) and 350–400 nm
(average transmission 25%) to ensure availability of all wavelengths
important for microalgae physiology. The natural solar light is com-
plemented with artificial light (Fig. 3) from custom-built arrays (Fu-
tureLED, Berlin, Germany) with 8 different types of LEDs (7 narrow-
band single-color LEDs, 1 white broadband LED). The LEDs emit light in
the wavelength range of 400–750 nm with a photosynthetic photon flux
density (PPFD, photon flux density in the range of photosynthetically
active radiation (PAR) of 400–700 nm [23]) of up to
Fig. 2. Climate simulation technology at TUM AlgaeTec Center, Munich, Germany. A
situation in local winter is depicted. By day, the halls are heated by mechanical venti-
lation and the windows are closed. At night, the mechanical ventilation is switched off
and the windows open to cool down the hall. The system can automatically select other
operation schemes (e.g. cooling by natural ventilation on local summer days).
383
Algal Research 25 (2017) 381–390
1900 μmol m−2 s−1 (at a distance of 25 cm from the LED lamp). The
peak wavelength, number, and power of the LEDs were chosen to allow
a faithful reproduction of the solar spectrum according to the solar
reference spectrum ASTM G173–03 Direct + circumsolar [24]. In
contrast to other climate simulation light sources for microalgae, the
TUM AlgaeTec light simulation shows a nearly perfect spectral match
all over the PAR wavelength range, even in the important spectral range
of the red absorption maximum of Chlorophyll a near 680 nm (Fig. 3).
The lamps are software-tunable to other spectra by independent
dimming of the 8 LED types. Dimming is performed with the Constant
Current Reduction method. This avoids any influence on microalgae by
the flashing light effect that could result from standard LED dimming
methods such as Pulse Width Modulation. The solar PPFD is measured
by a calibrated spectroradiometer directly under the glass roof and put
into an algorithm calculating the PPFD to be supplied by the LEDs to
correctly simulate the target climate PPFD on the reactor surface. The
LED-based PFFD adapts to a changing solar irradiance (e.g. a cloud)
in < 1 s. During periods of high solar irradiance and low LED irra-
diance, the LED lamps cast shadows on the reactor surface. A possible
influence of these shadows on microalgae growth was evaluated by
cultivation experiments with and without solar irradiance (black cover
above LED lamps) – the observed growth rates were identical, hence the
influence of the shadows can be assumed negligible.
Climate data were recorded in SQL databases in different temporal
resolutions: air data (temperature, humidity) every minute, light data
(solar and LED PPFD) every 5 s. New setpoints extracted from me-
teorological measurements of the target climate location were trans-
mitted to the climate simulation systems every minute.
2.3. Simulated climate
In this study, all climate simulation experiments were conducted
under identical climate conditions. The target climate for the climate
simulation was June 15, 2012 in Almería, Spain (Fig. 4), a sunny day
with high solar irradiance and air temperatures of ~17 °C at night and
~30 °C by day. The simulation of this target climate was repeated every
day. Almería was chosen because it is considered a particularly suitable
location for a prospective large-scale outdoor microalgae cultivation
site – situated at the southern coast of mainland Spain, the region is
characterized by high solar irradiance, warm air temperatures, access to
seawater, non-arable land, good infrastructure, a stable political cli-
mate, and experience with both large-scale greenhouse agriculture and
microalgae research. The measured environmental conditions in Al-
mería were the global radiation, W m−2, air temperature, °C, and re-
lative humidity of the air, %. The global radiation measured in Almería
covers all wavelengths of solar radiation while the climate simulation at
TUM AlgaeTec Center only covers wavelengths important for photo-
synthesis. To convert global radiation to PAR in energetic units (irra-
diance in 400–700 nm, W m−2), a factor of 40% was assumed [21]. To
convert energetic PAR (W m−2) to quantum flux PAR (PPFD,
μmol m−2 s−1), a factor of 4.6 μmol W−1 s−1 was used according to
literature data [23] validated by own measurements using a calibrated
spectroradiometer (Flame, Ocean Optics, Dunedin, USA).
2.4. Strain and seed culture
A seed culture of a microalgae strain broadly considered suitable for
large-scale saline microalgae production, Nannochloropsis salina, was
obtained from laboratory experiments in 1.8 L closed photobioreactors.
Seed culture preparation has been described elsewhere [25]. The seed
culture was grown in a 4 m2 thin-layer cascade reactor at pH 8.5 in the
simulated climate of Almería with two modifications: a reduction of the
PPFD by 80% (to avoid photodamage of the dilute cultures) and a
minimum air temperature of 22 °C (to avoid low reactor temperatures
in the morning; resulting minimal/maximal reactor temperatures:
night, 14 °C, day, 22 °C). The culture was kept in the exponential
A.C. Apel et al. Algal Research 25 (2017) 381–390

Fig. 3. TUM AlgaeTec Center sunlight simulation. Spectral

evaluation in the PAR wavelength range extended by 50 nm
(grey background) to wavelengths not covered by the PAR
definition, but possibly important for microalgae [21]. PA-
NELS A1 AND B1: PPFD spectra normalized to an identical

number of PAR photons. PANELS A2 AND B2: Percentage de-

viation from the solar reference spectrum. The panels
compare the solar reference spectrum ASTM G173–03 Di-
rect + circumsolar [24] ( ) to three spectra: a white LED in
a closed photobioreactor (Infors 5 Lux, Panel A, ), the
TUM AlgaeTec Center LED lamps (Panel A, , upper
photo), and a representative spectrum of the sunlight si-
mulation at TUM AlgaeTec Center where equal irradiance is
supplied by real sunlight and by the LED lamps (Panel B,
, lower photo). An excellent spectral match is achieved,
ensuring a realistic microalgal physiology.

Table 1
Summary of important parameters of the thin-layer cascade reactors in this study.

Parameter Value

Channel slope 1.7%

Volume flow rate 2.4 L s−1
Flow velocity on channel (calculated; measured) 0.48 m s−1; 0.44 m s−1
Layer thickness on channel (calculated; measured) 5.0 mm; 5.6 mm
Reynolds number Re 2347 (transitional flow)
Reactor area Areactor 8 m2
Alight Areactor−1 0.89
Surface-to-volume (Areactor Vcultivation−1) 160 m−1
Hydraulic mixing power on channel 0.4 W m−2
Approximate cultivation volume distribution during operation:
- Vcultivation 50 L 100%
- Inlet module 6L 12%
- Channels 32 L 64%
- Flow reversal module 4L 8%
- Retention tank 5L 10%
- Pumping circuit 3L 6%

bioprocess. Table 1 shows the reactor parameters used in this study.

Fig. 4. Target climate for the climate simulation: solar photosynthetic photon flux density
(PPFD), air temperature, and relative humidity of the air. Data recorded with a temporal
resolution of 1 min in Almería, Spain, on June 15, 2012, by DLR at Plataforma Solar de
Almería (www.dlr.de/sf/en/desktopdefault.aspx/tabid-8724/14207_read-35930). Data
courteously provided by S. Wilbert (DLR). All cultivation experiments in this publication
were performed at these repeated conditions.
growth phase by regular dilution with fresh inoculation medium.
2.5. Medium
Two media were used: inoculation medium (used at the start of a
new experiment) and feed medium (used to replenish nutrients during
an experiment). The inoculation medium was Artificial Seawater (ASW
[26]) without CaCl2 in the trace element solution as it is already present
in the main element solution. The feed medium was ASW excluding
NaCl, MgSO4 and CaCl2.
2.6. Operation of open thin-layer cascade photobioreactors
The algae suspension was circulated day and night. With the mod-
ular reactor design and the adjustable channel slope (0–3%), volume
flow (0–3.6 L s−1) and cultivation volume (50–140 L), a wide para-
meter range can be evaluated to identify optimal conditions for a
Using newly developed high-speed conductivity sensors [29], the cir-
culation period τ and mixing time θ99 (characterizing the duration of
nutrient concentration inhomogeneity in the algae suspension after
addition of feed medium) at different volume flows were determined to
be 28–48 s and 2.1–3.2 min, respectively. Comparing this with the
significantly higher time scale of algae growth and considering that no
pH or temperature gradients were observed in the reactor, the pilot-
scale thin-layer cascade reactor can be assumed to be perfectly mixed.
NOTES TO TABLE 1 Flow velocity and layer thickness were calculated
with the Manning formula for open channel flow by approximating the
hydraulic radius with the layer thickness and using a Manning's coef-
ficient of 0.008 s m−1/3 [30]. Flow velocity was measured with newly
developed high-speed minimal-invasive conductivity sensors using so-
dium chloride pulses [29]. Layer thickness was averaged from mea-
surements performed using a thin ruler on a regular measurement grid
with 8 sampling points per m2 of channel area. Minimal/maximal
measured layer thickness: 5/7 mm (rigid channel), 4/8 mm (pond liner
channel). Reynolds number of the algae suspension approximated using
the kinematic viscosity of seawater at 20 °C, 1 bar (1.05 · 10−6 m2 s−1,
[31]). Flow regime classified according to the open-channel flow clas-
sification by French [32]: Re ≤ 500, laminar; 500 ≤ Re ≤ 12,500,
transitional; Re ≥ 12,500, turbulent. Hydraulic mixing power [15]
calculated with the density of seawater (1.02 kg m−3). Areactor: Area
occupied by the entire reactor. Alight: Channel area (main photo-
synthetically active area). Vcultivation: cultivation volume.
All experiments were started with 50 L of the inoculation medium.

384
A.C. Apel et al. Algal Research 25 (2017) 381–390

The reactors were inoculated in the afternoon to an initial cell dry

weight concentration of ~0.3 g L−1. The cultivation volume was held
constant at 50 L by automatic addition of make-up water for evapora-
tion for 6 s via the solenoid valve (resulting in addition of ~ 550 mL tap
water) when the level of the algae suspension in the retention tank fell
below the level switch. Hence, the cultivation volume varied by only
1% due to evaporation, making precise determination of the intra-day
growth rate possible. To avoid nutrient depletion, 1 L of concentrated
feed medium was manually added as soon as the nitrate concentration
fell below 2 g L−1. The concentration of the added feed medium was
adapted using a Matlab process simulation tool to keep the nitrate
concentration above 2 g L−1 for at least 2 days (the tool predicts cell
growth based on the current growth rate and approximates the re-
sulting nitrate concentration based on the predicted cell growth and an
empirically determined nitrate-biomass yield coefficient). The pH/
temperature electrodes were calibrated before each experiment using 2-
point pH calibration at pH 7 and pH 10 (HI7007 and HI7010, Hanna
Instruments, Vöhringen, Germany) and a calibrated thermometer
(HI98509, Hanna Instruments, Vöhringen, Germany). During the ex-
periments, the pH of the algae suspension in the retention tank was
regularly crosschecked offline using a calibrated mobile pH meter
(HandyLab 100, SI Analytics, Mainz, Germany) to characterize the drift
of the pH probes that were immersed in the algae suspension for several Fig. 5. Visualization of the two different growth rate definitions in microalgae cultivation
weeks. A maximal deviation of pH 0.2 was observed between the online in a simulated climate, i.e. at realistic day and night conditions. A three-week process in a
and offline pH measurements. The pH of the algae suspension was thin-layer cascade reactor is shown. UPPER PANEL: The phases of standard microbial batch
controlled at pH 8.5 (except for one experiment, where the pH control growth processes can also be observed under climate simulation (considering only one
and CO2 supply were active by day but switched off at night). The sample per day at a fixed time of day). Cell dry weight concentration (CDW, ) de-
termined from OD750 samples (error bars omitted for visual clarity, average and max-
suspension temperature was measured but not controlled.
imum coefficient of variation 0.8% and 4.2%, respectively). The first CDW concentration
of each day is highlighted in the exponential (unlimited, ), linear (light-limited, ),
2.7. Analytical methods transition (Δ), and stationary ( ) phase. OD750-CDW correlation parameters: a = 0.235,
b = 0.268, c = 0.319. Fit functions in exponential ( , μmulti-day), linear ( ) and sta-
tionary ( ) phase. The data in the dashed rectangle are magnified in the LOWER PANEL:
Cell growth was measured as optical density at 750 nm (OD750) by
Separate exponential growth phases ( , μintra-day) occur on each day during the first days
diluting a sample in triplicates to an absorbance of 0.1–0.3 using tap of the process (error bars: ± 1 × standard deviation). Additionally, the exponential fit
water with 27 g L−1 NaCl (also served as blank). The cell dry weight function for the multi-day growth rate ( , μmulti-day) is shown. Later in the process, the
concentration (CDW) was determined once a day by vacuum filtration. daily growth curves cannot be reliably fitted with an exponential function anymore – they
Glass microfiber filters were pre-dried at 80 °C for at least 24 h and then rather follow a linear function or a step function (possibly due to light limitation and
sampling error at very high dilution ratios). The photosynthetic photon flux density
weighed (m1). A sample volume Vsample with at least 1 mgCDW was ap-
(PPFD, ) shows the daylight and night phases.
plied on the filter and washed during filtration with deionized water of
twice the sample volume to remove extracellular salts and soluble
separate exponential growth phases on each day, i.e. in each daylight
components. After the filtration, the loaded filter was dried at 80 °C for
phase. It was calculated by fitting all CDW samples of the respective day
at least 48 h and weighed again (m2). The cell dry weight concentration
with the CDW fit function (where t0 = tfirst CDW sample of day = 0).
was calculated from duplicate or triplicate samples as CDW =
The nitrate concentration was determined semiquantitatively with
(m2 − m1) ∙ Vsample−1.
test strips (Quantofix Nitrate/Nitrite, Macherey-Nagel, Düren,
A correlation factor α was calculated by linear regression of OD750
Germany) in the supernatant of a centrifuged algae suspension sample
and CDW. Cell growth data given in CDW was converted from OD750
(14,500 × g, 4 min, Espresso, Thermo Fisher, Waltham, USA).
measurements by CDW = α ∙ OD750. In extended experiments, the OD750-
CDW correlation factor was not constant, but decreased with increasing
2.8. Calculation of economic bioprocess metrics
cell density (α ≈ 0.45 g L−1 after inoculation, α ≈ 0.32 g L−1 in the
linear growth phase). In these cases, the OD750-CDW correlation factor
All metrics were calculated between two process times t1 , t2. Batch
was approximated using the correlation fit function α(t) = a ∙ exp(− b ∙ t)
metrics were referred to the beginning of the batch process (t1 = 0).
+ c with experiment-specific parameters a,b,c. Specific growth rates μ
Daily metrics were calculated based on the first optical density mea-
were calculated from CDW data using the CDW fit function CDW(t)
surement of each day and referred to the preceding ~ 24 h, i.e. plotted
= CDW(t0) ∙ eμ ∙ t. Curve fitting was performed by nonlinear least squares
at a process time t2. Hence, all metrics include both the biomass in-
fitting with the Levenberg-Marquardt algorithm (Matlab R2016b,
crease by day and the biomass loss at night and can therefore be used
Mathworks, Natick, USA).
for a realistic assessment of bioprocess economics. Daily metrics can
To account for the peculiarities of microalgae growth curves under
serve as an indication of the economics in a semi-continuous process
day-night-cycles, two different specific growth rates μmulti-day and μintra-
operated at a certain cell dry weight concentration.
day were calculated (visualized in Fig. 5). The growth rate in the ex-
Volumetric and areal productivities were determined from cell dry
ponential phase of the entire batch process μmulti-day was defined by
weight concentration data. The volumetric productivity PV and the
considering only one CDW sample per day, taken at a fixed time each
areal productivity PA were calculated as:
day. Then, it can be assumed that a batch process performed over
several days is characterized by an initial exponential phase (unlimited PV = (CDW (t2) − CDW (t1))⋅(t2 − t1)−1
growth) followed by a linear phase (light-limited growth). The multi-
PA = PV ⋅SV −1
day growth rate was calculated by fitting the first CDW sample of each
day with the CDW fit function (where t0 = tfirst CDW sample of experi- with the surface-to-volume ratio SV = (Areactor ∙ Vcultivation−1) where
ment = 0). The intra-day growth rate μintra-day was defined by assuming Areactor is the area occupied by the entire reactor and Vcultivation is the

385
A.C. Apel et al. Algal Research 25 (2017) 381–390

cultivation volume.
The CO2 utilization efficiency ηCO2 relates the mass of carbon fixed
in biomass mC,fixed to the mass of carbon supplied as CO2 mC,supplied:
ηCO2 = m(C, fixed) ⋅m(−C1, supplied)

mC,fixed = 0.5⋅(CDW (t2) − CDW (t1))⋅Vcultivation

(t2 )
mC,supplied = 0.273⋅ ∫(t ) 1
ṁ CO2 dt

where the mass of carbon fixed in biomass was calculated from the cell
dry weight concentration by assuming that 50% of the cell dry weight is
carbon [27]. The mass of carbon supplied was calculated from the in-
tegral of the CO2 mass flow rate and the mass ratio of carbon in a CO2
molecule (27.3%).
The photosynthetic conversion efficiency ηphot relates the energy
fixed in biomass Efixed to the energy supplied as photosynthetically
active radiation Esupplied:
−1
ηphot = Efixed⋅Esupplied

Efixed = (CDW (t2) − CDW (t1))⋅ΔHc0

t2
Esupplied = Areactor ∫t 1
IPAR dt

where Δ Hc0 is the standard enthalpy of combustion of dry microalgae

biomass (22.5 kJ g−1, [28]) and IPAR is the energetic PAR
(400–700 nm) irradiance in W m−2.

3. Results and discussion

Fig. 6. Parallel batch processes with Nannochloropsis salina in four thin-layer cascade
reactors. PANEL A: air temperature set point ( ) and actual value (▬), photosynthetic
3.1. Growth rates, channel variants, and reactor comparability
photon flux density (PPFD) of solar ( ) and entire (solar + LED, ) light on the reactor
surface. PANELS B–D show data from the reactors R1 (red), R2 (blue), R3 (green), and R4
Parallel batch processes with the saline microalga Nannochloropsis (grey). R1,2: rigid channel. R3,4: pond liner channel. PANEL B: Microalgae suspension
salina were performed in four thin-layer cascade reactors R1–R4 to temperature. PANEL C: Cell dry weight concentration (CDW), error bars: ± 1 × standard
deviation, OD750-CDW correlation factor α = 0.36 g L−1 (R2 > 0.95). The arrows in-
evaluate the maximal growth rates in the simulated Mediterranean
dicate the addition of 1 L of concentrated feed medium, the concentration factor is given
summer climate and the reactor comparability (Fig. 6). Reactors R1 and next to the arrows. PANEL D: Intra-day exponential growth rate, error bars: 95% confidence
R2 were equipped with rigid channels, reactors R3 and R4 were interval of the exponential fit. Multi-day growth rate (days 1–4): R1: 0.67 d−1, R2:
equipped with pond liner channels. The four reactors were situated in 0.71 d−1, R3: 0.60 d−1, R4: 0.61 d−1, Average (R1–R4): 0.65 d−1. (For interpretation of
the same hall and hence experienced the same environmental condi- the references to color in this figure legend, the reader is referred to the web version of
this article.)
tions (Fig. 6A). The air temperature followed the set points closely with
only few minor deviations. As the experiments were performed in
temperatures due to the lower thermal capacity of the channel.
German winter (max. solar elevation angle 20°), most photons were
The comparability of the four reactors can be evaluated by the cell
supplied by the LED system. On day 4, the irradiance and air tem-
growth (Fig. 6C, D). Temperature-dependent growth kinetics of N.
perature set points were reduced for 30 min to simulate passing clouds.
salina suggest that the suspension temperature deviation of 3 °C ob-
In the night between day 6 and 7, the windows did not open for five
served in this experiment could markedly influence the growth rate
hours due to rainfall. With the exception of these two singular events,
[36]. However, the cited growth kinetics were obtained under constant
the repeatedly simulated daily course of the environmental conditions
conditions and might therefore not be entirely applicable in processes
was well comparable during the experiment.
performed at a realistic day and night climate [34]. In fact, the data
Effects of the air temperature on the reactor temperature are buf-
from our study do not show a clear correlation of the suspension tem-
fered by the thermal capacity of the algae suspension. Hence, transient
perature and the growth rate. On day 2, despite a temperature differ-
peaks of the air temperature are barely visible in the reactor tempera-
ence of ~2 °C prevailing for several hours in the afternoon, the growth
ture (Fig. 6B). Due to the low cultivation volume and layer thickness,
rates in the reactors R1 and R4 were identical. On day 3, despite only
the reactor temperature cools down rapidly in the evening. This is
small temperature differences (< 1 °C) between all four reactors, dif-
considered a positive effect for minimization of the nightly biomass loss
ferent growth rates (1.6–1.9 d−1) were observed. If one reactor had
due to respiration [33]. For the same reason, the reactor temperature
offered better growth conditions than the other reactors, the growth
heats up rapidly in the morning. As photoinhibition is more likely to
rate in that reactor would have been consistently higher than in the
occur at sub-optimal temperatures [34,35], this effect helps to ensure
other reactors. However, this is not the case – on any given day, the
that the microalgae can efficiently utilize the available light. The
growth rates in the four parallel reactors do not vary significantly
temperature course in the four reactors was similar with a maximal
(based on the 95% confidence interval of the fit). It is therefore con-
deviation between two reactors of only ~ 3 °C. Possible reasons for
cluded that the variation of the growth rates within a single day results
temperature deviations in the reactors are varying offsets of the tem-
from the small but inevitable sampling error and biological variation
perature sensors or inhomogeneity in the air temperature and solar
which developed despite the fact that all reactors were inoculated from
irradiance at the reactor locations (e.g., reactors R3 and R4 are closer to
the same seed culture. The data also shows that the channel variant
the windows than the other two reactors, resulting in lower tempera-
(rigid, pond liner) does not have a significant effect on the growth rate.
tures by night). Additionally, the lower thickness of the pond liner
This means that cost-efficient thin-layer cascade reactors can be
channel (R3, R4) might have caused slightly lower suspension

386
A.C. Apel et al.
constructed with pond liner instead of the rigid channel materials that
have been used until now. The highest intra-day growth rates (up to
1.9 d−1) were observed below a cell dry weight concentration of 3 g
L−1. The average multi-day growth rate of all four reactors was
0.65 d−1. No literature data with comparable outdoor growth condi-
tions was found. Under constant conditions in a closed laboratory-scale
photobioreactor, a maximum growth rate of 1.3 d−1 was reported in
literature [36].
Intra-day exponential growth was observed below a cell dry weight
concentration of 10 g L−1. Above that concentration, the shape of the
growth curve and the increasing size of the 95% confidence interval of
the growth rates indicate that the intra-day growth did not follow an
exponential curve. Instead, the cell dry weight concentration was more
or less stable in the morning and in the afternoon, but showed an in-
crease around noon. The same effect was observed in other experi-
ments. It is presumed that at these high cell mass concentrations, only
the high mid-day irradiance is sufficient for growth. The best fit quality
was achieved with at least six OD750 samples per day. In a second
parallel experiment with the same simulated climate, the maximum
intra-day growth rate of 1.9 d−1 was confirmed, and intra-day growth
rates of ~1.5 d−1 were observed up to a cell dry weight concentration
of 7 g L−1 (data not shown). The short passage of clouds simulated on
day 4 of this experiment did therefore probably reduce the growth rate
by ~0.2 d−1. In summary, the growth rates in the novel pilot-scale
thin-layer cascade reactors were higher than literature data and the
growth conditions in the four parallel reactors were similar. It can be
concluded that the thin-layer cascade reactors promote better cell
growth than other reactors. Furthermore, the reactors at TUM AlgaeTec
Center can be used for parallel reaction engineering studies, e.g. for
evaluation of growth at different pH or channel slopes.
3.2. Maximum cell dry weight concentration
To evaluate the maximum attainable cell dry weight concentration
in the thin-layer cascade reactors, batch processes were conducted at
the physically simulated climate conditions of Almería and at constant
illumination with Nannochloropsis salina (Fig. 7). At constant illumina-
tion, a maximum cell dry weight concentration of 42 g L−1 was ob-
served. In this experiment, the PPFD was manually increased from 700
to 1800 μmol m−2 s−1 according to the increasing cell dry weight
concentration in the reactor. The air temperature was decreased to
counteract the increased radiative heat input to ensure a constant
Fig. 7. Maximal cell dry weight concentration (CDW) of Nannochloropsis salina in batch
processes in thin-layer cascade reactors at simulated climate conditions ( , vertical ar-
rows) and constant illumination ( , horizontal arrows). OD750‑CDW correlation para-
meters: a = 0.235, b = 0.268, c = 0.319. Error bars omitted for visual clarity, average
and maximum coefficient of variation: 0.8 % and 4.2 % (simulated climate), 1.1 % and
3.2 % (constant illumination). The arrows indicate the addition of 1 L of concentrated
feed medium, the concentration factor is given next to the arrows (at the 1500x addition,
the feed medium components were given into the reactor in solid form).
387
Algal Research 25 (2017) 381–390
reactor temperature of 26 °C (fluctuations of ± 3 °C). At the physically
simulated climate conditions, a cell dry weight concentration of
50 g L−1 was observed with N. salina after 24 days.
In literature, the highest reported outdoor cell dry weight con-
centrations of N. salina in open and closed cultivation systems were
around 1.7 g L−1 (calculated from the ash-free dry weight concentra-
tion of 1 g L−1 and the OD750-CDW correlations of the author's and this
study, [37]) and 6 g L−1 [38], respectively. At constant conditions, Zou
et al. [39] observed a maximal cell dry weight of Nannochloropsis sp. of
up to 19 g L−1 (ash-free dry weight concentration 18 g L−1, ash content
6%) in closed photobioreactors without medium exchange. When the
culture medium was regularly exchanged with the aim of preventing
build-up of inhibitory substances, up to 67 g L−1 were reached, leading
to the conclusion that medium exchange procedures might be a re-
quirement for high cell density cultivation of microalgae [40]. How-
ever, this study demonstrated that no removal of inhibitory substances
is required to grow N. salina to a cell dry weight concentration of
50 g L−1. High cell dry weight concentrations of N. salina – exceeding
the maximum concentration in other open cultivation systems by a
factor of 30 – were achieved in thin-layer cascade reactors under both
constant conditions and realistically simulated outdoor conditions.
Harvesting cost in large-scale outdoor cultivation can therefore be
markedly reduced with thin-layer cascade reactors.
3.3. Economic bioprocess metrics
The productivity, photosynthetic conversion efficiency, CO2 utili-
zation efficiency, and evaporation rate are important economic bio-
process metrics [21]. To characterize the novel thin-layer cascade re-
actor, these metrics were calculated from process data obtained in a
three-week batch cultivation with Nannochloropsis salina at simulated
climate conditions (Fig. 8).
A stable areal productivity of the entire batch process of 15–18 g
m−2 d−1 (volumetric productivity 2.4–2.9 g L−1 d−1) was obtained at
a process time of 6–16 days at cell dry weight concentrations in a wide
range of 15–40 g L−1 (Fig. 8A, D). Many large-scale microalgae mass
cultivation processes will probably be operated at a fixed cell dry
weight concentration in a continuous or semi-continuous (e.g., har-
vested once per day) process mode, where the daily productivity is a
decisive parameter. Daily areal productivities of up to 25 g m−2 d−1
(volumetric productivity 4 g L−1 d−1) were observed at a cell dry
weight concentration of ~17 g L−1. As Terry and Raymond [41] noted,
a direct comparison of productivities (or, more generally, any economic
bioprocess metrics) between different studies is rarely possible because
the process performance depends on the species, reactor design, en-
vironmental conditions, and culture management strategies. No litera-
ture data on cultivation of Nannochloropsis in open thin-layer cascade
reactors in a Mediterranean summer climate is available. However, de
Vree et al. recently reviewed outdoor processes with Nannochloropsis in
different photobioreactors [28]. This allows to put the results from this
study in context as long as no directly comparable data is available.
Under outdoor conditions, the highest reported productivities in open
and closed cultivation systems are 25 g m−2 d−1 (0.2 g L−1 d−1, Nan-
nochloropsis salina, raceway pond, Israel [26]) and 28 g m−2 d−1
(1.2 g L−1 d−1, Nannochloropsis sp., flat panel, The Netherlands [28]),
respectively. Compared to this best-available result for open cultivation
systems, the areal productivity obtained in this study is similar while
the volumetric productivity in this study is markedly higher. Compared
to this best-available result for closed cultivation systems, the areal
productivity observed in this study is slightly lower while the volu-
metric productivity in this study is higher again. Hence, the limited
comparability notwithstanding, it can be concluded that with Nanno-
chloropsis salina, the pilot-scale thin-layer cascade reactor developed in
this study allows to achieve areal productivities on a par with other
cultivation systems, and volumetric productivities that exceed data
from all other published open or closed photobioreactors. Furthermore,
A.C. Apel et al.
Fig. 8. Economic metrics of a batch process with Nannochloropsis salina in a thin-layer
cascade reactor at simulated climate conditions. PANEL A: Cell dry weight (CDW) con-
centration (●, error bars omitted for visual clarity, average and maximum coefficient of
variation: 0.8% and 4.2%) with highlighting of the first CDW concentration of each day
( ) used for determination of the daily metrics, photosynthetic photon flux density
(PPFD, ) of the simulated climate showing the day and night phases. Parameters of the
OD750-CDW correlation fit function: a = 0.235, b = 0.268, c = 0.319. The arrows in-
dicate the addition of 1 L of concentrated feed medium, the concentration factor is given
next to the arrows. PANEL B: CO2 mass flow rate set by the pH controller, referred to the
reactor area (non-optimized CO2 supply hose). PANEL C: pH. At different process times t,
three pH control regimes were used: 0 d < t < 0.6 d, pH control and CO2 addition
inadvertently deactivated; 0.6 d < t < 3.1 d, default regime (pH controlled day and
night at pH 8.5); t > 3.1 d, CO2-saving regime (pH controlled at pH 8.5 by day and
uncontrolled (no CO2 addition) at night). PANEL D: Batch (●) and daily (○) biomass
productivity. PANEL E: Batch (●) and daily (○) CO2 utilization efficiency, batch ( ) and
daily ( ) photosynthetic conversion efficiency. Due to the absence of notable growth in
the stationary phase, daily metrics were calculated only for the exponential and linear
growth phase.
the maximal photosynthetic conversion efficiency observed in this
study (4.6%, Fig. 8 E) exceeds the highest data reported for Nanno-
chloropsis sp. in open (1.5%, raceway pond, The Netherlands) and
closed cultivation systems (4.2%, vertical tubular photobioreactor, The
Netherlands) [28].
To minimize CO2 desorption from the reactor to the atmosphere, the
processes in this study were operated at pH 8.5, where carbon in the
aqueous phase is mainly present as hydrogen carbonate and carbonate.
In the batch process discussed here, the CO2 supply was switched off at
night, allowing the pH to float to its equilibrium value (Fig. 8 B, C). A
batch CO2 utilization efficiency of 75% was measured after 2 days at a
cell dry weight concentration of 3.4 g L−1 (Fig. 8 E). This is similar to
efficiencies reported in freshwater thin-layer cascade cultivation [15]
and raceway ponds [3,42] and shows that the CO2 supply system of the
thin-layer cascade reactors is generally capable of achieving high
388
Algal Research 25 (2017) 381–390
efficiencies. However, the batch CO2 utilization efficiency steadily de-
clined to a final value of 10% at the beginning of the stationary phase
after 20 days at a cell dry weight concentration of ~43 g L−1. Even at
high biomass concentrations of actively photosynthesizing cells, no pH
gradient was observed along the channels. Due to the high pH and
salinity as well as the absence of a pH gradient along the channels, it
can be assumed that the CO2 loss by desorption from the suspension to
the atmosphere was negligible and the supply of inorganic carbon to the
algae was similar at the beginning and the end of the channels. The
decline of the CO2 utilization efficiency was probably caused due to the
overshoot of pH control in the morning (Fig. 8 B) and the non-opti-
mized CO2 supply hose. High CO2 mass flow rates supplied via a barely
submerged CO2 supply hose of insufficient length resulted in large CO2
bubbles with a short residence time in the algae suspension before they
escaped into the atmosphere. A higher CO2 utilization efficiency was
achieved in further experiments by submerging the CO2 supply hose
deeper in the retention tank – this allowed the CO2 bubbles to be drawn
into the pump suction hose and resulted in an increased residence time
and better gas-liquid mass transfer due to the high turbulence in the
pumping circuit. Further improvement of the efficiency can be expected
by adjusting the PID parameters of the pH controller and by operating
the process at higher pH. A valuable side effect of pH control using
precise CO2 mass flow rate controllers is that the course of the CO2 mass
flow rate can be used to gain real-time information about the biopro-
cess. An irregular drop of the CO2 mass flow rate can reveal important
metabolic or water chemistry changes such as nutrient limitations long
before the effect is observed in the cell growth curve, allowing the
bioprocess operator to stabilize the process.
The evaporative water loss in open cultivation systems is an im-
portant physical effect that prevents overheating of the microalgae
suspension. At the same time, it influences the water footprint of the
bioprocess and the operational cost. The evaporation in the thin-layer
cascade reactors is characterized in Fig. 9. In the upper panel, en-
vironmental conditions determining the evaporation rate are shown. In
the lower panel, water replenishment events in four thin-layer cascade
reactors on four different days are plotted and fitted using a sigmoidal
function [43]. On average, 7.4 L m−2 d−1 evaporated. Using the
Fig. 9. Characterization of the evaporation in thin-layer cascade reactors under simulated
climate conditions. UPPER PANEL: The major environmental conditions that influence the
evaporation rate: air temperature ( ), microalgae suspension temperature (▬), and ir-
radiance ( ) on a representative day. The relative humidity of the air was 27–34% (day)
and 33–44% (night). LOWER PANEL: Cumulated volume of make-up water for evaporation
( , 16 curves, data from four thin-layer cascade reactors on four days, each dot re-
presents the addition of ~ 68 mL m−2). The black curve was fitted to the 16 evaporation
curves using a sigmoidal fit function [43]. Its derivative ( ) approximates the eva-
poration rate per unit area.
A.C. Apel et al.
derivative of the fit function, the minimal and maximal evaporation
rates were determined to be 80 mL m−2 h−1 at night and
620 mL m−2 h−1 in the afternoon. Hence, when 1 L of concentrated
feed medium was added during daytime, the increased reactor volume
evaporated within 12–30 min. Arithmetically speaking, the entire cul-
tivation volume of the thin-layer cascade reactor would evaporate in
20 h. This offers a convenient method for a rapid and controlled salinity
increase in the algae suspension to reduce contaminating species or
induce the accumulation of desired substances, e.g., carotenoids [44] or
lipids [45]. However, it also means that a freshwater source is required
if no major increase of the salinity is desired. An elegant approach to
reduce the freshwater consumption of thin-layer cascade reactors
would be to use the channel area to collect rainwater (while it rains, the
algae suspension is stored in a tank) and use it as make-up water for
evaporation. The evaporation rate measured in this study was obtained
with climate simulation of a warm day with very high solar irradiance
in Almería, Spain. Comparing this with actual data from Almería where
the average evaporation rate in raceway ponds and thin-layer cascade
reactors was determined to be 5–6 L m−2 d−1 [46,47], it can be con-
cluded that the climate simulation technology used in this study allows
a realistic simulation of the evaporation. Furthermore, the evaporation
rate measured in this study compares well with actual measured eva-
poration rates in large-scale outdoor thin-layer cascade reactors and
raceway ponds (6–20 L m−2 d−1 [21]).
4. Conclusion
Climate simulation – the physical indoor simulation of dynamic
outdoor environmental conditions – is an emerging tool in the devel-
opment of photobioreactors and bioprocesses. In this study, novel open
photobioreactors for saline microalgae were designed and evaluated
using a globally unique realistic climate simulation technology. In the
simulated Mediterranean summer climate of Almería (Spain) – a sui-
table location for a large-scale outdoor microalgae cultivation site –
high growth rates, biomass concentrations, and economic process me-
trics were observed with the saline microalga Nannochloropsis salina.
Precisely determined intra-day growth rates showed that cost-efficient
thin-layer cascade reactors with inexpensive pond liner channels are
feasible.
Thin-layer cultivation is a promising concept for open production of
microalgae and might replace the inefficient raceway ponds. It is an-
ticipated that the discussion of design guidelines and technical details
on thin-layer cascade reactors will further advance their development.
Realistic climate simulation will support this process and furthermore
enable reliable determination of microalgae growth kinetics for real
outdoor processes.
Author's contributions
All authors contributed to the conception and design of the study,
the interpretation of the data as well as the critical revision and final
approval of the manuscript; AA, CP, NB, NM, JG, JS performed the
experiments and analyzed the experimental data; AA, DWB wrote the
manuscript; TB, DWB obtained the funding.
Acknowledgments
Funding (grant number LaBay74A) provided by the Bavarian State
Ministry for Economic Affairs and the Media, Energy and Technology
(Munich, Germany), the Bavarian State Ministry of Education, Science
and the Arts (Munich, Germany) and Airbus Group (Leiden, The
Netherlands) is gratefully acknowledged. Andreas Apel and Christina
Pfaffinger were supported by the TUM Graduate School (Technical
University of Munich, Germany). The funding sources did not influence
the reported research. Stefan Wilbert (DLR, Plataforma Solar de
Almería, Almería, Spain) is gratefully acknowledged for providing
389
Algal Research 25 (2017) 381–390
meteorological data of Almería for the climate simulation. The authors
thank Anja Koller, Timm Severin, Hannes Löwe, Johannes Schmidt,
Matthias Glemser and Daniel Garbe (Technical University of Munich) as
well as Skye Thomas-Hall and Peer Schenk (The University of
Queensland, Brisbane, Australia) and John Benemann (MicroBio
Engineering, USA) for fruitful discussions on microalgae research. The
support by the TUM students Philipp Mayer-Ullmann, Michael
Neßlauer, Peter Sinner, Manuel Moos, Florian Böck, Elisabeth Gleis,
Tiasa Ghosh, Harish Reddy, Camila Castro Rico and Peter Kämpf in the
TUM AlgaeTec Center is acknowledged as well.
Appendix A. Supplementary data
Supplementary data to this article can be found online at http://dx.
doi.org/10.1016/j.algal.2017.06.004.
References
[1] R. Rosello Sastre, Products from microalgae: an overview, in: C. Walter, C. Posten
(Eds.), Microalgal Biotechnology: Integration and Economy, De Gruyter, Berlin,
2012, pp. 15–50.
[2] P.M. Schenk, S.R. Thomas-Hall, E. Stephens, U.C. Marx, J.H. Mussgnug, C. Posten,
et al., Second generation biofuels: high-efficiency microalgae for biodiesel pro-
duction, Bioenergy Res. 1 (2008) 20–43, http://dx.doi.org/10.1007/s12155-008-
9008-8.
[3] M.E. Huntley, Z.I. Johnson, S.L. Brown, D.L. Sills, L. Gerber, I. Archibald, et al.,
Demonstrated large-scale production of marine microalgae for fuels and feed, Algal
Res. 10 (2015) 249–265, http://dx.doi.org/10.1016/j.algal.2015.04.016.
[4] J. Sheehan, T. Dunahay, J.R. Benemann, P. Roessler, A Look Back at the U.S.
Department of Energy's Aquatic Species Program: Biodiesel From Algae, (1998).
[5] P.T. Pienkos, A. Darzins, The promise and challenges of microalgal-derived biofuels,
Biofuels Bioprod. Biorefin. 3 (2009) 431–440, http://dx.doi.org/10.1002/bbb.159.
[6] T.J. Lundquist, I.C. Woertz, N.W.T. Quinn, J.R. Benemann, A Realistic Technology
and Engineering Assessment of Algae Biofuel Production: Energy Biosciences
Institute Report, (2010).
[7] J.U. Grobbelaar, From laboratory to commercial production: a case study of a
Spirulina (Arthrospira) facility in Musina, South Africa, J. Appl. Physiol. 21 (2009)
523–527, http://dx.doi.org/10.1007/s10811-008-9378-5.
[8] M. Huesemann, T. Dale, A. Chavis, B. Crowe, S. Twary, A. Barry, et al., Simulation
of outdoor pond cultures using indoor LED-lighted and temperature-controlled ra-
ceway ponds and Phenometrics photobioreactors, Algal Res. 21 (2017) 178–190,
http://dx.doi.org/10.1016/j.algal.2016.11.016.
[9] P. Das, M.I. Thaher, M.A. Hakim, H.M. Al-Jabri, Sustainable production of toxin
free marine microalgae biomass as fish feed in large scale open system in the Qatari
desert, Bioresour. Technol. 192 (2015) 97–104, http://dx.doi.org/10.1016/j.
biortech.2015.05.019.
[10] Z. Cheng-Wu, O. Zmora, R. Kopel, A. Richmond, An industrial-size flat plate glass
reactor for mass production of Nannochloropsis sp. (Eustigmatophyceae),
Aquaculture 195 (2001) 35–49, http://dx.doi.org/10.1016/S0044-8486(00)
00533-0.
[11] J. Doucha, K. Lívanský, High Density Outdoor Microalgal Culture, in: R. Bajpai,
A. Prokop, M. Zappi (Eds.), Algal Biorefineries: Volume 1: Cultivation of Cells and
Products, Springer, Dordrecht, 2014, pp. 147–176.
[12] J. Doucha, K. Lívanský, Novel outdoor thin-layer high density microalgal culture
system: productivity and operational parameters, Arch. Hydrobiol. Algol. Stud.
Supplement Volumes 76 (1995) 129–147.
[13] I. Šetlík, V. Šust, I. Málek, Dual purpose open circulation units for large scale culture
of algae in temperate zones. I. Basic design considerations and scheme of a pilot
plant, Arch. Hydrobiol. Algol. Stud. Supplement Volumes 1 (1970) 111–164.
[14] J. Masojídek, Ivan Šetlík (1928–2009), J. Appl. Phycol. 21 (2009) 483–488, http://
dx.doi.org/10.1007/s10811-009-9450-9.
[15] J. Doucha, K. Lívanský, Productivity, CO2/O2 exchange and hydraulics in outdoor
open high density microalgal (Chlorella sp.) photobioreactors operated in a Middle
and Southern European climate, J. Appl. Physiol. 18 (2006) 811–826, http://dx.doi.
org/10.1007/s10811-006-9100-4.
[16] J. Doucha, K. Lívanský, Outdoor open thin-layer microalgal photobioreactor: po-
tential productivity, J. Appl. Phycol. 21 (2009) 111–117, http://dx.doi.org/10.
1007/s10811-008-9336-2.
[17] J. Masojídek, J. Kopecký, L. Giannelli, G. Torzillo, Productivity correlated to pho-
tobiochemical performance of Chlorella mass cultures grown outdoors in thin-layer
cascades, J. Ind. Microbiol. Biotechnol. 38 (2011) 307–317, http://dx.doi.org/10.
1007/s10295-010-0774-x.
[18] J. Masojídek, M. Sergejevová, J.R. Malapascua, J. Kopecký, Thin-layer systems for
mass cultivation of microalgae: flat panels and sloping cascades, in: A. Prokop,
R.K. Bajpai, M.E. Zappi (Eds.), Algal Biorefineries: Volume 2: Products and Refinery
Design, Springer International Publishing, Cham, 2015, pp. 237–261.
[19] C.G. Jerez, E. Navarro, I. Malpartida, R.M. Rico, J. Masojídek, R. Abdala, et al.,
Hydrodynamics and photosynthesis performance of Chlorella fusca (Chlorophyta)
grown in a thin-layer cascade (TLC) system, Aquat. Biol. 22 (2014) 111–122,
http://dx.doi.org/10.3354/ab00603.
A.C. Apel et al.
[20] V.V. Vieira, Microalgae Production Systems - The Cascade Raceways, European
Algae Biomass, Amsterdam, 2015.
[21] A.C. Apel, D. Weuster-Botz, Engineering solutions for open microalgae mass culti-
vation and realistic indoor simulation of outdoor environments, Bioprocess Biosyst.
Eng. 38 (2015) 995–1008, http://dx.doi.org/10.1007/s00449-015-1363-1.
[22] M. Glemser, M. Heining, J. Schmidt, A. Becker, D. Garbe, R. Buchholz, et al.,
Application of light-emitting diodes (LEDs) in cultivation of phototrophic micro-
algae: current state and perspectives, Appl. Microbiol. Biotechnol. 100 (2016)
1077–1088, http://dx.doi.org/10.1007/s00253-015-7144-6.
[23] W. Biggs, Radiation measurement, in: W.G. Gensler (Ed.), Advanced Agricultural
Instrumentation: Design and Use. Proceedings of the NATO Advanced Study
Institute on Advanced Agricultural Instrumentation, Il Ciocco (Pisa), Italy, May 27-
June 9, 1984, M. Nijhoff, Dordrecht, 1986, pp. 3–38.
[24] NREL, Terrestrial Reference Spectra for Photovoltaic Performance Evaluation
(ASTM G173–03), Available from: http://rredc.nrel.gov/solar/spectra/am1.5/,
(2012) [December 14, 2016].
[25] C.E. Pfaffinger, D. Schöne, S. Trunz, H. Löwe, D. Weuster-Botz, Model-based opti-
mization of microalgae areal productivity in flat-plate gas-lift photobioreactors,
Algal Res. 20 (2016) 153–163, http://dx.doi.org/10.1016/j.algal.2016.10.002.
[26] S. Boussiba, A. Vonshak, Z. Cohen, Y. Avissar, A. Richmond, Lipid and biomass
production by the halotolerant microalga Nannochloropsis salina, Biomass 12 (1987)
37–47.
[27] E.W. Becker, Microalgae: Biotechnology and Microbiology, Cambridge University
Press, Cambridge, 1993.
[28] J.H. de Vree, R. Bosma, M. Janssen, M.J. Barbosa, R.H. Wijffels, Comparison of four
outdoor pilot-scale photobioreactors, Biotechnol. Biofuels 8 (2015) 215, http://dx.
doi.org/10.1186/s13068-015-0400-2.
[29] T.S. Severin, S. Plamauer, A.C. Apel, T. Brück, D. Weuster-Botz, Rapid salinity
measurements for fluid flow characterisation using minimal invasive sensors, Chem.
Eng. Sci. 166 (2017) 161–167.
[30] W. Oswald, Large-scale algal culture systems (engineering aspects), in:
M.A. Borowitzka, L.J. Borowitzka (Eds.), Micro-algal Biotechnology, Cambridge
University Press, Cambridge, 1988, pp. 357–394.
[31] Laby Kaye, Tables of Physical & Chemical Constants, Available from: http://www.
kayelaby.npl.co.uk/general_physics/2_7/2_7_9.html, (February 14, 2017).
[32] R.H. French, Open-channel Hydraulics, McGraw-Hill, New York [u.a.], 1987.
[33] S.J. Edmundson, M.H. Huesemann, The dark side of algae cultivation: character-
izing night biomass loss in three photosynthetic algae, Chlorella sorokiniana,
Nannochloropsis salina and Picochlorum sp, Algal Res. 12 (2015) 470–476, http://
dx.doi.org/10.1016/j.algal.2015.10.012.
[34] E. Stengel, C.J. Soeder, Control of photosynthetic production in aquatic ecosystems,
in: J.P. Cooper (Ed.), Photosynthesis and Productivity in Different Environments,
Cambridge University Press, Cambridge, New York, 1975, pp. 645–660.
[35] M.A. Borowitzka, Culturing microalgae in outdoor ponds, in: R.A. Andersen (Ed.),
Algal culturing techniques, Elsevier/Academic Press, Burlington, 2005, pp.
390
Algal Research 25 (2017) 381–390
205–218.
[36] J. van Wagenen, T.W. Miller, S. Hobbs, P. Hook, B. Crowe, M.H. Huesemann, Effects
of light and temperature on fatty acid production in Nannochloropsis salina, Energies
5 (2012) 731–740, http://dx.doi.org/10.3390/en5030731.
[37] B. Crowe, S. Attalah, S. Agrawal, P. Waller, R. Ryan, J. van Wagenen, et al., A
comparison of Nannochloropsis salina growth performance in two outdoor pond
designs: conventional raceways versus the ARID pond with superior temperature
management, Int. J. Chem. Eng. 2012 (2012) 1–9, http://dx.doi.org/10.1155/
2012/920608.
[38] J.C. Quinn, T. Yates, N. Douglas, K. Weyer, J. Butler, T.H. Bradley, et al.,
Nannochloropsis production metrics in a scalable outdoor photobioreactor for
commercial applications, Bioresour. Technol. 117 (2012) 164–171, http://dx.doi.
org/10.1016/j.biortech.2012.04.073.
[39] N. Zou, C. Zhang, Z. Cohen, A. Richmond, Production of cell mass and eicosa-
pentaenoic acid (EPA) in ultrahigh cell density cultures of Nannochloropsis sp.
(Eustigmatophyceae), Eur. J. Phycol. 35 (2000) 127–133, http://dx.doi.org/10.
1080/09670260010001735711.
[40] A. Richmond, Biological principles of mass cultivation of photoautotrophic micro-
algae, in: A. Richmond, Q. Hu (Eds.), Handbook of Microalgal Culture: Applied
Phycology and Biotechnology, Wiley Blackwell, Chichester, 2013, pp. 171–204.
[41] K.L. Terry, L.P. Raymond, System design for the autotrophic production of micro-
algae, Enzym. Microb. Technol. 7 (1985) 474–487, http://dx.doi.org/10.1016/
0141-0229(85)90148-6.
[42] C.J. Warner, How Long Until Algal Biofuel is Commercially Relevant?: Achieving
Technology Readiness in the Near Term, (2012).
[43] S. Hintermayer, S. Yu, J.O. Krömer, D. Weuster-Botz, Anodic respiration of
Pseudomonas putida KT2440 in a stirred-tank bioreactor, Biochem. Eng. J. 115
(2016) 1–13, http://dx.doi.org/10.1016/j.bej.2016.07.020.
[44] L.J. Borowitzka, T.P. Moulton, M.A. Borowitzka, Salinity and the commercial pro-
duction of beta-carotene from Dunaliella salina, in: W.R. Barclay, R.P. McIntosh
(Eds.), Algal Biomass Technologies: An Interdisciplinary Perspective Proceedings of
a Workshop on the Present Status and Future Directions for Biotechnologies Based
on Algal Biomass Production, April 5–7, 1984, University of Colorado, Boulder, J.
Cramer, Berlin, 1986, pp. 224–229.
[45] K.K. Sharma, H. Schuhmann, P.M. Schenk, High lipid induction in microalgae for
biodiesel production, Energies 5 (2012) 1532–1553, http://dx.doi.org/10.3390/
en5051532.
[46] I. de Godos, J.L. Mendoza, F.G. Acién, E. Molina, C.J. Banks, S. Heaven, et al.,
Evaluation of carbon dioxide mass transfer in raceway reactors for microalgae
culture using flue gases, Bioresour. Technol. 153 (2014) 307–314, http://dx.doi.
org/10.1016/j.biortech.2013.11.087.
[47] M.D.M. Morales-Amaral, C. Gómez-Serrano, F.G. Acién, J.M. Fernández-Sevilla,
E. Molina-Grima, Outdoor production of Scenedesmus sp. in thin-layer and raceway
reactors using centrate from anaerobic digestion as the sole nutrient source, Algal
Research 12 (2015) 99–108, http://dx.doi.org/10.1016/j.algal.2015.08.020.