Nannochloropsis oceanica Cultivation in Pilot-Scale Raceway Ponds—From Design to Cultivation

: Raceways ponds are the microalgal production systems most commonly used at industrial scale. In this work, two di ﬀ erent raceway conﬁgurations were tested under the same processing conditions to compare their performance on the production of Nannochloropsis oceanica . Biomass productivity, biochemical composition of the produced biomass, and power requirements to operate those reactors were evaluated. Water depths of 0.20 and 0.13 m, and culture circulation velocities of 0.30 and 0.15 m s − 1 were tested. A standard conﬁguration, which had a full channel width paddlewheel, proved to be the most energy e ﬃ cient, consuming less than half of the energy required by a modiﬁed conﬁguration (had a half channel width paddlewheel). The later showed to have slightly higher productivity, not enough to o ﬀ set the large di ﬀ erence in energetic consumption. Higher ﬂow velocity (0.30 m s − 1 ) led to a 1.7 g m − 2 d − 1 improvement of biomass productivity of the system, but it increased the energy consumption twice as compared to the 0.15 m s − 1 ﬂow velocity. The latter velocity showed to be the most productive in lipids. A water depth of 0.20 m was the most suitable option tested to cultivate microalgae, since it allowed a 54% energy saving. Therefore, a standard raceway pond using a ﬂow velocity of 0.3 m s − 1 with a 0.20 m water depth was the most e ﬃ cient system for microalgal cultivation. Conversely, a ﬂow velocity of 0.15 m s − 1 was the most suitable to produce lipids.


Introduction
Microalgae are a diverse group of microorganisms that combine properties of higher plants with those of some prokaryotes [1,2]. They are ubiquitous, living either in aquatic (fresh and seawater) and terrestrial environments [3]. Microalgae occur as unicellular or simple colonial organisms, usually displaying rapid growth rates [4] and photosynthetic efficiencies between 9-10%, in general significantly higher than those of terrestrial plants [5][6][7]. They are key members of the biosphere, as they play an important role in the global carbon and nitrogen cycles. By using light energy to produce organic matter, algal photosynthesis contributes to around 50% of the total planetary primary production [8,9]. Cyanobacteria, previously called "blue-green algae", are even able to fix atmospheric nitrogen (N 2 ) into organic nitrogen and are the major contributors to global biological nitrogen fixation [10,11].
Being a biodiverse resource, microalgae can produce a wide range of value-added molecules such as proteins, lipids, carbohydrates, pigments and other compounds for various product raceway configuration with a minimum of three semicircular deflector baffles and an "island" at the end of the central divider, apart from being more energy-efficient than the standard configuration, can completely prevent the development of dead zones [32].
New design studies rely mostly on modeling and very few in actual experiments. Despite the evolution of the models and the associated technology, errors are always encountered when compared to experimental data [38,39]. In this context, this work aimed to compare the performance of two raceway ponds with different configurations based on three criteria: areal productivity, energy consumption, and the biochemical composition of biomass produced using Nannochloropsis oceanica as model microalga.

Materials and Methods
All experiments were performed between February 4th and July 28th of 2019 at Allmicroalgae's facilities at Pataias, Portugal. The microalga strain, N. oceanica CCAP 849/10, was obtained 'from Culture Collection of Algae and Protozoa and is kept at Allmicroalgae's culture collection.

Raceways Design, Commissioning and Construction
Based on the literature review and Allmicroalgae industrial experience in the growth of microalgae, two raceway pond prototypes were designed, as shown in Figure 1: one standard and one with a modified configuration. The standard raceway included the most common dimensions, configuration and construction materials reported, while the modified version considered several aspects previously reviewed from the literature and indicated as having a positive influence in the production of microalgal biomass. raceway configuration with a minimum of three semicircular deflector baffles and an "island" at the end of the central divider, apart from being more energy-efficient than the standard configuration, can completely prevent the development of dead zones [32].
New design studies rely mostly on modeling and very few in actual experiments. Despite the evolution of the models and the associated technology, errors are always encountered when compared to experimental data [38,39]. In this context, this work aimed to compare the performance of two raceway ponds with different configurations based on three criteria: areal productivity, energy consumption, and the biochemical composition of biomass produced using Nannochloropsis oceanica as model microalga.

Materials and Methods
All experiments were performed between February 4th and July 28th of 2019 at Allmicroalgae's facilities at Pataias, Portugal. The microalga strain, N. oceanica CCAP 849/10, was obtained `from Culture Collection of Algae and Protozoa and is kept at Allmicroalgae's culture collection.

Raceways Design, Commissioning and Construction
Based on the literature review and Allmicroalgae industrial experience in the growth of microalgae, two raceway pond prototypes were designed, as shown in Figure 1: one standard and one with a modified configuration. The standard raceway included the most common dimensions, configuration and construction materials reported, while the modified version considered several aspects previously reviewed from the literature and indicated as having a positive influence in the production of microalgal biomass.
The starting point was a typical raceway with two bends, as ponds with more than two bends have higher energy consumption [32]. To avoid the formation of dead zones as much as possible, three baffles were included in the circular endings of the ponds. Raceway ponds with around 28 m 2 were developed. They were 15 m long, 2 m wide and 1 m channel width. The depth was fixed at 0.375 m and other structural dimensions are listed in Table 1. The mixer consisted of a single eight-flatbladed paddlewheel installed right after one of the semicircular ends at the beginning of one of the channels.  The starting point was a typical raceway with two bends, as ponds with more than two bends have higher energy consumption [32]. To avoid the formation of dead zones as much as possible, three baffles were included in the circular endings of the ponds. Raceway ponds with around 28 m 2 were developed. They were 15 m long, 2 m wide and 1 m channel width. The depth was fixed at 0.375 m and other structural dimensions are listed in Table 1. The mixer consisted of a single eight-flat-bladed paddlewheel installed right after one of the semicircular ends at the beginning of one of the channels. The modified raceway model included a loop-shaped end of the central divider opposite to the paddlewheel. The 'loop' was intended to prevent the development of dead zones in the vicinity of this end of the divider. Besides, the paddlewheel hypothesized for the modified model had a width of 0.475 m, as compared to the standard model paddlewheel which measured 0.870 m. A picture of the raceways can be seen in Figure 2. Figure 3 shows a detailed view of the modified raceway.
Appl. Sci. 2020, 10, x FOR PEER REVIEW 4 of 14 The modified raceway model included a loop-shaped end of the central divider opposite to the paddlewheel. The 'loop' was intended to prevent the development of dead zones in the vicinity of this end of the divider. Besides, the paddlewheel hypothesized for the modified model had a width of 0.475 m, as compared to the standard model paddlewheel which measured 0.870 m. A picture of the raceways can be seen in Figure 2. Figure 3 shows a detailed view of the modified raceway.  The raw materials used for the construction of both raceways were the same. Polyethylene film and PVC waterproofing geomembrane were used in the base of the raceways to waterproof them. The structure was supported by stone powder and stainless steel and external walls were made of treated wood.

Trials
Each trial lasted 13 days and was performed simultaneously in both raceways. N. oceanica was inoculated at a 0.3 g L −1 concentration in Allmicroalgae's base medium, based on Guillard's F/2 medium [37], with 30 g L −1 NaCl and 0.31 g L −1 NO3 − . In all experiments, a fresh inoculum sequentially scaled up, was used. Initially, the cultures were grown in 5 L flasks during 7 to 15 days. When the culture reached 1.2 g L -1 , 20 × 5 L flasks of culture (100 L) were used to inoculate a single flat panel (FP), which was later divided into the other three FPs. These were used to inoculate the two raceway ponds. Throughout the assay, the culture was closely monitored. The pH was kept between 8.0-8.2, through the injection of pure CO2 to the culture. The NO3 − concentration was always kept above 0.12 g L −1 and water refilled whenever it was necessary. Regular microscopic observations were performed to monitor contaminations. At the end of the assay the culture was harvested and centrifuged at 4000 g for 30 min, and the resulting biomass and supernatant were biochemically analyzed. All the assays were made in duplicate.

Water Depth
Two different water depths were tested, 0.20 and 0.13 m, with a flow velocity of 0.30 m s −1 .

Flow Velocity
The flow velocity optimization trial was performed using the standard raceway. A velocity of 0.15 m s −1 was used with a 0.13 m water column.

Growth Assessment
Samples (50 mL) were taken on a daily basis and microscopic observation was performed in order to monitor contaminations, which were never above 2% of cell counts.
Optical density (OD) and dry weight (DW) were measured to follow microalgal growth. OD was performed at 540 nm using a spectrophotometer (4251/50, Zuzi, Spain). DW was determined using a 0.7 µ m glass fiber filter (VWR, Portugal) in a moisture analyzer (MA 50.R, RADWAG, Poland). In order to neutralize the salt weight, the culture was washed with an equivalent volume of ammonium formate solution at 35 g L −1 .
Areal biomass productivity, P (g L −1 d −1 ), was determined by Equation (1): The raw materials used for the construction of both raceways were the same. Polyethylene film and PVC waterproofing geomembrane were used in the base of the raceways to waterproof them. The structure was supported by stone powder and stainless steel and external walls were made of treated wood.

Trials
Each trial lasted 13 days and was performed simultaneously in both raceways. N. oceanica was inoculated at a 0.3 g L −1 concentration in Allmicroalgae's base medium, based on Guillard's F/2 medium [40], with 30 g L −1 NaCl and 0.31 g L −1 NO 3 − . In all experiments, a fresh inoculum sequentially scaled up, was used. Initially, the cultures were grown in 5 L flasks during 7 to 15 days. When the culture reached 1.2 g L -1 , 20 × 5 L flasks of culture (100 L) were used to inoculate a single flat panel (FP), which was later divided into the other three FPs. These were used to inoculate the two raceway ponds. Throughout the assay, the culture was closely monitored. The pH was kept between 8.0-8.2, through the injection of pure CO 2 to the culture. The NO 3 − concentration was always kept above 0.12 g L −1 and water refilled whenever it was necessary. Regular microscopic observations were performed to monitor contaminations. At the end of the assay the culture was harvested and centrifuged at 4000× g for 30 min, and the resulting biomass and supernatant were biochemically analyzed. All the assays were made in duplicate.

Water Depth
Two different water depths were tested, 0.20 and 0.13 m, with a flow velocity of 0.30 m s −1 .

Flow Velocity
The flow velocity optimization trial was performed using the standard raceway. A velocity of 0.15 m s −1 was used with a 0.13 m water column.

Growth Assessment
Samples (50 mL) were taken on a daily basis and microscopic observation was performed in order to monitor contaminations, which were never above 2% of cell counts.
Optical density (OD) and dry weight (DW) were measured to follow microalgal growth. OD was performed at 540 nm using a spectrophotometer (4251/50, Zuzi, Spain). DW was determined using a 0.7 µm glass fiber filter (VWR, Portugal) in a moisture analyzer (MA 50.R, RADWAG, Poland). In order to neutralize the salt weight, the culture was washed with an equivalent volume of ammonium formate solution at 35 g L −1 .
Areal biomass productivity, P (g L −1 d −1 ), was determined by Equation (1): where X 2 (g L −1 ) and X 1 (g L −1 ) represented the DW times t 1 (s) and t 2 (s), respectively, V (m 3 ) was the raceway volume and A (m 2 ) the ground area occupied by the bioreactor. Specific growth rate, µ (d −1 ), was expressed as a ratio between the difference final and initial logarithmic biomass concentrations (X 2 and X 1 , respectively) and the associated time range (t 2 and t 1 , respectively), as seen in Equation (2).
Global productivity and growth rate refer to the whole assay (13 days). Maximum productivity and growth rate were calculated based on the highest values obtained daily.
The lipid production/energy consumption ratio (L/kW) was calculated through the following equation: where L (%) was the lipid content, 13 days the total extension of the assay, and p the power consumption.
where rad (kW m −2 ) was the radiation that reaches the reactor, t (s) was the time between solar radiation measurement and HHV was the higher heating value, calculated based on elemental (CHN) analysis by Equation (5), as described by Callejón-Ferre et al. [42]: A spectrophotometric analysis (Genesys 10S UV-VIS spectrophotometer (Thermo Scientific, Massachusetts, USA)) was used to measure the nitrate concentration of the medium. Samples were centrifuged for 10 min at 2000 g (MiniStar Silverline, VWR, Portugal). The OD of the sample was calculated from the difference at 220 nm and twice the value at 275 nm and compared with a calibration curve with sodium nitrate [43]. The blank solution was made with 0.1 mol L −1 of HCl.

Proximate Composition
The ash content was determined in a furnace (J. P. Selecta, Sel horn R9-L) at 550 • C for 8 h. Total lipids were determined through a gravimetric analysis, using a previously modified method of Bligh and Dyer (1959) [44], and reported by Pereira et al. [45]. A biomass weight ranging 5-10 mg was solubilized in 0.8 mL of distilled water. A mixture of methanol (2 mL), chloroform (2 mL) and distilled water (1 mL) was used to extract the lipid fraction from the biomass [44]. After centrifugation for 10 min at 2500 g, 0.7 mL of the chloroform fraction was transferred to a pre-weighed tube. A heat treatment at 60 • C (QBD2, Grant Instruments, UK) to evaporate all the chloroform was performed and after 3 h in the desiccator, the final weight of the tube was measured in the precision balance (M5P, Sartorius, Germany).
CHN analysis was performed using a Vario el III (Vario EL, Elementar Analyser system, Germany), according to Quelhas et al. [41]. Total proteins were calculated by multiplying the nitrogen content by the standard conversion factor 6.25 [46] after CHN analysis. Carbohydrates were considered the remaining mass.

Fatty Acid Profile
Fatty acids were determined by means of its conversion to fatty acid methyl esters (FAME), according to a modified method of Lepage and Roy [47] described by Pereira et al. [48]. FAME were analyzed in a gas chromatography-mass spectrometer (GC-MS) (Bruker SCION 456/GC, SCION TQ MS). The gas chromatograph was equipped with a 30 m length column with an internal diameter of 0.25 mm and a film thickness of 0.25 µm (ZB-5MS, Phenomenex, Torrance, California, US). Helium was the carrier gas. The temperature program was 60 • C (1 min), 30 • C min −1 to 120 • C, 5 • C min −1 to 250 • C, and 20 • C min −1 to 300 • C (2 min). The total running time was 31.5 min. Injection temperature was 300 • C. Supelco ® 37 Component FAME Mix (Sigma-Aldrich, Sintra, Portugal) were used as standards.
After identified and quantified, the FAME profile was expressed as a relative percentage of the total fatty acids content present in the biomass.

Comparison of Two Different Raceway Configurations
N. oceanica growth experiments were performed under three different sets of process conditions ( Table 2) in order to determine which of the two raceway configurations were the most promising in terms of productivity, product quality, and energy consumption. Growth kinetics were evaluated separately for each set of process conditions (growth curves are plotted in Supplementary information, Figures 2 and 3). The global areal productivities obtained (Table 3) were higher in the modified raceway for process conditions 1 and 2. Global specific growth rates (Table 3) were similar in both raceways for both the process conditions tested. The range of productivities achieved was half of the value achieved in a demo-scale 530 m 2 raceway ponds using N. oceanica F&M-M24 strain (14 g m −2 d −1 ) [49]. de Vree et al. reported average areal productivity of 9.7 g m −2 d −1 for Nannochloropsis sp. CCAP 211/78 growth in a raceway pond similar to the one used during the present experiments (25.4 m 2 surface area) with 0.20 m culture depth [50]. In the same study, productivity ranged from 6.2 to 14 g m −2 d −1 . However, López et al. [36] reported a productivity of 6.0 ± 2.5 g m −2 d −1 for irradiation levels identical to the ones occurred during the present trials, for Nannochloropsis gaditana cultivated in raceways of 23.5 m long and 1.5 m wide channels, with a culture depth of 0.08 to 0.12 m (depending on the season) and 0.30 m s −1 circulation velocity. The same author reported considerably higher productivity (24.71 g m −2 d −1 ) for trials performed under higher average irradiation. In a similar experiment operating in continuous mode, 22.4 g m −2 d −1 were obtained for Nannochloropsis gaditana when grown in 7.2 m 2 raceway ponds (5.0 m long and 0.6 m wide channels) with a culture depth of 0.11 m and a circulation velocity of 0.30 m s −1 [51]. Besides, 20.37 g m −2 d −1 was the average areal productivity achieved in a culture of Nannochloropsis sp. cultivated in 25,000 L (dimension of 35 m by 3.5 m) outdoor raceway ponds situated in the Qatari desert [52]. The global productivities achieved are comparable with the average annual productivity of 20 g m −2 d −1 rarely exceeded in large-scale production in outdoor raceways, since the systems are affected by factors that cannot be controlled, e.g., weather conditions [36]. The specific growth rates obtained are lower than the range of previously reported data (0.11-0.21 d −1 ) [53]. Nevertheless, the values reported by Baohua et al. [54] for three different N. oceanica strains grown in 8000 L raceway ponds are identical to the ones obtained in this work (0.047 ± 0.001 d −1 , 0.066 ± 0.005 d −1 , and 0.051 ± 0.003 d −1 ).
Photosynthetic efficiency (PE) ( Table 4) was higher in the modified raceway for process conditions 1 and 2. The higher rotation frequency needed in the modified raceway to attain a given average flow velocity caused a higher turbulence level, enhancing the vertical mixing and the PE of the whole culture. This represents a possible cause for the higher productivities reached in the modified raceway.  There were minor differences in the proximate composition (Table 5) and fatty acid profile (Table 6) of the biomass produced in both raceways. The lipid content was lower than the values reported by Das et al. [52] (26.0% ± 1.1%) and Baohua et al. [54] (21.40% ± 1.53%-28.00% ± 1.00%), for Nannochloropsis sp. cultivated in outdoor raceway ponds and three different N. oceanica strains cultivated in indoor raceway ponds, respectively. On the other hand, the values reported by López et al. [36] (12.80% ± 3.54%-15.15% ± 4.16%) are in the range of the ones achieved in these trials. However, despite the lipid accumulation in Nannochloropsis sp. being strain and species-dependent, it is also conditioned by the culture conditions, varying both with nutritional and environmental factors such as irradiance, nitrogen concentration and salinity [20]. Regarding the fatty acid profile, C16 and C20 FAME were the most prevalent. Ma et al. [20] reported that C16, C18, and C20 FAME were the major fatty acids found in Nannochloropsis strains, contrary to the present study on which C18 were detected in very small amounts. EPA content ranged from 26.17% ± 3.32% to 29.45% ± 0.41% of total FAME, which was lower than the values reported by Baohua et al. [54] for three different N. oceanica strains cultivated in indoor raceways (35.58% ± 0.63%, 38.78% ± 2.49%, 40.13% ± 0.28%). The higher relative content of PUFA and omega-3 PUFA (ω-3) in comparison with saturated fatty acids (SFA) and omega-6 PUFA (ω-6), respectively, are interesting indicators of the quality of biomass produced by the N. oceanica strain used.
Concerning energy consumption, the modified raceway consumed, at least, twice the energy consumed by standard configuration ( Table 7). The non-uniform velocity pattern generated with a half channel width paddlewheel mixing was responsible for the discrepancy. The typical plug flow of raceways [34] was altered in the modified and zones wherein an inversion of flow direction occurred. Thus, a higher rotation frequency was needed to attain a given circulation velocity. Despite having much higher operational costs, the modified configuration presented slightly higher biomass productivity. Given the negligible difference in the productivity obtained, the standard configuration was found to be the best alternative of design.

Water Depth Optimization
Optimization of culture depth was based on the comparison of trials performed under process conditions 1 and 2. An increase of 0.8 g m −2 d −1 in the global biomass productivity with the 0.13 m water depth compared with the 0.20 m water depth was observed. These results are in agreement with those obtained by San Pedro et al. [51] and de Vree et al. [50], who reported that large pond depths lead to light-limited cells and the creation of large dark zones wherein microalgae do not photosynthesize, reducing the net productivity of the culture. Dark respiration can comprise as much as 30% of the strain's maximum specific growth rate [55] and can explain the small decrease obtained in the global specific growth rate with the increase of the water depth.
The quality of biomass produced in both process conditions remained the same in terms of quantity of lipids and EPA. PUFA/SFA and ω-6/ω-3 ratios were also identical for both process conditions. On the other hand, we verified a 54% increase in energy consumption with process condition 2, as a result of higher paddlewheel rotation speed. A power consumption increase between 0.20 m and 0.10 m water depth trials was reported by Mendoza et al. [56]. These authors showed that a 0.20 m water column is the optimum value for minimizing power consumption. Chiaramonti et al. described an increase in total head losses caused by a reduction in culture depths [57].
Thus, a decrease in water depth induced a small improvement of global biomass productivity, however, the power consumption was the differentiator factor. The lower energy required with a pond depth of 0.20 m made it the best option for microalgal cultivation.

Flow Velocity Optimization
Optimization of flow velocity was based on the comparison of trials performed under process conditions 2 and 3. An increase in global biomass productivity and specific growth rate was associated with an increase in the flow velocity. Accordingly, Moazami et al. reported that an increase in the paddlewheel rotation speed from 1.4 to 2.1 rad s −1 resulted in a biomass productivity increase [58]. However, Weissman et al. reported that different mixing velocities had no apparent effect on biomass productivity [59].
Despite the EPA content and, ω-6/ω-3 and PUFA/SFA ratios being similar, the same was not seen for the lipid content. This parameter was about 33% higher for a culture velocity of 0.30 m s −1 . According to Ma et al. (2016), lipid accumulation is favored under stressed conditions. Thus, 0.30 m s −1 may induce mechanical stress on the cells. Nevertheless, as previously mentioned, this is the most commonly used flow velocity for biomass production [20].
Specific power consumption was two-fold higher with a 0.30 m s −1 circulation velocity. This was due to the two-fold higher paddlewheel rotation speed needed to double the flow velocity. Mendoza et al. [56] and Weissman et al. [59] also reported an increase in the specific power consumption with liquid velocity. The former authors refer to an exponential relationship between these two parameters due to the exponential relationship between the liquid velocity and its kinetic energy [59].
In order to determine which was the most suitable flow velocity to cultivate N. oceanica microalgae, the ratio of lipid production/energy consumption was calculated. The lower velocity (0.15 m s −1 ) displayed a higher ratio, 17.1 kg kW −1 , when compared to 0.30 m s −1 flow velocity, 16.9 kg kW −1 . Therefore, 0.15 m s −1 was considered to be the best option tested to induce lipid production.

Conclusions
A comparison of two different pilot-scale raceway pond configurations and an optimization of the water depth and flow velocity were performed. These were evaluated in terms of the growth dynamics, biochemical composition of the used microalga, and power consumption requirements. The results obtained suggest that the improvement in the biomass productivity achieved in the modified when compared to the standard raceway was not significant to compensate for the much higher energy consumption needed. For the same reason, it was concluded that a 0.20 m was the best water depth tested for N. oceanica cultivation. The highest lipid production/energy consumption was obtained when using 0.15 m s −1 , being this velocity the most suitable for lipid production.
Supplementary Materials: The following are available online at http://www.mdpi.com/2076-3417/10/5/1725/s1, Figure S1: Calibration curve of optical density at 540 nm as a function of N. oceanica biomass dry weight, Figure S2: Growth curves of N. oceanica cultivated under process conditions 1. Values represent an average of two biological replicates, Figure S3: Growth curves of N. oceanica cultivated under process conditions 2. Values represent an average of two biological replicates, Figure S4: Growth curve of N. oceanica cultivated under process conditions 3. Values represent an average of two biological replicates, Figure S5: Temperature and instant radiation registered on site during first growth trial under process conditions 1, Figure S6: Temperature and instant radiation registered on site during second growth trial under process conditions 1, Figure S7: Temperature and instant radiation registered on site during first growth trial under process conditions 2, Figure S8: Temperature and instant radiation registered on site during second growth trial under process conditions 2, Figure S9: Temperature and instant radiation registered on site during first growth trial under process conditions 3, Figure S10: Temperature and instant radiation registered on site during second growth trial under process conditions 3, Figure S11