- Email: [email protected]
Nannochloropsis production metrics in a scalable outdoor photobioreactor for commercial applications
Bioresource Technology 117 (2012) 164–171
Contents lists available at SciVerse ScienceDirect
Bioresource Technology journal homepage: www.elsevier.com/locate/biortech
Nannochloropsis production metrics in a scalable outdoor photobioreactor for commercial applications Jason C. Quinn a, Tracy Yates b, Nathaniel Douglas b, Kristina Weyer b, Joel Butler b, Thomas H. Bradley a, Peter J. Lammers b,1,⇑ a b
Mechanical Engineering, 1374 Campus Delivery, Colorado State University, Fort Collins, Colorado 80523-1374, USA Solix BioSystems, Inc., 430 B North College Ave, Fort Collins, CO 80524, USA
h i g h l i g h t s 1 d 1 during calendar years of 2009 & 2010. d 1. 3 1 1 " Average bio-oil productivity of 10.7 m ha yr with peak of 36.3 m3 ha 1 yr 1. " Serial batch cultures maintained for >421 days of continuous cultivation. " Energy consumption sensitivity analysis for large-scale photobioreactor system.
" Growth rates averaged 0.16 g L " Peak growth rate of 0.37 L
a r t i c l e
1
i n f o
Article history: Received 7 January 2012 Received in revised form 18 April 2012 Accepted 20 April 2012 Available online 26 April 2012 Keywords: Microalgae Photobioreactor Nannochloropsis Biomass Lipids
a b s t r a c t Commercial production of renewable energy feedstocks from microalgae will require reliable and scalable growth systems. Two and one half years of biomass and lipid productivity data were obtained with an industrial-scale outdoor photobioreactor operated in Fort Collins, Colorado (USA). The annualized volumetric growth rates for Nannochloropsis oculata (CCMP 525) and Nannochloropsis salina (CCMP 1776) were 0.16 g L 1 d 1 (peak = 0.37 g L 1 d 1) and 0.15 g L 1 d 1 (peak = 0.37 g L 1 d 1) respectively. The collective average lipid production was 10.7 m3 ha 1 yr 1 with a peak value of 36.3 m3 ha 1 yr 1. Results from this study are unique based on publication of biomass and corresponding lipid content combined with demonstration of energy savings realized through analysis of gas delivery requirements, water recycling from successive harvests with no effect on productivity, and culture stability through serial batch lineage data and chemotaxonomic analysis of fatty acid contents. Ó 2012 Elsevier Ltd. All rights reserved.
1. Introduction Relative to first-generation biofuel feedstocks, microalgae are characterized by higher solar energy yield, the potential for yearround cultivation in many locations, the ability to grow in brackish and saline water as well as water produced from oil and gas extraction, higher areal productivities than oil seed crops and the utilization of non-arable land (Batan et al., 2010; Chisti, 2007; Li et al., Abbreviations: AGS, Algae Growth System; CCMP, Center for Culture of Marine Phytoplankton; DW, Dry weight; FAME, Fatty acid methyl ester; GC, Gas chromatograph; LCA, Life Cycle Analysis; VVM, Volume of air per volume of culture per minute. ⇑ Corresponding author. Tel.: +1 575 646 7458; fax: +1 575 646 5717. E-mail address: [email protected] (P.J. Lammers). 1 Present address: Energy Research Laboratory, Box 30001, MSC 3RES, New Mexico State University, Las Cruces, NM, 88003, USA. Tel.: +1 (575) 646-7458; fax: +1 (575) 646-5717. 0960-8524/$ - see front matter Ó 2012 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.biortech.2012.04.073
2008; Williams et al., 2009). Microalgae can grow in poor-quality water, utilize CO2 from point sources such as coal fired power plants and utilize nutrients from wastewater treatment plants with a uniquely high productivity potential (Chisti, 2008a; Li et al., 2008; Schenk et al., 2008; Wijffels and Barbosa, 2010). Current lab-scale experimental data have suggested a near-term realizable production an order of magnitude higher than the current productivity of ethanol from corn and bio-diesel from soy: 2533 liters hectare 1 yr 1 (271 gal acre 1 yr 1) of ethanol from corn and 584 L ha 1 yr 1 (62.5 gal acre 1 yr 1) of bio-diesel from soybeans, respectively (Chisti, 2007). Microalgae cultivation is typically done in, open raceway ponds or photobioreactor systems. The two primary advantages of photobioreactors are increased culture stability resulting from lower likelihood of pathogen or grazer infestations and higher volumetric productivities due to differences in light utilization. These advantages have led to an interest in photobioreactor technology for the
J.C. Quinn et al. / Bioresource Technology 117 (2012) 164–171
cultivation of inoculums for large raceway ponds and quality-controlled production of high-valued products such as nutraceuticals (Wijffels and Barbosa, 2010). The technical and economic evaluation of photobioreactors and open raceway pond technologies at commercial scales is difficult and very expensive to perform (Hu et al., 1996). Previous studies have reported monthly productivities from outdoor large-scale systems but inconsistencies in experimental setup, cultured species and time scales make utilization of these data sets challenging (Benemann and Oswald, 1996; Bosma et al., 2007; Sheehan et al., 1998; Weissman and Tillett, 1990). Large-scale data has been restricted to open raceway pond systems with a complete lack of large-scale productivity data from photobioreactors over a time period greater than one growing season. Due to limited productivity data at large-scale, major simplifying assumptions have been made in the assessment of microalgae productivity potential. Large-scale productivity data are needed to inform and advance the accuracy of life cycle analyses (LCA) and techno–economic modeling of second and third generation biofuel feedstocks which are the primary tools used to assess and inform policy makers and funding agencies. Recent LCA and techno–economic modeling efforts have examined the sustainability, economic, and environmental benefits of microalgae with conclusions from these studies varying significantly based on fundamental assumptions in the modeling efforts (Batan et al., 2010; Benemann and Oswald, 1996; Campbell et al., 2011; Clarens et al., 2010; Davis et al., 2011; Frank et al., 2011; Jorquera et al., 2010; Lardon et al., 2009; Stephenson et al., 2010). The microalgae assessments of Batan et al. (2010), Lardon et al.(2009), Hirano et al. (1998), Davis et al. (2011), Frank et al. (2011) and Campbell et al. (2011) used fixed biomass productivity models between 10 and 30 g m 2 d 1 (3.6104– 11.0 104 kg ha 1 yr 1) based on laboratory-scale data. Extrapolation of laboratory data has been dictated by the infancy of the microalgae biofuels industry and the uncertainty of the current large-scale biooil productivity potentials reported to range from 8.2 m3 ha 1 yr 1 (Scott et al., 2010) to 136.9 m3 ha 1 yr 1 (Chisti, 2007) and values between these two extremes (Chisti, 2008a, 2008b; Mata et al., 2010; Rodolfi et al., 2009; Schenk et al., 2008; Sheehan et al., 1998; Wijffels and Barbosa, 2010). In the present study biomass and lipid productivity data from a large-scale microalgae photobioreactor cultivation facility operated in Fort Collins, Colorado at the research and development facility of Solix BioSystems are presented. Two oleaginous microalgal strains were evaluated, Nannochloropsis oculata and Nannochloropsis salina. Furthermore, two generations of the Solix bioreactor, AGS 3 and 4 are compared. Detailed biomass and fatty acid methyl ester based lipid productivity data on peak and annual average are reported. These data facilitate techno–economic and life cycle modeling of microalgal based biofuel and bioproduct processes utilizing photobioreactor technology in stand-alone or integrated photobioreactor-open raceway pond configurations. Discussion focuses on a comparison of the performance of the system presented to previously published data, assessment of the stability of Nannochloropsis strains in long-term successive batch operations, energy requirements for the growth system with comparison to tradition open raceway ponds, and operational tradeoffs between maximizing growth rates versus bio-oil content.
2. Methods
165
1776). From September 2009 to September 2010 f/2 growth medium was modified to contain 5 mM NO3 L 1 and 0.368 mM PO4 at a salinity of 20 g L 1. The salinity of the growth media was increased to 27 g L 1 based on operational decisions made from October 2009 to January 2011 while all other components were held constant. The medium was filtered (0.2 micron filter) into a tank with the required inoculums where it was mixed to ensure homogeneity prior to inoculation of the photobioreactor system. Spent water recycling experiments were performed over four successive batch cultures by using 80% water from the harvest centrifuge with 20% fresh artificial seawater (Instant Ocean) at 27 g L 1 during the summer of 2010. N. oculata was cultivated in the AGS 3 photobioreactors from May 2008 to June 2009. N. salina was subsequently cultivated in AGS 3 and AGS 4 photobioreactors starting in January 2009. Operations were conducted in a serial batch mode with a portion of each harvest used to start the next batch. Harvest densities ranged between 2 and 3 gDW L 1. Some batches were harvested at lower densities due to low growth rates in low light periods during the winter. A select number of batches were inoculated with as low as 0.25 gDW L 1 and harvested at 6 gDW L 1 as detailed in the Section 3. The cultures were occasionally restarted from laboratory cultures using sterile techniques: Tube cultures (5 mL) were scaled through 125 and 250 mL Erlenmeyer flasks maintained in a 24 h illuminated incubator at 150 lmol photons m 2 s 1 at 25 °C. These cultures were transferred to 4.0 L Fernbach flasks on a shaker table under greenhouse conditions (non-sterile techniques from this scale onward) and transferred into a variable volume flat-panel bioreactor (20–60 L) maintained in a greenhouse. Final scale-up was then conducted in the AGS 3 or AGS 4 panels. Inocula for routine batch cultivation was derived from the previous harvest, after dilution with fresh nutrient media to the re-inoculation density of 1 gdw L 1. 2.2. Outdoor culture system 2.2.1. Photobioreactor geometry The AGS 3 and AGS 4 photobioreactor utilize a shallow water basin for thermal and structural support for vertically oriented, polyethylene growth panels, 0.05 m wide by 0.28 m high by 17.3 m long. Detailed photos of the outdoor facility are presented in supplementary data. The AGS 3 and AGS 4 photobioreactor systems maintained panel spacing of 0.15 m. The structure supporting the panels in the temperature-controlled basin was the only change from the AGS 3 to AGS 4 technology. The AGS 3 photobioreactors were deployed from April 2008 to December 2009. The AGS 4 support structure permits deployment in natural bodies of water via flotation which decreases the land grading requirement for terrestrial deployment. The AGS 4 photobioreactor panels were deployed in December 2009 and are currently being utilized in Fort Collins, Colorado at the Solix BioSystems headquarters and at the 0.4 hectare pilot plant facility outside of Durango, Colorado at Coyote Gulch Durango, Colorado. Both AGS 3 and AGS 4 photobioreactor panels were constructed with polyethylene. The standard inoculation volume corresponded to a volume to area ratio for the data presented of 49.6 L m 2 with a standard deviation of 19.9 L m 2. The large standard deviation is due to experimentation centered on the optimization of reactor volume. Schematic details of the growth chamber for the AGS3 and AGS 4 photobioreactor panels can be found in Fig. 1.
2.1. Organisms, culture conditions, and inoculation protocols 2.1.1. Nannochloropsis strains and growth medium Cultures were obtained from the Provasoli–Guillard National Center for Culture of Marine Phytoplankton (CCMP) by Solix BioSystems, specifically N. oculata (CCMP 525) and N. salina (CCMP
2.2.2. Photobioreactor mixing, pH, and temperature control Mixing for AGS 3 and AGS 4 photobioreactors was provided through a sparge air system that was operated during daylight hours. A mixing rate of 2.5 VVM for the AGS 3 photobioreactor and at 0.5 VVM for AGS 4 photobioreactors was used. The decrease
166
J.C. Quinn et al. / Bioresource Technology 117 (2012) 164–171
Fig. 1. Schematic drawing of the AGS 3 and AGS 4 photobioreactor growth panel. Differences between the AGS 3 and AGS 4 photobioreactors are restricted to the support structure which does not affect the optical properties of the photobioreactor.
in sparge rate from the AGS 3 to the AGS 4 photobioreactors was due to improvements in sparge-hole geometry supported by particle image velocimetry and growth experimentation. The pH of the system was monitored continuously with probes embedded in the AGS3 and AGS4 photobioreactors (Hach, model PC1R2A). The pH was maintained by CO2 supplied into the sparge air and delivered to the system with a duty cycle determined by pH feedback control (pH maintained at 7.3 ± 0.1) or by continually mixing 2% CO2 with sparge air (pH maintained at 7.3 ± 0.5). The temperature of the culture was maintained by the thermal mass of water basin which also supplied the structural support for the AGS 3 and AGS 4 photobioreactors. The temperature was continuously monitored and maintained between 19 and 26 °C via a Marley evaporative cooling system with a capacity of 270,000 BTU or a Jandy Lite2 pool heater with a capacity of 325,000 BTUH. Average temperatures were close to 25 °C for the majority of the batches. Microalgae biomass was harvested with a Sharpless AS26 continuous-flow centrifuge operating at 15,000 rpm and a flow rate of 9 L min 1. 2.3. Growth monitoring Samples of the cultures were monitored daily to track growth using optical density at 750 nm (OD750). Total fatty acid content (based on FAME determination), dry mass and salinity were measured at the end of each batch. Samples were drawn using a 10 mL syringe through sample lines attached to sample ports at the head of the reactors. Previous sampling experimentation showed that sampling location did not affect experimental results due to the homogeneity of the culture. Optical density measurements were performed on a Hach DR5000 spectrophotometer and the OD750 converted to gdw L 1 utilizing an empirically determined correlation factor (N > 100, R2 = 0.99989). Evaporative losses were integrated in final growth measurements through previously determined loss rates (Quinn et al., 2011). Nitrate and phosphate levels were routinely measured in culture filtrates using standard colorimetric methods and reagents (Hach). The level of both nutrients decreased to below detection limits within 4 days of inoculation in the AGS3 and AGS4 system regardless of season.
batch comparisons with other studies. Lipid fractions were determined using an in situ trans-esterification technique based on the methods of Schutter and Dick (2000): Five mg of microalgae sample centrifuged at 4000g for 5 min followed by removal of the supernatant. An auto-pipette was used to dispense 2.5 ml of 0.2 N KOH in methanol onto the 5 mg microalgae pellet. Samples were pipette mixed and transferred to a glass test tube previously washed in 1% HCl acid. An additional 2.5 ml of 0.2 N KOH in methanol was added and pipette mixed. Samples were aggressively mixed using a VWR Analog vortex mixer at a speed setting of 10 for 20 s followed by heating to 37 °C for 30 min. One ml of acetic acid and 2 ml of HPLC grade heptane was added and the samples were aggressively mixed by using a VWR Analog vortex mixer on a speed setting of 10 for 20 s and then centrifuged at 2000g for 5 min. The organic layer was removed and processed in a gas chromatograph (GC) to determine lipid content and composition. The single heptane extraction reproducibly removed 80% of the overall lipid content in the cells as determined by standard addition methods using C15:0 triacylglycerol. Samples were quantified based on normalization to recovery of a 23:0 FAME internal standard against a standard curve. Transesterified samples were prepared for GC analysis by first diluting the sample 1:10 with heptane. An internal standard (23:0) obtained from NU-CHEK PREP, Inc. was added to the sample and the head space was filled with nitrogen. Samples were analyzed with an Agilent Technologies 7890A GC instrument utilizing a 30 m 0.32 mm 0.25 lm Restek FAMEWAX column. A spitless injection was used requiring 1 lL of sample. Helium at 1.5 mL min 1 was used as carrier gas. The oven was operated at 90 °C for 0.5 min and then ramped to 208 °C at 70 °C min 1, then ramped to 230 °C at 3 °C min 1, and finally to 240 °C ramped at 2 °C min 1 and held for 1 min. Prior to running samples, a blank was run followed by the generation of a four-point standard curve using a GLC-461 standard obtained from NU-CHEK PREP, Inc. Areal productivity was calculated based on the productivity of the batch, FAME content, an assumed FAME density of 918 kg m 3, operation of 365 days a year, and with the area reported as photosynthetic area.
2.4. Fatty acid methyl ester (FAME) assay
2.5. Estimates of culture composition by fatty acid distributions and flow cytometry
Assays of lipid content based on FAME determination were performed on material collected at the end of each batch with the results extrapolated to calculate the overall annual oil production for
Compositional analysis of samples was based on comparison of fatty acid distributions with known values and the use of flow cytometry. Fatty acid distributions as percent of total FAME were re-
J.C. Quinn et al. / Bioresource Technology 117 (2012) 164–171
corded for 706 batch cultures of N. salina. The chemotaxonomic value of these data is widely appreciated (Khozin-Goldberg and Boussiba, 2011). Some samples were also analyzed by flow cytometry on a GuavaCyte instrument (Millipore) to determine chlorophyll-positive cell counts. The chlorophyll content per cell count in the instrument was estimated as the chlorophyll-fluorescence emission measured in the red-channel (646–692 nm) using the standard band-pass filters supplied with the instrument after laser excitation at 488 nm. A plot of chlorophyll-fluorescence versus forward scatter for 10,000 cells provided an estimate of culture composition. The smaller Nannochloropsis populations (3–4 micron diameter) map to a separate region of the plot from the signals associated with cells from a common invader species, Tetraselmis, that produced more forward scatter due to larger diameter (>8 microns) and higher in per-cell chlorophyll (data presented in supplementary data). 3. Results and discussion 3.1. Large-scale outdoor productivity The data shown corresponds only to the operation of the reactors in production mode using a single nutrient formulation and excludes data collected during periods of experimentation. The results reported are representative of those obtained with the same AGS3 and AGS4 panel designs (Fig. 1) deployed at a 174,000 L scale facility in southwest Colorado. 3.1.1. Biomass productivity The basic metric for biomass productivity in photobioreactor systems is the volumetric growth rate. The growth rate data from two and one half years of serial batch culturing in AGS3 and AGS4 panel designs is summarized in Fig. 2 in units of g L 1 d 1 for comparison to other published photobioreactor data. Peak growth rates were observed in summer and minimum rates in winter. A maximum batch-averaged productivity of 0.37 g L 1 d 1 was achieved in the summer of 2009 under normal operation cultivating N. oculata in production mode. The average productivity of N. oculata from May 2008 to June 2009 was 0.15 g L 1 d 1 which is nearly identical to the biomass productivity measured for N. salina from April 2009 to January 2011. The annual average productivity for the calendar year of 2009 was 0.16 g L 1 d 1 and for the calendar year of 2010, 0.15 g L 1 d 1. These data suggest that annual productivity numbers can still be improved.
167
An experiment testing the re-use of spent water (recycling of growth media from centrifuge in four successive batches) had no apparent effect on the productivity of the system (0.098 g L 1 d 1, standard deviation 0.049 g L 1 d 1 n = 5) as compared to the control (0.095 g L 1 d 1, standard deviation 0.031 g L 1 d 1 n = 5). It was observed that N. oculata was somewhat more sensitive to invasion by competing phytoplankton than N. salina. The most frequent competitor was tentatively identified as Tetraselmis sp. which only rarely exceeded 5% of total chlorophyll-positive cell counts as determined by flow cytometry. The appearance of Tetraselmis sp. in both Nannochloropsis species was correlated with warmer basin temperatures. An AGS4000 installation identical to the system presented in the materials and methods has been deployed at New Mexico State University operating with the same media and culture, N. salina (CCMP 1776). The lower range of required inoculation densities and upper range of harvest densities was investigated with this system. Inoculation at 0.25 gDW L 1 resulted in linear growth rates of 0.15 gDW L 1 d 1 with a standard deviation of 0.013 gDW L 1 d 1 (n = 10) during the month of July. Harvest densities as high as 6 gDW L 1 were observed in the same period. To achieve this growth range the system was fed a second batch of nutrients four days after inoculation. The ability of the system to operate at high linear growth rates over these density ranges supports the use of the system as an industrial-scale cultivation technology for both stand-alone production and as an inoculums source for large-scale integrated photobioreactor/open raceway pond systems. 3.1.2. Lipid productivity Lipid assays were performed on material collected at the end of each batch to calculate the overall annual oil production for the batches. Fig. 3 presents annualized bio-oil production as estimated from total fatty acid methyl ester content measurements. The biooil production data were determined from the same batch cultures presented in Fig. 2 to measure biomass productivity. The average lipid production was 10.7 m3 ha 1 yr 1. Average and peak lipid production was 7.04 and 21.1 m3 ha 1 yr 1 for N. oculata and 13.1 and 36.3 m3 ha 1 yr 1 for N. salina, respectively. Peak production for both species was attained close to the summer solstice. A direct relationship was established between microalgal biomass harvest density and total FAME content (Fig. 4). Higher cell densities at harvest typically resulted from longer periods of growth under limiting nitrogen conditions and hence more lipid
Fig. 2. Productivity data of Nannochloropsis oculata and Nannochloropsis salina on a g L 1 d 1 metric for batches cultivated in AGS 3 and AGS 4 photobioreactors from June 2008 to January 2011 at the research and development facility of Solix BioSystems located in Fort Collins, Colorado. Values represent the average biomass production during a batch divided by the batch length. Each batch consisted of a minimum of four and a maximum of 10 photobioreactors.
168
J.C. Quinn et al. / Bioresource Technology 117 (2012) 164–171
Fig. 3. Annualized bio-oil production data for batch cultivation in AGS 3 and AGS 4 photobioreactors from June 2008 to present cultivating Nannochloropsis oculata and Nannochloropsis salina. Values represent average total FAMEs obtained at harvest. Each batch average is derived from a minimum of four and a maximum of 10 photobioreactor panels (replicates).
Fig. 4. Total fatty acid content increases with harvested cell density. The total fatty acid content for different batches was measured as FAMEs then normalized to dry weight. Line minimizing least square error shows, R2 = 0.37. The mean% total FAME content for the full dataset (N = 706) was 34.7% (S.D. = 10.8).
accumulation. These observations are consistent with current theory regarding lipid accumulation in microalgae during nitrogenlimited growth (Hu et al., 2008). Nevertheless, harvest decisions do have an effect on estimates of lipid productivity and need to be taken into consideration as part of techno–economic evaluations. Note that all batch cultures in this study were started with identical levels of nutrients, with higher lipid productivity obtained in 2010 than in 2009 based on optimization of harvesting. The average oil production for data collected from calendar year 2009 was 9.92 and 13.1 m3 ha 1 yr 1 for calendar year 2010, a 32% increase in lipid production in 2010. A different harvest strategy was implemented in 2010 that contributed to the improved bio-oil productivity that year. Prior to 2010, microalgae were harvested when the cell density was typically between 2.0 and 2.5 gDW L 1, (average = 2.37, Std. Dev. = 0.54, n = 147), whereas during 2010, harvests were targeted for batch cultures at cell densities near 3.0 gDW L 1 (average = 3.13, Std. Dev. = 0.58, n = 101). The yearly average biomass productivity was indeed 6.25% lower in 2010 than in 2009 but the higher oil content per unit biomass
resulted in an overall bio-oil productivity improvement of 24% in 2010. Total FAME data from samples collected at harvest of 706 batch cultures of N. salina was analyzed to show the quantitative distribution of major fatty acid and a measure of variance in those distributions for batches grown during 2009 and 2010. Fatty acid distributions are also important indicators of downstream fuel process requirements, particularly with respect to unsaturated fatty acid contents that influence fuel characteristics (Bucy et al., 2012). The fatty acid distributions also provide chemotaxonomic information about the presence of other organisms in the cultivation system over time (Khozin-Goldberg and Boussiba, 2011). The experimental data shown in Table 2 are compared with the major fatty acid distribution for laboratory-grown N. salina cultures for comparison (Volkman et al., 1993). The data reveal that the culture systems described here afford a remarkable consistency in fatty acid distributions over extended cultivation periods. This consistency of the total fatty acid distributions as reflected in the standards deviations bodes well for commercial production of
169
J.C. Quinn et al. / Bioresource Technology 117 (2012) 164–171
Table 1 Lineage data for Nannochloropsis oculata and Nannochloropsis salina cultivated at Solix BioSystems research and development center in Fort Collins, CO and Coyote Gulch in Durango, Colorado. The number of generations here refers to the number of independent batch cultures rather than cell divisions. Location
Species
Days in operation
# Transfers
Start date
End date
Fort Collins Fort Collins Fort Collins Fort Collins Fort Collins Fort Collins Coyote Gulch Coyote Gulch
N. N. N. N. N. N. N. N.
176 288 131 161 122 144 260 421
25 22 16 29 41 11 10 20
4/3/2008 8/6/2008 5/6/2009 10/30/2009 3/17/2010 6/10/2010 7/16/2009 10/5/2009
9/26/2008 5/21/2009 9/14/2009 4/9/2010 7/7/2010 11/1/2010 4/2/2010 12/31/10
oculata oculata oculata salina salina salina oculata salina
Table 2 Fatty acid composition data for Nannochloropsis salina determined as fatty acid methyl esters cultivated at Solix BioSystems research and development center in Fort Collins. Average density at harvest was 2.7 gDW L 1 S.D. = 0.83 (n = 706). Fatty acid composition for a pure culture of Nannochloropsis salina (Volkman et al., 1993) is provided for comparison. The Volkman et al. (1993) culture was grown under low light at 20 °C and harvested at mid- or late-log phase. N. salina major fatty acid composition as% of total fatty acids
Volkman et al. (1993) This study Std. Dev.
14:0
16:0
16:1
18:0
18:1
18:2
18:3
20:4
20:5
5.0 3.1 0.74
27.8 36.6 3.8
31.8 33.5 2.7
1.0 1.1 0.34
8.3 10.4 3.1
1.5 0.95 0.54
0.2 1.23 1.2
4 2.76 0.62
24.2 8.9 4.1
high-value lipid products from microalgae that require precise quality control measures and reproducible results. Divergence in fatty acid content between the data generated by this study and that of Volkman et al. (1993) is most pronounced for the omega-3 eicosapentenoic acid (C20:5). While C20:5 content normalized to dry weight of biomass did not change during different stages of batch culture, the C20:5 content normalized to total FAME content dropped as cultures become nitrogen-depleted and begin to increase de novo fatty acid synthesis. Values for C20:5 normalized to total FAME content were measured at 1, 8 and 15 days after inoculation. The C20:5 content dropped from 12.8% (S.D. = 0.7%) of total FAMEs on day 1 to 6.5% (S.D. = 0.2%) on day 8 and 4.1% (S.D. = 0.2%) on day 15. The lower C20:5 fatty acid content in the current study was a likely function of the time of harvest differences between the two studies. The harvest point for this study was well past the mid- to late-log phase harvest point used by Volkman et al. (1993). The differences in C16:0 and C16:1 values between the two studies may be the result of three different mechanisms. Total fatty acid distributions were measured for the present study as a function of growth cycle. One day after inoculation, the C16:0 content in the system presented was typically 29.9% (S.D. = 0.1%) while the C16:1 content was 38.4% (S.D. = 0.6%). These early batch samples were thus more similar to the Volkman samples when the cell densities at sampling were similar. Genetic differences between the N. salina strains used in the two studies cannot be ruled out as contributing factors. Finally, fatty acid saturation is known to respond to temperature, so it is likely that some of the inconsistency between the data sets is a reflection of the growth temperatures averaging close to 25 °C in this study yielding higher C16:0 than C16:1 versus 20 °C used by Volkman et al. (1993) yielding higher C16:1 than C16:0. 3.1.3. Comparison to previously reported data Previous studies have reported productivities of open raceway ponds and photobioreactors using a variety of different microalgae but the scale of the systems and duration of data collection did not compare to the data reported here. No published reports on productivity from large-scale cultivation of Nannochloropis were found with the data presented unique in terms of overall duration and cultivated species. Bosma et al. (2007) presented data from Mono-
dus subterraneus grown in a single photobioreactor of 67 L operated for 3 months (18 July to 26 October 2001) with results from the study showing productivity ranging from 0.03 to 0.20 g L 1 d 1 which is similar to the productivity presented in the present study. A challenging component of comparing studies is the scalability of systems. Data collected from a single photobioreactor is not representative of the productivity of multiple systems based on shading from adjacent systems. Three studies were found that present productivity data from a variety of different microalgae strains grown in open raceway pond systems (Benemann and Oswald, 1996; Sheehan et al., 1998; Weissman and Tillett, 1990). Benemann and Oswald (1996) report monthly average biomass productivity data from two 0.1 ha ponds for a period of 14 months. Monthly productivity varied between 2.6 and 33.5 g m 2 d 1 with an average productivity of 14.1 g m 2 d 1 with an absolute average difference between the two ponds of 7.0 g m 2 d 1. Data was collected in two consecutive years during the month of September with an average productivity of 20.0 g m 2 d 1 with a standard deviation of 10.1 g m 2 d 1 (n = 4). The productivity reported is slightly higher than that from the current study but inconsistency in pond operation and lack of biological replicates makes utilization of this large-scale data in analysis challenging. Weissman and Tillett (1990) operated six 3 m2 open pond systems for 11 months averaging 17.1 g m 2 d 1 and two 0.1 ha open pond systems for 4 months averaging 8.45 g m 2 d 1. The productivity reported in the small ponds is higher than what is reported in the present study, but represents the average growth of five different algae species. Conclusions from the study reported a 60–80% decrease in productivity in the scale-up of the system from 3 m2 to 0.1 ha. The data reported by Weissman and Tillett (1990) are not in a form to be used in large-scale systems analysis based on the experimental setup and the dramatic decrease in productivity on scale-up. Sheehan et al. (1998) presented a variety of data collected from ponds ranging in size from 9.2 to 59 m2. Data collected from this facility was focused on the experimental evaluation of operational techniques. Even though the facility was operated for more than 7 years, no continuous reliable data was collected illustrating the annual productivity potential. The productivities reported in previous studies are similar or slightly higher than the productivities reported in the present study in terms of biomass with no data on lipid content reported.
170
J.C. Quinn et al. / Bioresource Technology 117 (2012) 164–171
Fig. 5. Productivity data from mixing experiment as a function of different sparge rates and duty cycle. Data has been normalized to the growth observed for the baseline sparge condition of 0.6 VVM and 100% duty cycle during photoactive periods. Duty cycle refers to the amount of time on as compared to the baseline case. Error bars represent 1 standard deviation based on biological triplicate.
Inconsistencies in previous experimental operation, operational time, system architecture and scale can account for the differences. The system presented here was operated over a three-year period illustrating the large-scale productivity and stability of the photobioreactor system presented. This combined with the added value of total fatty acid content for each harvested batch fills a current void in the critical assessment of the current near-term realizable productivity potential of microalgae at large-scale.
When compared with the published performance of open cultivation systems, these results are very encouraging. A primary limitation of open raceway cultivation is the widespread occurrence of culture failures and take-overs (Benemann and Oswald, 1996). The integration of photobioreactor cultivation systems with open raceway pond growth facilities is an obvious means of providing a reproducible and reliable source of clean culture inoculum in large-scale production facilities.
3.2. Culture stability
3.3. Energy consumption
The Nannochloropsis cultures described here were very resistant to culture failures or take-over by competing phytoplankton. An illustration of the longevity of cultures cultivated in the photobioreactors presented can be found in Table 1 which illustrates the obvious advantage of continuous cultivation without culture failures through multiple growing seasons. Serial batch cultures utilized a portion of each harvest to inoculate the next batch which produced the batch lineages shown in Table 1. None of the batch lineages were terminated due to culture failure. These results demonstrate highly reliable system performance over multiple years of production. Features of the cultivation system design and operation that are likely to have contributed to this stability include protection of the cultures from wind-blown particulates, filtering of the growth media and control of temperature variations to maintain the strains near their temperature optima. The cultures were not only stable but resistant to invasive takeover. The C18:3 fatty acid content (Table 2) is commonly used as a proxy for contamination as it is commonly found in green microalgae including Tetraselmis sp (Khozin-Goldberg and Boussiba, 2011). Those authors report a range of 0.2–1.0% C18:3 relative to total FAMEs for clean laboratory cultures of N. oculata and N. salina. For the non-axenic cultures in the AGS3 and AGS4 systems, a mean C18:3 content of 1.23% (S.D. = 1.2) across 706 batch cultures of N. salina was observed (Table 2). Further support for the stability of the Nannochloropsis cultures was obtained from estimates of Tetraselmis sp. cell counts by analysis of flow cytometry data. Tetraselmis and Nannochloropsis cells appear in a distinct, non-overlapping ‘‘cloud’’ pattern in chlorophyll-fluorescence versus forward-scatter plot, such that cells of each species can be counted separately within a single sample (Supplementary data). Tetraselmis cell counts ranged between 0.1% and 2%, consistent with the low levels of C18:3 fatty acid observed.
Mixing microalgae cultures can increase productivity by increasing the frequency of light to dark cycling of the cells; however, the mixing requirements for achieving a dramatic increase in productivity are multiple times higher than what can be tolerated economically (Qiang and Richmond, 1996). Most of the experimental data characterizing microalgae growth used extreme levels of mixing greater than 5 volume of air per volume of culture per minute (VVM) in order to remove mixing as a variable. However, in a low cell density, short optical-path reactor, low sparge rate (less than 5 VVM), mixing dynamics did not dramatically affect microalgae culture growth rates (Qiang and Richmond, 1996). Based on this finding and preliminary computational fluid dynamics and particle image velocimetry modeling, the sparge rate for the AGS 4 photobioreactor was decreased to 0.5 VVM relative to the generation 3 technology, which operated at a mixing rate of 2.5 VVM. The data presented in Figs. 2 and 3 demonstrate that the decrease in mixing did not have a statistical effect on the biomass or lipid productivity of the system based on a Student’s T-test with a 95% confidence interval. The relationship between growth and mixing energy was further explored through an outdoor growth experiment designed to look at the effect of duty cycle and sparge rate on productivity. Duty cycle was defined as the time on, compared to time off. For example a duty cycle of 5% corresponds to sparge air for 5% of the time as compared to the baseline system. Small scale AGS 4 photobioreactors were operated in biological triplicate at a variety of duty cycles and sparge rates ranging from the baseline of 0.6 VVM operated during daylight hours to 0.2 VVM operated on a 5% duty cycle (operation during daylight). Growth results at these extremes and five points in between are presented in Fig. 5. No significant differences in the growth rates of N. salina was observed as Student’s T-test determined all of the treatments failed to reject
J.C. Quinn et al. / Bioresource Technology 117 (2012) 164–171
the null hypothesis based on a 95% confidence interval as compared to the control data set. The lowest flow setting, 5% duty cycle at 0.2 VVM, yielded major energy savings through decreases in flow (67%) and operation time (95%) which resulted in a relatively small average decrease in growth (23%). Process engineering models illustrate the importance of energy reduction in all of the major unit operations in the microalgae-tobiofuels process as a prerequisite to commercialization (Batan et al., 2010; Davis et al., 2011). The energy consumption in the growth phase of photobioreactors compared to open raceway ponds is typically reported as more energy intensive based on sparge requirements compared to paddle wheel operation (Lehr and Posten, 2009; Stephenson et al., 2010). The data illustrate the efficiency of the AGS designs with respect to gas exchange as well as the magnitude of energy required to move CO2 and O2 through cultivation systems. The results presented here show that process improvements can result in a 98% decrease in gas-movement-related energy consumption. Comparing this energy consumption to that of traditional paddle wheels as reported by Benemann and Oswald (1996) of 10,750 kWhr ha 1 yr 1, the system presented uses dramatically less energy of 3510 kWhr ha 1 yr 1 based on the 5% duty cycle at 0.2 VVM. Detailed assumptions and energy calculations for all of the scenarios presented in Fig. 5 are presented in supplementary data. This result is in direct opposition of traditional thought in terms of energy requirements for mixing in photobioreactors (Benemann and Oswald, 1996; Chisti, 2007; Wijffels and Barbosa, 2010). 4. Conclusions N. oculata and N. salina were cultivated in large-scale outdoor photobioreactors from April 2008 to January 2011 with biomass and corresponding lipid content reported with serial batch cultures of both strains stable over 41 batch transfers. Results from energy consumption analysis show power consumption for mixing and gas transfer in photobioreactors can be similar to open raceway ponds without an effect on productivity. These results illustrate the need for techno–economic, environmental, and scalability assessment to revisit growth assumptions in an effort to more effectively evaluate the microalgae-to-biofuels process. Acknowledgements We gratefully acknowledge financial support provided by Solix BioSystems, Inc. and data collection and processing support from Chris Turner, Mark Machacek, Pete Hentges, John Walden, Greg Wardle, and Zach Turner. Work at New Mexico State University was additionally supported by the US Department of Energy under contract DE-EE0003046 awarded to the National Alliance for Advanced Biofuels and Bioproducts and a grant by the US Air Force Research Laboratory (FA8650-11-C-2127). Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j.biortech.2012. 04.073. References Batan, L., Quinn, J., Willson, B., Bradley, T., 2010. Net energy and greenhouse gas emission evaluation of biodiesel derived from microalgae. Environ. Sci. Technol. 44, 7975–7980.
171
Benemann, J.R., Oswald, W.J., 1996. Systems and economic analysis of microalgae ponds for conversion of CO2 to biomass. Final report: March 1996. Bosma, R., Van Zessen, E., Reith, J.H., Tramper, J., Wijffels, R.H., 2007. Prediction of volumetric productivity of an outdoor photobioreactor. Biotechnol. Bioeng. 97, 1108–1120. Bucy, H., Baumgardner, M., Marchese, A.J., 2012. Chemical and physical properties of algal methyl ester biodiesel containing varying levels of methyl eicosapentaenoate and methyl docosahexaenoate. Algal Research. http:// dx.doi.org/10.1016/j.algal.2012.02.001. Campbell, P.K., Beer, T., Batten, D., 2011. Life cycle assessment of biodiesel production from microalgae in ponds. Bioresour. Technol. 102, 50–56. Chisti, Y., 2007. Biodiesel from microalgae. Biotechnol. Adv. 25, 294–306. Chisti, Y., 2008a. Biodiesel from microalgae beats bioethanol. Trends Biotechnol. 26, 126–131. Chisti, Y., 2008b. Response to Reijnders: do biofuels from microalgae beat biofuels from terrestrial plants? Trends Biotechnol. 26, 351–352. Clarens, A.F., Resurreccion, E.P., White, M.A., Colosi, L.M., 2010. Environmental life cycle comparison of algae to other bioenergy feedstocks. Environ. Sci. Technol. 44, 1813–1819. Davis, R., Aden, A., Pienkos, P.T., 2011. Techno–economic analysis of autotrophic microalgae for fuel production. Appl. Energy 88, 3524–3531. Frank, E.D., Han, J., Palou-Rivera, I., Elgowainy, A., Wang, M.Q., 2011. Life-cycle analysis of algal lipid fuels with the greet model. Center for Transportation Research, Energy Systems Division, Argonne National Laboratory. Hirano, A., Hon-Nami, K., Kunito, S., Hada, M., Ogushi, Y., 1998. Temperature effect on continuous gasification of microalgal biomass: theoretical yield of methanol production and its energy balance. Catal. Today 45, 399–404. Hu, Q., Guterman, H., Richmond, A., 1996. A flat inclined modular photobioreactor for outdoor mass cultivation of photoautotrophs. Biotechnol. Bioeng. 51, 51–60. Hu, Q., Sommerfeld, M., Jarvis, E., Ghirardi, M., Posewitz, M., Seibert, M., Darzins, A., 2008. Microalgal triacylglycerols as feedstocks for biofuel production: perspectives and advances. Plant J. 54, 621–639. Jorquera, O., Kiperstok, A., Sales, E.A., Embirucu, M., Ghirardi, M.L., 2010. Comparative energy life-cycle analyses of microalgal biomass production in open ponds and photobioreactors. Bioresour. Technol. 101, 1406–1413. Khozin-Goldberg, I., Boussiba, S., 2011. Concerns over the reporting of inconsistent data on fatty acid composition for microalgae of the genus Nannochloropsis (Eustigmatophyceae). J. Appl. Phycol. 23, 933–934. Lardon, L., Helias, A., Sialve, B., Stayer, J.P., Bernard, O., 2009. Life-cycle assessment of biodiesel production from microalgae. Environ. Sci. Technol. 43, 6475–6481. Lehr, F., Posten, C., 2009. Closed photo-bioreactors as tools for biofuel production. Curr. Opin. Biotechnol. 20, 280–285. Li, Y., Horsman, M., Wu, N., Lan, C.Q., Dubois-Calero, N., 2008. Biofuels from microalgae. Biotechnol. Prog. 24, 815–820. Mata, T.M., Martins, A.A., Caetano, N.S., 2010. Microalgae for biodiesel production and other applications: a review. Renewable Sustainable Energy Rev. 14, 217– 232. Qiang, H., Richmond, A., 1996. Productivity and photosynthetic efficiency of Spirulina platensis as affected by light intensity, algal density and rate of mixing in a flat plate photobioreactor. J. Appl. Phycol. 8, 139–145. Quinn, J., de Winter, L., Bradley, T., 2011. Microalgae bulk growth model with application to industrial scale systems. Bioresour. Technol. 102, 5083–5092. Rodolfi, L., Zittelli, G.C., Bassi, N., Padovani, G., Biondi, N., Bonini, G., Tredici, M.R., 2009. Microalgae for oil: strain selection, induction of lipid synthesis and outdoor mass cultivation in a low-cost photobioreactor. Biotechnol. Bioeng. 102, 100–112. Schenk, P.M., Thomas-Hall, S.R., Stephens, E., Marx, U.C., Mussgnug, J.H., Posten, C., Kruse, O., Hankamer, B., 2008. Second generation biofuels: high-efficiency microalgae for biodiesel production. BioEnergy Res. 1, 20–43. Schutter, M.E., Dick, R.P., 2000. Comparison of fatty acid methyl ester (FAME) methods for characterizing microbial communities. Soil Sci. Soc. Am. J. 64, 1659–1668. Scott, S.A., Davey, M.P., Dennis, J.S., Horst, I., Howe, C.J., Lea-Smith, D.J., Smith, A.G., 2010. Biodiesel from algae: challenges and prospects. Curr. Opin. Biotechnol. 21, 277–286. Sheehan, J., Dunahay, T., Benemann, J., Roessler, P., 1998. A look back at the US department of energy’s aquatic species program: biodiesel from algae. NREL report, TP-580-24190. Stephenson, A.L., Kazamia, E., Dennis, J.S., Howe, C.J., Scott, S.A., Smith, A.G., 2010. Life-cycle assessment of potential algal biodiesel production in the United Kingdom: a comparison of raceways and air-lift tubular bioreactors. Energy Fuels 24, 4062–4077. Volkman, J.K., Brown, M.R., Dunstan, G.A., Jeffrey, S.W., 1993. The biochemicalcomposition of marine microalgae from the class eustigmatophyceae. J. Phycol. 29, 69–78. Weissman, J.C., Tillett, D.M., 1990. Design and Operation of an Outdoor Microalgae Test Facility: Large-Scale System Results. NREL, Golden, CO. Wijffels, R.H., Barbosa, M.J., 2010. An outlook on microalgal biofuels. Science 329, 796–799. Williams, P.R.D., Inman, D., Aden, A., Heath, G.A., 2009. Environmental and sustainability factors associated with next-generation biofuels in the US: what do we really know? Environ. Sci. Technol. 43, 4763–4775.
Contents lists available at SciVerse ScienceDirect
Bioresource Technology journal homepage: www.elsevier.com/locate/biortech
Nannochloropsis production metrics in a scalable outdoor photobioreactor for commercial applications Jason C. Quinn a, Tracy Yates b, Nathaniel Douglas b, Kristina Weyer b, Joel Butler b, Thomas H. Bradley a, Peter J. Lammers b,1,⇑ a b
Mechanical Engineering, 1374 Campus Delivery, Colorado State University, Fort Collins, Colorado 80523-1374, USA Solix BioSystems, Inc., 430 B North College Ave, Fort Collins, CO 80524, USA
h i g h l i g h t s 1 d 1 during calendar years of 2009 & 2010. d 1. 3 1 1 " Average bio-oil productivity of 10.7 m ha yr with peak of 36.3 m3 ha 1 yr 1. " Serial batch cultures maintained for >421 days of continuous cultivation. " Energy consumption sensitivity analysis for large-scale photobioreactor system.
" Growth rates averaged 0.16 g L " Peak growth rate of 0.37 L
a r t i c l e
1
i n f o
Article history: Received 7 January 2012 Received in revised form 18 April 2012 Accepted 20 April 2012 Available online 26 April 2012 Keywords: Microalgae Photobioreactor Nannochloropsis Biomass Lipids
a b s t r a c t Commercial production of renewable energy feedstocks from microalgae will require reliable and scalable growth systems. Two and one half years of biomass and lipid productivity data were obtained with an industrial-scale outdoor photobioreactor operated in Fort Collins, Colorado (USA). The annualized volumetric growth rates for Nannochloropsis oculata (CCMP 525) and Nannochloropsis salina (CCMP 1776) were 0.16 g L 1 d 1 (peak = 0.37 g L 1 d 1) and 0.15 g L 1 d 1 (peak = 0.37 g L 1 d 1) respectively. The collective average lipid production was 10.7 m3 ha 1 yr 1 with a peak value of 36.3 m3 ha 1 yr 1. Results from this study are unique based on publication of biomass and corresponding lipid content combined with demonstration of energy savings realized through analysis of gas delivery requirements, water recycling from successive harvests with no effect on productivity, and culture stability through serial batch lineage data and chemotaxonomic analysis of fatty acid contents. Ó 2012 Elsevier Ltd. All rights reserved.
1. Introduction Relative to first-generation biofuel feedstocks, microalgae are characterized by higher solar energy yield, the potential for yearround cultivation in many locations, the ability to grow in brackish and saline water as well as water produced from oil and gas extraction, higher areal productivities than oil seed crops and the utilization of non-arable land (Batan et al., 2010; Chisti, 2007; Li et al., Abbreviations: AGS, Algae Growth System; CCMP, Center for Culture of Marine Phytoplankton; DW, Dry weight; FAME, Fatty acid methyl ester; GC, Gas chromatograph; LCA, Life Cycle Analysis; VVM, Volume of air per volume of culture per minute. ⇑ Corresponding author. Tel.: +1 575 646 7458; fax: +1 575 646 5717. E-mail address: [email protected] (P.J. Lammers). 1 Present address: Energy Research Laboratory, Box 30001, MSC 3RES, New Mexico State University, Las Cruces, NM, 88003, USA. Tel.: +1 (575) 646-7458; fax: +1 (575) 646-5717. 0960-8524/$ - see front matter Ó 2012 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.biortech.2012.04.073
2008; Williams et al., 2009). Microalgae can grow in poor-quality water, utilize CO2 from point sources such as coal fired power plants and utilize nutrients from wastewater treatment plants with a uniquely high productivity potential (Chisti, 2008a; Li et al., 2008; Schenk et al., 2008; Wijffels and Barbosa, 2010). Current lab-scale experimental data have suggested a near-term realizable production an order of magnitude higher than the current productivity of ethanol from corn and bio-diesel from soy: 2533 liters hectare 1 yr 1 (271 gal acre 1 yr 1) of ethanol from corn and 584 L ha 1 yr 1 (62.5 gal acre 1 yr 1) of bio-diesel from soybeans, respectively (Chisti, 2007). Microalgae cultivation is typically done in, open raceway ponds or photobioreactor systems. The two primary advantages of photobioreactors are increased culture stability resulting from lower likelihood of pathogen or grazer infestations and higher volumetric productivities due to differences in light utilization. These advantages have led to an interest in photobioreactor technology for the
J.C. Quinn et al. / Bioresource Technology 117 (2012) 164–171
cultivation of inoculums for large raceway ponds and quality-controlled production of high-valued products such as nutraceuticals (Wijffels and Barbosa, 2010). The technical and economic evaluation of photobioreactors and open raceway pond technologies at commercial scales is difficult and very expensive to perform (Hu et al., 1996). Previous studies have reported monthly productivities from outdoor large-scale systems but inconsistencies in experimental setup, cultured species and time scales make utilization of these data sets challenging (Benemann and Oswald, 1996; Bosma et al., 2007; Sheehan et al., 1998; Weissman and Tillett, 1990). Large-scale data has been restricted to open raceway pond systems with a complete lack of large-scale productivity data from photobioreactors over a time period greater than one growing season. Due to limited productivity data at large-scale, major simplifying assumptions have been made in the assessment of microalgae productivity potential. Large-scale productivity data are needed to inform and advance the accuracy of life cycle analyses (LCA) and techno–economic modeling of second and third generation biofuel feedstocks which are the primary tools used to assess and inform policy makers and funding agencies. Recent LCA and techno–economic modeling efforts have examined the sustainability, economic, and environmental benefits of microalgae with conclusions from these studies varying significantly based on fundamental assumptions in the modeling efforts (Batan et al., 2010; Benemann and Oswald, 1996; Campbell et al., 2011; Clarens et al., 2010; Davis et al., 2011; Frank et al., 2011; Jorquera et al., 2010; Lardon et al., 2009; Stephenson et al., 2010). The microalgae assessments of Batan et al. (2010), Lardon et al.(2009), Hirano et al. (1998), Davis et al. (2011), Frank et al. (2011) and Campbell et al. (2011) used fixed biomass productivity models between 10 and 30 g m 2 d 1 (3.6104– 11.0 104 kg ha 1 yr 1) based on laboratory-scale data. Extrapolation of laboratory data has been dictated by the infancy of the microalgae biofuels industry and the uncertainty of the current large-scale biooil productivity potentials reported to range from 8.2 m3 ha 1 yr 1 (Scott et al., 2010) to 136.9 m3 ha 1 yr 1 (Chisti, 2007) and values between these two extremes (Chisti, 2008a, 2008b; Mata et al., 2010; Rodolfi et al., 2009; Schenk et al., 2008; Sheehan et al., 1998; Wijffels and Barbosa, 2010). In the present study biomass and lipid productivity data from a large-scale microalgae photobioreactor cultivation facility operated in Fort Collins, Colorado at the research and development facility of Solix BioSystems are presented. Two oleaginous microalgal strains were evaluated, Nannochloropsis oculata and Nannochloropsis salina. Furthermore, two generations of the Solix bioreactor, AGS 3 and 4 are compared. Detailed biomass and fatty acid methyl ester based lipid productivity data on peak and annual average are reported. These data facilitate techno–economic and life cycle modeling of microalgal based biofuel and bioproduct processes utilizing photobioreactor technology in stand-alone or integrated photobioreactor-open raceway pond configurations. Discussion focuses on a comparison of the performance of the system presented to previously published data, assessment of the stability of Nannochloropsis strains in long-term successive batch operations, energy requirements for the growth system with comparison to tradition open raceway ponds, and operational tradeoffs between maximizing growth rates versus bio-oil content.
2. Methods
165
1776). From September 2009 to September 2010 f/2 growth medium was modified to contain 5 mM NO3 L 1 and 0.368 mM PO4 at a salinity of 20 g L 1. The salinity of the growth media was increased to 27 g L 1 based on operational decisions made from October 2009 to January 2011 while all other components were held constant. The medium was filtered (0.2 micron filter) into a tank with the required inoculums where it was mixed to ensure homogeneity prior to inoculation of the photobioreactor system. Spent water recycling experiments were performed over four successive batch cultures by using 80% water from the harvest centrifuge with 20% fresh artificial seawater (Instant Ocean) at 27 g L 1 during the summer of 2010. N. oculata was cultivated in the AGS 3 photobioreactors from May 2008 to June 2009. N. salina was subsequently cultivated in AGS 3 and AGS 4 photobioreactors starting in January 2009. Operations were conducted in a serial batch mode with a portion of each harvest used to start the next batch. Harvest densities ranged between 2 and 3 gDW L 1. Some batches were harvested at lower densities due to low growth rates in low light periods during the winter. A select number of batches were inoculated with as low as 0.25 gDW L 1 and harvested at 6 gDW L 1 as detailed in the Section 3. The cultures were occasionally restarted from laboratory cultures using sterile techniques: Tube cultures (5 mL) were scaled through 125 and 250 mL Erlenmeyer flasks maintained in a 24 h illuminated incubator at 150 lmol photons m 2 s 1 at 25 °C. These cultures were transferred to 4.0 L Fernbach flasks on a shaker table under greenhouse conditions (non-sterile techniques from this scale onward) and transferred into a variable volume flat-panel bioreactor (20–60 L) maintained in a greenhouse. Final scale-up was then conducted in the AGS 3 or AGS 4 panels. Inocula for routine batch cultivation was derived from the previous harvest, after dilution with fresh nutrient media to the re-inoculation density of 1 gdw L 1. 2.2. Outdoor culture system 2.2.1. Photobioreactor geometry The AGS 3 and AGS 4 photobioreactor utilize a shallow water basin for thermal and structural support for vertically oriented, polyethylene growth panels, 0.05 m wide by 0.28 m high by 17.3 m long. Detailed photos of the outdoor facility are presented in supplementary data. The AGS 3 and AGS 4 photobioreactor systems maintained panel spacing of 0.15 m. The structure supporting the panels in the temperature-controlled basin was the only change from the AGS 3 to AGS 4 technology. The AGS 3 photobioreactors were deployed from April 2008 to December 2009. The AGS 4 support structure permits deployment in natural bodies of water via flotation which decreases the land grading requirement for terrestrial deployment. The AGS 4 photobioreactor panels were deployed in December 2009 and are currently being utilized in Fort Collins, Colorado at the Solix BioSystems headquarters and at the 0.4 hectare pilot plant facility outside of Durango, Colorado at Coyote Gulch Durango, Colorado. Both AGS 3 and AGS 4 photobioreactor panels were constructed with polyethylene. The standard inoculation volume corresponded to a volume to area ratio for the data presented of 49.6 L m 2 with a standard deviation of 19.9 L m 2. The large standard deviation is due to experimentation centered on the optimization of reactor volume. Schematic details of the growth chamber for the AGS3 and AGS 4 photobioreactor panels can be found in Fig. 1.
2.1. Organisms, culture conditions, and inoculation protocols 2.1.1. Nannochloropsis strains and growth medium Cultures were obtained from the Provasoli–Guillard National Center for Culture of Marine Phytoplankton (CCMP) by Solix BioSystems, specifically N. oculata (CCMP 525) and N. salina (CCMP
2.2.2. Photobioreactor mixing, pH, and temperature control Mixing for AGS 3 and AGS 4 photobioreactors was provided through a sparge air system that was operated during daylight hours. A mixing rate of 2.5 VVM for the AGS 3 photobioreactor and at 0.5 VVM for AGS 4 photobioreactors was used. The decrease
166
J.C. Quinn et al. / Bioresource Technology 117 (2012) 164–171
Fig. 1. Schematic drawing of the AGS 3 and AGS 4 photobioreactor growth panel. Differences between the AGS 3 and AGS 4 photobioreactors are restricted to the support structure which does not affect the optical properties of the photobioreactor.
in sparge rate from the AGS 3 to the AGS 4 photobioreactors was due to improvements in sparge-hole geometry supported by particle image velocimetry and growth experimentation. The pH of the system was monitored continuously with probes embedded in the AGS3 and AGS4 photobioreactors (Hach, model PC1R2A). The pH was maintained by CO2 supplied into the sparge air and delivered to the system with a duty cycle determined by pH feedback control (pH maintained at 7.3 ± 0.1) or by continually mixing 2% CO2 with sparge air (pH maintained at 7.3 ± 0.5). The temperature of the culture was maintained by the thermal mass of water basin which also supplied the structural support for the AGS 3 and AGS 4 photobioreactors. The temperature was continuously monitored and maintained between 19 and 26 °C via a Marley evaporative cooling system with a capacity of 270,000 BTU or a Jandy Lite2 pool heater with a capacity of 325,000 BTUH. Average temperatures were close to 25 °C for the majority of the batches. Microalgae biomass was harvested with a Sharpless AS26 continuous-flow centrifuge operating at 15,000 rpm and a flow rate of 9 L min 1. 2.3. Growth monitoring Samples of the cultures were monitored daily to track growth using optical density at 750 nm (OD750). Total fatty acid content (based on FAME determination), dry mass and salinity were measured at the end of each batch. Samples were drawn using a 10 mL syringe through sample lines attached to sample ports at the head of the reactors. Previous sampling experimentation showed that sampling location did not affect experimental results due to the homogeneity of the culture. Optical density measurements were performed on a Hach DR5000 spectrophotometer and the OD750 converted to gdw L 1 utilizing an empirically determined correlation factor (N > 100, R2 = 0.99989). Evaporative losses were integrated in final growth measurements through previously determined loss rates (Quinn et al., 2011). Nitrate and phosphate levels were routinely measured in culture filtrates using standard colorimetric methods and reagents (Hach). The level of both nutrients decreased to below detection limits within 4 days of inoculation in the AGS3 and AGS4 system regardless of season.
batch comparisons with other studies. Lipid fractions were determined using an in situ trans-esterification technique based on the methods of Schutter and Dick (2000): Five mg of microalgae sample centrifuged at 4000g for 5 min followed by removal of the supernatant. An auto-pipette was used to dispense 2.5 ml of 0.2 N KOH in methanol onto the 5 mg microalgae pellet. Samples were pipette mixed and transferred to a glass test tube previously washed in 1% HCl acid. An additional 2.5 ml of 0.2 N KOH in methanol was added and pipette mixed. Samples were aggressively mixed using a VWR Analog vortex mixer at a speed setting of 10 for 20 s followed by heating to 37 °C for 30 min. One ml of acetic acid and 2 ml of HPLC grade heptane was added and the samples were aggressively mixed by using a VWR Analog vortex mixer on a speed setting of 10 for 20 s and then centrifuged at 2000g for 5 min. The organic layer was removed and processed in a gas chromatograph (GC) to determine lipid content and composition. The single heptane extraction reproducibly removed 80% of the overall lipid content in the cells as determined by standard addition methods using C15:0 triacylglycerol. Samples were quantified based on normalization to recovery of a 23:0 FAME internal standard against a standard curve. Transesterified samples were prepared for GC analysis by first diluting the sample 1:10 with heptane. An internal standard (23:0) obtained from NU-CHEK PREP, Inc. was added to the sample and the head space was filled with nitrogen. Samples were analyzed with an Agilent Technologies 7890A GC instrument utilizing a 30 m 0.32 mm 0.25 lm Restek FAMEWAX column. A spitless injection was used requiring 1 lL of sample. Helium at 1.5 mL min 1 was used as carrier gas. The oven was operated at 90 °C for 0.5 min and then ramped to 208 °C at 70 °C min 1, then ramped to 230 °C at 3 °C min 1, and finally to 240 °C ramped at 2 °C min 1 and held for 1 min. Prior to running samples, a blank was run followed by the generation of a four-point standard curve using a GLC-461 standard obtained from NU-CHEK PREP, Inc. Areal productivity was calculated based on the productivity of the batch, FAME content, an assumed FAME density of 918 kg m 3, operation of 365 days a year, and with the area reported as photosynthetic area.
2.4. Fatty acid methyl ester (FAME) assay
2.5. Estimates of culture composition by fatty acid distributions and flow cytometry
Assays of lipid content based on FAME determination were performed on material collected at the end of each batch with the results extrapolated to calculate the overall annual oil production for
Compositional analysis of samples was based on comparison of fatty acid distributions with known values and the use of flow cytometry. Fatty acid distributions as percent of total FAME were re-
J.C. Quinn et al. / Bioresource Technology 117 (2012) 164–171
corded for 706 batch cultures of N. salina. The chemotaxonomic value of these data is widely appreciated (Khozin-Goldberg and Boussiba, 2011). Some samples were also analyzed by flow cytometry on a GuavaCyte instrument (Millipore) to determine chlorophyll-positive cell counts. The chlorophyll content per cell count in the instrument was estimated as the chlorophyll-fluorescence emission measured in the red-channel (646–692 nm) using the standard band-pass filters supplied with the instrument after laser excitation at 488 nm. A plot of chlorophyll-fluorescence versus forward scatter for 10,000 cells provided an estimate of culture composition. The smaller Nannochloropsis populations (3–4 micron diameter) map to a separate region of the plot from the signals associated with cells from a common invader species, Tetraselmis, that produced more forward scatter due to larger diameter (>8 microns) and higher in per-cell chlorophyll (data presented in supplementary data). 3. Results and discussion 3.1. Large-scale outdoor productivity The data shown corresponds only to the operation of the reactors in production mode using a single nutrient formulation and excludes data collected during periods of experimentation. The results reported are representative of those obtained with the same AGS3 and AGS4 panel designs (Fig. 1) deployed at a 174,000 L scale facility in southwest Colorado. 3.1.1. Biomass productivity The basic metric for biomass productivity in photobioreactor systems is the volumetric growth rate. The growth rate data from two and one half years of serial batch culturing in AGS3 and AGS4 panel designs is summarized in Fig. 2 in units of g L 1 d 1 for comparison to other published photobioreactor data. Peak growth rates were observed in summer and minimum rates in winter. A maximum batch-averaged productivity of 0.37 g L 1 d 1 was achieved in the summer of 2009 under normal operation cultivating N. oculata in production mode. The average productivity of N. oculata from May 2008 to June 2009 was 0.15 g L 1 d 1 which is nearly identical to the biomass productivity measured for N. salina from April 2009 to January 2011. The annual average productivity for the calendar year of 2009 was 0.16 g L 1 d 1 and for the calendar year of 2010, 0.15 g L 1 d 1. These data suggest that annual productivity numbers can still be improved.
167
An experiment testing the re-use of spent water (recycling of growth media from centrifuge in four successive batches) had no apparent effect on the productivity of the system (0.098 g L 1 d 1, standard deviation 0.049 g L 1 d 1 n = 5) as compared to the control (0.095 g L 1 d 1, standard deviation 0.031 g L 1 d 1 n = 5). It was observed that N. oculata was somewhat more sensitive to invasion by competing phytoplankton than N. salina. The most frequent competitor was tentatively identified as Tetraselmis sp. which only rarely exceeded 5% of total chlorophyll-positive cell counts as determined by flow cytometry. The appearance of Tetraselmis sp. in both Nannochloropsis species was correlated with warmer basin temperatures. An AGS4000 installation identical to the system presented in the materials and methods has been deployed at New Mexico State University operating with the same media and culture, N. salina (CCMP 1776). The lower range of required inoculation densities and upper range of harvest densities was investigated with this system. Inoculation at 0.25 gDW L 1 resulted in linear growth rates of 0.15 gDW L 1 d 1 with a standard deviation of 0.013 gDW L 1 d 1 (n = 10) during the month of July. Harvest densities as high as 6 gDW L 1 were observed in the same period. To achieve this growth range the system was fed a second batch of nutrients four days after inoculation. The ability of the system to operate at high linear growth rates over these density ranges supports the use of the system as an industrial-scale cultivation technology for both stand-alone production and as an inoculums source for large-scale integrated photobioreactor/open raceway pond systems. 3.1.2. Lipid productivity Lipid assays were performed on material collected at the end of each batch to calculate the overall annual oil production for the batches. Fig. 3 presents annualized bio-oil production as estimated from total fatty acid methyl ester content measurements. The biooil production data were determined from the same batch cultures presented in Fig. 2 to measure biomass productivity. The average lipid production was 10.7 m3 ha 1 yr 1. Average and peak lipid production was 7.04 and 21.1 m3 ha 1 yr 1 for N. oculata and 13.1 and 36.3 m3 ha 1 yr 1 for N. salina, respectively. Peak production for both species was attained close to the summer solstice. A direct relationship was established between microalgal biomass harvest density and total FAME content (Fig. 4). Higher cell densities at harvest typically resulted from longer periods of growth under limiting nitrogen conditions and hence more lipid
Fig. 2. Productivity data of Nannochloropsis oculata and Nannochloropsis salina on a g L 1 d 1 metric for batches cultivated in AGS 3 and AGS 4 photobioreactors from June 2008 to January 2011 at the research and development facility of Solix BioSystems located in Fort Collins, Colorado. Values represent the average biomass production during a batch divided by the batch length. Each batch consisted of a minimum of four and a maximum of 10 photobioreactors.
168
J.C. Quinn et al. / Bioresource Technology 117 (2012) 164–171
Fig. 3. Annualized bio-oil production data for batch cultivation in AGS 3 and AGS 4 photobioreactors from June 2008 to present cultivating Nannochloropsis oculata and Nannochloropsis salina. Values represent average total FAMEs obtained at harvest. Each batch average is derived from a minimum of four and a maximum of 10 photobioreactor panels (replicates).
Fig. 4. Total fatty acid content increases with harvested cell density. The total fatty acid content for different batches was measured as FAMEs then normalized to dry weight. Line minimizing least square error shows, R2 = 0.37. The mean% total FAME content for the full dataset (N = 706) was 34.7% (S.D. = 10.8).
accumulation. These observations are consistent with current theory regarding lipid accumulation in microalgae during nitrogenlimited growth (Hu et al., 2008). Nevertheless, harvest decisions do have an effect on estimates of lipid productivity and need to be taken into consideration as part of techno–economic evaluations. Note that all batch cultures in this study were started with identical levels of nutrients, with higher lipid productivity obtained in 2010 than in 2009 based on optimization of harvesting. The average oil production for data collected from calendar year 2009 was 9.92 and 13.1 m3 ha 1 yr 1 for calendar year 2010, a 32% increase in lipid production in 2010. A different harvest strategy was implemented in 2010 that contributed to the improved bio-oil productivity that year. Prior to 2010, microalgae were harvested when the cell density was typically between 2.0 and 2.5 gDW L 1, (average = 2.37, Std. Dev. = 0.54, n = 147), whereas during 2010, harvests were targeted for batch cultures at cell densities near 3.0 gDW L 1 (average = 3.13, Std. Dev. = 0.58, n = 101). The yearly average biomass productivity was indeed 6.25% lower in 2010 than in 2009 but the higher oil content per unit biomass
resulted in an overall bio-oil productivity improvement of 24% in 2010. Total FAME data from samples collected at harvest of 706 batch cultures of N. salina was analyzed to show the quantitative distribution of major fatty acid and a measure of variance in those distributions for batches grown during 2009 and 2010. Fatty acid distributions are also important indicators of downstream fuel process requirements, particularly with respect to unsaturated fatty acid contents that influence fuel characteristics (Bucy et al., 2012). The fatty acid distributions also provide chemotaxonomic information about the presence of other organisms in the cultivation system over time (Khozin-Goldberg and Boussiba, 2011). The experimental data shown in Table 2 are compared with the major fatty acid distribution for laboratory-grown N. salina cultures for comparison (Volkman et al., 1993). The data reveal that the culture systems described here afford a remarkable consistency in fatty acid distributions over extended cultivation periods. This consistency of the total fatty acid distributions as reflected in the standards deviations bodes well for commercial production of
169
J.C. Quinn et al. / Bioresource Technology 117 (2012) 164–171
Table 1 Lineage data for Nannochloropsis oculata and Nannochloropsis salina cultivated at Solix BioSystems research and development center in Fort Collins, CO and Coyote Gulch in Durango, Colorado. The number of generations here refers to the number of independent batch cultures rather than cell divisions. Location
Species
Days in operation
# Transfers
Start date
End date
Fort Collins Fort Collins Fort Collins Fort Collins Fort Collins Fort Collins Coyote Gulch Coyote Gulch
N. N. N. N. N. N. N. N.
176 288 131 161 122 144 260 421
25 22 16 29 41 11 10 20
4/3/2008 8/6/2008 5/6/2009 10/30/2009 3/17/2010 6/10/2010 7/16/2009 10/5/2009
9/26/2008 5/21/2009 9/14/2009 4/9/2010 7/7/2010 11/1/2010 4/2/2010 12/31/10
oculata oculata oculata salina salina salina oculata salina
Table 2 Fatty acid composition data for Nannochloropsis salina determined as fatty acid methyl esters cultivated at Solix BioSystems research and development center in Fort Collins. Average density at harvest was 2.7 gDW L 1 S.D. = 0.83 (n = 706). Fatty acid composition for a pure culture of Nannochloropsis salina (Volkman et al., 1993) is provided for comparison. The Volkman et al. (1993) culture was grown under low light at 20 °C and harvested at mid- or late-log phase. N. salina major fatty acid composition as% of total fatty acids
Volkman et al. (1993) This study Std. Dev.
14:0
16:0
16:1
18:0
18:1
18:2
18:3
20:4
20:5
5.0 3.1 0.74
27.8 36.6 3.8
31.8 33.5 2.7
1.0 1.1 0.34
8.3 10.4 3.1
1.5 0.95 0.54
0.2 1.23 1.2
4 2.76 0.62
24.2 8.9 4.1
high-value lipid products from microalgae that require precise quality control measures and reproducible results. Divergence in fatty acid content between the data generated by this study and that of Volkman et al. (1993) is most pronounced for the omega-3 eicosapentenoic acid (C20:5). While C20:5 content normalized to dry weight of biomass did not change during different stages of batch culture, the C20:5 content normalized to total FAME content dropped as cultures become nitrogen-depleted and begin to increase de novo fatty acid synthesis. Values for C20:5 normalized to total FAME content were measured at 1, 8 and 15 days after inoculation. The C20:5 content dropped from 12.8% (S.D. = 0.7%) of total FAMEs on day 1 to 6.5% (S.D. = 0.2%) on day 8 and 4.1% (S.D. = 0.2%) on day 15. The lower C20:5 fatty acid content in the current study was a likely function of the time of harvest differences between the two studies. The harvest point for this study was well past the mid- to late-log phase harvest point used by Volkman et al. (1993). The differences in C16:0 and C16:1 values between the two studies may be the result of three different mechanisms. Total fatty acid distributions were measured for the present study as a function of growth cycle. One day after inoculation, the C16:0 content in the system presented was typically 29.9% (S.D. = 0.1%) while the C16:1 content was 38.4% (S.D. = 0.6%). These early batch samples were thus more similar to the Volkman samples when the cell densities at sampling were similar. Genetic differences between the N. salina strains used in the two studies cannot be ruled out as contributing factors. Finally, fatty acid saturation is known to respond to temperature, so it is likely that some of the inconsistency between the data sets is a reflection of the growth temperatures averaging close to 25 °C in this study yielding higher C16:0 than C16:1 versus 20 °C used by Volkman et al. (1993) yielding higher C16:1 than C16:0. 3.1.3. Comparison to previously reported data Previous studies have reported productivities of open raceway ponds and photobioreactors using a variety of different microalgae but the scale of the systems and duration of data collection did not compare to the data reported here. No published reports on productivity from large-scale cultivation of Nannochloropis were found with the data presented unique in terms of overall duration and cultivated species. Bosma et al. (2007) presented data from Mono-
dus subterraneus grown in a single photobioreactor of 67 L operated for 3 months (18 July to 26 October 2001) with results from the study showing productivity ranging from 0.03 to 0.20 g L 1 d 1 which is similar to the productivity presented in the present study. A challenging component of comparing studies is the scalability of systems. Data collected from a single photobioreactor is not representative of the productivity of multiple systems based on shading from adjacent systems. Three studies were found that present productivity data from a variety of different microalgae strains grown in open raceway pond systems (Benemann and Oswald, 1996; Sheehan et al., 1998; Weissman and Tillett, 1990). Benemann and Oswald (1996) report monthly average biomass productivity data from two 0.1 ha ponds for a period of 14 months. Monthly productivity varied between 2.6 and 33.5 g m 2 d 1 with an average productivity of 14.1 g m 2 d 1 with an absolute average difference between the two ponds of 7.0 g m 2 d 1. Data was collected in two consecutive years during the month of September with an average productivity of 20.0 g m 2 d 1 with a standard deviation of 10.1 g m 2 d 1 (n = 4). The productivity reported is slightly higher than that from the current study but inconsistency in pond operation and lack of biological replicates makes utilization of this large-scale data in analysis challenging. Weissman and Tillett (1990) operated six 3 m2 open pond systems for 11 months averaging 17.1 g m 2 d 1 and two 0.1 ha open pond systems for 4 months averaging 8.45 g m 2 d 1. The productivity reported in the small ponds is higher than what is reported in the present study, but represents the average growth of five different algae species. Conclusions from the study reported a 60–80% decrease in productivity in the scale-up of the system from 3 m2 to 0.1 ha. The data reported by Weissman and Tillett (1990) are not in a form to be used in large-scale systems analysis based on the experimental setup and the dramatic decrease in productivity on scale-up. Sheehan et al. (1998) presented a variety of data collected from ponds ranging in size from 9.2 to 59 m2. Data collected from this facility was focused on the experimental evaluation of operational techniques. Even though the facility was operated for more than 7 years, no continuous reliable data was collected illustrating the annual productivity potential. The productivities reported in previous studies are similar or slightly higher than the productivities reported in the present study in terms of biomass with no data on lipid content reported.
170
J.C. Quinn et al. / Bioresource Technology 117 (2012) 164–171
Fig. 5. Productivity data from mixing experiment as a function of different sparge rates and duty cycle. Data has been normalized to the growth observed for the baseline sparge condition of 0.6 VVM and 100% duty cycle during photoactive periods. Duty cycle refers to the amount of time on as compared to the baseline case. Error bars represent 1 standard deviation based on biological triplicate.
Inconsistencies in previous experimental operation, operational time, system architecture and scale can account for the differences. The system presented here was operated over a three-year period illustrating the large-scale productivity and stability of the photobioreactor system presented. This combined with the added value of total fatty acid content for each harvested batch fills a current void in the critical assessment of the current near-term realizable productivity potential of microalgae at large-scale.
When compared with the published performance of open cultivation systems, these results are very encouraging. A primary limitation of open raceway cultivation is the widespread occurrence of culture failures and take-overs (Benemann and Oswald, 1996). The integration of photobioreactor cultivation systems with open raceway pond growth facilities is an obvious means of providing a reproducible and reliable source of clean culture inoculum in large-scale production facilities.
3.2. Culture stability
3.3. Energy consumption
The Nannochloropsis cultures described here were very resistant to culture failures or take-over by competing phytoplankton. An illustration of the longevity of cultures cultivated in the photobioreactors presented can be found in Table 1 which illustrates the obvious advantage of continuous cultivation without culture failures through multiple growing seasons. Serial batch cultures utilized a portion of each harvest to inoculate the next batch which produced the batch lineages shown in Table 1. None of the batch lineages were terminated due to culture failure. These results demonstrate highly reliable system performance over multiple years of production. Features of the cultivation system design and operation that are likely to have contributed to this stability include protection of the cultures from wind-blown particulates, filtering of the growth media and control of temperature variations to maintain the strains near their temperature optima. The cultures were not only stable but resistant to invasive takeover. The C18:3 fatty acid content (Table 2) is commonly used as a proxy for contamination as it is commonly found in green microalgae including Tetraselmis sp (Khozin-Goldberg and Boussiba, 2011). Those authors report a range of 0.2–1.0% C18:3 relative to total FAMEs for clean laboratory cultures of N. oculata and N. salina. For the non-axenic cultures in the AGS3 and AGS4 systems, a mean C18:3 content of 1.23% (S.D. = 1.2) across 706 batch cultures of N. salina was observed (Table 2). Further support for the stability of the Nannochloropsis cultures was obtained from estimates of Tetraselmis sp. cell counts by analysis of flow cytometry data. Tetraselmis and Nannochloropsis cells appear in a distinct, non-overlapping ‘‘cloud’’ pattern in chlorophyll-fluorescence versus forward-scatter plot, such that cells of each species can be counted separately within a single sample (Supplementary data). Tetraselmis cell counts ranged between 0.1% and 2%, consistent with the low levels of C18:3 fatty acid observed.
Mixing microalgae cultures can increase productivity by increasing the frequency of light to dark cycling of the cells; however, the mixing requirements for achieving a dramatic increase in productivity are multiple times higher than what can be tolerated economically (Qiang and Richmond, 1996). Most of the experimental data characterizing microalgae growth used extreme levels of mixing greater than 5 volume of air per volume of culture per minute (VVM) in order to remove mixing as a variable. However, in a low cell density, short optical-path reactor, low sparge rate (less than 5 VVM), mixing dynamics did not dramatically affect microalgae culture growth rates (Qiang and Richmond, 1996). Based on this finding and preliminary computational fluid dynamics and particle image velocimetry modeling, the sparge rate for the AGS 4 photobioreactor was decreased to 0.5 VVM relative to the generation 3 technology, which operated at a mixing rate of 2.5 VVM. The data presented in Figs. 2 and 3 demonstrate that the decrease in mixing did not have a statistical effect on the biomass or lipid productivity of the system based on a Student’s T-test with a 95% confidence interval. The relationship between growth and mixing energy was further explored through an outdoor growth experiment designed to look at the effect of duty cycle and sparge rate on productivity. Duty cycle was defined as the time on, compared to time off. For example a duty cycle of 5% corresponds to sparge air for 5% of the time as compared to the baseline system. Small scale AGS 4 photobioreactors were operated in biological triplicate at a variety of duty cycles and sparge rates ranging from the baseline of 0.6 VVM operated during daylight hours to 0.2 VVM operated on a 5% duty cycle (operation during daylight). Growth results at these extremes and five points in between are presented in Fig. 5. No significant differences in the growth rates of N. salina was observed as Student’s T-test determined all of the treatments failed to reject
J.C. Quinn et al. / Bioresource Technology 117 (2012) 164–171
the null hypothesis based on a 95% confidence interval as compared to the control data set. The lowest flow setting, 5% duty cycle at 0.2 VVM, yielded major energy savings through decreases in flow (67%) and operation time (95%) which resulted in a relatively small average decrease in growth (23%). Process engineering models illustrate the importance of energy reduction in all of the major unit operations in the microalgae-tobiofuels process as a prerequisite to commercialization (Batan et al., 2010; Davis et al., 2011). The energy consumption in the growth phase of photobioreactors compared to open raceway ponds is typically reported as more energy intensive based on sparge requirements compared to paddle wheel operation (Lehr and Posten, 2009; Stephenson et al., 2010). The data illustrate the efficiency of the AGS designs with respect to gas exchange as well as the magnitude of energy required to move CO2 and O2 through cultivation systems. The results presented here show that process improvements can result in a 98% decrease in gas-movement-related energy consumption. Comparing this energy consumption to that of traditional paddle wheels as reported by Benemann and Oswald (1996) of 10,750 kWhr ha 1 yr 1, the system presented uses dramatically less energy of 3510 kWhr ha 1 yr 1 based on the 5% duty cycle at 0.2 VVM. Detailed assumptions and energy calculations for all of the scenarios presented in Fig. 5 are presented in supplementary data. This result is in direct opposition of traditional thought in terms of energy requirements for mixing in photobioreactors (Benemann and Oswald, 1996; Chisti, 2007; Wijffels and Barbosa, 2010). 4. Conclusions N. oculata and N. salina were cultivated in large-scale outdoor photobioreactors from April 2008 to January 2011 with biomass and corresponding lipid content reported with serial batch cultures of both strains stable over 41 batch transfers. Results from energy consumption analysis show power consumption for mixing and gas transfer in photobioreactors can be similar to open raceway ponds without an effect on productivity. These results illustrate the need for techno–economic, environmental, and scalability assessment to revisit growth assumptions in an effort to more effectively evaluate the microalgae-to-biofuels process. Acknowledgements We gratefully acknowledge financial support provided by Solix BioSystems, Inc. and data collection and processing support from Chris Turner, Mark Machacek, Pete Hentges, John Walden, Greg Wardle, and Zach Turner. Work at New Mexico State University was additionally supported by the US Department of Energy under contract DE-EE0003046 awarded to the National Alliance for Advanced Biofuels and Bioproducts and a grant by the US Air Force Research Laboratory (FA8650-11-C-2127). Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j.biortech.2012. 04.073. References Batan, L., Quinn, J., Willson, B., Bradley, T., 2010. Net energy and greenhouse gas emission evaluation of biodiesel derived from microalgae. Environ. Sci. Technol. 44, 7975–7980.
171
Benemann, J.R., Oswald, W.J., 1996. Systems and economic analysis of microalgae ponds for conversion of CO2 to biomass. Final report: March 1996. Bosma, R., Van Zessen, E., Reith, J.H., Tramper, J., Wijffels, R.H., 2007. Prediction of volumetric productivity of an outdoor photobioreactor. Biotechnol. Bioeng. 97, 1108–1120. Bucy, H., Baumgardner, M., Marchese, A.J., 2012. Chemical and physical properties of algal methyl ester biodiesel containing varying levels of methyl eicosapentaenoate and methyl docosahexaenoate. Algal Research. http:// dx.doi.org/10.1016/j.algal.2012.02.001. Campbell, P.K., Beer, T., Batten, D., 2011. Life cycle assessment of biodiesel production from microalgae in ponds. Bioresour. Technol. 102, 50–56. Chisti, Y., 2007. Biodiesel from microalgae. Biotechnol. Adv. 25, 294–306. Chisti, Y., 2008a. Biodiesel from microalgae beats bioethanol. Trends Biotechnol. 26, 126–131. Chisti, Y., 2008b. Response to Reijnders: do biofuels from microalgae beat biofuels from terrestrial plants? Trends Biotechnol. 26, 351–352. Clarens, A.F., Resurreccion, E.P., White, M.A., Colosi, L.M., 2010. Environmental life cycle comparison of algae to other bioenergy feedstocks. Environ. Sci. Technol. 44, 1813–1819. Davis, R., Aden, A., Pienkos, P.T., 2011. Techno–economic analysis of autotrophic microalgae for fuel production. Appl. Energy 88, 3524–3531. Frank, E.D., Han, J., Palou-Rivera, I., Elgowainy, A., Wang, M.Q., 2011. Life-cycle analysis of algal lipid fuels with the greet model. Center for Transportation Research, Energy Systems Division, Argonne National Laboratory. Hirano, A., Hon-Nami, K., Kunito, S., Hada, M., Ogushi, Y., 1998. Temperature effect on continuous gasification of microalgal biomass: theoretical yield of methanol production and its energy balance. Catal. Today 45, 399–404. Hu, Q., Guterman, H., Richmond, A., 1996. A flat inclined modular photobioreactor for outdoor mass cultivation of photoautotrophs. Biotechnol. Bioeng. 51, 51–60. Hu, Q., Sommerfeld, M., Jarvis, E., Ghirardi, M., Posewitz, M., Seibert, M., Darzins, A., 2008. Microalgal triacylglycerols as feedstocks for biofuel production: perspectives and advances. Plant J. 54, 621–639. Jorquera, O., Kiperstok, A., Sales, E.A., Embirucu, M., Ghirardi, M.L., 2010. Comparative energy life-cycle analyses of microalgal biomass production in open ponds and photobioreactors. Bioresour. Technol. 101, 1406–1413. Khozin-Goldberg, I., Boussiba, S., 2011. Concerns over the reporting of inconsistent data on fatty acid composition for microalgae of the genus Nannochloropsis (Eustigmatophyceae). J. Appl. Phycol. 23, 933–934. Lardon, L., Helias, A., Sialve, B., Stayer, J.P., Bernard, O., 2009. Life-cycle assessment of biodiesel production from microalgae. Environ. Sci. Technol. 43, 6475–6481. Lehr, F., Posten, C., 2009. Closed photo-bioreactors as tools for biofuel production. Curr. Opin. Biotechnol. 20, 280–285. Li, Y., Horsman, M., Wu, N., Lan, C.Q., Dubois-Calero, N., 2008. Biofuels from microalgae. Biotechnol. Prog. 24, 815–820. Mata, T.M., Martins, A.A., Caetano, N.S., 2010. Microalgae for biodiesel production and other applications: a review. Renewable Sustainable Energy Rev. 14, 217– 232. Qiang, H., Richmond, A., 1996. Productivity and photosynthetic efficiency of Spirulina platensis as affected by light intensity, algal density and rate of mixing in a flat plate photobioreactor. J. Appl. Phycol. 8, 139–145. Quinn, J., de Winter, L., Bradley, T., 2011. Microalgae bulk growth model with application to industrial scale systems. Bioresour. Technol. 102, 5083–5092. Rodolfi, L., Zittelli, G.C., Bassi, N., Padovani, G., Biondi, N., Bonini, G., Tredici, M.R., 2009. Microalgae for oil: strain selection, induction of lipid synthesis and outdoor mass cultivation in a low-cost photobioreactor. Biotechnol. Bioeng. 102, 100–112. Schenk, P.M., Thomas-Hall, S.R., Stephens, E., Marx, U.C., Mussgnug, J.H., Posten, C., Kruse, O., Hankamer, B., 2008. Second generation biofuels: high-efficiency microalgae for biodiesel production. BioEnergy Res. 1, 20–43. Schutter, M.E., Dick, R.P., 2000. Comparison of fatty acid methyl ester (FAME) methods for characterizing microbial communities. Soil Sci. Soc. Am. J. 64, 1659–1668. Scott, S.A., Davey, M.P., Dennis, J.S., Horst, I., Howe, C.J., Lea-Smith, D.J., Smith, A.G., 2010. Biodiesel from algae: challenges and prospects. Curr. Opin. Biotechnol. 21, 277–286. Sheehan, J., Dunahay, T., Benemann, J., Roessler, P., 1998. A look back at the US department of energy’s aquatic species program: biodiesel from algae. NREL report, TP-580-24190. Stephenson, A.L., Kazamia, E., Dennis, J.S., Howe, C.J., Scott, S.A., Smith, A.G., 2010. Life-cycle assessment of potential algal biodiesel production in the United Kingdom: a comparison of raceways and air-lift tubular bioreactors. Energy Fuels 24, 4062–4077. Volkman, J.K., Brown, M.R., Dunstan, G.A., Jeffrey, S.W., 1993. The biochemicalcomposition of marine microalgae from the class eustigmatophyceae. J. Phycol. 29, 69–78. Weissman, J.C., Tillett, D.M., 1990. Design and Operation of an Outdoor Microalgae Test Facility: Large-Scale System Results. NREL, Golden, CO. Wijffels, R.H., Barbosa, M.J., 2010. An outlook on microalgal biofuels. Science 329, 796–799. Williams, P.R.D., Inman, D., Aden, A., Heath, G.A., 2009. Environmental and sustainability factors associated with next-generation biofuels in the US: what do we really know? Environ. Sci. Technol. 43, 4763–4775.