An economic, sustainability, and energetic model of biodiesel production from microalgae

Bioresource Technology 111 (2012) 191–200

Contents lists available at SciVerse ScienceDirect

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

An economic, sustainability, and energetic model of biodiesel production from microalgae F. Delrue a,⇑, P.-A. Setier a, C. Sahut b,c,d, L. Cournac b,c,d, A. Roubaud a, G. Peltier b,c,d, A.-K. Froment a,⇑ a

CEA, Direction de la Recherche Technologique, Laboratoire des Technologies de la Biomasse, 17 Rue des Martyrs, F-38054 Grenoble, France CEA, Direction des Sciences du Vivant, Institut de Biologie Environnementale et de Biotechnologie, Laboratoire de Bioénergétique et Biotechnologie des Bactéries et Microalgues, CEA Cadarache, Saint-Paul-lez-Durance 13108, France c CNRS, UMR 6191 Biologie Végétale et Microbiologie Environnementale, Saint-Paul-lez-Durance 13108, France d Aix Marseille Université, UMR 6191 Biologie Végétale et Microbiologie Environnementale, Saint-Paul-lez-Durance, Saint-Paul-lez-Durance 13108, France b

a r t i c l e

i n f o

Article history: Received 25 October 2011 Received in revised form 3 February 2012 Accepted 4 February 2012 Available online 14 February 2012 Keywords: Algal biodiesel GHG emission rate Global sensitivity analysis Net energy ratio Production cost Water footprint

a b s t r a c t A new process evaluation methodology of microalgae biodiesel has been developed. Based on four evaluation criteria, i.e. the net energy ratio (NER), biodiesel production costs, greenhouse gases (GHG) emission rate and water footprint, the model compares various technologies for each step of the process, from cultivation to oil upgrading. An innovative pathway (hybrid raceway/PBR cultivation system, belt filter press for dewatering, wet lipid extraction, oil hydrotreating and anaerobic digestion of residues) shows good results in comparison to a reference pathway (doubled NER, lower GHG emission rate and water footprint). The production costs are still unfavourable (between 1.94 and 3.35 €/L of biodiesel). The most influential parameters have been targeted through a global sensitivity analysis and classified: (i) lipid productivity, (ii) the cultivation step, and (iii) the downstream processes. The use of low-carbon energy sources is required to achieve significant reductions of the biodiesel GHG emission rate compared to petroleum diesel. Ó 2012 Elsevier Ltd. All rights reserved.

1. Introduction Microalgae have been considered recently as a feedstock with great potential for biofuel production since they present many advantages (e.g. high productivity, lipid accumulation, and ability to grow on waste). However, major drawbacks limit the industrial development of microalgae biodiesel. One of these technical challenges is related to the dilution of microalgae biomass, typically around 0.5–1 kg/m3 in open ponds and 5–10 kg/m3 in photobioreactors (PBRs). Extremely high volumes of water need to be processed during cultivation, harvesting and especially drying. This leads to a high electricity and heat consumption. After the first studies from Weissman and Goebel (1987) or Benemann and Oswald (1996), numerous techno-economic analyses and life cycle analyses (LCA) have been published especially over the last two or three years with the principal objective to evaluate the real potential of microalgae biodiesel production at fullscale. Techno-economic analyses generally focus on production costs only. They usually point out the high production cost of both ⇑ Corresponding authors. Tel.: +33 4 38 78 34 07; fax: +33 4 38 78 52 51 (F. Delrue), tel.: +33 4 38 78 48 93; fax: +33 4 38 78 52 51 (A.-K. Froment). E-mail addresses: fl[email protected] (F. Delrue), [email protected] (A.-K. Froment). 0960-8524/$ - see front matter Ó 2012 Elsevier Ltd. All rights reserved. doi:10.1016/j.biortech.2012.02.020

microalgae biodiesel (from transesterification of oil) and green diesel (from hydrotreating of oil) from 1.06 €/L in the best cases (in open ponds for Lundquist et al., 2010) up to 7.44 €/L (Rosenberg et al., 2011). Despite some discrepancies in the results, Sun et al. (2011) showed that this variability can be greatly reduced if a normalised set of input assumptions is used in the models. LCA studies investigate energy balances and GHG emissions. The energy balance of microalgae diesel production seems to be very fragile (net energy ratio, NER, ratio between the energy produced and the primary energy consumed in the process, of 1.08 for Batan et al., 2010). The NER depends on the type of cultivation process used (NER 1 for raceway ponds and flat-plate PBRs, data from Jorquera et al., 2010) or on the combination of both cultivation and conversion processes used (Clarens et al., 2011). These analyses point out the necessity to recover the energy content of the microalgae residue after lipid extraction (Lardon et al., 2009) which can help in reducing the biodiesel production cost as well as the carbon emission (by 33% and 75%, respectively for Harun et al., 2011). Another important criterion for evaluating a process is its water consumption which is rarely considered but can significantly impact the environmental sustainability of the process. Subhadra and Edwards (2011) calculated algal biodiesel water footprints of 23 and 62 L/MJ (920 and 2480 L of water/L of biodiesel at 40 MJ/L of LHV, low heating

192

F. Delrue et al. / Bioresource Technology 111 (2012) 191–200

value) for different scenarios (i.e. two or three co-products and biomass concentrations of 0.5 or 1.0 g/L). These results highlight the need to use non-potable water. The three criteria altogether (energetic, economic, and environmental) have their importance in the evaluation of a process, especially for the biofuel industry. This present work aims at developing an improved model for microalgae diesel production that simultaneously evaluates these three criteria. Since both the transesterification and hydrotreating pathways have been considered, the term biodiesel has been used thereafter for both biodiesel and green diesel non-discriminately. A major disadvantage of usual techno-economic analysis is that it generally considers one technology for each step of the process which does not allow the comparison of the multiple available technologies and their optimised process pathways. For these reasons, this study proposes a model with various technologies for each step of the process. Also, to account for the uncertainties on the microalgae biodiesel production industry, the model parameters are varying within a range. Only local sensitivity analyses (SA) have been published so far. For example, Davis et al. (2011) defined three scenarios and concluded that the lipid content is the most influential parameter for both open ponds and PBRs. However, to provide relevant results, local SA involves defining a base case where a model simulation is run with all parameters set to a fixed value. Each parameter is perturbed in turn, keeping all other parameters at their fixed value. The main disadvantage of local SA for microalgae biofuel production is the lack of public data regarding full-scale optimised model parameters (e.g. lipid and biomass productivity, operating conditions) needed to define the base case. A global SA differs in that all the model parameters vary simultaneously with no need for a base case. To overcome the drawback of local SA, a global SA has been performed in this study to determine the most influential model parameters. The proposed model is also used to define an innovative pathway for microalgae biodiesel production, and show which targets should be considered with priority for future research and development efforts.

2. Methods 2.1. System description The system considered by the model includes all the process steps from microalgae cultivation to biodiesel (Fig. 1). The production of algal biodiesel is assumed to be located in a sunny area in France (South-East), with sources of CO2 from flue gases and wastewater at disposal. These assumptions may be considered as optimistic but have been chosen to simplify the model. The production infrastructures are included in the economic estimations but are not considered in the energy balance and in the GHG emission rate calculations. The total volume of microalgae cultivation is set at 1000,000 m3 which is equivalent to a cultivation surface of 333.3 ha of 30 cm-depth raceways. This volume has been considered as a reasonable prospection of what would be an industrial facility of microalgae cultivation dedicated to biofuel production. Mass balance calculations for nutrients are based on the microalgae formula from Grobbelaar (2004): CO0.48H1.83N0.11P0.01. On this particular point, various elemental analyses have been conducted by different research teams with similar results (Supplemental Table 1). The water refill is assumed to be done using wastewater in order to limit the impact of water on the three criteria (economic, energetic and environmental) and to benefit from the wastewater treatment potential of microalgae (Park et al., 2011). The exceeding nutrient demand is delivered by additional

nutrients (ammonium diphosphate for phosphorus and anhydrous ammonia for nitrogen). Water is recycled from the harvesting, dewatering and drying steps to the cultivation step (Fig. 1). The recycled fraction is estimated by the model (see Section 2.5 for details). 2.2. Model description The model is based on the mass and energy balances equations coming from process evaluation of each technology. Model parameters listed in Table 1 have been optimally reduced in order to minimise possible degrees of freedom and to avoid any dependence between model parameters (required for the global sensitivity analysis). A single parameter for the lipid content, defined as convertible lipids (CLs), has been selected (Table 1). CLs account for lipids that are effectively extracted from the harvested biomass and converted into biofuel. This principally concerns triacylglycerides (TAGs) since phospholipids are to be removed (by a degumming process not considered in the present model) in order to avoid phosphorus in the biofuel (Lu et al., 2009). 2.2.1. Monte Carlo sampling method Since technologies considered in this study are not yet mature, a Monte-Carlo sampling method has been adopted in order to account for the variability and the uncertainty of the parameter value. According to this method, each parameter is defined by its minimum and maximum values between which they vary in a random manner. The average values of the model parameters (for mass and energy balances and for economic evaluation) have been determined based on a detailed and critical literature review. The data used originates from peer-reviewed literature as well as technical reports. Then, the minima and maxima have been defined using methods depending on the data availability: (i) from the literature, if consistent minima and maxima are available (ii) from the value found in the literature plus or minus 50% (iii) from the results of the ProSimPlusÒ simulations for lipid extraction since its energy demand is driven by the varying solvent to biomass ratio. 2.2.2. Capital cost estimation Assuming an overnight construction, the capital costs of each process have been evaluated based on the literature survey for well-documented processes (raceways, PBRs, harvesting, drying, anaerobic digestion and gasification) and on estimations based on the method described by Chauvel et al. (2001) for lipid extraction technologies and conversion processes. All prices have been updated to 2011 Euros. Conversions of dollar into euro have been made based on an exchange rate of 1.42 €/$ (data from April 2011). The capital cost estimations using the method from Chauvel et al. (2001) have been assumed to be in the range [50%; +50%] (values in Supplemental Table 8). 2.2.3. Technologies considered by the model This section describes all the technologies that have been considered for the model. The calculations made to estimate the energy demands and the capital costs as well as the references used in these calculations have not been included but are available in the Supplemental Data together with comparison of results to other techno-economic analyses and LCA. Concerning the process choices, the aim here is not to be exhaustive but rather to select reference processes (such as n-hexane lipid extraction or alkaline transesterification) together with innovative ones, representative of promising process approach: for example hybrid PBR/raceway cultivation systems, low-cost dewatering technologies (e.g. belt filter press or solar drying) and wet extraction processes (like di-methyl ether, DME, lipid extraction, from Kanda and Li, 2011).

193

F. Delrue et al. / Bioresource Technology 111 (2012) 191–200

Fig. 1. System boundaries: five-steps process train including all technologies considered and associated inputs (energy and chemicals).

Table 1 Model parameters used for mass balances and system definition.

a

Parameter

Unit

Minimum

Maximum

PBR/raceway ratio CO2 input concentration PBR productivity Raceway productivity Cultivation step HRTa Raceway circulation rate Raceway depth Raceway evaporation rate Convertible lipid contentb Harvesting duration Water refill duration Water recycling fraction Clarification HRT Biomass concentration after clarification Thickening HRT Biomass concentration after thickening Biomass concentration after 1st drying step (if centrifugation or belt filter press) Biomass concentration after 1st drying step (if bed drying) Biomass concentration after solar drying Lipid extraction efficiency Use of methanol for transesterification Use of NaOH for transesterification Use of H3PO4 for transesterification Use of H2 for hydrotreating Anaerobic digestion yieldd Percentage of nitrogen and phosphorus recirculated through the digestate of the anaerobic digestione Natural gas boiler thermal efficiency Microalgae LHV Biodiesel LHV Biodiesel density

m3/m3 % kg/m3  day g/m2  day day m/s m m/year % h/day h/day – h % DWB h % DWB % DWB % DWB % DWB – kg/ton of CLc kg/ton of CL kg/ton of CL kg/ton of CL m3 of CH4/kg of algal residue – – MJ/kg MJ/kg kg/L

0.3 5 0.65 20 2 0.2 0.2 1.8 20 1 1 0.8 0.5 1.0 12 2.5 20 25 60 0.9 50 4.3 3.5 7.5 0.2 0.3 0.8 15 38 0.87

0.7 15 1.95 30 6 0.3 0.3 2.4 50 8 8 0.99 2 2.0 48 3.5 30 35 80 0.99 150 12.9 10.5 22.5 0.4 0.6 0.89 22 45 0.92

Hydraulic retention time. Lipids in the harvested microalgae that can be extracted and then converted into biodiesel, estimations from microalgae lipid content and triacylglycerides (TAGs) percentage in these lipids. c CL: convertible lipids. d Based on calculations on data from Sialve et al. (2009). e Based on calculations on data from Ras et al. (2011). b

194

F. Delrue et al. / Bioresource Technology 111 (2012) 191–200

The model is designed to include raceways, PBRs and also hybrid systems (e.g. PBR for biomass growth and raceway for lipid accumulation) that are defined by the PBR to raceway ratio. The raceway ponds design is based on Lundquist et al. (2010): 4-ha ponds mixed using paddlewheels (velocities between 0.2 and 0.3 m/s). An evaporation rate between 1.8 and 2.4 m/year has been assumed (Lundquist et al., 2010) which is well adapted for South of France (see Supplemental Data). The biomass productivity in raceways has been assumed to vary between 20 and 30 g/m2/d, in comparison with the 25 g/m2/d raceway productivity for Davis et al. (2011). Their capital cost range has been evaluated at the large cost range of [8–50] €/m3 of raceway (due to high variations in the literature). Concerning PBRs, no specific design has been developed in this work since the authors believe that PBR’s design will greatly improve in the future and they did not want to limit the extent of this study on one PBR design only. Therefore, global assumptions have been made to allow mass and energy balances as well as the economic calculations. The biomass productivity in PBRs is based on the assumption of 1.25 kg/m3/d from Davis et al. (2011) and the range of [0.65–1.9] kg/ m3/d has been chosen to account for conservative (0.65 kg/m3/d) and very optimistic productivity (1.9 kg/m3/d). PBRs’ energy requirements are limited to the electricity needed for CO2 blowers and the pumps (wastewater inlet and recycled water) which is already the case in flat-plate PBR (Posten, 2009). This implies that the PBR design is optimised such that sufficient mixing occurs. PBR capital cost has been estimated to vary in the range [625– 1875] €/m3. Cooling is done by sprinklers at 1 L/m3/min (12 h/ day with a 5% blowdown, adapted from Davis et al., 2011). The CO2 input (from flue gas) is estimated based on the biomass needs, on the CO2 input concentration and on the assumption that 50% of the CO2 is effectively converted into biomass. The remaining 50% is lost to the atmosphere (directly in raceways and through degassing in PBRs). Verrecht et al. (2010) estimated that the energy demand for pumping wastewater sludge with a total head loss of 3 m and a pump efficiency of 50% is 0.016 kW h/m3. They also evaluated the energy demand to be 0.025 kW h/m3 for a blower with an outlet pressure of 106,300 Pa, a sump depth of 5 m (a conservative value since raceway sump depth is generally lower, example of 1 m in Lundquist et al., 2010), a blower efficiency of 60% and allowing for losses incurred in the pipework. Since these estimations correspond well with data from manufacturers, they were used to estimate electricity for water and slurry pumping and for CO2 aeration. All the references used for the raceways and PBR capital cost estimations (see Section 3. of the Supplemental Data) include the cost of a CO2 delivery system, then, a specific estimation has not been performed in this work. The extra-energy needed for pumping for PBRs due to a lower volume to area ratio than raceways has not been accounted by the model. Harvesting is done by a coagulation-flocculation and decantation step (capital cost range of [25–85] €/m3 for a settler). Autoflocculation and/or low cost organic coagulant would be profitable for the biodiesel microalgae production, but no up-scalable data is available yet. Therefore, ferric chloride has been considered for the calculations ([3.7–6.1] € per ton of dry weight biomass, TDWB). No electricity demand has been accounted for mixing during coagulation. After this decantation step, the algae slurry is processed to the thickeners (capital cost range of [200–450] €/m3 of thickeners). After 12–48 h, the slurry is pumped to the drying stage. Four drying processes have been adapted from wastewater treatment technologies (solar drying, belt filter press, centrifugation and bed drying, see Fig. 1). Solar drying is accomplished in a greenhouse with a capital cost range of [50–150] €/ton of water evaporated (TWE)/year. The only energy requirements considered is for the automated machine that mix and convey the biomass from one

end of the greenhouse to the other in 30 days (20–40 kW h/TWE). Belt filter press and centrifugation require electricity ([0.18–0.55] and [4–12] kW h/TWE, respectively) and flocculant (polymer at a cost of 30–40 €/TDWB). Their capital cost ranges have been estimated to be [0.25–0.75] €/TWE/year for belt filter press and [0.5–1.5] €/TWE/year for centrifugation. For the drying beds, electricity needed for biomass pumping has been considered ([0.25–0.75] kW h/TWE). However, the additional labour cost for handling the biomass after drying has not been considered in the model. The capital cost range of belt filter press has been evaluated at [3.6–9.8] €/TWE/year. Additional drying when needed (i.e. when dry lipid extraction is selected by the model) is achieved by a natural gas thermal dryer (capital cost range of [35–100] €/TWE/year). An optimistic heat consumption of 300 kW h/TWE has been assumed as the minimum (belt thermal dryer) and a non-optimised value of 900 kW h/TWE has been defined as the maximum (drum dryer). A natural gas boiler is used for delivering heat for lipid extraction, transesterification, hydrotreating and anaerobic digestion (thermal efficiency between 0.8 and 0.89, EPA, 2008). The capital cost range for this natural gas boiler has been estimated at [8600–23,900] €/MW. Hexane and DME extraction processes have been simulated in ProSimPlusÒ using an extraction column with consideration for solvent recovery (see Supplemental Data for flowsheets and more information). Solvent loss has been assumed at 1% of the extracted oil. Similarly, the extracted lipids have been assumed to be 99% pure. The minimum and maximum energy requirements have been estimated by varying the solvent to biomass ratio between 10 and 20 for n-hexane (for a 90% Dry Weight Biomass, DWB) and 20–30 for DME (for a 20–35% DWB). DME lipid extraction is done at 500 kPa and 20 °C by liquid DME (Kanda and Li, 2011) and the oil/DME mixture is then heated at 60 °C in order to recover DME. The use of a heat exchanger between the recovered and re-compressed DME and the biomass entering the extraction column may significantly reduce the natural gas consumption in the boiler. With the adequate heat exchange area (700 m2 for a 400-ha farm producing 80 TDWB per day), the reduction vary from 50% in the most conservative case up to 100%. Therefore, the heat demand for DME lipid extraction has been set at [0–1.13] kW h/kg of DWB and at [0.87–1.74] kW h/kg of DWB for n-hexane through the ProSimPlusÒ simulations. Also, the electricity demand for lipid extraction has been evaluated at [0.00024–0.00045] kW h/kg of DWB for n-hexane and [0.41–0.61] kW h/kg of DWB for DME. The capital cost of the equipments for lipid extraction has been estimated at [40–120] €/TDWB/year for n-hexane and at [75–225] €/TDWB/year for DME. Transesterification and hydrotreating have also been simulated in ProSimPlusÒ to estimate the electricity demand (range of [0.00019–0.00057] kW h/kg of CL for transesterification and [0.0008–0.0024] kW h/kg of CL for hydrotreating), the heat demand (range of [0.34–1.01] kW h/kg of CL for transesterification and [0.16–0.47] kW h/kg of CL for hydrotreating) as well as the chemicals needed (methanol, NaOH and H3PO4 for transesterification, H2 for hydrotreating). Capital cost ranges for this conversion step have been estimated at [170–510] €/ton of CL per year for transesterification and at [70–210] €/ton of CL/year for hydrotreating. Hydrogen (used for hydrotreating) is assumed to be obtained from steam reforming. After this conversion step, the algal biodiesel is assumed to be ready for use. The microalgae residues after lipid extraction have been considered for energy recovery using either anaerobic digestion (electricity demand of [0.05–0.2] kW h/kg of residue and heat demand of [0.1–0.3] kW h/kg of residue) or gasification (energy efficiency from 0.727 up to 0.824, Brown et al., 2009). The capital costs have been estimated to range between 65 and 190 €/ton of residue/year for anaerobic digestion and between 100 and 300 €/ton of residue/

F. Delrue et al. / Bioresource Technology 111 (2012) 191–200

year for gasification. The gas produced by either gasification or anaerobic digestion is directly used in the boiler as replacement of natural gas. When anaerobic digestion is selected by the model, the digestate (residue of the digestion) is recycled to the growth medium. Based on calculations on data from Ras et al. (2011), we assumed that between 30% and 60% of the nitrogen and the phosphorus content of the digestate are biologically available to the microalgae (Table 1). This supposes that the non-biologically available minerals of the digestate are eliminated by using separation techniques (filtration or chemical precipitation) in order to keep a reasonable mineral load in the process. 2.3. Net energy ratio The net energy ratio (NER) is defined as the ratio between the ‘‘Energy produced’’ and the ‘‘Primary energy input’’. The term ‘‘Energy produced’’ comprises both the energy contained in the biodiesel and in the gas from gasification or anaerobic digestion of the algal residues. The term ‘‘Primary energy input’’ accounts for the electricity and the natural gas needed by the processes to produce biofuel as well as the primary energy required to produce the main chemicals used (the nutrients, the extraction solvents, H2 for hydrotreating, methanol, caustic soda and H3PO4 for transesterification). 2.4. Production cost estimation The biodiesel production cost is estimated using the method detailed in Supplemental Table 9 (Chauvel et al., 2001). This is the ratio between the total operating cost over a year and the annual production of biodiesel. The total operating cost is the sum of the operating cost (utilities, labour and other costs) and the fixed cost. The fixed cost is calculated from the depreciable capital cost (overnight construction) that includes 55% of the capital cost for the general maintenance, storage, engineering and spare parts costs, licence fees (fixed at 0.5 M€), initial expenses at 2% of the capital cost and process start-up cost at 25% of the operating cost. The calculation is based on annuities of 20 years, discount rate of 8% and 7% of the capital cost per year for maintenance cost, taxes, insurance and business expenses. The production cost includes a carbon credit in the range [10–50] €/ton of CO2. A credit range of [0.04– 0.12] €/m3 of wastewater is also attributed. Wastewater treatment cost has been estimated at 0.16 €/m3 with data from Michel and Mirgon (2008) and a 25–75% credit rate is assumed since additional treatment would be needed. These carbon and wastewater credit values are very conservative but the reason for their implementation in the model is conceptual since their exact values are difficult to evaluate. These credit values could be modified in a future version of the model depending on the evolution of public policies.

195

during the dewatering and drying steps and the recycle fraction (ranging between 0.8 and 0.99). The maximum value of 0.99 for the water recycling fraction has been assumed as an optimal recycle fraction with very few losses. The minimum conservative value of 0.8 (Subhadra and Edwards, 2011) also considers the need to withdraw some of the water out of the system in order to limit the concentrations of the micropollutants and salts accumulating from the incoming wastewater. 2.6. Sensitivity analysis Usual local SA uses variations around a base case of a model. However, such baseline conditions are not yet established by the microalgae biofuel production industry. No data of industrial scale algae-to-biofuel process exists to date making extrapolation very difficult. Major improvements are also expected in every aspect of the process. For these reasons, a global SA has been preferred as it does not need a base case and model parameters can vary on a large range (which is already the case in the model). The Saltelli-Sobol indices Si (Saltelli, 2002) are used to estimate the influence of each parameter on the four outputs which are: the NER, the production cost, the GHG emission rate and the water footprint. Given a model Y = f(X1, X2, . . . Xk), with Y a scalar and (X1, X2, . . . Xk) uncorrelated input model factors, the first order sensitivity coefficient Si for a generic factor Xi is written (Eq. (1)).

Si ¼

VðEðY jX i ÞÞ VðYÞ

ð1Þ

V measures the variance and E(Y|Xi) is the mean of Y taken over all factors except Xi. The sum of Si is equal to 1 and they represent the influence over the output variable Y. These indices are evaluated with a numerical method requiring a Monte-Carlo sampling (10,000 samples for each run of the model). The purpose of this article is not to detail this calculation method which can be found in the article cited above (Saltelli, 2002). 3. Results and discussion 3.1. Impact of the technology choice on the process evaluation The impact of the technologies on the evaluation of the whole process has been estimated using the process evaluation model. A mixed scenario has been defined as follows: (i) a PBR to raceway ratio varying from 0.3 to 0.7 in order to account for the mixed PBR/raceway cultivation option where PBRs and raceways are both taking an important part of the cultivation step (>30%) and (ii) a random choice of technology for each step. The results of a model run for this scenario is presented in Fig. 2. The results are ranked by class (500) and form a pseudo-Gauss curve. The parameters Y25% and Y75% are defined as follow (Fig. 2):

2.5. Environmental assessment The GHG emission rates of producing and using microalgae biodiesel are calculated by the model. Similarly to the NER, all the GHG emissions of producing natural gas, electricity and main chemicals used in the process are accounted in the calculation of the GHG emission rate. It is expressed in kgCO2-eq for 100 km of transportation. For comparison purposes, calculations using data from Cherubini et al. (2009) show an emission rate of 20.2 kgCO2-eq/100 km for diesel. The parameters that are used in the estimation of the GHG emission rate are listed in Supplemental Table 10. The model can also estimate the water footprint (in litres of water consumed by litre of biodiesel produced) by using the evaporation rate, the water used for cooling the PBRs, the water loss

– Y25% is the output value under which 25% of the output values is contained. – Y75% is the output value under which 75% of the output values is contained. The range [Y25%–Y75%] that contains 50% of the output values has been used afterwards in order to characterise the results of the model. For each of the 25,000 Monte Carlo samples of one run, every process step is allocated randomly one of the previously evaluated technologies. Scenarios have been made by forcing the model to select one technology (e.g. the model automatically select transesterification for the process conversion step in the ‘‘transesterification’’ scenario). For each step of the process, the technologies are compared based on the Y25% and Y75% of the four model

196

F. Delrue et al. / Bioresource Technology 111 (2012) 191–200

Fig. 2. Definition of Y25% and Y75%, practical parameters for technology comparison and evaluation.

outputs of their specific simulation. Table 2 shows the results of this comparison. The water footprint is based on two parameters: the evaporation rate (for raceways) and the water recycling fraction. Therefore, the change in water footprint from one simulation to the other is only driven by the cultivation step. It does not include the water footprint associated with material inputs to the process (nutrients, extraction solvents, . . .) and the water used for cooling the PBRs is the same for each simulation (1 L/m3/min, 12 h/day with a 5% blowdown). The water footprint of Table 2 should be equal except for the raceway and PBR scenarios. The relative standard deviations (RSD) of these water footprints of 0.4% and 0.5% have been calculated for Y25% and Y75%, respectively. These low RSD demonstrate the excellent repeatability of the model. For the mixed scenario, 83% of the results have a NER above 1 and still 58% of the results have an NER higher than the NER of ethanol from corn grain dry-milling (NER = 1.25) calculated by Hill et al. (2006). However, only 14% of the results has a NER higher than the estimation of Hill et al. for soybean biodiesel (NER = 1.93). The production cost is very high when comparing to petroleum diesel (0.58 €/L after refining, without any taxes, EIA, 2011). Then, the GHG emission rate in the mixed scenario is reduced for 80% of the results compared to diesel GHG emission

rate (20.2 kgCO2-eq/100 km) with a reduction comprised between 8% and 50% relative to petroleum diesel (based on the Y25% and Y75% values). In comparison, Hill et al. (2006) calculated that biodiesel from soybean led to a reduction in GHG emission of 41% (compared to petroleum diesel) and 12% for corn bioethanol (compared to gasoline). Water footprint values are comprised between 230 and 410 L of water/L of biodiesel for the mixed scenario comparing favourably with other biodiesel (Table 2). Petroleum diesel has a far lower water footprint (12.9 L of water/L of diesel, King and Webber, 2008). However, the water used for the petroleum extraction process has to be treated after being used, increasing the environmental impact of petroleum diesel. From Table 2, it can be noted that raceways and PBRs have similar NER and that the GHG emission rates are slightly lower for raceways. However, the water footprint is five-times higher in raceways since the high liquid surface exposed to air leads to large amount of water evaporated. The biodiesel production cost is around 80% higher for PBRs. Actually, the capital cost of PBRs is responsible for 58% of the biodiesel production cost in the PBR scenario on average when it is 29% for raceways in the raceway scenario (Fig. 3). Total capital investment is responsible of more than 60% of the total production cost in the raceway simulation and more than 70% in the PBR simulation. Our biodiesel production cost values for raceways ([1.30–2.53] €/L of biodiesel) are very close to the value (1.83 €/L of biodiesel) calculated by Davis et al. (2011). These authors also estimated the production cost of algal biodiesel using PBR as the cultivation system and the cost increased to 3.82 €/L (Davis et al., 2011) which is also in the range estimated in this study ([2.29–4.34] €/L). Based on the four evaluation criteria, bed drying seems to be the more competitive drying process (Table 2). However, solar-driven processes (solar drying and bed drying) are suspected to cause damage to lipids through oxidation due to long drying cycle times (Lardon et al., 2009) and therefore, belt filter press drying has been preferred in the innovative pathway (Table 3). Concerning solar drying, low-cost options are needed for the greenhouses since their cost is prohibitive at present (Table 2). DME lipid extraction consumes less energy and emits less GHG than n-hexane lipid extraction (Table 2). Thermal drying is needed additionally to the first drying step for n-hexane lipid extraction

Table 2 Y25% and Y75% of the NER, the production cost, the GHG emission rate and the water consumption for the mixed scenario and for the scenarios imposing one technology at a time. Technology fixed, others randomly varying

NER ()

Production cost (€/L of biodiesel)

GHG emission rate (kgCO2-eq/ 100 km)

Water consumption (L of water/L of biodiesel)

Y25%

Y75%

Y25%

Y75%

Y25%

Y75%

Y25%

Mixed scenario

1.09

1.69

2.20

4.18

10.0

18.7

228

409

Cultivation step Raceway PBR

1.08 1.08

1.68 1.69

1.30 2.29

2.53 4.34

9.2 10.1

17.9 18.9

1020 166

1860 276

Dewatering/drying step Centrifugation Belt filter Press Solar drying Bed drying

1.00 1.05 1.12 1.13

1.69 1.82 1.48 1.83

2.07 2.04 3.68 1.97

3.56 3.49 5.95 3.36

10.1 9.7 10.6 9.6

20.6 20.1 16.5 18.5

228 228 229 230

410 405 405 408

Lipid extraction step n-Hexane DME

0.95 1.43

1.26 1.97

2.30 2.19

4.17 4.18

14.8 8.2

23.1 12.8

228 229

411 411

Conversion process Transesterification Hydrotreating

1.06 1.12

1.63 1.76

2.25 2.18

4.20 4.13

10.5 9.5

19.3 18.0

230 227

410 411

Algal residues energy use Anaerobic digestion Gasification

1.18 1.01

1.84 1.55

2.19 2.25

4.13 4.24

9.6 10.4

18.3 19.1

228 228

409 409

Y75%

197

F. Delrue et al. / Bioresource Technology 111 (2012) 191–200

or agricultural wastewaters may also be considered for microalgae cultivation, provided adequate pre-treatments are performed to increase the biodisponibility of nutrients and avoid the toxicity of pollutants. 3.2. Comparison of the two pathways

Fig. 3. Biodiesel production cost allocation for the raceway/PBR mixed, raceway and PBR scenarios. The total production costs indicated on this figure are simulated average values.

since a 90% DWB is required. Large amounts of natural gas are used in the thermal drying leading to higher emission of GHG and more unfavourable energy balance. Wet lipid extraction is a very promising process but its applicability at full-scale is still to be proven. For the conversion processes, hydrotreating shows slightly better evaluation criteria than transesterification: 7% higher NER, 3% lower production cost and 8% lower GHG emission rate (Table 2). These differences might be statistically significant since they are higher than the RSD calculated previously on the water footprints (0.4% and 0.5%). Anaerobic digestion proves to have a better potential than gasification for the algal residues conversion process with higher NER and much lower GHG emission rate (Table 2). The model has also confirmed the importance of using wastewater as nutrients and water supply. In the mixed scenario, wastewater accounted for 7% of the nitrogen and phosphorus input (minimum of 3% and maximum of 30%). These results are identical for both nitrogen and phosphorus since the Redfield ratio (N:P) of wastewater is similar to that of microalgae (Section 1. of Supplemental Data for more details). Additional nutrients supplied for microalgae growth are responsible (on average) of 5% of the biodiesel production cost (minimum of 1% and maximum of 10%), 50% of the energy consumption of the cultivation step (minimum of 18% and maximum of 70%) and 5% of the GHG emission rate (minimum of 1% and maximum of 9%). A basic mass balance calculation based on the nitrogen and phosphorus content of wastewaters, shows that for France, one year of municipal wastewater could support 40% of the annual gasoil and diesel consumption. It clearly shows the potential of using wastewater for biofuel production. Industrial

From these results, two different pathways have been defined. First, the reference pathway, based on processes that have been considered in most of previous techno-economic studies: raceways, centrifugation, thermal drying, n-hexane lipid extraction, transesterification and anaerobic digestion. Secondly, an innovative pathway has been defined gathering promising technologies based on the results from Table 2: PBRs and raceways, bed drying, wet lipid extraction, hydrotreating and anaerobic digestion. The technologies with the best NER, water footprint and GHG emission rate have been chosen for the innovative pathway with less regards towards cost since incentive policies and low-cost equipments (especially PBRs) are expected. The results of the model simulations for these two pathways are detailed in Table 3. The NER doubles from the reference to the innovative pathway and now compares very advantageously to the NER of other biofuels already reported in Section 3.1 (Hill et al., 2006). The inclusion of PBR increases the capital cost and therefore the biodiesel production cost which is clearly noticeable on the values of the Y25% and Y75% of the innovative pathway in Table 3. Thermal drying is not needed in the innovative pathway since the biomass concentration required for wet lipid extraction by DME is 20–30%. Less natural gas is used for the drying step and the GHG emission rate drops significantly. The water footprint for the reference pathway is much higher than for the innovative pathway since the cultivation step only relies on raceways. 3.3. Parameters influence As a result of the SA, Fig. 4 shows the Saltelli-Sobol indices (Si, i.e. the relative influence of the parameter on the model output) of the most influencing parameters for the four evaluation criteria. Low influence parameters (Si < 2.5%) have not been selected. The sums of the Saltelli-Sobol indices are comprised between 73% and 98% indicating that few relations of the second and higher order exist (influence of the combination of two or more parameters). For both pathways, the NER is mainly influenced by the CL content and the heat consumption (Fig. 4a). The anaerobic digestion methane yield is influencing NER for both pathways, which highlighted the necessity to recover the energy of the algal

Table 3 Simulation results for the reference pathway (raceways, centrifugation, thermal drying, n-hexane lipid extraction, transesterification and anaerobic digestion) and the innovative pathway (PBRs and raceways, belt filter press, DME lipid extraction, hydrotreating and anaerobic digestion) in comparison with petroleum diesel, soybean biodiesel and corn ethanol. Pathway

NER () Y25%

Reference Innovative Petroleum diesel Soybean biodiesel Corn bioethanol a b c d e f g h

Y75%

0.93 1.20 1.81 2.42 4.9a 1.43 – 2.5d 1.18 – 2.0g

Production cost (€/L of biodiesel)

GHG emission rate (kgCO2-eq/100 km)

Water footprint (L of water/L of biodiesel)

Y25%

Y75%

Y25%

Y75%

Y25%

Y75%

1.24 1.94 0.58b 0.44e 0.37e

1.95 3.35

16.2 7.0 20.18a 12.01e 19.55e

24.8 11.1

1010 229 12.9c 7500f 1400 – 9800h

1860 410

Cherubini et al. (2009). EIA (2011). King and Webber (2008). Biodiesel from soy, rape seed or sunflower, from Cherubini et al. (2009). Hill et al. (2006). Gerbens-Leenes et al. (2009). Bioethanol from corn, sugar beat or wheat, from Cherubini et al. (2009). Bioethanol from sugar beet to sorghum, from Gerbens-Leenes et al. (2009).

198

F. Delrue et al. / Bioresource Technology 111 (2012) 191–200

Fig. 4. Saltelli-Sobol indices Si for major influencing parameters (i.e. relative parameter influence on the model output, sum of Si equals 1) for the NER (a), the production cost (b), the GHG emission rate (c) and the water consumption (d), reference (left) and innovative pathways (right).

residues (Fig. 4a). Indeed, methane production contributes to 33% of the total energy production on average for both reference and innovative pathways. The CO2 input concentration is also an influencing factor for the NER (Si of 2%, but not appearing in Fig. 4a, and 5% for the reference and innovative pathways, respectively). The energetic demand of the aeration is inversely proportional to the CO2 content in the aeration flux. Aeration blowers have a high energy demand: the model calculated that 53% and 54% of the total

cultivation energy demand is dedicated on average to the CO2 blowers for the reference and innovative pathways, respectively. We found similar parameters influencing the production cost as Davis et al. (2011): the lipid content and the growth rate (Fig. 4b). Additionally, the capital cost of raceways for the reference pathway and PBRs for the innovative pathway appeared to significantly influence the production cost. Similarly to NER, the GHG emission rates of both pathways are directly related to two types

F. Delrue et al. / Bioresource Technology 111 (2012) 191–200

of parameters (Fig. 4c): the CL content and the heat demand. As expected, the water recycling fraction is a crucial parameter for the water footprint for both routes (Fig. 4d). 3.4. Influences by group of parameters Since there are two environmental criteria, the total influence of one parameter (Si,tot.) is calculated as the sum of the influence on the NER (Si,NER), the influence on the biodiesel production cost (Si,cost) and the mean value between the influence on the GHG emission rate (Si,GHG) and the influence on the water footprint (Si,WFP) (see Eq. (2)).

Si;tot: ¼ Si;NER þ Si;cost þ

Si;GHG þ Si;WFP 2

ð2Þ

These total influences have been calculated for every parameters on both pathways (Si,tot-ref and Si,tot-innov). The parameters have been arranged by classes: biology (Si,tot-ref = 32% and Si,tot-innov = 48%), cultivation (Si,tot-ref = 26% and Si,tot-innov = 21%), downstream processes (Si,tot-ref = 19% and Si,tot-innov = 17%), conversion process (Si,tot-ref = 0.03% and Si,tot-innov = 0.2%), biodiesel characteristics (Si,tot-ref = 3.5% and Si,tot-innov = 1.1%) and anaerobic digestion (Si,tot-ref = 3.6% and Si,tot-innov = 1.2%). The biology group (namely the CL content) appears to be very influential for the three criteria (energetic, economic and environmental). Research on lipid accumulation and productivity through strain selection and improvements of biological lipid accumulation pathways is very important and may help significantly in the algal biodiesel industrial development. Major improvements may also be performed on the cultivation step especially in engineering and design. Actually, PBR and raceway productivities and costs as well as HRT are very influential parameters that may be optimised through innovative design and improved cultivation strategies. The downstream processes, namely drying and lipid extraction, are also considered as a very influencing group of parameters. Not only thermal drying should be avoided at all costs but the drying step has to be reduced as much as possible. In this perspective, wet processes (such as DME lipid extraction from Kanda and Li, 2011 and sonication from Woods et al., 2011) have great potential. More generally, harvesting processes avoiding heating water should be favored. Interestingly, conversion processes for biodiesel production seem to have very little influence (less than 1% of the total influence). These technologies are well known and optimised as they are already commonly used. Therefore, research on the biodiesel conversion process from algal oil should be mainly focused on the adaptation to algae biomass of these already well-known technologies. Anaerobic digestion has a non-negligible influence on the process criteria. Improving the methane yield and reducing the energy consumption are the two areas to be investigated (less influence on the economics and the environmental criteria). Second (and higher) order effects have not been taken into account in this SA, it represents between 12% (reference pathway) and 16% (innovative pathway) of the total influence. 3.5. Perspectives The production has been assumed to be located in France which leads to specific values for some parameters (principally CO2 emission from electricity and prices of electricity and natural gas). Therefore, simulations on the mixed scenario have been done using the values for these three parameters in the United States and in the European Union. The results show an average increase on the GHG emission rate of 67% for the United States (range of [19.0– 30.2]) and of 39% for the European Union (range of [15.6–25.0]). The production cost also varies slightly during these simulations (average increase of 2.2% for the European Union and average

199

decrease of 2.8% for the United States). Collet et al. (2011) also found that the mode of electricity generation was the main environmental burden which they propose to overcome by using on-site renewable energy sources such as solar panels or wind turbine. Indeed, microalgae biodiesel production should be associated with low-carbon energy sources in order to limit its environmental impact. The innovative pathway has been defined by limiting the environmental impact of microalgae biodiesel production and by optimising its energy efficiency. However, cost is essential for the development of the microalgae biodiesel industry. Low-cost designs (for PBR essentially), higher biomass and lipid productivity as well as incentive policies are the possible ways of improving biodiesel production cost. Carbon credit might be an interesting driver for algal biodiesel production since it represents 5% of the biodiesel production cost in the mixed scenario on average (minimum of 0.7% and maximum of 14%). Using CO2 from flue gas significantly helped in reducing the GHG emission rates as already stated by Soratana and Landis (2011). However, the authors also noticed that the avoided CO2 release may often be outweighed by impacts from the construction materials. A more complete environmental analysis through a LCA would certainly benefit to the model accuracy. The wastewater credit proposed in this study to account for the price of usual wastewater treatment processes decreased the production cost by about 0.7% on average (minimum of 0.1% and maximum of 3.5%). Since our wastewater and CO2 credits are conservative, voluntary policies on waste recycling and carbon credit should certainly promote the development of the algal biofuel industry. However, the availability of these resources is a major issue that Geographical Information System (GIS) studies should contribute in sizing its full extent. A recent study by Pate et al. (2011) demonstrated that the principal challenge is to provide an adequate CO2 source. On a different level, higher engine efficiency and limited use of motorised transportation may significantly contribute in the future to increase the part of algal biodiesel in energy supply for transportation. 4. Conclusion An innovative pathway optimising the environmental and energetic criteria has been defined for microalgae biodiesel production. However, the production cost is still too high for economic viability and should be optimised. The global sensitivity analysis has identified three major bottlenecks of microalgae biodiesel production: (i) lipid accumulation by microalgae, (ii) cultivation steps (lowcost PBR and optimisation of the productivity needed), and (iii) downstream processes (effective wet biomass technologies required). Also, low-carbon energy sources are needed to achieve substantial GHG emission rates reduction. This process evaluation methodology can be used as a decision tool to determine future priorities for R&D. Acknowledgements The authors of this work would like to thank Florent Montignac for his help with the Monte Carlo sampling method, Geert Haarlemmer for proofreading this paper and the reviewers for their useful comments. Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at doi:10.1016/j.biortech.2012.02.020.

200

F. Delrue et al. / Bioresource Technology 111 (2012) 191–200

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. Benemann, J.R., Oswald, W.J., 1996. Systems and Economic Analysis of Microalgae Ponds for Conversion of CO2 to Biomass, Report No.: DOE/PC/93204–T5 Contract No.: FG22-93PC93204. Sponsored by the Department of Energy, Department of, Civil Engineering, Berkeley (CA). Brown, D., Gassner, M., Fuchino, T., Maréchal, F., 2009. Thermo-economic analysis for the optimal conceptual design of biomass gasification energy conversion systems. Appl. Therm. Eng. 29, 2137–2152. Chauvel, A., Fournier, G., Raimbault, C., 2001. Manuel d’évaluation économique des procédés. IFP-TECHNIP, France. Cherubini, F., Bird, N.D., Cowie, A., Jungmeier, G., Schlamadinger, B., Woess-Gallasch, S., 2009. Energy- and greenhouse gas-based LCA of biofuel and bioenergy systems: key issues, ranges and recommendations. Resour. Conserv. Recy. 53, 434–447. Clarens, A.F., Nassau, H., Resurreccion, H.P., White, M.A., Colosi, L.M., 2011. Environmental impacts of algae-derived biodiesel and bioelectricity for transportation. Environ. Sci. Technol. 45, 7554–7560. Collet, P., Hélias, A., Lardon, L., Ras, M., Goy, R.A., Steyer, J.P., 2011. Life-cycle assessment of microalgae culture coupled to biogas production. Bioresour. Technol. 102, 207–214. Davis, R., Aden, A., Pienkos, P.T., 2011. Techno-economic analysis of autotrophic microalgae for fuel production. Appl. Energy 88, 3524–3531. EIA. Gasoline and Diesel Fuel Update. April 2011 [cited 2011 June 21]. Available from: . EPA. 2008. Industrial Boiler Efficiency (Industrial process Applications, Office of Air and Radiations, Report No.: EPA 400-S-08-001, United States. Gerbens-Leenes, W., Hoekstraa, A.Y., van der Meer, T.H., 2009. The water footprint of bioenergy. Proc. Natl. Acad. Sci. USA 106, 10219–10223. Grobbelaar, J.U., 2004. Algal nutrition. In: Richmond, A. (Ed.), Handbook on Microalgal Culture: Biotechnology and Applied Phycology. Blackwell Science, pp. 97–115. Harun, R., Davidson, M., Doyle, M., Gopiraj, R., Danquah, M., Forde, G., 2011. Technoeconomic analysis of an integrated microalgae photobioreactor, biodiesel and biogas production facility. Biomass Bioenergy 35, 741–747. Hill, J., Nelson, E., Tilman, D., Polasky, S., Tiffany, D., 2006. Environmental, economic, and energetic costs and benefits of biodiesel and ethanol biofuels. Proc. Natl. Acad. Sci. USA 103, 11206–11210. Jorquera, O., Kiperstock, A., Sales, E.A., Embiruçu, M., Ghirardi, M.L., 2010. Comparative energy life-cycle analyses of microalgal biomas production in open ponds and photobioreactors. Bioresour. Technol. 101, 1406–1413. Kanda, H., Li, P., 2011. Simple extraction method of green crude from natural bluegreen microalgae by dimethyl ether. Fuel 90, 1264–1266.

King, C.W., Webber, M.E., 2008. The water intensity of the plugged-in automobile economy. Environ. Sci. Technol. 42, 4305–4311. Lardon, L., Hélias, A., Sialve, B., Steyer, J.P., Bernard, O., 2009. Life-cycle assessment of biodiesel production from microalgae. Environ. Sci. Technol. 43, 6475–6481. Lu, H., Liu, Y., Zhou, H., Yang, Y., Chen, M., Liang, B., 2009. Production of biodiesel from Jatropha curcas L. oil. Comput. Chem. Eng. 33, 1091–1096. Lundquist, T.J., Woertz, I.C., Quinn, N.W.T., Benemann, J.R., 2010. A Realistic Technology and Engineering Assessment of Algae Biofuel Production. Energy Biosciences Institute, Berkeley, CA. Michel, F., Mirgon, C., 2008. Etude à caractère économique portant sur les coûts d’épuration et la valeur patrimoniale des stations d’épuration des établissements industriels du bassin Rhin-Meuse. BIPE, Issy-les-Moulineaux, France. Park, J.B.K., Craggs, R.J., Shilton, A.N., 2011. Wastewater treatment high rate algal ponds for biofuel production. Bioresour. Technol. 102, 35–42. Pate, R., Klise, G., Wu, B., 2011. Resource demand implications for US algae biofuels production scale-up. Appl. Energy 88, 3377–3388. Posten, C., 2009. Design principles of photo-bioreactors for cultivation of microalgae. Eng. Life Sci. 9, 165–177. Ras, M., Lardon, L., Sialve, B., Bernet, N., Steyer, J.P., 2011. Experimental study on a coupled process of production and anaerobic digestion of Chlorella vulgaris. Bioresour. Technol. 102, 200–206. Rosenberg, J.N., Mathias, A., Korth, K., Betenbaugh, M.J., Oyler, G.A., 2011. Microalgal biomass production and carbon dioxide sequestration from an integrated ethanol biorefinery in Iowa: a technical appraisal and economic feasibility evaluation. Biomass Bioenergy 35, 3865–3878. Saltelli, A., 2002. Making best use of model valuations to compute sensitivity indices. Comput. Phys. Commun. 145, 280–297. Sialve, B., Bernet, N., Bernard, O., 2009. Anaerobic digestion of microalgae as a necessary step to make microalgal biodiesel sustainable. Biotechnol. Adv. 47, 409–416. Soratana, K., Landis, A.E., 2011. Evaluating industrial simbiosis and algae cultivation from a life cycle perspective. Bioresour. Technol. 102, 6892–6901. Subhadra, B.G., Edwards, M., 2011. Coproduct market analysis and water footprint of simulated commercial algal biorefineries. Appl. Energy 88, 3515–3523. Sun, A., Davis, R., Starbuck, M., Ben-Amotz, A., Pate, R., Pienkos, P.T., 2011. Comparative cost analysis of algal oil production for biofuels. Energy 36, 5169– 5179. Verrecht, B., Maere, T., Nopens, I., Brepols, C., Judd, S., 2010. The cost of a large-scale hollow fibre MBR. Wat. Res. 44, 5274–5283. Weissman, J.C., Goebel, R.P., 1987. Design and Analysis of Open Pond Systems for the Purpose of Producing Fuels: A Subcontract Report, Report No.: SERI/STR231-2840. Contract No.: DE-AC02-83CH10093. Sponsored by the Department of Energy, Solar Energy Research Institute, Darmstadt, Germany. Woods, L., Riccobono, M., Mehan, N., Hestekin, J., Beitle, R., 2011. Synergistic effect of abrasive and sonication for release of carbohydrate and protein from algae. Sep. Sci. Technol. 46, 601–604.