Quantitative Uncertainty Analysis of Life Cycle Assessment for Algal

Policy Analysis pubs.acs.org/est

Quantitative Uncertainty Analysis of Life Cycle Assessment for Algal Biofuel Production Deborah L. Sills,*,†,‡ Vidia Paramita,† Michael J. Franke,† Michael C. Johnson,§ Tal M. Akabas,† Charles H. Greene,‡ and Jefferson W. Tester† †

Cornell Energy Institute, Cornell University, Ithaca, New York 14853, United States Department of Earth and Atmospheric Sciences, Cornell University, Ithaca, New York 14853, United States § Department of Chemical Engineering, Massachusetts Institute of Technology, Cambridge, Massachusetts 02139, United States ‡

S Supporting Information *

ABSTRACT: As a result of algae’s promise as a renewable energy feedstock, numerous studies have used Life Cycle Assessment (LCA) to quantify the environmental performance of algal biofuels, yet there is no consensus of results among them. Our work, motivated by the lack of comprehensive uncertainty analysis in previous studies, uses a Monte Carlo approach to estimate ranges of expected values of LCA metrics by incorporating parameter variability with empirically specified distribution functions. Results show that large uncertainties exist at virtually all steps of the biofuel production process. Although our findings agree with a number of earlier studies on matters such as the need for wet lipid extraction, nutrients recovered from waste streams, and high energy coproducts, the ranges of reported LCA metrics show that uncertainty analysis is crucial for developing technologies, such as algal biofuels. In addition, the ranges of energy return on (energy) invested (EROI) values resulting from our analysis help explain the high variability in EROI values from earlier studies. Reporting results from LCA models as ranges, and not single values, will more reliably inform industry and policy makers on expected energetic and environmental performance of biofuels produced from microalgae.

■

INTRODUCTION

choices made by authors cited in Figure 1a is presented in Table S1 in the Supporting Information (SI). While a number of previous LCA studies on algal biofuels examined multiple scenarios (Figure 1a and Table S1) and conducted sensitivity analyses to measure the effects of varying single parameters on model outcomes,5,7−9,18,20 there were limitations associated with their results. Specifically, these studies presented results as point values, ignoring the combined effects of variability and uncertainties of input parameters. Two studies included uncertainty analyses that simultaneously varied process parameter inputs using Monte Carlo simulations,8,20 but the effects of uncertainty on performance of individual unit processes were not presented. The use of LCA to quantify environmental performance of a fuel assumes that such models produce single value results with minimal uncertainty. This assumption, however, is questionable due to modeling limitations;21−23 scenario uncertainty, which results from choices in model boundaries; functional units; coproduct allocation methods; and parameter uncertainty caused by the lack of performance data.24−26 Previous studies that quantitatively measured uncertainty of LCAs of rape-seed

1,2

Despite algae’s potential as a renewable energy feedstock, it is not yet clear if algal biofuels can be produced economically in an environmentally sustainable manner.3 One metric used to measure the economic viability and environmental performance of a fuel is the Energy Return on (Energy) Invested (EROI) ratio, defined as the energy contained in one unit of fuel divided by the total nonrenewable energy required to produce one unit of fuel. Numerous studies have used Life Cycle Assessment (LCA) to estimate EROI and global warming potential (GWP) of algal biofuels.4−18 However, Figure 1a shows the high variability of estimated EROI ratios from previous algal biofuel LCA studies, which range from 0.09 to 4.3. This enormous range of predicted values span the “break even” value of 1.0, in which the energy content of produced energy just equals the total nonrenewable energy consumed, demonstrating that there is no agreement as to whether algal fuels are expected to yield net gains in energy. Differences in published EROI values reflect differences in model scope, boundaries, and functional units; coproduct allocation methods; and choices of model parameters and other assumptions. 19 Since there are numerous options for cultivation, dewatering, and processing of algal feedstocks, researchers have modeled a wide array of unit-process options, further complicating comparisons.19 A summary of modeling © 2012 American Chemical Society

Received: Revised: Accepted: Published: 687

July 18, 2012 December 12, 2012 December 13, 2012 December 13, 2012 dx.doi.org/10.1021/es3029236 | Environ. Sci. Technol. 2013, 47, 687−694

Environmental Science & Technology

Policy Analysis

Figure 1. (a) Energy Return on Energy Invested (EROI) values from previously published LCA studies of algal biofuel production. (b) EROI values resulting from the “worst” (Low Productivity + Dry Extraction + anaerobic digestion) and “best” (High Productivity + Wet Extraction + anaerobic digestion) cases of this study. Center lines represent median values, edges of boxes represent 25th and 75th percentiles, and error bars represent 5th and 95th percentiles of the distributions resulting from 10 000 Monte Carlo simulations.

Figure 2. Alternative unit processes modeled for each of the five stages of the LCA model. Orange and purple arrows represent the “wet extraction” and “dry extraction” process trains, respectively.

biodiesel,27 electricity produced in the U.S.,28 and electronic products22 showed that the variability in model results can be large, highlighting the importance of presenting results as ranges and not single values. Because no industrial-scale algal biofuels production facilities exist and technologies are still being developed, data on fullscale cultivation and processing of algal biomass is scarce and often proprietary. These factors lead to high variability in model parameter values and hence large uncertainties in model predictions. Our work was motivated by the lack of comprehensive uncertainty analysis in earlier LCA studies. To obtain a better understanding of the expected performance of proposed algae to biofuel processes, LCA models should include an uncertainty analysis that quantifies the effect of simultaneously varying all model parameters on results. Such an improvement would more reliably inform industry and policy

makers on expected EROI values and environmental sustainability of algal biofuels. Goal and Scope Definition. The main goal of this study is to quantify the effect of parameter uncertainty on a “well-towheels” LCA of algal biofuel that includes analyses of individual unit processes. Although nonrenewable energy demand, EROI, and GWP are estimated over a wide range of cultivation and processing conditions, we will not conclude if algal biofuel is energetically viable or environmentally sustainable. A key objective is to demonstrate the need for employing comprehensive uncertainty analysis at each process stage before drawing conclusions. In addition, a sensitivity analysis that identifies the parameters to which the model is most sensitive, which is also relevant but not the focus of this study, is presented in Section 4 of the SI. Most of the results we present estimate nonrenewable energy demand, because it correlated 688

dx.doi.org/10.1021/es3029236 | Environ. Sci. Technol. 2013, 47, 687−694


Policy Analysis

which water required for inoculum growth was pumped and filtered to fill the PBRs; ponds were filled via gravity flow. Airlift mixing was used in the PBRs to retain adequate suspension of the biomass and for oxygen removal, and paddlewheels were used to mix gas and liquid suspensions in the raceway ponds. We assumed that the facility was located adjacent to a source of waste CO2 (e.g., fossil fuel fired power plant), and only energy requirements for CO2 gas compression needed to inject CO2 into the PBRs and raceway ponds were assumed. Ammonium nitrate and calcium triple phosphate, used as nutrients, were supplied during inoculum growth in the PBRs and the first two days of cultivation in the open ponds to optimize high growth rates. Nitrogen and phosphorus were assumed to not be supplied during the last two days of cultivation to optimize lipid production.33 A number of factors determine aerial productivity of biomass, including choice of species, solar insolation, temperature, presence of competing organisms, culture density, and nutrient inputs. Instead of constructing a model from first principles, we used a wide range of productivity values from the literature. Additionally, since model results were highly sensitive to the productivity parameter (see Figure S3), which is in agreement with previous studies,9,35 we divided this parameter into three ranges, referred to as low, base, and high productivity. The values included in the low (2.4−16 g·m−2·day−1), base (17−33 g·m−2·day−1), and high (34−50 g·m−2·day−1) productivities were based on existing values reported in the literature,33,36−43 values that may be achieved in the near term,33,43 and values that are theoretically possible and may be achieved with scientific and technological breakthroughs,2,44−46 respectively. Harvesting and Dewatering. We assumed in-pond sedimentation with autoflocculation as the first harvesting step. After the supernatant was transported back to the ocean via gravity flow, the settled biomass, with a concentration range of 1−2% TS, was removed with a traveling bridge and pumped to a primary dewatering unit. We modeled two unit processescentrifugation and a belt filter pressto dewater the biomass slurry to 20% TS. Lipid Extraction. Previous LCA studies have shown that dry lipid extraction methods result in extremely high fossil energy requirements from thermal drying5,12,17,18 that are most likely to be commercially impractical. Thus, a number of authors modeled solvent extraction of algal lipids with biomass concentrations of approximately 20% TS4−6,9,13,15,17,18 assuming that such processes are under development. However, to the best of our knowledge, no published reports in the open literature provide data from a solvent-based process that extracts lipids from wet algal biomass. Thus, the dry and wet lipid extraction processes modeled here are conventional hexane extraction with biomass at 90% TS, and hydrothermal liquefaction with biomass at 20% TS, respectively. The dry extraction model was based on the soybean mill model in the Ecoinvent database47 coupled with a drum dryer powered by natural gas to dry biomass to 90% TS.48 Hydrothermal liquefaction was modeled based on results of bench-scale experiments and process modeling conducted in Aspen Plus and Matlab,49 which were used to compute electricity and heating requirements. We assumed hydrothermal liquefaction conditions of 300 °C, 200−250 bar, and 30 min. Lipid Conversion to Biofuel. Although previous LCA studies focus almost exclusively on algal-triacylglyceride (TAG) conversion to fatty acid methyl esters (FAME)i.e., biodieselusing transesterification, we also modeled con-

closely to GWP for all unit processes except for coproduct production. This correlation has also been shown previously for several fuels and products.29

■

MODEL CONSTRUCTION We assumed that marine algae were cultivated at a coastal location in a 1210-ha production facility with access to sea− water. The f unctional unit for this study is 1 MJ of liquid biof uel (biodiesel or “green” diesel). The model was divided into five unit processes: cultivation, harvesting and dewatering, lipid extraction, lipid conversion to a liquid transportation fuel, and coproduct production from defatted algae. Alternative technologies, presented in Figure 2, were modeled for each of the five stages. Each process stage was analyzed separately and assembled with the other stages to create six alternative case studies. Life Cycle Inventories. All relevant energy and material inputs, with a 5% cutoff, were included in life cycle inventories (LCIs) for each unit process, following ISO standards.30 The lifetime of processing infrastructure was assumed to be 20 years, except for photobioreactors and open pond liners, which were assumed to be 5 and 10 years, respectively. We excluded labor and transport of produced biofuel to the user. We used the Eco-Indicator 2002+ method31 to quantify nonrenewable energy consumption, including upstream fossil energy demands, and GWP in units of mass of CO2 equivalents per functional unit. Most model parameters were input as probability distribution functions (PDFs) to quantitatively represent each parameter’s inherent variability and uncertainty. Due to the scarcity of data, which is often the case in LCA studies, we were not able to capture the true underlying distribution functions for a number of parameters.23 Therefore, we often used triangular distributions with a mode and minimum/maximum based on an average and limiting values of available data, respectively. When sufficient data were available (e.g., for hydrothermal processes simulated in Aspen Plus), we fit normal and log-normal distributions in Matlab. Parameter uncertainties, depicted by PDFs, were simultaneously propagated through the model using 10,000 Monte Carlo simulations within SimaPro 7.3.2 coupled with the Ecoinvent 2.2 database32 to compile LCIs for each unit process. Results for two processing trains, focused on dry and wet lipid extractions (Figure 2), combined with three ranges of algal productivity during cultivation (a total of six cases) are presented. The dry extraction route included cultivation, centrifugation, thermal drying + hexane extraction, and transesterification to biodiesel. The wet extraction route included cultivation, belt filter press, hydrothermal liquefaction, and hydrotreating to produce “green diesel.” Since previous LCA studies have shown that coproducts (e.g., methane from anaerobic digestion of defatted algae) are required for net gains in energy,5−18 coproducts were included in all six cases. Unit Processes. Each unit process is briefly described in this section; detailed descriptions are included in the SI. Cultivation. Cultivation of marine algae was modeled using a hybrid growth system described previously.33,34 Briefly, each growth cycle included one day of inoculum cultivation in horizontal tubular photobioreactors (PBRs) and four days of growth in geotextile-lined open raceway ponds, with a ratio of inoculum to pond volumes of 1:3; culture crashes were assumed to occur in the open ponds with probabilities of 0− 6%. Water was pumped from the ocean to an intake canal, from 689



Policy Analysis

version of algal lipids to “green” diesel using hydrotreatment. Hydrotreatment, the process used in petroleum refineries to remove sulfur and nitrogen from petroleum crudes, can be used to deoxygenate algal lipids and form straight-chain alkanes, often referred to as “green” diesel, that are very similar to petroleum diesel.50−52 Alkaline transesterification of algal lipids to FAME was modeled based on a previous study.47 The modeled process was operated at 60 °C; it required methanol, electricity, heat, construction materials, water, phosphoric acid, potassium hydroxide, and sulfuric acid (Table S9), and produced FAME and glycerol. Hydrotreatment was modeled with Aspen Plus and Matlab as a stand-alone hydrotreater operated at 325 °C with pressures that ranged from 36 to 50 bar. The relative amounts of lipids deoxygenated through the hydrodeoxygenation, decarbonylation, and decarboxylation pathways were modeled stochastically with assumptions based on the literature50,51 to compute a range of H2 demands. A range of H2 demands needed to break the TAG backbone, saturate double bonds, and remove N and S present in the lipid fraction after hydrothermal liquefaction were also estimated, along with subsequent ranges of electricity and heat requirements. Coproducts. We incorporated two coproducts made from defatted algal biomassanaerobic digestion-derived methane (for energy) and animal feedinto the LCA model using the system expansion method,30 whereby we assumed that anaerobic digestion followed by methane combustion in a combined heat and power (CHP) unit displaced electricity (from the U.S. grid) and heat (produced from natural gas), and defatted algae displaced soy meal. Nonrenewable energy consumption and GHG emissions that were avoided due to the displacement of electricity and heat or soymeal were credited to the primary productalgal biofuel. Values from the literature were used to estimate methane yields from anaerobic digestion,53,54 electricity (for mixing) and heating demands for the digesters,8,9,54 and efficiencies for the CHP unit.9,55 The soy-meal, animal-feed model was based on an existing Ecoinvent process model.48

Figure 3. Deterministic (a) energy balances and (b) Global Warming Potential for biofuel production using the base productivity (25 g/m2/ day); wet and dry lipid extraction, respectively; and anaerobic digestion as a coproduction process.

Global Warming Potential. As has been shown in earlier LCA studies of other fuels,29 results presented in Figure 3a and 3b demonstrate that processes with high nonrenewable energy demands were associated with high GWP values (e.g., thermal drying and algal cultivation). Therefore, most of the results from the uncertainty analysis presented in the following section are for nonrenewable energy demand. Uncertainty Analysis. Ranges of nonrenewable energy demands associated with each unit process for the dry and wet extraction process trains, using the base productivity (17−33 g·m−2·day−1), are presented in Figure 4. The 5th, 25th, 50th (median), 75th, and 95th percentiles of the distributions resulting from Monte Carlo simulations are presented. While the focus of this section is comprehensive uncertainty analysis, we report results from a sensitivity analysis, which are similar to those presented in previous studies,5,7−9 in Figure S3. The nonrenewable energy demand associated with thermal drying (Figure 4) ranged from 1.5 to 2.2 MJ per MJ of biofuel produced, which is in agreement with the deterministic results presented in Figure 3, in that drying makes up the highest fraction of fossil energy consumption. However, nonrenewable energy requirements for cultivation (modeled with the base productivity), which ranged from 1.1 to 1.7 MJ per MJ of biofuel, may be as large as those for thermal drying (Figure 4). In other words, the distributions of thermal drying and cultivation overlap, suggesting that, while wet lipid extraction methods are crucial to yield net gains in energy, decreasing energy requirements for cultivating algae are also needed. Consequently, we analyzed the cultivation process in more detail and report results in the following section.

■

RESULTS Base Case. Energy balances and GWP on a “well to wheels” life cycle basis for algal biofuel are presented in Figure 3, as single values (without uncertainty analysis) for the dry and wet extraction routes using the mode of the Base Productivity parameter (25 g·m−2·day−1) and anaerobic digestion as a coproduction process. Our first objective was to provide base case results to emphasize the importance of the uncertainty analysis discussed later. Since thermal drying demanded 1.8 MJ of nonrenewable energy per MJ of biofuel (Figure 3a), it is clear that for algal biofuels to yield net gains in energy, lipid extraction methods for wet biomass (e.g., 20% TS) must be developed. A number of previous studies also demonstrated that wet lipid extraction is crucial for algal biofuels to yield net gains in energy.12,16−18 In addition, since biofuel production using the wet extraction process train consumed 1.6 MJ of nonrenewable energy to produce 1 MJ of biofuel (Figure 3a), we conclude that high-energy coproducts, such as methane from anaerobic digestion, must be incorporated into integrated algal biorefineries to yield EROI values that are higher than 1. This conclusion is also in agreement with previous studies.5,7,8,18 690



Policy Analysis

Cultivation. Fossil energy demands for cultivation of algal feedstocks per 1 MJ of biofuel for three productivity ranges low (2.4−16 g·m−2·day−1), base (17−33 g·m−2·day−1), and high (34−50 g·m−2·day−1)are presented in Figure 5.

Figure 4. Nonrenewable energy demand per 1 MJ of biofuel produced for individual unit processes that make up the (a) dry and (b) wet extraction process trains. Center lines represent median values, edges of boxes represent 25th and 75th percentiles, and limiting bars represent 5th and 95th percentiles of the distributions resulting from 10 000 Monte Carlo simulations. Figure 5. Nonrenewable energy demand (MJ/MJ biofuel produced) for the cultivation subprocesses (“reactor construction” includes photobioreactors and open raceway ponds). Center lines represent median values, edges of boxes represent 25th and 75th percentiles, and limiting bars represent 5th and 95th percentiles of the distributions resulting from 10 000 Monte Carlo simulations.

The results presented in Figure 4 show that the distribution of energy demands required for centrifugation is higher than that required for the belt filter press. This result agrees with comments made by Starbuck, which state that energetic demands of centrifugation are too high, and more energy efficient dewatering processes (e.g., belt filtration) should be used.56 However, the results presented in Figure 4 assume that belt filtration is as effective at separating solids as is centrifugation, which may not be true. Other authors maintain that a belt filter press may not effectively dewater suspensions with smaller algal cells.57,58 In summary, even though belt filtration is more energy efficient than centrifugation, additional studies are needed to assess the effectiveness of belt filtration. With respect to the two lipid extraction processes, results show that hydrothermal liquefaction has lower fossil energy demands than does hexane extraction, and the ranges of energy demand for these two processes do not overlap (Figure 4). Specifically, the 95th percentile of hydrothermal liquefaction is 0.15 MJ less than the 5th percentile of the hexane extraction distribution per 1 MJ of biofuel. Note that thermal drying, with an energy demand of 1.5−2.1 MJ per 1 MJ of biofuel (Figure 4), is usually coupled with hexane extraction, which results in an energy demand for dry solvent extraction that is about 1 order of magnitude higher than that for hydrothermal liquefaction. The range of values for hydrotreatment and transesterification, presented in Figure 4, overlap, suggesting that the choice of lipid conversion technology should be based on considerations beyond nonrenewable energy demand. Hydrotreatment may have advantages over transesterification, including the use of existing infrastructure and a fuel that is more compatible with existing engines.52 Furthermore, transesterification produces glycerol as a byproduct, which in previous studies was assumed to be a coproduct;5,17,52 however, large-scale biodiesel production is likely to exceed market demand for glycerol as a coproduct, which will turn this byproduct into a liability.59 Additionally, renewable H2producing technologies (e.g., hydrothermal gasification of defatted algae) may be developed as part of an integrated biorefinery, which might reduce demands of nonrenewable energy for hydrotreatment.

Nonrenewable energy demands for algae cultivation ranged from 1.7 to 4.9, 0.94 to 1.8, and 0.7 to 1.3 MJ per MJ biofuel produced for the low, base, and high productivity ranges, respectively. Only the 5th percentile of the range for the high productivity (0.7 MJ) is lower than the energy content of the biofuel, indicating need to reduce energy demands of cultivation at all productivity ranges. Because increasing productivity results in higher concentrations of algae at harvest, but does not change reactor size or amount of water transported and mixed, energetic demands associated with reactor construction, water transport, and mixing are reduced per unit biofuel as values of aerial productivity increase (Figure 5). On the other hand, energetic costs associated with CO2 compression and nutrient demands remain constant per unit biofuel produced (Figure 5). Therefore, even when high productivities are assumed, upstream energetic burdens of nutrients remain at 0.32−0.59 MJ per MJ of biofuel (Figure 5), demonstrating the need for N and P recovered from waste streams. Coproducts. Results from the deterministic analysis presented in Figure 3 show that incorporating anaerobic digestion-derived energy as a coproduct resulted in avoiding a net consumption of 0.8 MJ of electricity and heat (including upstream energy costs, which are 3.4 MJ/MJ for U.S. grid electricity), as well as N and P with upstream fossil energy demands of 0.2 MJ per MJ of biofuel produced. The results of uncertainty analysis, presented in Figure 6, compare the effect of incorporating anaerobic digestion-derived methane or animal feed as coproducts. From the standpoint of nonrenewable energy demand (Figure 6a), the data suggest that anaerobic digestion may be superior to the production of animal feed, as demonstrated by the approximately 45% decrease in the each percentile (over the base productivity and wet extraction case alone) compared to an approximately 10% decrease for animal feed. However, since all three distributions have large 691



Policy Analysis

Figure 7. Energy return on energy invested (EROI) for algal biofuel production (including anaerobic digestion as a coproduct) with low, base, and high productivity parameters, dry and wet lipid extraction process trains. Center lines represent median values, edges of boxes represent 25th and 75th percentiles, and limiting bars represent 5th and 95th percentiles of the distributions resulting from 10 000 Monte Carlo simulations.

Figure 6. (a) Nonrenewable energy demand (MJ) and (b) Global Warming Potential (GWP) (g CO2-eq) for algal biofuel using the Base Productivity Case plus wet extraction alone, and with two coproducts: anaerobic digestion or animal feed. Center lines represent median values, edges of boxes represent 25th and 75th percentiles, and limiting bars represent 5th and 95th percentiles of the distributions resulting from 10 000 Monte Carlo simulations.

coupled with the dry and wet extraction routes and anaerobic digestion as a coproduction process. As expected, the dry extraction route coupled with each of the three productivity ranges resulted in EROI values lower than 1.0, demonstrating that the development of wet lipid extraction processes is crucial to achieve gains in net energy. In addition, algae cultivation with low productivity (2.4−16 g·m−2·day−1) coupled with dry or wet lipid extraction resulted in EROI values lower than 1.0, showing that it is also crucial to increase aerial productivity of algal feedstocks and lower energy demands associated with cultivation. Seventy five percent of the Monte Carlo simulations for the case of base productivity (17−33 g·m−2·day−1) coupled with wet lipid extraction resulted in EROI values higher than 1.0 (Figure 7). But only high productivity (34−50 g·m−2·day−1) combined with wet extraction resulted in EROI values higher than 1.0 over the entire range of results. Unfortunately, it is not clear if the range of growth rates associated with the highproductivity parameter are feasible.60 Nonetheless, in addition to developing wet lipid extraction techniques, research directed toward increasing productivity and lowering energetic demands of all subprocesses associated with algae cultivation is needed to increase EROI values of biofuels produced from microalgae.

overlapping regions (Figure 6a), it is not possible to conclude whether anaerobic digestion will result in a realistic improvement in performance. The uncertainty in all three ranges of energy demand shown in Figure 6a highlight once more that reporting single values is misleading. Since GWP did not closely correlate with the nonrenewable energy demand associated with anaerobic digestion and animal feed production as it did for all other unit processes, we presented the “well to wheels” GWP for algal biofuel alone and combined with each coproduct in Figure 6b. The GWP for biofuel alone using the base productivity and wet lipid extraction route was higher than the “well to wheels” GWP of petroleum-based diesel fuel (90 g CO2 eq per MJ) for the entire range of results. Adding animal feed production or anaerobic digestion reduced the median GWP of algal biofuel by 21 g CO2 eq per MJ of biofuel, compared to biofuel alone, suggesting that incorporating coproducts may reduce the GWP of algal biofuel to a value lower than the GWP of petroleum diesel (Figure 6b). However, GWP values associated with biofuel production plus anaerobic digestion are quite uncertain as is evident from the large range of values that span from 30 to 110 g CO2 eq per MJ of biofuel. Furthermore, it is not clear whether incorporating anaerobic digestion-derived energy as a coproduct will result in a greater reduction in GWP than animal feed, which has values that range from 70 to 105 g CO2 eq per MJ of biofuel produced (Figure 6b). Vasudevan et al. showed that animal feed production and anaerobic digestion resulted in GWP reductions of approximately 80 and 50 g CO2 eq per MJ of biofuel, respectively.18 Although their calculations are not in question, there are limitations as their model produces only one result among a range of possible results. The range of results presented in Figure 6 highlights that uncertainty analysis is essential to provide decision makers with realistic estimates of nonrenewable energy consumption and GWP. Net Energy Balances. EROI ratios are presented for six cases in Figure 7: cultivation with three productivity ranges

■

DISCUSSION Results of this study are generally in agreement with previous reports that identified the role of critical processing steps in algae to biofuel schemes. For example, the need to develop viable wet lipid extraction technologies, incorporate highenergy coproducts, and reduce energy consumption of algae cultivation was clearly demonstrated. The major contribution of the present work was to extend previous LCA studies by using a Monte Carlo approach to quantify the role of uncertainty associated with process parameters in determining outcomes of LCA metrics, such as EROI and GWP. Results show that uncertainties exist at all stages of biofuel production from microalgae, from cultivation to dewatering to conversion processes and production of coproducts. Figure 1a and 1b show that the highly variable EROI data reported in previous LCA papers fall within the ranges of the “worst” (low productivity with dry extraction and anaerobic digestion) and “best” (high productivity with wet extraction and anaerobic 692



Policy Analysis

(8) Clarens, A. F.; Nassau, H.; Resurreccion, E. P.; White, M. A.; Colosi, L. M. Environmental Impacts of Algae-Derived Biodiesel and Bioelectricity for Transportation. Environ. Sci. Technol. 2011, 45 (17), 7554−7560. (9) Frank, E. D.; Palou-Rivera, I.; Elgowainy, A.; Wang, M. Q. LifeCycle Analysis of Algal Lipid Fuels with the GREET Model; Argonne National Laboratory: Argonne, IL, 2011. (10) Jorquera, O.; Kiperstok, A.; Sales, E. A.; Embiruçu, M.; Ghirardi, M. L. Comparative energy life-cycle analyses of microalgal biomass production in open ponds and photobioreactors. Bioresour. Technol. 2010, 101 (4), 1406−1413. (11) Khoo, H. H.; Sharratt, P. N.; Das, P.; Balasubramanian, R. K.; Naraharisetti, P. K.; Shaik, S. Life cycle energy and CO2 analysis of microalgae-to-biodiesel: Preliminary results and comparisons. Bioresour. Technol. 2011, 102 (10), 5800−5807. (12) Lardon, L.; Hélias, A.; Sialve, B.; Steyer, J.-P.; Bernard, O. LifeCycle Assessment of Biodiesel Production from Microalgae. Environ. Sci. Technol. 2009, 43 (17), 6475−6481. (13) Razon, L. F.; Tan, R. R. Net energy analysis of the production of biodiesel and biogas from the microalgae Haematococcus pluvialis and Nannochloropsis. Appl. Energy 2011, 88 (10), 3507−3514. (14) Sander, K.; Murthy, G. Life cycle analysis of algae biodiesel. Int. J. Life Cycle Assess. 2010, 15 (7), 704−714. (15) Sawayama, S.; Minowa, T.; Yokoyama, S.-Y. Possibility of renewable energy production and CO2 mitigation by thermochemical liquefaction of microalgae. Biomass Bioenergy 1999, 17 (1), 33−39. (16) Shirvani, T.; Yan, X.; Inderwildi, O. R.; Edwards, P. P.; King, D. A. Life cycle energy and greenhouse gas analysis for algae-derived biodiesel. Energy Environ. Sci. 2011, 4 (10), 3773−3778. (17) Stephenson, A. L.; Kazamia, E.; Dennis, J. S.; Howe, C. J.; Scott, S. A.; Smith, A. G. Life-Cycle Assessment of Potential Algal Biodiesel Production in the United Kingdom: A Comparison of Raceways and Air-Lift Tubular Bioreactors. Energy Fuels 2010, 24 (7), 4062−4077. (18) Vasudevan, V.; Stratton, R. W.; Pearlson, M. N.; Jersey, G. R.; Beyene, A. G.; Weissman, J. C.; Rubino, M.; Hileman, J. I. Environmental Performance of Algal Biofuel Technology Options. Environ. Sci. Technol. 2012, 46 (4), 2451−2459. (19) Liu, X.; Clarens, A. F.; Colosi, L. M. Algae biodiesel has potential despite inconclusive results to date. Bioresour. Technol. 2012, 104, 803−806. (20) Clarens, A. F.; Resurreccion, E. P.; White, M. A.; Colosi, L. M. Environmental Life Cycle Comparison of Algae to other Bioenergy Feedstocks. Environ. Sci. Technol. 2010, 44 (5), 1813−1819. (21) Lloyd, S. M.; Ries, R. Characterizing, propagating, and analyzing uncertainty in life-cycle assessment: A survey of quantitative approaches. J. Ind. Ecol. 2007, 11 (1), 161−179. (22) Weber, C. L. Uncertainty and Variability in Product Carbon Footprinting. J. Ind. Ecol. 2012, 16 (2), 203−211. (23) Williams, E. D.; Weber, C. L.; Hawkins, T. R. Hybrid framework for managing uncertainty in life cycle inventories. J. Ind. Ecol. 2009, 13 (6), 928−944. (24) Huijbregts, M. A. J.; Gilijamse, W.; Ragas, A. M. J.; Reijnders, L. Evaluating uncertainty in environmental life-cycle assessment. A case study comparing two insulation options for a Dutch one-family dwelling. Environ. Sci. Technol. 2003, 37 (11), 2600−2608. (25) Huijbregts, M. Application of uncertainty and variability in LCA. Int. J. Life Cycle Assess. 1998, 3 (5), 273−280. (26) Huijbregts, M. Part II: Dealing with parameter uncertainty and uncertainty due to choices in life cycle assessment. Int. J. Life Cycle Assess. 1998, 3 (6), 343−351. (27) Malça, J.; Freire, F. Uncertainty analysis in biofuel systems. J. Ind. Ecol. 2010, 14 (2), 322−334. (28) Weber, C. L.; Jaramillo, P.; Marriott, J.; Samaras, C. Life cycle assessment and grid electricity: What do we know and what can we know? Environ. Sci. Technol. 2010, 44 (6), 1895−1901. (29) Huijbregts, M. A. J.; Rombouts, L. J. A.; Hellweg, S.; Frischknecht, R.; Hendriks, A. J.; van de Meent, D.; Ragas, A. M. J.; Reijnders, L.; Struijs, J. Is cumulative fossil energy demand a useful

digestion) cases. This indicates that the values reported in earlier studies (Figure 1a) are not incorrect, but, instead, each represents one specific case, which should not be used to conclude whether algal biofuels are expected to be energetically viable or environmentally sustainable. Instead, LCA results, especially those associated with developing technologies such as algal biofuel, should be reported as ranges of expected values to provide decision makers with reliable results. Limitations of this study include the assumption of no correlations among process parameters, and, for some parameters, not accurately knowing their underlying distribution functions used for Monte Carlo sampling. Therefore, future work should focus on elucidating sources of uncertainty and reducing uncertainty of the most sensitive process parameters. It should also be noted that this study measured the uncertainty of only two LCA metrics: EROI and GWP. Clearly, uncertainty analyses of additional environmental performance metrics, such as water demand and land use, are needed to comprehensively assess the sustainability of algal biofuel.

■

ASSOCIATED CONTENT

S Supporting Information *

Figures S1−S3, Tables S1−S19, and text. This material is available free of charge via the Internet at http://pubs.acs.org.

■

AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected]; phone: 607-277-5609; fax: 607255-8313. Notes

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS We thank Mark Huntley, Beth Ahner, and Ruth Richardson of Cornell University and Ian Archibald of Algae Architects for providing insights into algae cultivation and conversion processes; Xingen Lei of Cornell University for sharing his knowledge about animal feed; and Jeremy Luterbacher of Cornell University for his advice on LCA modeling. This study was supported with partial funding from the USDA.

■

REFERENCES

(1) Chisti, Y. Biodiesel from Microalgae. Biotechnol. Adv. 2007, 25 (3), 294−306. (2) Wijffels, R. H.; Barbosa, M. J. An Outlook on Microalgal Biofuels. Science 2010, 329, 796−799. (3) USDOE. National Algal Biofuels Technology Roadmap; U.S. Department of Energy: College Park, MD, 2010. (4) Batan, L.; Quinn, J.; Willson, B.; Bradley, T. Net Energy and Greenhouse Gas Emission Evaluation of Biodiesel Derived from Microalgae. Environ. Sci. Technol. 2010, 44 (20), 7975−7980. (5) Brentner, L. B.; Eckelman, M. J.; Zimmerman, J. B. Combinatorial Life Cycle Assessment to Inform Process Design of Industrial Production of Algal Biodiesel. Environ. Sci. Technol. 2011, 45 (16), 7060−7067. (6) Campbell, P. K.; Beer, T.; Batten, D. Life cycle assessment of biodiesel production from microalgae in ponds. Bioresour. Technol. 2011, 102, 50−56. (7) Chowdhury, R.; Viamajala, S.; Gerlach, R. Reduction of environmental and energy footprint of microalgal biodiesel production through material and energy integration. Bioresour. Technol. 2012, 108, 102−111. 693



Policy Analysis

indicator for the environmental performance of products? Environ. Sci. Technol. 2005, 40 (3), 641−648. (30) Guinée, J. Handbook on Life Cycle Assessment: Operational Guide to the ISO Standards; Kluwer: Dordrecht, The Netherlands, 2002. (31) Jolliet, O.; Margni, M.; Charles, R.; Humbert, S.; Payet, J.; Rebitzer, G.; Rosenbaum, R. IMPACT 2002+: A new life cycle impact assessment methodology. Int. J. Life Cycle Assess. 2003, 8 (6), 324− 330. (32) Frischknecht, R.; Jungbluth, N.; Althaus, H.-J.; Doka, G.; Dones, R.; Heck, T.; Hellweg, S.; Hischier, R.; Nemecek, T.; Rebitzer, G.; Spielmann, M. The Ecoinvent Database: Overview and Methodological Framework. Int. J. Life Cycle Assess. 2005, 10 (1), 3−9. (33) Huntley, M.; Redalje, D. CO2 Mitigation and renewable oil from photosynthetic microbes: A new appraisal. Mitig. Adapt. Strat. Glob. Change 2007, 12 (4), 573−608. (34) Paramita, V. A. Life Cycle Analysis of Algal BiofuelTreating Uncertainty and Variability. M. S. Thesis, Cornell Univeristy: Ithaca, NY, 2012. (35) Davis, R.; Aden, A.; Pinkos, P. T. Tecno-economic analysis of autotrophic microalgae for fuel production. Appl. Energy 2011, 88 (10), 3524−3531. (36) Benemann, J. R.; Goebel, R. P.; Weissman, J. C.; Augenstein, D. C. Microalgae as a Source of Liquid Fuels; U.S. Department of Energy: Washington, DC, 1982; pp 202. (37) García-González, M.; Moreno, J.; Cañavate, J. P.; Anguis, V.; Prieto, A.; Manzano, C.; Florencio, F. J.; Guerrero, M. G. Conditions for open-air outdoor culture of Dunaliella salina in southern Spain. J. Appl. Phycol. 2003, 15 (2−3), 177−184. (38) Laws, E. A.; Bidigare, R. R.; Popp, B. N. Effect of growth rate and CO2 concentration on carbon isotopic fractionation by the marine diatom Phaeodactylum tricornutum. Limnol. Oceanogr. 1997, 42 (7), 1552−1560. (39) Moheimani, N.; Borowitzka, M. The long-term culture of the coccolithophore Pleurochrysis carterae (Haptophyta) in outdoor raceway ponds. J. Appl. Phycol. 2006, 18 (6), 703−712. (40) Olaizola, M. Commercial development of microalgal biotechnology: From the test tube to the marketplace. Biomol. Eng. 2003, 20 (4−6), 459−466. (41) Olguín, E. J.; Galicia, S.; Mercado, G.; Pérez, T. Annual productivity of Spirulina (Arthrospira) and nutrient removal in a pig wastewater recycling process under tropical conditions. J. Appl. Phycol. 2003, 15 (2−3), 249−257. (42) Pushparaj, B.; Pelosi, E.; Tredici, M.; Pinzani, E.; Materassi, R. An integrated culture system for outdoor production of microalgae and cyanobacteria. J. Appl. Phycol. 1997, 9 (2), 113−119. (43) Richmond, A.; Cheng-Wu, Z. Optimization of a flat plate glass reactor for mass production of Nannochloropsis sp. outdoors. J. Biotechnol. 2001, 85 (3), 259−269. (44) Griffiths, M.; Harrison, S. Lipid productivity as a key characteristic for choosing algal species for biodiesel production. J. Appl. Phycol. 2009, 21 (5), 493−507. (45) Goldman, J. C. Outdoor algal mass culturesII. Photosynthetic yield limitations. Water Res. 1979, 13 (2), 119−136. (46) Weyer, K.; Bush, D.; Darzins, A.; Willson, B. Theoretical Maximum Algal Oil Production. BioEnergy Res. 2010, 3 (2), 204−213. (47) Jungbluth, N.; Chudacoff, M.; Dauriat, A.; Dinkel, F.; Doka, G.; Faist Emmenegger, M.; Gnansounou, E.; Kljun, N.; Sohleiss, K.; Spielmann, M.; Stettler, C.; Sutter, J. Life Cycle Inventories of Bioenergy; Swiss Center for Life Cycle Inventories: Dübendorf, 2007. (48) Nemecek, T.; Kägi, T. Life Cycle Inventories of Swiss and European Agricultural Production Systems; Final Report Ecoinvent v2.0 No. 15a; Swiss Centre for Life Cycle Inventories: Zürich, Dübendorf, 2007. (49) Johnson, M. C. Hydrothermal processing of high-lipid biomass to fuels. Ph.D. Dissertation, Massachusetts Institute of Technology, Cambridge, MA, 2012. (50) Huber, G. W.; O’Connor, P.; Corma, A. Processing biomass in conventional oil refineries: Production of high quality diesel by

hydrotreating vegetable oils in heavy vacuum oil mixtures. Appl. Catal., A 2007, 329, 120−129. (51) Marker, T. L. Opportunities for Biorenewables in Oil Refineries; U.S. Department of Energy: Des Plains, IL, 2005. (52) Kalnes, T. N.; Koers, K. P.; Marker, T.; Shonnard, D. R. A technoeconomic and environmental life cycle comparison of green diesel to biodiesel and syndiesel. Environ. Prog. Sust. Energy 2009, 28 (1), 111−120. (53) Sialve, B.; Bernet, N.; Bernard, O. Anaerobic digestion of microalgae as a necessary step to make microalgal biodiesel sustainable. Biotechnol. Adv. 2009, 27 (4), 409−416. (54) Collet, P.; Hélias, A.; Lardon, L.; Ras, M.; Goy, R. A.; Steyer, J. P. Life-cycle assessment of microalgae culture coupled to biogas production. Bioresour. Technol. 2011, 102 (1), 207−214. (55) Report Prepared for U.S. Environmental Protection Agency Combined Heat and Power Partnership. Biomass Combined Heat and Power Catalog of Technologies; 2007. http://www.epa.gov/chp/ documents/biomass_chp_catalog.pdf. (56) Starbuck, C. M. Comment on “Environmental life cycle comparison of algae to other bioenergy feedstocks”. Environ. Sci. Technol. 2011, 45, 833. (57) Molina Grima, E.; Belarbi, E.-H.; Acién Fernández, F.; Robles Medina, A.; Chisti, Y. Recovery of microalgal biomass and metabolites: Process options and economics. Biotechnol. Adv. 2003, 20 (7−8), 491−515. (58) Wiley, P. E.; Campbell, J. E.; McKuin, B. Production of biodiesel and biogas from algae: A review of process train options. Water Environ. Res. 2011, 83 (4), 326−338. (59) Johnson, D. T.; Taconi, K. A. The glycerin glut: Options for the value-added conversion of crude glycerol resulting from biodiesel production. Environ. Prog. 2007, 26 (4), 338−348. (60) Stephens, E.; Ross, I. L.; King, Z.; Mussgnug, J. H.; Kruse, O.; Posten, C.; Borowitzka, M. A.; Hankamer, B. An economic and technical evaluation of microalgal biofuels. Nat. Biotechnol. 2010, 28 (2), 126−128.

694


Quantitative Uncertainty Analysis of Life Cycle Assessment for Algal

Recommend Documents