Integrated Evaluation of Cost, Emissions, and Resource Potential for

Apr 21, 2014 - The cost benefit of algal technology for combined CO2 mitigation and nutrient abatement. S.J. Judd , F.A.O. Al Momani , H. Znad , A.M.D...
9 downloads 9 Views 450KB Size
Article pubs.acs.org/est

Integrated Evaluation of Cost, Emissions, and Resource Potential for Algal Biofuels at the National Scale Ryan E. Davis,§ Daniel B. Fishman,‡ Edward D. Frank,*,† Michael C. Johnson,† Susanne B. Jones,∥ Christopher M. Kinchin,§ Richard L. Skaggs,∥ Erik R. Venteris,∥ and Mark S. Wigmosta∥ †

Center for Transportation Research, Argonne National Laboratory, 9700 South Cass Avenue, Argonne, Illinois 60439, United States U.S. Department of Energy, 1000 Independence Avenue, SW, Washington, DC 20585, United States § National Renewable Energy Laboratory, 15013 Denver West Parkway, Golden, Colorado 80401, United States ∥ Pacific Northwest National Laboratory, Richland, Washington 99354, United States ‡

S Supporting Information *

ABSTRACT: Costs, emissions, and resource availability were modeled for the production of 5 billion gallons yr−1 (5 BGY) of renewable diesel in the United States from Chlorella biomass by hydrothermal liquefaction (HTL). The HTL model utilized data from a continuous 1-L reactor including catalytic hydrothermal gasification of the aqueous phase, and catalytic hydrotreatment of the HTL oil. A biophysical algae growth model coupled with weather and pond simulations predicted biomass productivity from experimental growth parameters, allowing site-by-site and temporal prediction of biomass production. The 5 BGY scale required geographically and climatically distributed sites. Even though screening down to 5 BGY significantly reduced spatial and temporal variability, siteto-site, season-to-season, and interannual variations in productivity affected economic and environmental performance. Performance metrics based on annual average or peak productivity were inadequate; temporally and spatially explicit computations allowed more rigorous analysis of these dynamic systems. For example, 3-season operation with a winter shutdown was favored to avoid high greenhouse gas emissions, but economic performance was harmed by underutilized equipment during slow-growth periods. Thus, analysis of algal biofuel pathways must combine spatiotemporal resource assessment, economic analysis, and environmental analysis integrated over many sites when assessing national scale performance.



INTRODUCTION Algae-based transportation fuels are being developed to achieve energy independence, fossil fuel use reduction, and greenhouse gas (GHG) emissions reduction.1 Algal biofuel production must meet several criteria simultaneously to achieve these ends: It must produce fuel at an acceptable price with acceptable financial and engineering risk, must require realizable capital levels, and must consume land, water, and nutrients at levels consistent with national resources. This paper examines this broad set of criteria for an algal renewable diesel (RD) pathway at the 5 billion gallons yr−1 (BGY) national scale (19 billion L yr−1), roughly 10% of the U.S. diesel demand. This work studied renewable diesel production via hydrothermal liquefaction (HTL) of algal biomass grown in open raceway ponds. HTL is a thermal process that uses hot, pressurized, water in a condensed subcritical phase to convert whole wet biomass to gas and oil products. Aqueous and solid phases are also produced from which nutrients and additional energy (biogas) may be recovered. Although many reports exist for HTL treatment of algal biomass, those studies are largely batch reactor studies, usually in small quantities.2−11 Until now, there have been few public data for upgrading algal HTL oils © 2014 American Chemical Society

through hydrotreating to make fungible fuels with suitable oxygen and nitrogen levels as well as desirable hydrogen to carbon (H/C) ratio or for HTL aqueous-phase treatment.12,13 This study utilized data obtained from a continuous 1-L HTL reactor,13 catalytic hydrothermal gasification of the aqueous phase, and catalytic hydrotreatment of the HTL oil intermediate for production of renewable gasoline and dieselrange fuels. Costs were obtained via techno−economic analysis (TEA), an engineering costing method that determines selling prices to evaluate and quantify economic implications for technology options.14,15 Emissions were studied via life-cycle analysis (LCA), a methodology that sums direct process emissions with those from upstream supporting operations such as electricity provisioning, nutrient manufacturing, and transportation of intermediates and products. Biomass production potential was assessed using a new algae growth model developed for the Received: Revised: Accepted: Published: 6035

December 16, 2013 April 2, 2014 April 21, 2014 April 21, 2014 dx.doi.org/10.1021/es4055719 | Environ. Sci. Technol. 2014, 48, 6035−6042

Environmental Science & Technology

Article

Figure 1. Process flow diagram for modeled system. Percentages are ash-free dry algae weight percent of total flow. The model considered outdoor raceway ponds and a 3-stage harvesting and dewatering process that produced algae slurry with 20 wt % ash-free solids content for HTL processing and hydrotreating based upon the assumptions described in the SI. DAF = dissolved air flotation.

phase, and hydrotreatment of the HTL oil are those performed on Nannochloropsis sp..13 Even these data are preliminary results based on a limited number of experiments under nonoptimized conditions. A growth model was not available for Nannochloropsis, but one was available for the strain we examine here, NAABB 2412,24 also known as strain DOE 1412, a highly productive Chlorella strain isolated by Dr. J. Polle (Brooklyn College). This approach allowed the HTL mass balances to be applied to the modeled biomass and enabled preliminary investigation of key sensitivities among RA, TEA, and LCA for an HTL pathway utilizing the best experimental information available. We assumed colocation of HTL processing, algae growth, nutrient recycling, and HTL oil upgrading. Algal HTL oil is expected to be more stable than fast pyrolysis oils and may be suitable for transporting to a regional upgrading facility. Nevertheless, to be conservative, we colocated the operations in part to ensure that nitrogen recovered as ammonia in produced waters, e.g., from CHG and from hydrodenitrification, can be returned to the algae culture. Resource Assessment and Growth Model Methods. Suitable land in the coterminous U.S. was identified using highresolution topography and land-cover data.25 Suitable lands had low slope (≤1%), were nonagricultural, undeveloped, or lowdensity developed, and were not environmentally sensitive. Site selection of 485 ha unit-farms found 88 692 candidates. Each was assumed to have 405 ha of cultivation pond area.26 The TEA and LCA studies assumed 10-fold aggregation to 4050 ha cultivation area for sake of process scale. Water cost and availability modeling followed a modified version of that presented previously.26 Sites were screened for adequate freshwater with a cap of 5% of the mean annual flow for each HUC_627 watershed and groundwater having salinity below 2000 mg L−1 and no deeper than 300 m. Sites with available freshwater were then ranked by the value of the produced algal fuels minus the water cost as determined by a cost−distance GIS model based on geostatistical estimates of the depth and salinity of well water supplies. Feedstock water demand equaled net evaporative loss (after precipitation) plus deliberate discharge (“blowdown”) to limit accumulation of salts to 4000 mg L−1.28 Water consumption from downstream processes is small compared to evaporation-driven demands and was not included in the estimate. Growth and water-demand models were driven by the Cligen site-level stochastic weather generator29 coupled to a mass and energy balanced hydrodynamic pond model,30 including water temperature and net evaporative loss after precipitation. Model

PNNL Biomass Assessment Tool (BAT) that allows high spatiotemporal resolution (30−100 m, hourly) of pond state and algae growth rate driven by a stochastic weather generator. Many large cultivation facilities were required to achieve 5 BGY. Thus, we examined profitability, sustainability, and capacity of a large ensemble of geographically distributed sites. Experimental algae growth data were available for only a few locations and for short time periods, yet site-specific performance data are required for modeling purposes. To close this gap, local climate data were combined with a biophysicsbased growth model that used experimentally determined parameters for promising species. Many TEAs and LCAs have been published for algae,16−21 but these studies focus on single sites rather than large ensembles and none evaluate the implications of seasonal climate variation. The present work fills a gap in the literature by evaluating site-to-site, seasonal, and intra-annual variations in production and their implications for cost and emissions. Consideration of these variations was required when assessing national-scale potential: Using only peak or lumped data gave incomplete results. We also found that economic and sustainability objectives can conflict and therefore a unified analysis of economics and sustainability driven by spatiotemporal resource assessment (RA) was required and is suggested as the proper methodology for research, policy, and financial communities to use when making algal biofuel development decisions. This work also extends the literature by integrating harmonized TEA, LCA, and RA models based on a mass and energy-balanced Aspen Plus model.



MATERIALS AND METHODS The algae growth and harvest model is based upon earlier work that evaluated lipid extraction and hydrotreatment. The model considers algae grown in raceway ponds22 and harvested by several stages of wastewater treatment processes that were selected by their industrial maturity, applicability to algae, and availability of data for performance with algae.23 Nutrient supply with diammonium phosphate and ammonia meets demand after nutrient recycling on-site. CO2 is supplied via low-pressure flue-gas pipelines. Details are provided in the Supporting Information (SI). Currently, experimental data for algal growth, harvesting, and continuous-HTL processing, when available, are for different species and conditions. The system model therefore required assumptions and extrapolations and combined data from different species, typical of the current state of the art for whole-pathway algal biofuel modeling. The only data known that included continuous HTL, treatment of HTL aqueous 6036

dx.doi.org/10.1021/es4055719 | Environ. Sci. Technol. 2014, 48, 6035−6042

Environmental Science & Technology

Article

Figure 2. Map showing the sites selected to meet the 5 BGY production target and the median sites for each study group. The groups are the original analysis areas defined for our initial lipid extraction study.23 Three regions used in the previous study (Groups 4−6) were empty because of the higher fuel yields per farm in the current study.

separate Aspen simulations. Consistency, especially with regard to recycle loops and heat integration, was possible because the process itself divides naturally along these lines. Thermal processes occur only within the conversion and upgrading section, so heat integration could be addressed directly within that single model. The simulations were iterated to mass balance closure of nutrient, recycle, and water streams. Key whole-process mass and energy balances were used as the basis for analysis (TEA and LCA). Life-Cycle Analysis Methods. LCA was performed with the Greenhouse gases, Regulated Emissions, and Energy use in Transportation model (GREET). GREET is a publicly available LCA tool that investigates numerous fuel and vehicle cycles.32−34 GREET has been extended to include algal biofuel production.35,36 GREET computes fossil, petroleum, and total energy use, emissions of greenhouse gases (CO2, CH4, and N2O), and emissions of six criteria pollutants. See Figure SI-4 in the SI for the LCA system boundary definition. LCA utilized the same approach as in our previous work.23 One MJ of fuel on a lower heating value (LHV) basis was used as the functional unit. The naphtha and diesel yields in the conversion and upgrading model were estimated from a SimDis curve (ASTM D2887), which determines the weight fraction vaporized at a given temperature. The heavier than diesel-range material from the SimDis curve was assumed to be cracked predominately to additional diesel-range material plus a small amount of naphtha-range material. The approach is approximate; therefore, both naphtha and diesel were treated as a single liquid fuel in the LCA by combining them on an energy basis. Note that the naphtha cut accounts for only 13 wt % of the sum of the two. Techno−Economic Analysis Methods. TEA methodologies for process modeling and cash flow calculations were based on details presented in refs 15 and 23, with consistent system assumptions as discussed above for LCA. Material and energy balance outputs from the Aspen simulations determined

outputs were computed on an hourly basis for 30 simulated years. Site-specific biomass yield had a strong effect on RA screening and was computed using a species-specific growth model.24 This approach calculated the specific growth rate in discretized culture volume slices at specified depths. The biomass light absorption coefficient (Beer−Lambert’s Law) and the specific growth rate as a function of light intensity and water temperature were determined experimentally for strain NAABB 2412. Simulated harvest was triggered at 0.5 g/L algae concentration. RA did not consider CO2 supply, which is likely to affect site selection.31 Mass and Energy Balance Methods. The algae growth and harvest model (Figure 1) was based upon earlier work that evaluated lipid extraction and subsequent hydroprocessing.23 That model used assumptions vetted by several stakeholders in the algae community and harmonized across the TEA, LCA, and RA studies. The HTL processing conditions and yields for the current model were taken from experimental work conducted at PNNL as part of the National Alliance for Advanced Biofuels and Bioproducts (NAABB) consortium.13 The feed was Nannochloropsis oceanica grown under conditions to promote high productivity. The HTL liquid product was gravity-separated (no solvent extraction) into organic and aqueous phases which were fed to a hydrotreater and catalytic gasification, respectively. All reactors were operated with continuous flow. The experiments were run at 35 wt % solids (including ash), but the model assumed 20 wt % solids (ash-free basis) and corrected for the higher heat demand with the additional water. Heat integration was considered in the model. Superheated steam from the steam cycle boiler and steam recuperated from the steam reformer was used to power compressors and provide steam. The (upstream) production and harvest model and the (downstream) conversion and upgrading model were based on 6037

dx.doi.org/10.1021/es4055719 | Environ. Sci. Technol. 2014, 48, 6035−6042

Environmental Science & Technology

Article

Figure 3. Variation in productivity (A) and GHG emissions (B) over 30 years for the five representative sites in the five site groups. Whiskers show the maxima and minima over 30 years of monthly sampling, the boxes show the 10th and 90th percentiles, and the bars show the means. The 3season (4-season) average productivity averaged over all sites was 17.9 (14.6) g m−2 d−1.. Summer = Jun−Aug; Fall = Sep−Nov; Winter = Dec−Feb; Spring = Mar−May. Note: Emissions for petroleum diesel are 95 gCO2e MJ−1.



RESULTS Resource Assessment Results and Representative Sites. The production of 5 BGY of RD required 1671 of the 485-ha sites (Figure 2). An additional 479 sites were selected to support sensitivity studies up to 6.25 BGY. All sites were located on the Gulf of Mexico coast and the Florida peninsula. Selection was not forced into this area; rather, the site ranking and selection were influenced strongly by productivity and freshwater availability. Because the Aspen models were run and converged manually for each case, TEA studies could not examine every site. Therefore, sites were grouped geographically by productivity and a median-rank site was determined for each group. These median sites (Figure 2) were used as representative sites for each region in TEA and LCA studies and group numbers were assigned consistently with our prior analyses.23 Figure 3A displays variability in productivity over the 30 simulated years for each representative site. Figure 3B shows corresponding changes in GHG emissions that result primarily from reduced algae yield per MJ of energy expended mixing and from pumping to supply water. Consequences of this variability on TEA and LCA are described below. Table SI-6 in the SI displays the direct energy use per process, which largely determines the GHG emissions. The largest direct energy demand is for pond mixing and, second, for dewatering and conversion. Techno−Economic Analysis Results. When the productivity averaged over all sites, seasons, and years defined the plant in a single fixed-productivity scenario (annualized scenario), the plant produced 4 million gallons per year of naphtha and 27 million gallons per year of diesel. Assuming a $2.60 gal−1 (U.S. dollars gallon−1), i.e., $0.69 L−1, credit for

the size and number of capital equipment items. After equipment costs were determined, direct and indirect overhead cost factors were applied to determine a total capital investment (TCI). The TCI, along with the facility operating expenses (also developed using Aspen models flow rates), were used in a discounted cash flow rate of return (DCFROR) analysis to determine the minimum diesel selling price (MDSP). The MDSP is the price required to obtain a net present value of zero for a specified internal rate of return (IRR) after taxes, set here at a 10% target IRR. Process and cost modeling methodologies are based on “nth-plant” assumptions, which assumes a sufficient number of similar facilities have already been built to mitigate early entry “risk” costs such as higher target IRR metrics, equipment overdesign, and lower on-stream factors (e.g., operating days per year). Further information on nth-plant assumptions and DCFROR methodologies can be found in ref 15. Purchased equipment cost estimates were based on vendor quotations, costing software, and standard engineering literature. The estimated purchase cost for each piece of equipment was adjusted to 2011 U.S. dollars and scaled to the size called for by Aspen modeling. Capital cost values were set for all unit operations based on maximum seasonal throughput for a given facility location. Variable operating costs, which are incurred only when the process is operating, and largely scale with capacity, were also based on material and energy balance results from the Aspen simulation. Fixed operating costs, e.g., labor, maintenance, and insurance, are incurred whether or not the process is operating and were based on operating at maximum capacity at a given site. 6038

dx.doi.org/10.1021/es4055719 | Environ. Sci. Technol. 2014, 48, 6035−6042

Environmental Science & Technology

Article

Figure 4. Effect of variability on minimum diesel selling price (MDSP). Sites with higher seasonal variability in algae production (green curve), e.g., Group 1, had higher CAPEX costs per annual gallon (lines without markers) because of less efficient utilization of capital from idling overcapacity relative to peak season (see text). MDSP (table at bottom) varied site to site. Annualized, single-train, and four-train MDSP values in the table account for seasonal variation, each with increasing fidelity. The weighted-average MDSP for the total 5 BGY ensemble of sites was $9.8/gal for the single-point annualized scenario, but became $11/gal for the single-train case, and $12/gal for the four-train case because of variability. Thus, the fuel price increased as seasonal variability was treated with increasing level of detail because of cost penalties from increasing facility capacity allowances. CAPEX included land cost. Weighting was based on the number of individual unit farms located in each designated group shown in Figure 2.

Figure 5. Sensitivity to total RD production cutoff. GHG emissions for three-season average (A) and corresponding productivity (B), as well as land and water consumption (C). In A and B, site by site values are shown in blue, sorted by priority with most promising sites on the left, e.g., the blue value plotted at 2 BGY enters the ensemble to expand the total production to 2 BGY. Wiggles in the blue curve indicate site-to-site variation. The ensemble average (red) sums over all site values (blue) less than the abscissa value. Below 4BGY, the site values fluctuate around the ensemble average. Above 4 BGY, site-to-site variation increases and the higher per-site values pull the ensemble average upward. 6039

dx.doi.org/10.1021/es4055719 | Environ. Sci. Technol. 2014, 48, 6035−6042

Environmental Science & Technology

Article

production and harvesting system followed the maximum productivity, e.g., Groups 2 and 3 exhibited the highest summertime production rate (Figure 4), and thus required the highest installed capital costs. Nevertheless, even though Group 3 exhibited the highest summertime productivity, it also exhibited the lowest wintertime productivity, and this seasonal variation translated into the highest selling price of all sites considered. Alternatively, the lowest MDSP occurred for Groups 7 and 8, both in Florida, which had the greatest winter productivity, and had the least amount of seasonal variability and lowest degree of under-utilized capital costs during off-peak seasons (e.g., CAPEX per annual gallon in Figure 4). When considering economic results for all representative sites with a single-train conversion model, location and seasonal variability effects increased the weighted-average MDSP by $1.2 gal−1 to $11.0 gal−1 compared to the annualized scenario that neglected these variations ($9.8 gal−1). For a four-train conversion scenario that would be more robust for 10-fold turn-down, the MDSP increased to $12.4 gal−1, with the additional cost increase due to economy of scale losses in switching to multiple smaller-size equipment operations. Although operating a single train might be possible, these scenarios bracket the possible turn-down needs for the HTL and upgrading plant and show that the costing implications of production variation require further analysis than has been presented to date. The scenarios presented here are illustrative rather than definitive. The 3-season (4-season) average productivity, fossil energy use, and GHG emissions, averaged over all sites, were 17.9 (14.6) g m−2 d−1, 0.41 (0.44) MJ MJ−1, and 36 (38) gCO2e MJ−1, respectively. Summarized results like these are typical in the literature but fail to convey the variability associated with the spatiotemporal sample comprising a national scale assessment. Even if uncertainties are stated, information is missed. For example, seasonal LCA analysis favored 3-season operation with a winter shutdown to avoid high GHG emissions and showed both large site-to-site and interannual variations. Threeseason operation is contrary to TEA results, which previously have shown economic benefit from operating through all four seasons.23 Correlations with total aggregated production (Figure 5) suggest that expanding to larger total fuel production will exacerbate these variations. Plant construction could add significantly to the GHG emissions, but was not considered here.37 This study progressed from no consideration of variability (annualized scenario) up through increasingly refined assessment, working from site-to-site and seasonal variation of 30year averages (TEA) to full consideration of interannual variations (LCA). Each step led to further insights and further consequences for costs and emissions. The changes in MDSP may seem small as percentages, but the spread in cost across the selected sites (Figure 4) is $2 gal−1 even for the RA selection of the most ideal locations in the U.S. When R&D reduces total costs closer to a $3 gal−1 target range, spatiotemporal variations of this scale will be important. Although cost reductions required to achieve $3 gal−1 should also reduce the size of the MDSP variations described here, analysis of spatiotemporal variations should be part of cost and emissions analyses. The effects discussed above ultimately derive from the nonlinear response of the system cost and emissions to the algae productivity (Figure 6). Because the 4-season average productivity (14.6 g m−2 d−1) lies near the nonlinear portion of

naphtha (ICIS 2011 spot price average for paraffinic naphtha), the MDSP was $9.8 gal−1 of diesel. The annualized scenario does not account for spatial or seasonal variations in biomass productivity. The productivity can vary 10-fold between seasons at some sites (Figure 4). A single train design for conversion and upgrading may not robustly accommodate these large changes in flow. One option is to implement multiple processing trains and idle them as needed (while harvest and dewatering operations already utilize many parallel units and do not require seasonal design modifications). Capital costs for the conversion and upgrading equipment in a four-train scenario were estimated by scaling, using single-train costs (CAPEX set via the most productive season) raised to a power of 0.6. Handling turn-down in this way increased the weighted-average MDSP from $11.0 gal−1 (single-train) to $12.4 gal−1 (4-train), when both scenarios introduce consideration of seasonal variability among representative site groups. See Figure 4. Life-Cycle Results. The Excel-based LCA model allowed variation to be studied in more detail because GHG emissions could be calculated for every representative site and every season over 30 years, avoiding 30-year averaging. During the winter, three of the five sites exceeded the emissions for petroleum diesel (95 gCO2e MJ−1) (Figure 3B) and none of the sites achieved a 2-fold reduction in emissions. If GHG emissions reduction were the focus, LCA results imply 3-season operation and winter shutdown may be required if winter productivities are not increased, but this is contrary to TEA requirements. For example, the weighted average in the sitedependent single train TEA scenario increased from $11.0/ gallon diesel (4-season operation) to $13.3/gallon of diesel (3season operation) with a 90% stream factor for each operating season. The 5 BGY fuel target was an arbitrary cutoff corresponding to roughly one quad of energy, a minimum for a significant contribution to the national energy mix. We studied the sensitivity to this cutoff for the 3-season GHG emissions averaged over 30 years (3-season results are used as a proxy for 4-season results after improvement of strains for winter performance). LCA results were computed for the 1671 baseline sites plus an additional 479 sites to reach 6.25 BGY and the fuel-weighted average emissions were determined for subsamples producing between zero and 6.25 BGY. The results, Figure 5, show that the cumulative ensemble average is relatively insensitive to the 5 BGY cutoff near 5 BGY, but the site-to-site variation increased substantially as the RA selection pushed to less ideal sites. The sites around 2.3 BGY have higher water demands from blowdown and evaporation, leading to higher site GHG emissions associated with pumping power. The results are equivalent for a 4-season average. Land consumption trends were nearly linear, but water consumption was nonlinear with a break point around 4 BGY due to inclusion of less water-efficient sites. These trends arose from productivity dominating the selection ranking, but, if water efficiency had dominated the ranking, the trends would differ.



DISCUSSION RA modeling showed spatial, seasonal, and year-to-year variability in productivity and water demand. In TEA studies of seasonal variations after averaging the 30-year data period, seasonality affected economics strongly because of cost penalties incurred for under-utilization of equipment during off-peak seasons. For example, the total capital costs for the 6040

dx.doi.org/10.1021/es4055719 | Environ. Sci. Technol. 2014, 48, 6035−6042

Environmental Science & Technology

Article

locations. When assessing national-scale fuel production, resource assessment, including weather and site-level production performance, must be used to drive LCA and TEA studies. The variation in algae growth rate across seasons, years, and sites will affect overall economics and emissions and must be considered for an analysis to be comprehensive. The combination of seasonal productivity variations and consideration of an ensemble of sites large enough for a national assessment affected economic and environmental performance adversely. Algal biofuel modeling is at an early stage because of missing process data and because the pathway is a complex industrial pathway involving many entities, including algae production, processing, and possibly regional upgrading facilities, refineries, and fuel blending. Although we believe the models reported here represent the state of the art, it was not our intention to provide algal fuel prices and emissions for use in policy making. Rather, it is our intention to show that spatiotemporal effects must be included in national scale analyses and to identify key R&D needs, such as reducing seasonal variation in productivity, e.g., via improved strains and crop rotation.

Figure 6. Sensitivity of minimum diesel selling price and GHG emissions to productivity for HTL pathway with average water use from all sites. The nonlinear response to the climate-dependent productivity drives the sensitivity of TEA and LCA results to climatic variation.



the response curves, seasonal and annual reductions in productivity push the system to poor performance rapidly. Note that high productivity gives only asymptotic returns. Increasing productivity is not sufficient by itself to achieve realizable systems: To take advantage of increases in productivity and oil yield required for algal biofuel viability, the asymptotes in Figure 6 must be driven down by improving the energy efficiency and by reducing the costs of the unit operations themselves. The LCA and TEA results for emissions and fuel costs were presented to allow quantitative discussion of the effects of seasonal and site-to-site variation, but the results themselves are not meant to definitively establish performance of commercial algal systems as so many questions remain about scale-up, industrial process integration, and cost reduction. The scope of this work does not assess the ultimate cost, sustainability, or resource potential for improved cultivation and processing conditions moving forward into the future, but a number of improvements that would improve viability are as follows. Doubling productivity would reduce MDSP to $6 gal−1. Other costs savings are possible if pond liners can be avoided, on the order of 25%. Avoiding DAF, e.g., through higher performance upstream, or otherwise making use of more energy- and/or cost-efficient dewatering operations would reduce cost and energy demand. The algae community is pursuing these topics, but it should be noted that pond cost (CAPEX) continues to dominate the TEA results. Also, this analysis considered just one possible pathway, while others are possible, such as lipid extraction and upgrading to RD. We assumed on-site upgrading of HTL oil to blend stock to simplify the analysis, but future work must carefully examine trade-offs between scale, transportation, and cost. See ref 35 for nutrient recycling implications of refinery integration. Recent studies on nutrient31 and infrastructure38 implications for site selection should be incorporated into future work. The U.S research, policy, and financial communities require metrics for managing algal biofuel development. Decisionmaking requires accurate assessment of the state of the art from which a baseline system is defined and requires a uniform method for comparing options. The results reported here show that both annual-average and peak algae growth rate are poor metrics when considered alone for selecting cultivation

ASSOCIATED CONTENT

S Supporting Information *

Details of the model arranged in parallel with the structure of the main article. This material is available free of charge via the Internet at http://pubs.acs.org.



AUTHOR INFORMATION

Corresponding Author

*Phone: 630-252-5875; fax: 630-252-1342; e-mail: efrank@anl. gov. Notes

The authors declare no competing financial interest.



ACKNOWLEDGMENTS This work was sponsored by the U.S. Department of Energy (DOE). Argonne National Laboratory is a DOE laboratory managed by UChicago Argonne, LLC under contract DEAC02-06CH11357. NREL is a national laboratory of the U.S. Department of Energy, Office of Energy Efficiency & Renewable Energy, operated by the Alliance for Sustainable Energy, LLC, under contract DE-AC36-08GO28308. Pacific Northwest National Laboratory is operated by Battelle for DOE under contract DE-AC05-76RL01830.



REFERENCES

(1) U.S. Department of Energy. National Algal Biofuels Technology Roadmap; Washington, DC, 2010; p 140. (2) Minowa, T.; Yokoyama, S.-y.; Kishimoto, M.; Okakura, T. Oil production from algal cells of Dunaliella tertiolecta by direct thermochemical liquefaction. Fuel 1995, 74 (12), 1735−1738. (3) Brown, T. M.; Duan, P.; Savage, P. E. Hydrothermal liquefaction and gasification of Nannochloropsis sp. Energy Fuels 2010, 24 (6), 3639−3646. (4) Jena, U.; Das, K. Comparative evaluation of thermochemical liquefaction and pyrolysis for bio-oil production from microalgae. Energy Fuels 2011, 25 (11), 5472−5482. (5) Valdez, P. J.; Dickinson, J. G.; Savage, P. E. Characterization of product fractions from hydrothermal liquefaction of Nannochloropsis sp. and the influence of solvents. Energy Fuels 2011, 25 (7), 3235− 3243. (6) Vardon, D. R.; Sharma, B. K.; Scott, J.; Yu, G.; Wang, Z.; Schideman, L.; Zhang, Y.; Strathmann, T. J. Chemical properties of

6041

dx.doi.org/10.1021/es4055719 | Environ. Sci. Technol. 2014, 48, 6035−6042

Environmental Science & Technology

Article

biocrude oil from the hydrothermal liquefaction of Spirulina algae, swine manure, and digested anaerobic sludge. Bioresour. Technol. 2011, 102 (17), 8295−8303. (7) Garcia Alba, L.; Torri, C.; Samorì, C.; van der Spek, J.; Fabbri, D.; Kersten, S. R. A.; Brilman, D. W. F. Hydrothermal treatment (HTT) of microalgae: Evaluation of the process as conversion method in an algae biorefinery concept. Energy Fuels 2011, 26 (1), 642−657. (8) Jazrawi, C.; Biller, P.; Ross, A. B.; Montoya, A.; Maschmeyer, T.; Haynes, B. S. Pilot plant testing of continuous hydrothermal liquefaction of microalgae. Algal Res. 2013, 2 (3), 268−277. (9) López Barreiro, D.; Prins, W.; Ronsse, F.; Brilman, W. Hydrothermal liquefaction (HTL) of microalgae for biofuel production: State of the art review and future prospects. Biomass Bioenergy 2013, 53 (0), 113−127. (10) Biller, P.; Ross, A. B. Potential yields and properties of oil from the hydrothermal liquefaction of microalgae with different biochemical content. Bioresour. Technol. 2011, 102 (1), 215−225. (11) Yu, G.; Zhang, Y.; Schideman, L.; Funk, T.; Wang, Z. Distributions of carbon and nitrogen in the products from hydrothermal liquefaction of low-lipid microalgae. Energy Environ. Sci. 2011, 4 (11), 4587−4595. (12) Zhu, Y.; Albrecht, K. O.; Elliott, D. C.; Hallen, R. T.; Jones, S. B. Development of hydrothermal liquefaction and upgrading technologies for lipid-extracted algae conversion to liquid fuels. Algal Res. 2013, 2, 455−464. (13) Elliott, D. C.; Hart, T. R.; Schmidt, A. J.; Neuenschwander, G. G.; Rotness, L. J.; Olarte, M. V.; Zacher, A. H.; Albrecht, K. O.; Hallen, R. T.; Holladay, J. E. Process development for hydrothermal liquefaction of algae feedstocks in a continuous-flow reactor. Algal Res. 2013, 2, 445−454. (14) Aden, A.; Foust, T. Technoeconomic analysis of the dilute sulfuric acid and enzymatic hydrolysis process for the conversion of corn stover to ethanol. Cellulose 2009, 16 (4), 535−545. (15) Humbird, D.; Davis, R.; Tao, L.; Kinchin, C.; Hsu, D.; Aden, A.; Schoen, P.; Lukas, J.; Olthof, B.; Worley, M. Process Design and Economics for Biochemical Conversion of Lignocellulosic Biomass to Ethanol; NREL Report TP-5100-47764; National Renewable Energy Laboratory: Golden, CO, 2011. (16) Nagarajan, S.; Chou, S. K.; Cao, S.; Wu, C.; Zhou, Z. An updated comprehensive techno-economic analysis of algae biodiesel. Bioresour. Technol. 2013, 145 (0), 150−156. (17) Sun, A.; Davis, R.; Starbuck, M.; Ben-Amotz, A.; Pate, R.; Pienkos, P. T. Comparative cost analysis of algal oil production for biofuels. Energy 2011, 36 (8), 5169−5179. (18) Amer, L.; Adhikari, B.; Pellegrino, J. Technoeconomic analysis of five microalgae-to-biofuels processes of varying complexity. Bioresour. Technol. 2011, 102 (20), 9350−9359. (19) Ma, J. Techno-Economic Analysis and Engineering Design Consideration of Algal Biofuel in Southern Nevada; Nevada Renewable Energy Consortium, 2011. (20) Richardson, J. W.; Johnson, M. D.; Zhang, X.; Zemke, P.; Chen, W.; Hu, Q. A financial assessment of two alternative cultivation systems and their contributions to algae biofuel economic viability. Algal Res. 2014, in press, corrected proof. (21) Handler, R. M.; Canter, C. E.; Kalnes, T. N.; Lupton, F. S.; Kholiqov, O.; Shonnard, D. R.; Blowers, P. Evaluation of environmental impacts from microalgae cultivation in open-air raceway ponds: Analysis of the prior literature and investigation of wide variance in predicted impacts. Algal Res. 2012, 1 (1), 83−92. (22) Lundquist, T.; Woertz, I.; Quinn, N.; Benemann, J. R. A Realistic Technology and Engineering Assessment of Algae Biofuel Production; Energy Biosciences Institute: Berkeley, CA, 2010, p 1. (23) Davis, R.; Fishman, D.; Frank, E. D.; Wigmosta, M. S.; Aden, A.; Coleman, A. M.; Pienkos, P. T.; Skaggs, R. J.; Venteris, E. R.; Wang, M. Q. Renewable Diesel from Algal Lipids: An Integrated Baseline for Cost, Emissions, and Resource Potential from a Harmonized Model; Technical Report ANL/ESD/12-4; PNNL-21437; NREL/TP-510055431; Prepared for U.S. Department of Energy Biomass Program; Argonne National Laboratory: Argonne, IL; National Renewable

Energy Laboratory: Golden, CO; Pacific Northwest National Laboratory: Richland, WA, 2012; p 85. (24) Huesemann, M. H.; Van Wagenen, J.; Miller, T.; Chavis, A.; Hobbs, S.; Crowe, B. A screening model to predict microalgae biomass growth in photobioreactors and raceway ponds. Biotechnol. Bioeng. 2013, 110 (6), 1583−1594. (25) Wigmosta, M. S.; Coleman, A. M.; Skaggs, R. J.; Huesemann, M. H.; Lane, L. J. National microalgae biofuel production potential and resource demand. Water Resour. Res. 2011, 47 (1), W00H04. (26) Venteris, E. R.; Skaggs, R. L.; Coleman, A. M.; Wigmosta, M. S. A GIS cost model to assess the availablity of freshwater, seawater, and saline groundwater for algal biofuel production in the United States. Environ. Sci. Technol. 2013, 4840−4849. (27) NHD. National Hydrography Dataset, Medium Resolution Watershed Boundary Dataset. http://nhd.usgs.gov/wbd.html. (28) Venteris, E. R.; Wigmosta, M. S. Water Cost and Availabilty for Algae Cultivation- Salinity Issues; ABO Webinars; Algae Biomass Organization: Preston, MN, 2013. (29) Nicks, A.; Gander, G. In CLIGEN: A weather generator for climate inputs to water resource and other models. In Proceedings of Fifth International Conference on Computers in Agriculture; 1994; pp 903−909. (30) Perkins, W. A.; Richmond, M. C.; McMichael, G. A. In TwoDimensional Modeling of Time-Varying Hydrodynamics and Juvenile Chinook Salmon Habitat in the Hanford Reach of the Columbia River; World Water and Environmental Resources Congress; ASCE: Reston, VA, 2004; pp 1−8. (31) Venteris, E. R.; Skaggs, R. L.; Wigmosta, M. S.; Coleman, A. M. A national-scale comparison of resource and nutrient demands for algae-based biofuel production by lipid extraction and hydrothermal liquefaction. Biomass Bioenergy 2014, in press, corrected proof. (32) Wang, M. The Greenhouse Gases, Regulated Emissions, and Energy Use in Transportation (GREET) Model: Version 1.5; Center for Transportation Research, Argonne National Laboratory: Argonne, IL, 2008. (33) Wang, M. Q. GREET 1.5 - Transportation Fuel-Cycle Model - Vol. 1: Methodology, Development, Use, and Results; ANL/ESD-39 Vol. 1; Center for Transportation Research, Argonne National Laboratory: Argonne, IL, 1999. (34) GREET Argonne Greet Model. http://greet.es.anl.gov/main (11/ 30/2013). (35) Frank, E.; Elgowainy, A.; Han, J.; Wang, Z. Life cycle comparison of hydrothermal liquefaction and lipid extraction pathways to renewable diesel from algae. Mitig. Adapt. Strateg. Global Change 2013, 18 (1), 137−158. (36) Frank, E.; Han, J.; Palou-Rivera, I.; Elgowainy, A.; Wang, M. Life-Cycle Analysis of Algal Lipid Fuels with the GREET Model; Center for Transportation Research, Energy Systems Division, Argonne National Laboratory: Argonne, IL, 2011. (37) Canter, C. E.; Davis, R.; Urgun-Demirtas, M.; Frank, E. D. Infrastructure associated emissions for renewable diesel production from microalgae. Algal Res. 2014, in press. (38) Venteris, E. R.; McBride, R. C.; Coleman, A. M.; Skaggs, R. L.; Wigmosta, M. S. Siting algae cultivation facilities for biofuel production in the United States: Trade-offs between growth rate, site constructability, water availability, and infrastructure. Environ. Sci. Technol. 2014, 48 (6), 3559−3566.

6042

dx.doi.org/10.1021/es4055719 | Environ. Sci. Technol. 2014, 48, 6035−6042