1-s2.0-S0301421512003163-main

Energy Policy 46 (2012) 498–510

Contents lists available at SciVerse ScienceDirect

Energy Policy

0301-42

http://d

n Corr

E-m

journal homepage: www.elsevier.com/locate/enpol

Microalgal biodiesel and the Renewable Fuel Standard’s greenhousegas requirement

Kullapa Soratana n, Willie F. Harper Jr., Amy E. Landis

Department of Civil and Environmental Engineering, University of Pittsburgh, 949 Benedum Hall, Pittsburgh, PA 15261, USA

a r t i c l e i n f o

Article history:

Received 31 August 2011

Accepted 9 April 2012Available online 23 April 2012

Keywords:

Microalgal biodiesel

Life cycle assessment

Renewable Fuel Standard

15/$ - see front matter & 2012 Elsevier Ltd. A

x.doi.org/10.1016/j.enpol.2012.04.016

esponding author. Tel.: þ1 412 805 5055; fax

ail address: [email protected] (K. Soratana).

a b s t r a c t

The Renewable Fuel Standard (RFS2) under the Energy Independence and Security Act of 2007 requires

15.2 billion gallons of domestic alternative fuels per year by 2012, of which 2 billion gallons must be

from advanced biofuel and emit 50% less life-cycle greenhouse gas (GHG) emissions than petroleum-

based transportation fuels. Microalgal biodiesel, one type of advanced biofuel, has the qualities and

potential to meet the RFS’s requirement. A comparative life cycle assessment (LCA) of four microalgal

biodiesel production conditions was investigated using a process LCA model with Monte Carlo

simulation to assess global warming potential (GWP), eutrophication, ozone depletion and ecotoxicity

potentials. The four conditions represent minimum and maximum production efficiencies and different

sources of carbon dioxide and nutrient resources, i.e. synthetic and waste resources. The GWP results of

the four CO2 microalgal biodiesel production conditions showed that none of the assumed production

conditions meet the RFS’s GHG requirement. The GWP results are sensitive to energy consumption in

harvesting process. Other impacts such as eutrophication, ozone depletion and ecotoxicity potentials,

are sensitive to percent lipid content of microalgae, service lifetime of PBRs and quantity of hexane in

extraction process, respectively. Net energy ratio and other emissions should be included in future RFS

for a more sustainable fuel.

& 2012 Elsevier Ltd. All rights reserved.

1. Introduction

The Renewable Fuel Standard (RFS) under the Energy Inde-pendence and Security Act (EISA) of 2007 requires domesticalternative fuels to meet 15.2 billion gallons by 2012, of which2 billion gallons must be from advanced biofuels. Advancedbiofuels, which include cellulosic biofuel, biomass-based dieseland other advanced biofuel, are the renewable fuels other thancorn ethanol (U.S. Environmental Protection Agency, 2010). Inaddition, life-cycle greenhouse gas (GHG) emissions fromadvanced biofuels must be at least 50% less than GHG emissionsfrom petroleum-based transportation fuels distributed in 2005(Office of Transportation and Air Quality, 2010a, b). Microalgalbiodiesel, an advanced biofuel, has the potential to support theU.S. transportation fuel and meet the RFS’s advanced biofuelsrequirement (U.S. Department of Energy, 2010). Microalgae havebeen investigated for the production of a number of differentproducts including methane, ethanol, electricity and biodiesel(Batan et al., 2010; Li Q., et al., 2008; Sander and Murthy, 2010;Stephenson et al., 2010).

ll rights reserved.

: þ1 412 624 1168.

Microalgae as biodiesel feedstock have a high growth rate,high productivity, and high photosynthetic efficiency (Avagyan,2008; Bruce, 2008; Lehr and Posten, 2009; Li Y., et al., 2008; U.S.Department of Energy, 2010). These characteristics comply withthe needs established by the Roadmap for Bioenergy and BiobasedProducts in the U.S. which are that it is easy to grow, exhibits highyields, and provides good quality fuel (Avagyan, 2008; BiomassResearch and Development Technical Advisory Committee andBiomass Research and Development Initiative, 2007). The qualityof microalgal biodiesel meets American Society for Testing andMaterials (ASTM) Biodiesel Standard D6751, thus can substitutefor petroleum diesel (Bruce, 2008; Chisti, 2007). Microalgalcultivation has been shown to consume limited land and lesswater resources than terrestrial biofuel crops. The study by Chistiin 2007 suggested that the land for microalgal cultivation requiresonly 1–3% of the total agricultural area in the U.S. for the same oil-crop diesel yield (Chisti, 2007).

Microalgal cultivation considered in this study was assumed tooccur in a closed photobioreactor (PBR). Compared to open ponds,the PBR has a better control of cultivation conditions such as masstransfer, water loss by evaporation, and contamination (Li Y.,et al., 2008; Posten, 2009). The PBR system is suitable for sensitivestrains since contamination can be controlled more easily thanin an open pond. The cell mass productivity of PBRs is aboutthree times higher than the productivity of open ponds; hence

Fig. 1. System boundaries of microalgal biodiesel production conditions. (a) The low-efficiency production with synthetic resources; (b) the high-efficiency production with

synthetic resources; (c) the low-efficiency production with waste resources; (d) the high-efficiency production with waste resources.

K. Soratana et al. / Energy Policy 46 (2012) 498–510 499

harvesting costs can be significantly reduced. Although thevolume of industrial PBR is 5–10 m3, some designs can be scaledto larger volume of 10–100 m3, and the most practical method toincrease the PBR volume is by adding more PBR units (Carvalhoet al., 2006; Janssen et al., 2003). While the closed PBR is a viablealternative for large scale production of microalgae biomass, itsoperation is still more costly than open ponds (Carvalho et al.,2006; Posten, 2009).

Although various advantages support the potential of usingmicroalgae as biodiesel feedstock, due to certain limitations, notmany applications have reached the industrial scale (Carvalhoet al., 2006). The limitations of cultivation techniques include thelow yield from open ponds and the high cost of PBRs (Lehr andPosten, 2009). High harvesting costs have been observed due tothe lighting limitations of the cultivating systems and due to thelow concentration of biomass in the systems, which result fromthe relatively small cell-size of microalgae. Drying is also anenergy-consuming process due to the large water content of theharvested biomass. In addition, microalgal cultivation facilitiesrequire higher capital cost and more operation and maintenancecompared to conventional agricultural activities. However, thedevelopment of new technologies is expected to overcome theselimitations (Li Y., et al., 2008).

Life cycle assessment (LCA) is a tool that can be used toexamine the resource consumption and potential impacts of any

product or service (International Organization for Standardization,2006; Udo de Haes and van Rooijen, 2005). LCA consists offour main steps: (1) goal and scope definition, (2) life-cycleinventory (LCI), (3) life cycle impact assessment (LCIA) and(4) interpretation. LCA is applied to this study to quantify resourceconsumption and environmental and human health impacts frompond to wheel or from microalgal cultivation to microalgal bio-diesel consumption.

The objective of this study was to conduct a comparative LCAon four conditions of microalgal biodiesel productions to evaluatetheir potential to meet the RFS2 and then to identify processes orinputs that could be targeted to minimize the overall environ-mental impact of microalgal biodiesel production. A commonperception is that high-efficiency production with synthetic

resources (condition HS) might consume more energy with bettersystem control, while a production scenario utilizing natural andwaste resources (i.e. conditions HW and LW) might consumemore energy in preparing and cleaning resources from wastestreams with unpromising yield. LCA enables researchers andpolicy makers to quantify the impacts of these systems andinvestigate tradeoffs. In addition, we use LCA results from thisstudy to evaluate policies such as the RFS and to improve uponthe production of microalgal biodiesel. Co-product allocation wasnot conducted due to uncertainties related to yield and quality ofco-products and by-products.

K. Soratana et al. / Energy Policy 46 (2012) 498–510500

2. Methods

2.1. Environmental impact assessment using LCA

Four microalgal biodiesel production conditions were definedbased on the combination of two different efficiencies of produc-tions and two different sources of resources used during cultiva-tion. The two efficiencies of production examined were high-efficiency production (H) and low-efficiency production (L),described in more detail in Sections 2.1.1 and 2.1.2. The twosources of resources examined were synthetic resources (S),which included synthetic CO2 and synthetic fertilizers, andnatural and waste resources (W), such as sunlight, CO2 frompower plant flue gas and nutrients from municipal wastewater(Soratana and Landis, 2011). Thus, four conditions for producingmicroalgal biodiesel were evaluated: HS, HW, LS, and LW.

The four microalgal biodiesel production conditions wereexamined and compared for their resource consumption andenvironmental impacts using LCA; the boundaries and para-meters for each of the four conditions are presented in Fig. 1.The input parameters for each of the conditions were evaluatedfor different design production conditions using Monte Carlosimulations. Monte Carlo simulation has been used in conjunctionwith LCA to construct a distribution of output parameters thatallows the LCA practitioner to evaluate the range of possiblevalues for a system, as opposed to a single point estimate (Landiset al., 2007; Soratana and Marriott, 2010). Sensitivity analysis wasconducted using Tornado correlation diagrams, and the majorfactors influencing global warming (GWP), eutrophication, ozonedepletion and ecotoxicity potentials were identified.

2.1.1. High-efficiency production scenarios

In this study, the high-efficiency production (H) was defined asthe production scenario that cultivated genetically modifiedmicroalgae with 70% lipid content. This production methodemployed high-efficiency technology, which provided 90% har-vesting, 98% extraction, and 87% conversion efficiencies (Lardonet al., 2009; Mata et al., 2010; Vyas et al., 2010). The high-efficiency production scenario was combined with the use ofsynthetic resources to create the first production condition (HS);synthetic resource parameters included utilization of freshwaterand all synthetic resources such as urea, superphosphate andpotassium chloride fertilizers, synthetic CO2, and light providedby compact fluorescent bulbs. High-efficiency production wasalso combined with the utilization of waste resources (HW) toevaluate the second production condition. Waste resources(W) included natural and waste resources such as natural light,nitrogen (N) and phosphorus (P) waste streams from municipalwastewater and CO2 from flue gas of power plant. The systemboundaries for the different production conditions are shownin Fig. 1, while the associated data is presented in ElectronicAnnex 1.

2.1.2. Low-efficiency production scenarios

Low-efficiency production (L) was defined as the productionthat cultivated indigenous microalgae strains with 50% lipidcontent; employed harvesting, extraction and conversion whichwere assumed to have lower efficiency than the high-efficiencyproduction by 20%, or with 72%, 78% and 70% efficiencies,respectively. The resultant values corresponded to the possibleranges given by several other studies (Lardon et al., 2009; Mataet al., 2010; Vyas et al., 2010). Low-efficiency productionwas combined with synthetic resources (LS) and waste resources(LW) whose definitions were identical to those presented for

combination with the high-efficiency production conditions(Section 2.1.1).

2.1.3. System boundary

The boundaries for the LCA of microalgal biodiesel in thisstudy are presented in Fig. 1. Energy consumption and environ-mental and human health impacts from the four microalgalbiodiesel production conditions were compared on the samefunctional unit basis which was 8.94�1010 MJ/year of biodieselor 0.67 billion gallon of biodiesel/year which is equal to onebillion gallon of ethanol/year or 50% of the EISA 2007 renewablefuel volume requirement from the advanced biofuel by 2012(Office of Transportation and Air Quality, 2010a). The RFS setsenergy requirements for biofuels in terms of bioethanol. Since thisstudy focused on biodiesel, the functional unit was chosen torepresent the basis of the RFS, which is MJ of bioethanol. Due tothe lower productivity of the low-efficiency production condi-tions, LS and LW, they required more cultivation units (15 millionunits of 10 m3 PBR for each L condition) in order to provide thesame functional unit as the high-efficiency production conditions,HS and HW (3.15 million units of 10 m3 PBR for each H condition).

This study did not include the construction of biodieselrefining facilities since the properties of microalgal oil wereassumed to be compatible with the existing technology of otherbiodiesels. Transportation between facilities was omitted basedon the assumption that each facility was established on the samelocation. Furthermore, no valuable co-products or by-products(e.g. fertilizer, biodegradable plastic, reusable non-potable water,or wastes from cultivation) were included because of uncertain-ties related to yield and quality (Anderson and Dawes, 1990;Avagyan, 2008; Braunegg et al., 1998; Singh et al., 2011).Biological activity is highly dynamic and there is a need tounderstand the temporal variation in the yield and quality ofthese products in order to adequately model them within an LCA(Daigger and Grady Jr., 1982). There is also a need to betterunderstand the tradeoffs that are introduced when these addi-tional products are collected. More long-term data is neededbefore LCA practitioners can properly incorporate these co-pro-ducts and by-products into microalgal biodiesel LCAs.

2.1.4. Life cycle inventory (LCI)

Inventories of each process, from strain selection, cultivation,harvesting/dewatering, drying, cell disruption, extraction andconversion to combustion, were primarily collected from peer-reviewed literature and LCI databases, as presented in ElectronicAnnexes 1 and 2, respectively. Inventories collected from peer-reviewed literature included the resource quantities required toproduce one functional unit (Electronic Annex 1). Some inputssuch as number of PBR units and quantity of flocculant requiredto produce one functional unit were calculated separately, calcu-lations for which are also described in Electronic Annex 1.

Another set of inventories included impacts from the acquisitionand production of resources, e.g. PBR construction material, fertili-zers, chemicals and energy, required to produce one functional unit.The inventories were extracted from ETH-ESU, ecoinvent version2.0, BUWAL250 and IDEMAT (Delft University of Technology, 2001;Frischknecht and Jungbluth, 2004; Frischknecht et al., 2007;Spriensma, 2004), which are all European databases, as presentedin Electronic Annex 2. U.S. databases were not available for everyvariable considered in this work; therefore, to avoid inconsistencies,European databases were employed. Another approach to construct-ing the LCI is to use a mix of databases to construct the LCI;however, this approach has drawbacks in that there would be manyinconsistencies related to temporal, spatial and system boundariesamong the inventory data.

K. Soratana et al. / Energy Policy 46 (2012) 498–510 501

Emissions from combustion of microalgal biodiesel were6.89�109 kg CO2 eq, 1.21�105 kg N eq and 1.89�105 kg NOx eqper functional unit, the same value as soybean diesel. This assumptionwas based on the properties of both microalgal and soybeanbiodiesels, which are similar (Demirbas and Fatih Demirbas, 2011).The emission values were taken from soybean biodiesel combustionin GREET 1.8d (Argonne National Laboratory, 2010). More details arepresented in Electronic Annexes 1 and 2.

2.1.5. Life cycle impact assessment (LCIA)

LCIA is the third step in an LCA where environmental impactsare calculated from the LCI. In this study, the U.S. EPA’s Tool forthe Reduction and Assessment of Chemical and other Environ-mental Impacts (TRACI) and IMPACT 2002þ methods were used(Bare et al., 2003; Jolliet et al., 2003). Nine TRACI impactcategories were evaluated in this study: GWP, acidification,carcinogenics, non-carcinogenics, respiratory effects, eutrophica-tion, ozone depletion, ecotoxicity and smog. Non-renewableenergy use (NREU) was calculated using IMPACT 2002þ .

2.2. Sensitivity analysis using Monte Carlo analysis

A sensitivity analysis was performed using Monte Carlosimulations, and results from the sensitivity analysis are high-lighted for four TRACI impact categories: GWP, eutrophication,ozone depletion and ecotoxicity potentials (the analysis for theremaining categories is presented in Electronic Annexes 7–9). Thesensitivity analysis focused on: (1) examining the uncertainty ofthe results and (2) investigating how sensitive the process withthe highest impact is to the total impact of the microalgalbiodiesel production. Input variables were assigned triangulardistributions, described in Electronic Annex 1 in the onlineversion of this article. Monte Carlo simulations were conductedfor 10,000 iterations using @Risk (Palisade Corporation, 2010).Probability distributions for the outputs were determined by theChi-squared goodness of fit test. From this data, the minimum andmaximum values and the 95% confidence interval of the fourimpacts were obtained. Tornado diagrams were created from the

-0.1

0.0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1.0

LS HS LWHW LS HS LWHW LS HS LWHW LS HS LWHW LS HS LWHW

kg CO2 eq H+ moles eq kg benzene eq kg toluene eq kg PM2.5 eq

Global warming potential

Acidification Carcinogenics Non-carcinogenics

Respiratoryeffects

Cultivation Harvesting Drying Cell disr

Fig. 2. LCIA results for the four production conditions normalized to the LS condition. LCIA

from IMPACT 2002þ .

Monte Carlo simulations to determine the parameters of theprocesses that contributed the highest impact in each category.The tornado diagrams show how much the results will changegiven changes to the input parameters. For each impact category,the three activities with the highest correlation coefficient werereported and selected for further sensitivity analysis.

3. Results

3.1. Environmental impacts from the four microalgal biodiesel

production conditions

Normalized impacts from four different microalgal biodieselproduction conditions are depicted in Fig. 2. More details of theimpact results by process are available in Electronic Annex 3. The HWproduction condition contributed the lowest impacts while the LScondition contributed the highest impacts in all impact categories. Onthe other hand, the LW condition contributes higher impacts than theHS condition, with the exception of its carcinogenic potential due tothe avoided impact from the utilization of nutrients in wastewater. Ascan be seen from the GWP results, HW, HS and LW conditionscontribute approximately 6.9�1010, 8.5�1010 and 1.2�1011 kg CO2

eq per functional unit, or 40%, 48% and 69% of GWP of LS condition,respectively.

The filtration and screening in the harvesting process con-tribute significantly to the impacts in half of the categoriesexamined: GWP, acidification, respiratory effects, eutrophicationpotentials, and NREU. The GWP mainly results from the release ofCO2 to the atmosphere, acidification from SOx, while respiratoryeffects and eutrophication potentials result from the release ofNO2 from the production of energy used during harvestingprocess. For instance, energy consumption of these portions ofthe harvesting process contributes 9.4�1010 and 5.4�1010 kgCO2 eq per functional unit, which were up to 54% and 79% of thetotal GWP of the LS and LW conditions, respectively. The utiliza-tion of CO2 and nutrients from industrial wastes in LW and HW

LS HS LWHW LS HS LWHW LS HS LWHW LS HS LWHW LS HS LWHW

kg N eq kg CFC-11 eq kg 2,4-D eq kg NOx eq MJ primary

Eutrophication Ozone depletion Ecotoxicity Smog NREU

uption Extraction Conversion Combustion

categories were calculated using TRACI with the exception of the NREU category

K. Soratana et al. / Energy Policy 46 (2012) 498–510502

conditions does not offset those impacts contributed from theenergy intensive harvesting process.

Despite an assumed lifetime of 15 years, the HDPE materials usedto construct the PBR contribute predominantly to the total carcino-genic potential of the LW and HW conditions, non-carcinogenicpotential of the LS, LW and HW conditions and ecotoxicity and smogformation potentials of all four conditions. The production of PBRconstruction material of the LS condition contributes 5.2�106 kgbenzene eq per functional unit to carcinogenic potential, whichaccounted for 20% of the total carcinogenic potential of the LScondition, whereas the production of the PBR construction materialof the HW condition contributes 1.4�106 kg benzene eq per func-tional unit or 68% of the total carcinogenic potential of the HWcondition, which contributes the lowest carcinogenic potential amongthe four production conditions. HDPE’s highest impacts result fromthe release of lead in water to carcinogenic and non-carcinogenicpotentials, the contribution of aluminum in water to ecotoxicity andthe release of NOx to the atmosphere to smog formation potentialfrom the production of HDPE (Frischknecht et al., 2007).

The consumption of hexane in the extraction process contri-butes both to ozone depletion potential and carcinogenics. Hex-ane is the main contributor to ozone depletion and contributesapproximately 44% and 70% of the total impact in the HS and LWconditions, respectively. This is primarily due to the release ofbromotrifluoromethane from the production of hexane. Eventhough the production of hexane for the LS condition is equal tothat of the LW condition (7.4�102 kg CFC-11 eq per functionalunit) and similarly hexane production in the HS condition is thesame as the HW condition (2.7�102 kg CFC-11 eq per functionalunit), the production of hexane contributes higher impact to thetotal carcinogenic potential of the LW condition (70%) as com-pared to the LS condition (45%). The production of hexanecontributes higher impact to the HW condition (65%) as comparedto the HS condition (44%) in the carcinogenic category.

Fig. 3. Tornado correlation coefficient of impacts from the LS condition. Input parameters

percent change to the impact category when the input parameter is changed. (a) GW

condition; (d) ecotoxicity of LS condition.

3.2. Sensitivity analysis of impacts from microalgal biodiesel

The sensitivity of GWP, eutrophication, ozone depletion, andecotoxicity potentials of the four production conditions wereobtained from Monte Carlo simulations. The sensitivity resultsshow the effect of the input parameters on the impact results.

3.2.1. Tornado correlation coefficient results

The GWP and eutrophication potential of the LS condition arepresented in Fig. 3(a) and (b). The results indicate that both environ-mental impacts would change significantly given changes to the lipidcontent of microalgae and energy consumption of microstrainer andbelt filter. In this case, energy consumption of the microstrainer andbelt filter influences the GWP results more than the lipid content. Thelipid content has an inverse effect on both GWP and eutrophicationpotentials and influences the eutrophication potential more than theenergy consumption of the microstrainer and belt filter. The lipidcontent of microalgae is also the most sensitive parameter with aninverse effect on ozone depletion and ecotoxicity potentials of the LScondition. Ozone depletion potential is sensitive to the quantity ofhexane consumed in the extraction process and quantity of urea or Nfertilizer consumed in the cultivation process as presented in Fig. 3(c).Ecotoxicity is sensitive to two input parameters: number and servicelifetime of PBR as presented in Fig. 3(d). The Tornado correlationcoefficients of the LS condition and another three microalgal biodieselproduction conditions are presented in Electronic Annexes 6 and 7.

3.2.2. Probability distribution of impacts from microalgal biodiesel

Fig. 4 shows the resultant probability distributions for GWP. TheChi-squared method was used to determine best fit distributions: LSand HW were best fit to the inverse Gaussian distribution, while theHS and LW were best fit to the log normal distribution. The ‘RFSBaseline’ in Fig. 4 represents the RFS’s requirement in order for a

are shown on the y-axis while the coefficient value on the x-axis represents the

P of LS condition; (b) eutrophication of LS condition; (c) ozone depletion of LS

2 2 211 11 11

10 10 11

11 11 11

10 10 11

0

1E-11

2E-11

3E-11

4E-11

5E-11

6E-11

4E+0

9

2.4E

+10

4.4E

+10

6.4E

+10

8.4E

+10

1.04

E+1

1

1.24

E+1

1

1.44

E+1

1

1.64

E+1

1

1.84

E+1

1

2.04

E+1

1

2.24

E+1

1

Fre

quen

cy

GWP (kg CO eq) per 8.9 10 MJ of biodiesel in a year

RFS Baseline

LS HSLW HW

Fig. 4. The probability distribution of GWP from four microalgal biodiesel production conditions.

K. Soratana et al. / Energy Policy 46 (2012) 498–510 503

biofuel to be classified as biomass-based diesel or advanced biofuel,defined as 50% of the life cycle GHGs from petroleum fuels distributedin 2005 (Office of Transportation and Air Quality, 2010b; Skone andGerdes, 2008). The remainder of the results can be found in ElectronicAnnex 4. Results of the other three impact potentials, eutrophication,ozone depletion and ecotoxicity, from the four microalgal biodieselproduction conditions are presented in Electronic Annex 5.

The range of GWP results for the LS and LW conditions wassimilar. However, the result of the LS production was best fit tothe Inverse Gaussian distribution with slightly higher maximumprobability density, while the result of the LW condition was bestfit to the log normal distribution. The range of GWP of the HS andHW productions was similar. The results of the HS and HWproductions were best fit to log normal and Inverse Gaussiandistributions, respectively. The mean value of the HW productionwas lower than that of the HS production approximately by1.7�1010 kg CO2 eq per functional unit with similar maximumprobability density. Yet, even for the HW condition with thelowest GWP, approximately a 94% reduction in GHGs is necessaryto meet the RFS Baseline. Other studies reported that allocation ofimpacts to co-products, such as bioethanol and algae meal, canreduce CO2 by 108%, while has little impact on the total GWP ofmicroalgal biodiesel when co-product is glycerol (Sander andMurthy, 2010; Stephenson et al., 2010).

3.2.3. Comparison of the four conditions

The minimum, mean and maximum values of the GWP,eutrophication, ozone depletion and ecotoxicity potentials fromthe LS, HS, LW and HW microalgal biodiesel production condi-tions in Annex 5 are reported with a 95% confidence interval, asillustrated in Fig. 5, and are discussed in more detail in thefollowing paragraphs. The RFS sets a baseline only for life-cyclegreenhouse gas emissions, and as such no baseline comparisonswere made for other impacts.

GWP. The LS and LW conditions contribute higher GWP thanthe HS and HW conditions. The LS condition contributes approxi-mately 1.7�1011 kg CO2 eq per functional unit, which is abouttwo times higher than that of the HS condition. The different

results were from the lower efficiencies of harvesting, extractionand conversion processes in the LS and LW conditions, whichwere assumed to be 20% lower than the HS and HW conditions,respectively, as shown in Electronic Annex 1. Based on the samesynthetic resources, by switching from lower-efficiency produc-tion to the high-efficiency production, or from LS condition to HScondition, GWP can be reduced by 9.3�1010 kg CO2 eq perfunctional unit or approximately 54% of the LS condition. Theimpact reductions are mainly from consuming less energy duringlighting and harvesting processes. For the utilization of wasteresources, switching from LW condition to HW condition, GWPcan be reduced by 5.5�1010 kg CO2 eq per functional unit orGWP from the HW condition was less than that of the LWcondition by 46% from consuming less energy during harvestingprocess. On the other hand, the utilization of synthetic resourcesinstead of natural and waste resources can reduce the GWP of theLS condition by 32%, and can reduce the GWP of the HS conditionby 20%. The GWP results of the LS, HS, LW and HW conditionswith 95% confidence interval can be seen in Fig. 5(a).

Eutrophication potential. The eutrophication potential results ofthe four microalgal biodiesel production conditions follow thesame trend as the GWP results, since both impacts are sensitive tothe same set of parameters which are energy consumption ofmicrostrainer and belt filter in harvesting process and lipidcontent of microalgae, as illustrated in Fig. 5(b). The LS conditioncontributes the highest eutrophication potential compared toother productions. It contributes approximately 2.6�107 kg Neq per functional unit. By changing from the LS condition to theLW, HS and HW conditions, eutrophication potential can bereduced by 20%, 52% and 67%, respectively.

Ozone depletion potential. Fig. 5(c) depicts that ozone depletionpotentials are contributed from the four microalgal biodiesel produc-tion conditions. According to Fig. 3(c), the results suggested that thelipid content of microalgae and quantity of hexane consumed inextraction process are the two most sensitive parameters to ozonedepletion potential. The LS condition, which has the largest ozonedepletion potential, contributes approximately 1.5�103 kg CFC-11eq per functional unit, while the HW condition, which has thesmallest ozone depletion potential, contributes approximately

Table 1Eight sensitivity analysis scenarios.

SAscenario

Parametersa Description of scenarios Environmental impactevaluated

Base case (BC)b

SA1 Energy consumptions of microstrainer and belt

filter

50% from the base cases GWP LS, HS, LW and

HW

Eutrophication LS and HW

SA2 Microstrainer and belt filter Parameter increased by 50% from the base

cases

GWP LS, HS, LW and

HW

Eutrophication LS and HW

SA3 Microalgae Parameter reduced by 50% from the base cases Eutrophication LS and HS

Ozone depletion LS and HS

Ecotoxicity LS and HS

SA4 Lipid content of microalgae Parameter increased by 50% from the base

cases

Eutrophication LS and HS

Ozone depletion LS and HS

Ecotoxicity LS and HS

SA5 Quantity of hexane used during extraction Parameter reduced by 50% from the base cases Ozone depletion LW and HW

SA6 Quantity of hexane used during extraction Parameter increased by 50% from the base

cases

Ozone depletion LW and HW

SA7 Service lifetime of PBR Parameter reduced by 50% from the base cases Ecotoxicity LW and HW

SA8 Service lifetime of PBR Parameter increased by 50% from the base

cases

Ecotoxicity LW and HW

SA—sensitivity analysis scenarios, BC—base case (the four microalgal diesel production conditions) to compare with other SA scenarios, LS—lower-efficiency production with

synthetic resources, HS—high-efficiency production with synthetic resources, LW—lower-efficiency production with natural and waste resources, HW—high-efficiency production

with natural and waste resources, and ‘Description of scenarios’ column describes how the SA differs from the BC.a Parameters evaluated are those with the highest Tornado correlation coefficient shown in Electronic Annex 7.b Input values of BCs can be seen in Electronic Annex 1.

4.0E+09

4.4E+10

8.4E+10

1.2E+11

1.6E+11

2.0E+11

LS HS LW HW

GW

P (

kg C

Oeq

) pe

r F

U

0.0E+00

5.0E+06

1.0E+07

1.5E+07

2.0E+07

2.5E+07

3.0E+07

LS HS LW HW

Eut

roph

icat

ion

(kg

N e

q) p

er F

U

0.0E+00

2.0E+02

4.0E+02

6.0E+02

8.0E+02

1.0E+03

1.2E+03

1.4E+03

1.6E+03

LS HS LW HW

Ozo

ne D

eple

tion

(kg

CF

C-1

1 eq

) pe

r F

U

0.0E+00

2.0E+09

4.0E+09

6.0E+09

8.0E+09

1.0E+10

1.2E+10

LS HS LW HW

Eco

toxi

city

(kg

2,4

-D e

q) p

er F

U

RFS Baseline

Fig. 5. Total impacts with a 95% confidence interval from the four microalgal biodiesel production conditions. (a) GWP; (b) eutrophication; (c) ozone depletion;

(d) ecotoxicity. Means of impacts from the four productions are represented by the line, while the edge of the bar represents the 95% confidence interval. The RFS Baseline is

the Renewable Fuel Standard requirement in order for a biofuel to be classified as biomass-based diesel or advanced biofuel, defined as 50% of the life cycle GHGs from petroleum

fuels distributed in 2005 (Office of Transportation and Air Quality, 2010b; Skone and Gerdes, 2008).

K. Soratana et al. / Energy Policy 46 (2012) 498–510504

Fig. 6. Sensitivity analysis scenarios for the low-efficiency production and high-efficiency production conditions with synthetic resources (LS and HS). Sensitivity analysis was

conducted for four input parameters: (a) the energy consumption of microstrainer and belt filter on the GWP, and (b, c, d) the lipid content of microalgae on the eutrophication,

ozone depletion and ecotoxicity potentials, respectively. Description of the SA scenarios can be seen in Table 1.

K. Soratana et al. / Energy Policy 46 (2012) 498–510 505

Fig. 7. Environmental impact potentials from the low-efficiency production and high-efficiency production with natural and waste resources (LW and HW) sensitivity analysis

scenarios. The parameters considered are energy consumption of microstrainer and belt filter on GWP and eutrophication potential in (a) and (b), quantity of hexane in

extraction process on ozone depletion potential in (c) and service lifetime of PBR on ecotoxicity potential in (d). Description of the SA scenarios can be seen in Table 1.

K. Soratana et al. / Energy Policy 46 (2012) 498–510506

K. Soratana et al. / Energy Policy 46 (2012) 498–510 507

3.3�102 kg CFC-11 eq per functional unit. Although, ranges of theimpact from the HS and LW conditions are overlapping, the impactfrom the HS condition is slightly lower. The HS and LW conditionscontribute approximately 6.7�102 and 7.7�102 kg CFC-11 eq perfunctional unit, respectively. For the same source of resources, byswitching from lower-efficiency to higher-efficiency production, fromLS condition to HS condition and from LW condition to HW condition,ozone depletion potential of the lower-efficiency production can beoffset by 54% and 57%, respectively. In addition, for the sameproduction efficiency, the utilization of natural and waste resourcesinstead of synthetic resources can offset ozone depletion potentials ofLS and HS conditions by 47% and 50%, respectively.

Ecotoxicity potential. As can be seen in Fig. 5(d), the LScondition contributes approximately 1.1�1010 kg 2,4-D eq perfunctional unit to ecotoxicity potential, which is higher than theHS condition by 63%. The ecotoxicity potential of the LW condi-tion is 5.3�109 kg 2,4-D eq per functional unit, which is higherthan that of the HW condition by 73%. The explanation for thisresult is that the ecotoxicity potential of the conditions dependson the lipid content of microalgae, as shown in Fig. 3(d), and alsoon the number of PBRs. The lower efficiency of LS and LWconditions, as presented in Electronic Annex 1, requires morePBR units, therefore more HDPE is needed to construct PBRs in theLS and LW conditions than the HS and HW conditions in order toproduce enough microalgae for one functional unit.

3.2.4. Sensitivity analysis of the parameters with the highest

environmental impact

Four parameters were found to have the most impact on thesystem (see results from the Tornado correlation coefficientspresented in Electronic Annex 7). A sensitivity analysis of thesefour parameters is conducted; they include: (1) lipid content ofmicroalgae, (2) service lifetime of PBR, (3) energy consumptionduring the harvesting process, which includes energy used byboth the microstrainer and belt filter and (4) quantity of hexaneconsumed in the extraction process. These parameters primarilyaffect four environmental impacts: GWP, eutrophication, ozonedepletion and ecotoxicity. The GWP was evaluated in order tocompare the result with the RFS’s requirement. The other threeenvironmental impacts were selected and investigated since eachimpact was influenced by different parameters. The sensitivityanalysis (SA) was conducted by increasing and decreasing the

Table 2The percent of impacts and energy required of the eight microalgal biodiesel producti

Scenario GWP % of BC Eutrophication % of BC Ozone d

SA1-LS 4BC by 28–45% – –

SA2-LS oBC by 28–45% – –

SA3-LS – 4BC by 76–92% oBC by

SA4-LS – oBC by 27–29% 4BC by

SA1-HS 4BC by 31–56% – –

SA2-HS oBC by 24–48% – –

SA3-HS – 4BC by 85–95% oBC by

SA4-HS – oBC by 28–34% 4BC by

SA1-LW oBC by 34–63% oBC by 26–39% –

SA2-LW 4BC by 30–59% 4BC by 22–37% –

SA5-LW – – oBC by

SA6-LW – – 4BC by

SA7-LW – – –

SA8-LW – – –

SA1-HW oBC by 35–69% oBC by 26–42% –

SA2-HW 4BC by 44–58% 4BC by 29–46% –

SA5-HW – – oBC by

SA6-HW – – 4BC by

SA7-HW – – –

SA8-HW – – –

values of the four parameters by 50% of the base case scenario, asdescribed in Table 1. Eight sensitivity analysis scenarios wereevaluated.

The sensitivity analyses of LS and HS production conditionsand LW and HW production conditions are presented inFigs. 6 and 7, respectively. The GWP, eutrophication, ozonedepletion and ecotoxicity potential results were plotted againstthe net energy ratio (NER), which is a ratio of energy produced toenergy consumed, to illustrate the change of the impacts as theenergy consumption changes (Jorquera et al., 2010; Sander andMurthy, 2010). The results of the SA were compared to theoriginal results from the BC scenarios. Percent of impacts andenergy required for the eight microalgal diesel sensitivity analysisscenarios compared to the four base cases are listed in Table 2.

The sensitivity analysis shows how the four different inputparameters affect the resultant impacts from the LS and HS basecase conditions (BC-LS and BC-HS). GWP, eutrophication, ozonedepletion and ecotoxicity potentials of the BC-LS and BC-HS arepresented in Fig. 6. When the LS and HS conditions are changedaccording to SA1 and SA2 (changing energy consumption of themicrostrainer and belt filter), the resultant GWP overlaps with thebase case (Fig. 6(a.1) and (a.2)) which indicates that changing theenergy consumption of the microstrainer and belt filter in SA1 orSA2 does not always change the resultant GWP compared to BC-LS and BC-HS.

By changing energy consumption of the microstrainer and beltfilter of the BC-LS by 50%, the total GWP result of SA1-LS increasesand SA2-LS decreases. Likewise, the energy required to produceone functional unit of SA1-LS increases while that of SA2-LSdecreases. Therefore, decreasing the energy consumption for themicrostrainer and belt filter shown in SA1-LS is one way tosignificantly reduce the overall GWP. The plots of SA1, SA2 andBC of the HS are similar to those of the LS. However, by decreasingthe energy consumption of the microstrainer and belt filter ofthe BC-HS by 50% (SA1-HS), the total GWP result and energyconsumed to produce one functional unit significantly decreasemore than the decreases of the SA1-LS. In addition, the BC-HSrequires less energy to produce one functional unit; therefore, ithas higher potential to achieve the RFS requirement compared tothe BC-LS.

Sensitivity of the input parameter affecting lipid content (SA3and SA4) for the LS and HS conditions were evaluated for theireffect on the eutrophication potential and the NER (Fig. 6(b.1) and

on scenarios compared to the four base cases.

epletion % of BC Ecotoxicity % of BC Energy required % of BC

– oBC by 32–60%

– 4BC by 33–61%

87–89% 4BC by 49–50% 4BC by 95–98%

29–30 % oBC by 17–20% oBC by 29–35%

– oBC by 33–68%

– 4BC by 27–56%

81–83% 4BC by 61–68% 4BC by 92–102%

25–28% oBC by 21–22% oBC by 25–38%

– oBC by 37–73%

– 4BC by 31–67%

48–49% – oBC by 3–5%

46–49% – oBC by 2–4%

4BC by 100–105% oBC by 4–8%

oBC by 32–37% oBC by 2%, 4BC by 0.4%

– oBC by 37–76%

– 4BC by 45–64%

39–41% – oBC by 16%, 4BC by 3%

40–42% – 4BC by 1–4%

4BC by 92–105% 4BC by 1–2%

oBC by 28–33% 4BC by 1–2%

K. Soratana et al. / Energy Policy 46 (2012) 498–510508

(b.2)). When lipid content is reduced by 50% (SA3) in both the HSand LS conditions, the NER decreases significantly and eutrophi-cation increases significantly from the base case, with littleoverlap. Decreasing the lipid content of the BC-LS and BC-HSconditions by 50% (SA3) increases their eutrophication potentialand energy required to produce one functional unit. However,when lipid content is increased by 50% (SA4) in both the HS andLS conditions, the NER only slightly increases while eutrophica-tion slightly decreases. These results suggest that microalgae withhigh lipid content is more desirable to achieve a higher NER andslight decreases in eutrophication potential. However, with lowerlipid content, there is a significant impact on eutrophicationpotential and NER.

Ozone depletion potentials of the LS and HS conditions (BC-LSand BC-HS) are sensitive to lipid content of microalgae. The trendsof ozone depletion potential are similar to the trends of eutro-phication potential. Although, when the lipid content of the BCsare decreased by 50% (SA3), the ozone depletion potentials aremore clustered than the eutrophication potentials, as shown inFig. 6(c.1) and (c.2). Decreasing the lipid content of the BC by 50%(SA3) drastically increases ozone depletion potentials of SA3-LSand SA3-HS. Conversely, increasing the lipid content of the BC by50% (SA4) decreases ozone depletion potential of SA4-LS and SA4-HS. The partial overlapping of the NER of SA4 and the base casessuggests that the increase of lipid content from the base case by50% (SA4) decreases ozone depletion potential but does notnecessarily decrease the energy required to produce one func-tional unit. The NER can be reduced to decrease life-cycle GHGemissions, thus achieving the RFS requirement; however micro-algal biodiesel production still contributes to significant ozonedepletion potential.

Similar to the ozone depletion potential, the ecotoxicitypotential is influenced by the lipid content of microalgae(Fig. 6(d.1) and (d.2)). The trends of the ecotoxicity potential alsofollow the trends of the ozone depletion potential; the resultsfrom decreasing the lipid content of the base case by 50% (SA3) donot overlap with other plots, while the results from increasing thelipid content of the base case by 50% (SA4) partly overlap with thebase case. By increasing the lipid content of the LS and HSconditions by 50% (SA3-LS and SA3-HS), the ecotoxicity potentialincreases, while decreasing the lipid content of the same condi-tions by 50% (SA4-LS and SA4-HS), the ecotoxicity potentialdecreases. These results show that increasing the lipid contentof microalgae potentially decreases energy required to produceone functional unit of microalgal biodiesel without any change toits ecotoxicity potential.

The four resultant impacts from the LW and HW base caseconditions (BC-LW and BC-HW) are influenced by three differentparameters, as presented in Fig. 7. Energy consumed during theharvesting process by both the microstrainer and belt filter (SA1and SA2) affects GWP and eutrophication potential; quantity ofhexane consumed during the extraction process (SA5 and SA6)affects ozone depletion potential; service lifetime of the PBR (SA7and SA8) affects ecotoxicity. The partial overlapping of GWP fromthe base case and the scenario with lower and higher energyconsumptions during the harvesting process than the base casecondition by 50% (SA1 and SA2) in Fig. 7(a.1) and (a.2) indicatethat changing the energy consumption during harvesting processdoes not always change the resultant GWP compared to the basecase conditions. By decreasing energy consumption during theharvesting process of the base case by 50% (SA1), the total GWPresults of SA1-LW and SA1-HW decrease by 34–63% and 35–69%,respectively. These results indicate that SA1-HW has a higherpotential to achieve the RFS requirement compared to theSA1-LW. Approximately a 71–93% reduction in GHG emissionsof SA1-HW is necessary to meet the RFS requirement.

Similarly, when changing the energy consumption during theharvesting process (SA1 & SA2), the resultant eutrophicationpotentials of the LW and HW conditions overlap with thesensitivity analysis scenarios as can be seen in Fig. 7(b.1) and(b.2). The eutrophication potential results suggest that changingthe energy consumption during the harvesting process does notalways change the resultant eutrophication potential. Comparedto the base cases, BC-LW and BC-HW, the total eutrophication ofSA1-LW decreases by 26–39% and SA1-HW decreases by 26–42%,respectively. On the other hand, the total eutrophication potentialof SA2-LW increases by 22–37%, and of SA2-HW increases by 29–46% compared to the base case conditions. Based on these results,decreasing the energy consumption during harvesting process by50%, the 69% reduction of life cycle GHG emissions of the BC-HWcan be achieved, while the maximum reduction of eutrophicationpotential is only 46% of the BC-HW. Therefore, even the micro-algal biodiesel production condition that can reduce GHG emis-sions to meet the RFS’s requirement still has significant impact oneutrophication potential.

The total ozone depletion potential is sensitive to the quantityof hexane and the ecotoxicity potential is sensitive to the servicelifetime of the PBR for both the LW and HW conditions. Decreas-ing the quantity of hexane consumed during the extractionprocess of the LW condition by 50% (SA5) results in a decreasein the total ozone depletion potential, while increasing thequantity of hexane of the LW condition by 50% (SA6) increasesthe total ozone depletion potential, as can be seen from the plotsin Fig. 7(c.1) and (c.2). A similar trend is also observed in SA5, SA6and the BC-HW conditions. The NER of SA5 and SA6 (changing thequantity of hexane during extraction process) overlap with thebase cases: BC-LW and BC-HW. The results showed that changinghexane quantity during extraction process only changes the totalozone depletion potential, but has little impact on the NER.

Changing the service lifetime of PBR (SA7 and SA8) affects thetotal ecotoxicity potential, whereas there is no effect on the NERas shown in Fig. 7(d.1) and (d.2). The microalgal biodieselproduction using a PBR with system lifetime longer than theBC-LW by 50% (SA8-LW) decreases the total ecotoxicity potential,while the production using a PBR with shorter lifetime than theBC-LW by 50% (SA7-LW) increases the total ecotoxicity potential.Likewise, increasing a PBR’s lifetime by 50% of the HW condition(SA8-HW) decreases the total ecotoxicity potential, whiledecreasing the PBR’s lifetime by 50% of the HW conditionsignificantly increases ecotoxicity potential. Hence, the PBR’slifetime should be considered when designing the microalgalbiodiesel production system or evaluating life cycle impacts ofthe microalgal biodiesel production.

The reduction of energy required to produce one functionalunit in some cases can reduce GHG emissions and meet the RFSrequirement, however other environmental impacts, such asozone depletion and ecotoxicity potentials, are also importantand future RFSs should also take them into consideration.

4. Discussion

GWP results of the four production conditions from the pondto pump LCA suggested that GWP of the HW and LS productionconditions are 6.2�1010 and 1.7�1011 kg CO2 eq per functionalunit or 0.69–1.88 kg CO2 eq per MJ, respectively, while the pond-to-wheel GWP from the two production conditions are 0.71 and1.93 kg CO2 eq per MJ. According to the Argonne NationalLaboratory’s Greenhouse gases, Regulated Emissions, and Energyuse in Transportation (GREET) Model version 1.8d.1, the well-to-wheel GWP is approximately 0.09–0.10 kg CO2 eq per MJ (ArgonneNational Laboratory, 2010; Huang et al., 2010; Independent Statistics

K. Soratana et al. / Energy Policy 46 (2012) 498–510 509

and Analysis, 2011; Research and Innovative TechnologyAdministration, 2010). The GWP from the HW condition, which isthe condition with the lowest impact among the four productionconditions, is approximately 8 times higher than the GWP from theconventional diesel production. This does not meet the RFS’s require-ment for advanced biofuels, which requires 50% fewer life-cycle GHGemissions than that from the petroleum-based transportation fuelsdistributed in 2005 (Office of Transportation and Air Quality, 2010a,b). Other environmental impacts from conventional diesel andbiodiesels can be compared to results from other studies anddatabases such as ecoinvent (Batan et al., 2010; Clarens et al., 2011;Frischknecht et al., 2007; Sheehan et al., 1998; Stephenson et al.,2010; Wang et al., 2008). Allocation can be conducted on varioustypes of co-products, such as bioethanol, algae meal, methane,electricity, glycerol and potassium sulfate. Previous research showsthat environmental impacts such as GWP can be allocated from morethan 100% to very little depending on quantity and quality of the co-products as well as the specific conditions and system boundaries ofthe systems that microalgal biodiesel production is compared to(Sander and Murthy, 2010; Stephenson et al., 2010). Due to theuncertainty and variation among possible co-products and by-pro-ducts and due to their respective lack of data availability, allocationwas not implemented in this study.

Sensitivity analysis was conducted on the parameters that weredetermined to have the most influence on the resulting environ-mental impacts. The GWP and eutrophication potential of the basecase conditions, energy consumptions of the microstrainer and beltfilter reduced by 50% from the base case condition (SA1) and energyconsumptions of the microstrainer and belt filter increased by 50%from the base case condition (SA2) resulted in a power functioncurve. The results suggest that the effect of the improvement of theenergy consumption of the microstrainer and belt filter on GWP andeutrophication potential has a diminishing return rate. The eutrophi-cation, ozone depletion and ecotoxicity potentials of the base caseconditions productions and the two SA scenarios evaluating thechange in lipid content (SA3 and SA4), resulted in natural logarithmcurves, and a diminishing return rate also occurs. For the ozonedepletion of SA5 and SA6, by switching from the scenario consumingmore hexane during the extraction process than the base casecondition by 50% (SA6) to the BC-LW or from the BC-LW to thescenario consuming less hexane during the extraction process thanthe base case condition by 50% (SA5), ozone depletion potential canbe decreased, however, NER remains the same. The trends of theecotoxicity potentials of SA7 and SA8 are similar to the trends of theozone depletion of SA5 and SA6. By switching from the scenario withshorter service lifetime of PBR than the base case condition by 50%(SA7) to the BC-LW or from the BC-LW to the scenario with longerservice lifetime of PBR by 50% (SA8), ecotoxicity potential can bedecreased, while NER remains the same.

The NER results (shown in Figs. 2, 6, and 7, and presented inmore detail in Annexes 11–13) of microalgal biodiesel are verylow compared to conventional fuels and other first-generationbiofuels such as corn ethanol (Argonne National Laboratory,2010). Microalgal biodiesel production modeled in this studybased on current technologies and without allocation to co-products does not meet the RFS’s GHG requirement for the US’senergy security purpose at present. The NER of microalgalbiodiesel can be increased through the improvement of produc-tion technologies and the use of industrial symbiosis approaches,which are the two strategies investigated in this study. Reducingenergy consumption during the harvesting process by 50% (SA1)alone can increase NER of the BC-HS and BC-HW by 3 to 6 and 4 to11 times, respectively. Under the SA1 condition, changing fromsynthetic to natural and waste resources can reduce the energyrequired to produce microalgal biodiesel to half. Another strategyto increase the NER is to develop valuable co-products, such as

microalgal bioethanol, methane or electricity, which may improvethe yield and the potential of microalgal biodiesel production toachieve the RFS’s GHG emission reduction requirement.

The GWP results from the four microalgal biodiesel productionconditions indicate that none of the assumed conditions in this studymeet the RFS’s requirement for advanced biofuel. However, theresults from this study suggest potential for improvement of theproduction. The GWP and eutrophication potential results of theprocess LCA model indicate that the harvesting process is an energyintensive process. Based on the results from Tornado correlationcoefficients, energy consumption of the microstrainer and belt filter,which are the methods used in harvesting process of this study, areone of the main contributors to GWP in the four productionconditions and are also the main contributor to eutrophicationpotential in the LW and HW production conditions. Another para-meter of importance is the lipid content of microalgae; it inverselyinfluences eutrophication, ozone depletion and ecotoxicity potentialsof the LS and HS production conditions. In addition, the quantity ofhexane consumed during the extraction process and the servicelifetime of PBRs are the major contributor to ozone depletion andecotoxicity potentials, respectively, for both the LW and HW produc-tion conditions. Therefore, decreasing the energy consumption of themicrostrainer and belt filter during the harvesting process, therecycling of hexane from the extraction process, and/or increasingof the lipid content of microalgae can result in an overall decrease inthe life-cycle environmental impact of microalgal biodiesel produc-tion and can help move them toward fuels that meet the RFS2.

This study shows that there are other factors such as NER(Shaw et al., 2010), nutrient use and reuse (U.S. Department ofEnergy, 2010) and eutrophication potential, that are important toconsider when regulating environmental impacts of biofuels.When developing policy for renewable fuels, and microalgalbiodiesel in particular, policies that focus on greenhouse gasesalone will fall short of providing sustainable fuel alternatives.Future polices should consider setting baselines for multipleenvironmental impacts, like smog, water quality, and globalwarming. These policies should also consider encouraging strate-gies such as industrial symbiosis, nutrient recycling and reuse,and development of valuable co-products.

5. Conclusion

Microalgal biodiesel was evaluated from four different produc-tion conditions: LS, HS, LW and HW. None of the four differentconditions investigated in this study meet the RFS’s requirement,which requires that advanced biofuels exhibit 50% less life-cycle GHGemissions than that from the petroleum-based transportation fuelsdistributed in 2005. Monte Carlo analysis was conducted to calculateprobability distributions for the environmental impacts of microalgalbiodiesel production. The Tornado correlation coefficient was used toidentify the parameters with high contribution to the total impacts.The four parameters that had the largest impact on the results werelipid content of microalgae, service lifetime of PBRs, energy consump-tion of the microstrainer and belt filter in the harvesting process andquantity of hexane consumed in the extraction process. Improvingthese parameters can reduce GWP, eutrophication, ozone depletionand ecotoxicity potentials of microalgal biodiesel production. In orderto meet the RFS, the high-efficiency production with the utilization ofwaste resources under the condition where energy consumption ofthe microstrainer and belt filter in harvesting process is reduced by50% of the HW condition (SA1-HW) must reduce GHG emissions byan additional 71–93% to achieve the RFS baseline. Future policies forrenewable, sustainable fuels should also consider other factors suchas NER and other emissions, in addition to GHG emissions.

K. Soratana et al. / Energy Policy 46 (2012) 498–510510

Acknowledgments

This work is based upon work supported by the National ScienceFoundation (NSF Grant no. 0932606). Any opinions, findings andconclusions or recommendations expressed in this material are thoseof the authors and do not necessarily reflect the views of NSF. Theauthors thank the anonymous reviewers for their suggestions.

Appendix A. Supporting information

Supplementary data associated with this article can be found inthe online version at http://dx.doi.org/10.1016/j.enpol.2012.04.016.

References

Anderson, A., Dawes, E., 1990. Occurence, metabolism, metabolic role, and industrialuses of bacterial polyhydroxyalkanoates. Microbiological Reviews 54, 450–472.

Argonne National Laboratory, 2010. GREET Model. U.S. Department of Energy.Avagyan, A., 2008. A contribution to global sustainable development: inclusion of

microalgae and their biomass in production and bio cycles. Clean Technologiesand Environmental Policy 10, 313–317.

Bare, J.C., Norris, G.A., Pennington, D.W., McKone, T., 2003. TRACI: The Tool for theReduction and Assessment of Chemical and Other Environmental Impacts.Journal of Industrial Ecology 6, 49–78.

Batan, L., Quinn, J., Bradley, T., Willson, B., 2010. Net energy and greenhouse gasemissions evaluation of biodiesel derived from microalgae. EnvironmentalScience & Technology 45, 7975–7980.

Biomass Research and Development Technical Advisory Committee and BiomassResearch and Development Initiative, 2007. Roadmap for Bioenergy andBiobased Products in the United States.

Braunegg, G., Lefebvre, G., Genser, K.F., 1998. Polyhydroxyalkanoates, biopolye-sters from renewable resources: physiological and engineering aspects.Journal of Biotechnology 65, 127–161.

Bruce, E.R., 2008. Opportunities for Renewable Bioenergy Using Microorganisms.(pp. 203–212).

Carvalho, A.P., Meireles, L.A., Malcata, F.X., 2006. Microalgal reactors: a review ofenclosed system designs and performances. Biotechnology Progress 22,1490–1506.

Chisti, Y., 2007. Biodiesel from microalgae. Biotechnology Advances 25, 13.Clarens, A.F., Nassau, H., Resurreccion, E.P., White, M.A., Colosi, L.M., 2011.

Environmental impacts of algae-derived biodiesel and bioelectricity fortransportation. Environmental Science & Technology, 7554–7560.

Daigger, G.T., Grady Jr., C.P.L., 1982. The dynamics of microbial growth on solublesubstrates: a unifying theory. Water Research 16, 365–382.

Delft University of Technology, 2001. IdeMat Online.Demirbas, A., Fatih Demirbas, M., 2011. Importance of algae oil as a source of

biodiesel. Energy Conversion and Management 52, 163–170.Frischknecht, R., Jungbluth, N., 2004. SimaPro Database Manual, The ETH-ESU 96

Libraries version 2.1. ESU-services.Frischknecht, R., Jungbluth, N., Althaus, H.-J.R., Bauer, C., Doka, G., Dones, R.,

Hischier, R., Nemecek, T., Primas, A., Wernet, G., 2007. Ecoinvent Report no. 2.Code of Practice Data v2.0. Swiss Federal Institute of Technology Zurich(ETHZ), Dubendorf.

Huang, G., Chen, F., Wei, D., Zhang, X., Chen, G., 2010. Biodiesel production bymicroalgal biotechnology. Applied Energy 87, 38–46.

Independent Statistics and Analysis, 2011. Voluntary Reporting of Greenhouse GasesProgram: Fuel Emission Coefficients. U.S. Energy Information Administration.

International Organization for Standardization, 2006. ISO 14040 Environmental Man-agement – Life Cycle Assessment – Principles and Framework. Switzerland.

Janssen, M., Tramper, J., Mur, L.R., Wijffels, R.H., 2003. Enclosed outdoor photo-bioreactors: light regime, photosynthetic efficiency, scale-up, and futureprospects. Biotechnology and Bioengineering 81, 193–210.

Jolliet, O., Margni, M., Charles, R., Humbert, S., Payet, J., Rebitzer, G., Rosenbaum, R.,2003. IMPACT 2002þ: a new life cycle impact assessment methodology. TheInternational Journal of Life Cycle Assessment 8, 324–330.

Jorquera, O., Kiperstok, A., Sales, E.A., EmbiruAu, M., Ghirardi, M.L., 2010.Comparative energy life-cycle analyses of microalgal biomass production in

open ponds and photobioreactors. Bioresource Technology 101, 1406–1413.Landis, A.E., Miller, S.A., Theis, T.L., 2007. Life cycle of the corn–soybean agroeco-

system for biobased production. Environmental Science & Technology 41,1457–1464.

Lardon, L., Helias, A., Sialve, B., Steyer, J.-P., Bernard, O., 2009. Life-cycle assessment of

biodiesel production from microalgae. Environmental Science & Technology, 7.Lehr, F., Posten, C., 2009. Closed photo-bioreactors as tools for biofuel production.

Current Opinion in Biotechnology 20, 280–285.Li, Q., Du, W., Liu, D., 2008. Perspectives of microbial oils for biodiesel production.

Applied Microbiology & Biotechnology 80, 749–756.Li, Y., Horsman, M., Wu, N., Lan, C.Q., Dubois-Calero, N., 2008. Articles: biocatalysts

and bioreactor design, biofuels from microalgae. Biotechnology Progress 24,815–820.

Mata, T.M., Martins, A.A., Caetano, N.S., 2010. Microalgae for biodiesel productionand other applications: a review. Renewable and Sustainable Energy Reviews14, 217–232.

Office of Transportation and Air Quality, 2010a. EPA Finalizes Regulations for theNational Renewable Fuel Standard Program for 2010 and Beyond. United

States Environmental Protection Agency.Office of Transportation and Air Quality, 2010b. EPA Lifecycle Analysis of Green-

house Gas Emissions from Renewable Fuels. United States EnvironmentalProtection Agency.

Palisade Corporation, 2010. @Risk Industrial.Posten, C., 2009. Design principles of photo-bioreactors for cultivation of micro-

algae. Engineering in Life Sciences 9, 165–177.Research and Innovative Technology Administration, 2010. Average Fuel Efficiency

of U.S. Passenger Cars and Light Trucks.Sander, K., Murthy, G., 2010. Life cycle analysis of algae biodiesel. The Interna-

tional Journal of Life Cycle Assessment 15, 704–714.Shaw, D., Hung, M.-F., Lin, Y.-H., 2010. Using net energy output as the base to

develop renewable energy. Energy Policy 38, 7504–7507.Sheehan, J., Camobreco, V., et al., 1998. Life Cycle Inventory of Biodiesel and

Petroleum Diesel for Use in an Urban Bus. National Renewable Energy

Laboratory, U.S. Department of Energy and U.S. Department of Agricultural,Colorado.

Singh, A., Nigam, P.S., Murphy, J.D., 2011. Mechanism and challenges in commer-cialisation of algal biofuels. Bioresource Technology 102, 26–34.

Skone, T.J., Gerdes, K., 2008. Development of Baseline Data and Analysis of LifeCycle Greenhouse Gas Emissions of Petroleum-Based Fuels.

Soratana, K., Landis, A.E., 2011. Evaluating industrial symbiosis and algae cultiva-tion from a life cycle perspective. Bioresource Technology 102, 6892–6901.

Soratana, K., Marriott, J., 2010. Increasing innovation in home energy efficiency:Monte Carlo simulation of potential improvements. Energy and Buildings 42,828–833.

Spriensma, R., 2004. SimaPro Database Manual, The BUWAL 250 Library version2.0. PRe! Consultants.

Stephenson, A.L., Kazamia, E., Dennis, J.S., Howe, C.J., Scott, S.A., Smith, A.G., 2010.Life-cycle assessment of potential algal biodiesel production in the United

Kingdom: a comparison of raceways and air-lift tubular bioreactors. Energy &Fuels 24, 4062–4077.

U.S. Department of Energy, 2010. National Algal Biofuels Technology Roadmap: ATechnology Roadmap Resulting from the National Algal Biofuels Workshop(DOE/EE-0332). College Park, Maryland, pp. 140.

U.S. Environmental Protection Agency, 2010. Renewable Fuel Standard Program(RFS2) Regulatory Impact Analysis.

Udo de Haes, H., van Rooijen, M., 2005. Life Cycle Approaches: The Road FromAnalysis to Practice, first edition UNEP/SETAC Life Cycle Initiative.

Vyas, A.P., Verma, J.L., Subrahmanyam, N., 2010. A review on FAME productionprocesses. Fuel 89, 1–9.

Wang, B., Li, Y., Wu, N., Lan, C., 2008. CO2 bio-mitigation using microalgae. AppliedMicrobiology and Biotechnology 79, 707–718.