Probabilistic Lifecycle Assessment of Butanol Production from Corn

Article Cite This: Environ. Sci. Technol. 2018, 52, 14528−14537

pubs.acs.org/est

Probabilistic Lifecycle Assessment of Butanol Production from Corn Stover Using Different Pretreatment Methods Nawa Raj Baral,†,‡ Carlos Quiroz-Arita,§ and Thomas H. Bradley*,§ †

Joint BioEnergy Institute, Lawrence Berkeley National Laboratory, Berkeley, California 94720, United States Biological Systems and Engineering Division, Lawrence Berkeley National Laboratory, Berkeley, California 94720, United States § Colorado State University, Department of Mechanical Engineering, Campus Delivery 1374, Fort Collins, Colorado 80523-1374, United States ‡

Environ. Sci. Technol. 2018.52:14528-14537. Downloaded from pubs.acs.org by MIDWESTERN UNIV on 01/13/19. For personal use only.

S Supporting Information *

ABSTRACT: The recalcitrant nature of lignocelluloses requires a pretreatment process before the fermentative butanol production. The commonly used pretreatment processes, such as steam explosion, sulfuric acid, ammonia fiber explosion, ionic liquid (IL), and biological, require different quantities and types of process chemicals, and produce different quality and quantity of fermentable sugars. This study determines life-cycle greenhouse gas emissions (GHG) these pretreatment methods by developing a system-level process model including corn stover feedstock supply system and the downstream butanol production process. This study further evaluates the uncertainty associated with energy use and GHG emissions for each stage of the entire butanol production chain and provide the future optimization opportunities. Probabilistic results of these analyses describe a distribution of GHG emissions with an average of 18.09−1056.12 gCO2e/MJ and a 95% certainty to be less than 33.3−1888.3 gCO2e /MJ. The highest GHG emissions of IL-pretreatment of 1056.12 gCO2e/MJ reaches to 89.8 gCO2e/MJ by switching IL-recovery from 80 to 99 wt %, which is the most influential parameter for IL-pretreatment. Additionally, credits from excess electricity, butanol yield, nitrogen replacement, and diesel fuel for transportation and harvesting were the most influential parameters. Based on the current state of technologies, apart from ionic liquid and biological pretreatments, other pretreatment processes have similar metrics of sustainability.

■

INTRODUCTION Biobutanol is a valuable industrial chemical,1 a precursor to synthetic rubber,2 and a potential transportation biofuel.3 To date, biobutanol production has been investigated through an acetone-butanol-ethanol (ABE) fermentation process wherein Clostridium species metabolizes fermentable sugars such as glucose and xylose to produce ABE (typically in the ratio of 3:6:1).4 ABE-derived butanol production from lignocellulosic biomass requires a pretreatment process.5 The pretreatment process disrupts or destroys the cellulose-hemicellulose-lignin structure and creates pores, resulting in accessible surface area for enzymatic hydrolysis.6 The successive pretreatment and enzymatic hydrolysis processes transform cellulose and hemicellulose into fermentable sugars. Steam explosion, sulfuric acid, ammonia fiber explosion (AFEX), ionic liquid, and biological are the commonly used pretreatment processes for lignocellulosic biomass.7 These pretreatment processes require different process chemicals and produce different quantities and qualities of fermentable sugars.8 The quality and quantity of fermentable sugar effects the downstream ABE fermentation process, resulting in different quantity of biobutanol. Differences in the type of process chemicals and the required quantity of utilities © 2018 American Chemical Society

(heating, cooling, and electricity) results in different lifecycle impacts for each pretreatment technology. To date, lifecycle assessments (LCAs) of butanol production from sugar cane9 corn grain,9−11 wheat straw,11 switchgrass,10,11 and corn stover3,11 have been investigated. Studies on cellulosic biomass have only considered sulfuric acid pretreatment processes and many of these studies present deterministic results. One notable study10 discussed uncertainty associated with the overall greenhouse gas (GHG) emissions for both ethanol and butanol production from corn grain and switchgrass feedstocks; but the uncertainty associated with each stage of the production chain (including feedstock supply logistics and the downstream conversion processes) is not discussed. This study seeks to bridge these research and modeling gaps and develops a comparison of the GHG emissions of the five most commonly considered pretreatment methods using a probabilistic lifecycle assessment method. Received: Revised: Accepted: Published: 14528

September 13, 2018 November 2, 2018 November 16, 2018 November 16, 2018 DOI: 10.1021/acs.est.8b05176 Environ. Sci. Technol. 2018, 52, 14528−14537

Article

Environmental Science & Technology

Figure 1. Overall butanol production chain from corn stover with the major production stages. This figure is only an example with some of the major component of the overall production process, which is created by using a process modeling software-SuperPro Designer.12 In this figure, the process chemicals for different pretreatment processes are (1) steam explosion pretreatment (none); (2) sulfuric acid pretreatment (sulfuric acid); (3) ammonia fiber explosion (ammonia); (4) ionic liquid pretreatment (1-ethyl-3-methylimidazolium acetate (EMIM Ac), and tripotassium phosphate); and (5) biological pretreatment (fungal media).

■

Corn Stover Feedstock Supply. The feedstock supply model was adapted from the authors’ recent work.16 This study adds custom created feedstock harvesting and collecting models (nutrient replacement, windrowing, baling and stacking) using publicly available data gathered from literature, which is summarized in SI Table S1.17−20 The required amount of feedstock for each pretreatment process (Table 1)

MATERIALS AND METHODS

Scope and System Boundary. The primary goal of this study was to quantify and compare the GHG emissions of butanol production systems, considering the five most commonly used pretreatment methods. The butanol production system model includes all of the required processes or operations associated with corn stover feedstock supply and biochemical conversion processes to produce biobutanol at “biorefinery gate” (Figure 1). Figure 1 illustrates all the major stages of the overall butanol production chain including materials and external energy inputs. Functional Unit. The functional unit of 1 MJ-butanol produced, at biorefinery gate, was considered for this study. Feedstock and Process Models. Modeling Approach. The butanol production chain was divided into two main sections: (1) feedstock supply; and (2) the biochemical conversion process. Each of these sections was further divided into several process units (Figure 1). The modeling software SuperPro Designer-V1012 and Microsoft Excel were used to develop the detailed process models and to link all the unit processes (Supporting Information (SI) Tables-S1, S2, and S3). GaBi was used to develop most of the unit processes and determine lifecycle energy use and GHG emissions associated with material and energy flow reflective of U.S. conditions. The required material and energy data to develop the process model in GaBi13 were gathered from the U.S. Life Cycle Inventory database14 and GREET model15 developed by Argonne National Laboratory. The following sections discuss methods, data sources, and assumptions considered in this study to develop process models of each life-cycle stage of the butanol production system.

Table 1. Comparison among Feedstock Requirement, Make-up Water, And Direct CO2 Emissions pretreatment process steam explosion sulfuric acid AFEX ionic liquid biological

feedstock consumption (kg/L-butanol)

make-up water consumption (L/ L-butanol)

direct biogenic CO2 emissions (kg/Lbutanol)

6.95

14.09

6.86

5.99 6.02 7.08 13.86

11.36 11.28 10.83 28.00

5.41 5.40 5.21 11.46

(to produce the same quantity of butanol of 94.64 million liters/year (25 million gallons/year)) was estimated using the structural composition of the feedstock and its conversion rate during pretreatment, enzymatic hydrolysis, and fermentation (SI Tables S2 and S3). The feedstock system model quantifies the energy and materials consumed to cultivate, harvest and supply the required feedstock from the field to the biorefinery gate. The required energy and materials include those associated with N, P, and K replacement due to the removal of corn stover from the field. N2O emissions from nitrogen fertilizer (1.325% of nitrogen), and N2O credits due to corn stover removal (0.69% of N per unit mass and 1.25% of N2O per unit of nitrogen in corn stover) were considered in this 14529

DOI: 10.1021/acs.est.8b05176 Environ. Sci. Technol. 2018, 52, 14528−14537

Article

Environmental Science & Technology study.20 Windrowing, baling, stacking, and transportation (includes loading, transport, and unloading) operations include the energy and emissions associated with diesel fuel consumed for these operations. Feedstock storage includes energy and emissions associated with tarp and gravel required to prepare and maintain an outdoor storage facility at the biorefinery site.16 Biochemical Conversion Process. The biochemical conversion process was divided into six life-cycle stages: (1) preprocessing; (2) pretreatment; (3) neutralization; (4) hydrolysis and fermentation; (5) recovery and separation; and (6) stillage (mainly lignin and wastewater) utilization, as shown in Figure 1. Not every life-cycle stage is applicable to every pretreatment process. The pretreatment unit includes the recovery of ionic liquid and ammonia (SI Table S2) for ionic liquid and AFEX pretreatment processes, respectively. Previous techno-economic studies with AFEX8 and ionic liquid21 pretreatments have discussed the detailed recovery methods. Among the pretreatment methods considered in this study, steam explosion and particularly sulfuric acid pretreatment processes generate microbial inhibitors such as sugar and lignin degradation compounds.22 The formation and concentration of these process inhibitors is dependent on pretreatment severity (time, temperature, and catalyst loading if required).22 This study used mild pretreatment condition (SI Table S2) to reduce the concentration of process inhibitors.23 The process inhibitors are then partly removed by using the flash condensation23 following the pretreatment resulting in a lower concentration of inhibitors in the fermentation broth. Butanol fermentation, recovery, and separation stages occurs at different concentrations and temperatures than do ethanol processes, and therefore requires rigorous consideration of the associated energy inputs and process equipment.3 The study therefore redesigned the fermentation, products recovery, and separation sections in contrast to a widely used ethanol fermentation model23 developed by National Renewable Energy Laboratory (NREL). The modifications and similarities are briefly discussed in the following paragraph. The preprocessing model is derived from the literature on cellulosic ethanol production.23,24 The model for pretreatment processes was developed following a set of recent publications.8,21 The key operating conditions and conversion rates associated with the considered pretreatment processes were gathered from recent publications, which is summarized in SI Table S2. For the ammonia-based neutralization process model is derived from the literature on cellulosic ethanol production.23 Fermentation, recovery and separation models are derived from detailed studies of butanol fermentation.3,25 These units model a process of vacuum fermentation26 and recovery followed by distillation and decantation system3,25 to separate the mixture of water, acetone, butanol, and ethanol. The vacuum fermentation allows for the recovery of acetone, butanol and ethanol along with water in a vapor phase, which reduces the toxicity of butanol to microbes. This study considered fermentation of both C6 and C5 sugars, and the liquid phase after pretreatment (for steam explosion, sulfuric acid, and AFEX pretreatments) as the Clostridium species effectively utilize both sugars3 when compared to the yeast. While the sugar hydrolysate is not produced during biological pretreatment,8 the ionic liquid pretreatment excludes the fermentation of the liquid fraction after pretreatment27 due to the toxic nature of EMIM Ac to enzyme and microbes. Operating data for the wastewater treatment and lignin

combustion process were assigned from the process models developed by NREL;23 however, the process model developed in this study is fully flexible and sensitive to feedstock/data changes. The required process chemicals, energy (electricity, heating, and cooling), and coproduct quantities per unit product (butanol) were output from the process model to provide a lifecycle materials inventory to the LCA process (SI Tables S4−S8). Product Displacement, Allocation, and Environmental Impacts. Acetone and ethanol are the material coproducts of the ABE fermentation system and can be considered direct replacements for acetone from petroleum feedstock and corn ethanol, respectively. In addition to the displacement method, mass, energy and economic allocation approaches are used to allocate co product credits (SI Table S9). Process steam and electricity are the energy coproducts of the butanol production system. While process steam is not generally exportable from the biorefinery, this study uses excess electricity to displace U.S. average electricity. This assumption is included in sensitivity cases as the market for electricity from cellulosic biorefineries is largely speculative.3 The environmental impacts of the butanol production system with different pretreatment processes was measured using Global Warming Potential (GWP) and Net Energy Ratio (NER), defined as the ratio of the available energy outputs and all the required energy inputs. Sensitivity and Uncertainty Analysis. The baseline comparison of the pretreatment options for cellulosic butanol is performed using average values of the input parameters (summarized in the SI Tables S1, S2, and S3). Sensitivity analysis allows us to rank the most influential input parameters for the output metrics of GWP and NER and was performed considering ±20% variation in the baseline value for a set of input parameters. Uncertainty analysis was performed to document the uncertainty in the outputs of life-cycle energy use, GHG emissions, and NER as a function of the uncertainty in the input parameters. For each input uncertainty considered, the minimum and maximum values, and standard deviation were used to develop a probability distribution to represent uncertainty in the input parameters (SI Tables S1, S2, and S3). This study considered four different probability distribution functions: (i) uniform; (ii) triangular; (iii) normal; and (iv) log-normal. The uncertainty analysis was populated through 10 000 Monte Carlo trials (SI Tables S4−S8).

■

RESULTS AND DISCUSSION Feedstock Requirements, Makeup Water, And Direct Biogenic CO2 Emissions. Table 1 summarizes the comparison among feedstock requirements, makeup water requirements, and direct biogenic CO2 emissions. While the theoretical conversion rate of butanol to fermentable sugar is about 20% less than that of ethanol,3 the butanol pretreatment and conversion technologies considered in this study require ∼2 times the quantity of feedstock25 over state of the art of ethanol production (which requires 3 kg/L-ethanol).23 This study’s derived feedstock requirement for butanol with sulfuric acid pretreatment is comparable to the value reported in literature of 5 kg/L-butanol.3 Biological pretreatment processes are less efficient at transforming feedstock to butanol, resulting in ∼2 times more feedstock than the sulfuric acid pretreatment technology. This is mainly due to the loss of cellulose and hemicellulose during biological pretreatment.28 Other pretreatment processes require a similar level of feedstock when compared to the sulfuric acid pretreatment method. 14530


Article


Table 2. Comparison of GHG Emissions for the Various Pretreatment Options and Lifecycle Stages of a Butanol Production System Including Coproducts Allocationsa GHG emissions (g-CO2e/MJ-butanol) life-cycle stage

steam explosion

sulfuric acid

AFEX

ionic liquid

biological

subtotal for feedstock supply nutrient replacement windrowing baling stacking biorefinery transport storage subtotal for biorefinery preprocessing pretreatment neutralization α SSFR separation stillage utilization β PSE recovery total without coproducts credits acetone and ethanol credits total with acetone and ethanol credits excess electricity credits Net GHG emissions (with excess electricity credits) total using mass allocation for coproducts total using energy allocation for coproducts total using economic allocation for coproducts

22.75 11.67 2.04 2.71 0.60 5.33 0.40 31.74 3.08 14.05 8.68 21.72 13.20 20.23 −49.21 54.49 −13.35 41.14 −23.05 18.09 45.28 46.40 46.99

19.21 10.06 1.76 2.34 0.51 4.20 0.34 36.91 3.07 6.73 18.42 17.02 11.52 17.62 −37.46 56.12 −13.22 42.90 −14.84 28.06 46.73 47.86 48.44

19.41 10.11 1.77 2.37 0.52 4.30 0.35 40.85 3.00 32.55 0.00 17.21 8.42 12.06 −32.38 60.26 −13.61 46.64 −19.76 26.88 49.93 51.16 51.78

23.29 11.90 2.08 2.77 0.62 5.52 0.41 1058.96 3.07 1078.72 0.00 15.18 3.09 6.15 −47.26 1082.25 −13.44 1068.81 −12.69 1056.12* 899.00 920.61 931.27

50.41 23.28 4.06 5.40 1.18 15.70 0.80 29.40 3.44 4.44 0.00 26.44 4.32 51.98 −61.23 79.81 −13.74 66.07 −41.84 24.23 66.04 67.66 68.47

Note: α (Simultaneous saccharification, fermentation and recovery); β (“PSE” refers to process steam and electricity generated from the stillage utilization process). *This study considered ionic liquid (IL) recovery of 80%. The optimiztic value of IL recovery of 99%29 reduces the GHG emissions below the emissions of gasoline of 93 g CO2e/MJ.18 Alternatively, alternative ILs such as cholinium lysinate29 and triethylammonium hydrogen sulfate30 could be other promising options to reduce GHG emissions of IL pretreatment. a

lifecycle. Similar contributions to life-cycle energy use are presented in SI Table S10. Although this study compares lifecycle energy consumption, energy streams with different exergy can have differing utilities, values, and impact on environment. This study only accounts for the impact on the environment through CO2 equivalent emissions. Except for the biological pretreatment stage, emissions, and energy, GHG emissions and lifecycle energy use from the feedstock supply logistics stage are similar across the pretreatment methods considered, due to their similar levels of required feedstock. About 2 times the quantity of feedstock is required for the biological pretreatment process when compared to the other methods evaluated, resulting in a higher quantity of feedstock, larger supply radius, and more field machinery required. This results in a corresponding 2-fold increase in GHG emissions and lifecycle energy used for the biological pretreatment process. Across all of the pretreatment methods evaluated, nutrient replacement was a major contributor to the GHG emissions (∼50%) and lifecycle energy use during the feedstock supply stage, followed by transportation, baling, windrowing, stacking and storage. Previous studies18 also found that nutrient application accounts for 32−56% of the net GHG emissions for starch and cellulosic feedstocks. As illustrated in Table 2, supplying the feedstock accounts for 63% of the total GHG emissions for the biological pretreatment process. Apart from biological pretreatment process, the biochemical conversion process was the major contributor to the overall GHG emissions and energy use. The significantly higher GHG emissions and energy use for ionic liquid pretreatment process is mainly due to energy and

Wastewater was assumed to be recycled after wastewater treatment, thereby reducing the required quantity of fresh makeup water. A previous study3 reported makeup water requirements of about 9 L/L-butanol using the sulfuric acid pretreatment process, which is a lower water requirement than the required water found for this study. Differences are due to the differing feedstock requirements and butanol conversion rate. Butanol production with biological pretreatment requires about 2 times more makeup water than other pretreatments, mainly due to the 2-fold feedstock requirement and the need to maintain a constant solid loading ratio of 30 wt % during pretreatment. The coproduct of the wastewater treatment process is methane (mainly from the unutilized fermentable sugars after fermentation) which is combusted together with lignin in a boiler to generate process heat, electricity, and CO2. The variations in the direct biogenic CO2 emissions (CO2 from renewable biomass and its derivatives) from the butanol production system with different pretreatment process are mainly due to differences in the lignin fraction and the quantity of unutilized sugars available in the stillage. The direct biogenic CO2 emissions estimated in this study with sulfuric acid pretreatment are similar to the direct CO2 emissions with the same pretreatment process reported in previous studies (about 5.7 kg/L-butanol).3 Biobutanol conversion does not require any makeup fossil fuel, therefore, all the direct CO2 from the butanol production system are biogenic. Baseline Greenhouse Gas Emissions and Life-Cycle Energy Use. Table 2 summarizes the baseline GHG emissions contributions from each stage of the butanol production 14531


Article


the elimination of process chemicals and materials during pretreatment. For comparison, the estimated GHG emissions and energy use without and with considering excess electricity from their reported values were 54.8 and 35.8 g CO2e/MJbutanol, and 0.71 and 0.54 MJ/MJ-butanol, respectively.3 However, these results also considered energy and emissions from biofuel distribution,3 which was excluded in this study. Table 3 and SI Table S10 summarize the net energy ratio with coproducts displacement under mass, energy and economic allocations. Biological pretreatment required less fossil energy input due to the elimination of process chemicals during pretreatment, followed by sulfuric acid, steam explosion, AFEX, and ionic liquid. Ionic liquid pretreatment is not energy efficient with the 80% ionic liquid recovery considered in the baseline scenario. The results for the ionic liquid pretreatment could be technologically improved with higher rates of ionic liquid recovery or by using ionoSolv (uses ionic liquid and water) pretreatment.30,33 Apart from ionic liquid pretreatment, which generates excess electricity, the NER is significantly higher for the other pretreatments resulting in biobutanol that could be used as an alternative to fossil fuel-based butanol. Comparison of Greenhouse Gas Emissions Associated with Butanol from Previous Studies. Figure 2 depicts a comparison of GHG emissions associated with the butanol production processes studied here with those of different biomass feedstocks and fossil energy. Except for the ionic liquid-pretreatment, GHG emissions from biomass sources are lower than those of fossil energy-derived butanol (for instance, isobutanol from propylene with syngas from coal or natural gas). However, the biobutanol production process with ionic liquid-pretreatment requires improved ionic liquidrecovery or more environment friendly ionic liquids (when compared to EMIM Ac) to be competitive with fossil energy pathways. Butanol production with most of the evaluated pretreatment process achieves net GHG emissions below the emissions from gasoline of 93 g CO2e/MJ.18 These results indicate that biobutanol with an efficient conversion pathway can contribute to the decarbonization of the transportation and industrial sectors where butanol could be used as a fuel or solvent. The overall results obtained in this study closely correlates to the results found in previous studies with sulfuric acid pretreatment and corn stover feedstock of 35.8 gCO2e/MJ and with the same pretreatment and switchgrass feedstock10 of 48 g CO2e/MJ. The differences between this study and results from previous studies3,10 may be due to differences in system boundary, feedstocks, and the different levels of lifecycle inventory parameters. For example, a previous study with corn stover feedstock3 accounts for emissions from biofuel distribution, and another study with switchgrass feedstock10 includes emissions from indirect land use changes, while both are excluded in this study, as the main focus of this study was to compare the impact of different pretreatment methods. Greenhouse Gas Emissions with Different Biorefinery Sizes. This study further evaluates the impact of biorefinery size on GHG emissions, which is shown in Figure 3. The required quantity of biomass feedstock increases with an increase in the size of biorefinery, resulting in a larger feedstock collection area. This increases truck trips and the quantity of required diesel fuel. The required fuel per tonne of dry biomass for field operations, including windrowing, baling and stacking, was less influential when compared to the fuel required for feedstock transportation. Although an increase in GHG

emissions contribution from ionic liquid (1-ethyl-3-methylimidazolium acetate)31 and antisolvent (potassium phosphate).15 For this pretreatment option, ionic liquid accounts for about 93% of the total GHG emissions. This study considered 80% recycle of both ionic liquid and antisolvent. Previous studies21,32 find that recovery of more than 97% of ionic liquid and 90% of antisolvent are required to make the ionic liquid pretreatment cost competitive to other pretreatments. In this study, assuming 99% of ionic liquid and 90% of antisolvent recovery reduces the total GHG emissions by 11 times (Figure 2). Other variations across the pretreatment

Figure 2. Greenhouse gas emissions associated with butanol production chain with different feedstocks and pretreatment methods. The horizontal dashed lines (- × - - × - - × -) and (-○- - -○-) refer to GHG emissions of isobutanol production from fossil energy with syngas from coal of 83.42 gCO2e/MJ and syngas from natural gas of 71.52 gCO2e/MJ, respectively.16 The sensitivity bar refers to the range of GHG emissions with ionic liquid recovery of 80−99 wt %.

processes are mainly due to the different type of process chemicals and water requirements for upstream processes (i.e., pretreatment and fermentation). Less amount of water required for pretreatment and fermentation (i.e., with ionic liquid and AFEX) also reduces GHG emissions and the required energy for separation and water recovery. Additionally, the lower separation energy for biological pretreatment processes are mainly due to supplying the wastewater directly from the pretreatment section to the stillage utilization section. Another important contributor to the reduction in GHG emissions and energy use is the recovery of waste heat and the recycle of waste steam from the steam turbine. Use of electricity generated from steam turbine-generator system further reduces the GHG emission and energy use. In addition to waste steam and electricity recycling, coproduct displacements significantly reduce the GHG emissions and energy use of the butanol production process. Displacement of grid electricity with the excess electricity produced by the cellulosic biorefinery was the major contributor to reduce GHG emissions and energy use when compared to two main coproducts, i.e., acetone and ethanol displacement (Table 2 and SI Table S10). Among the evaluated pretreatment methods, steam explosion pretreatment was found to have the lowest GHG emissions. This is due to 14532


Article

Environmental Science & Technology Table 3. Uncertainties Associated with GHG Emissions and NER with Allocation and System Expansion GHG emissions (gCO2e/MJ-butanol) pretreatment process steam explosion

sulfuric acid

AFEX

ionic liquid

biological

net energy ratio (NER)

mass allocation

energy allocation

μ±σ mode, frequency 95% certainty

47.45 ± 6.96 45.8, 469

48.62 ± 7.11 46.9, 478

49.23 ± 7.2 47.5,487

0.94 ± 0.07 0.94, 84

0.92 ± 0.07 0.91, 84

0.91 ± 0.07 0.89, 81

1.76 ± 0.5 1.6, 152

55.7

57.5

58.2

1.06

1.03

1.02

2.6


48.1 ± 3.55 48.6,99

49.26 ± 3.63 49.3,96

49.86 ± 3.37 48.6,98

1.06 ± 0.08 1.08,58

1.03 ± 0.08 1.03,60

1.02 ± 0.08 1.0,54

1.77 ± 0.29 1.7,58

54.1

55.4

56

1.18

1.15

1.14

1.88


51.27 ± 6.79 46.4,56

52.53 ± 6.95 50.9,61

53.16 ± 7.03 51.5,55

0.98 ± 0.1 0.97,41

0.95 ± 0.1 0.93,40

0.94 ± 0.1 0.95,44

1.77 ± 0.48 1.5,79

63.5

65.1

65.9

1.14

1.11

1.1

2.6


900.0 ± 358.9 741.1,70

921.6 ± 367.7 721.7,74

932.2 ± 371.7 768.0,71

0.42 ± 0.09 0.42,44

0.41 ± 0.09 0.39,45

0.41 ± 0.08 0.39,44

0.44 ± 0.11 0.44,50

1589.9

1629.4

1645

0.57

0.56

0.55

0.64


71.6 ± 18.8 67.9,270

73.3 ± 19.3 68.7,161

74.2 ± 19.5 68.7,427

0.91 ± 0.14 0.98,53

0.89 ± 0.14 0.86,61

0.88 ± 0.13 0.85,56

1.53 ± 5.88 1.36,427

97.5

98.7

11.7

1.11

1.09

1.08

11.68

statistics

economic allocation

mass allocation

energy allocation

economic allocation

system expansiona

System expansion includes displacement of acetone, corn ethanol, and grid electricity. “μ” refers to mean and “σ” refers to standard deviation.

a

bioconversion process becomes more efficient due to the economy of scale. These effects trade-off to minimize the net GHG emissions at a biorefinery size (Figure 3) in the range of 56.8−94.6 million liters of butanol per year (15−25 million gallons of butanol/year). However, this will change with improved conversion processes in the future. For instance, except for the comparatively less efficient biological pretreatment process, the net GHG emissions per MJ of butanol does not change greatly with the size of biorefinery. Therefore, coselection of the appropriate location of the biorefinery (close to the biomass growing field) and the biorefinery size are the most important factors to reduce the system GHG emissions. Sensitivity for GHG Emissions and Lifecycle Energy Use. The results of the sensitivity analysis are presented in Figure 4 as a ranking of the most influential input parameters for GHG emissions from cellulosic butanol production with the different pretreatment methods. This figure presents only the 15 most influential input parameters among the 80 input parameters evaluated. Apart from ionic liquid pretreatment, (where ionic liquid embedded GHG emissions dominate the response) excess electricity, butanol yield, process chemicals mainly ammonia and enzymes, and nitrogen replacement were the most influential input parameters. Most of the ionic liquids are green solvents; however, their multiple production steps produce the higher GHG emissions of about 7.62 kg-CO2e/kg of ionic liquid31 when compared to the GHG emission of 0.045 kg-CO2e/kg of sulfuric acid.15 This sensitivity analysis shows that a cost-effective and energy-efficient ionic liquid recovery method is essential to reduce the GHG emissions impact of this technology. This sensitivity analysis shows that coproducts credits play a pivotal role to reduce GHG emissions impact of butanol production system regardless of

Figure 3. Greenhous gas emissions from feedstock supply (a) and overall butanol production chain including coproducts credits (b) with different biorefinery sizes and pretreatment methods. “IL-PT” refers to ionic liquid pretreatment.

emissions was found for the feedstock supply process with an increase in the size of the biorefinery, the downstream 14533


Article


Figure 4. Sensitivity analysis results illustrating the 15 most influential input parameters for GHG emissions under each pretreatment methods. .

Figure 5. Uncertainty associated with lifecycle energy use for butanol production with different pretreatment methods.

(under the same baseline values of other parameters), the net GHG emissions of butanol production chain is reduced by 8.4, 6.2, 7.1, 6.4, and 5.8% with steam explosion, sulfuric acid, AFEX, ionic liquid, and biological pretreatment processes, respectively. This reduction is not only due to increase in product yield but also due to the reduction in the required recovery and separation energy for ABE. An effective pretreatment process would reduce enzyme consumption during the simultaneous saccharification and fermentation

pretreatment process evaluated. Therefore, a market for acetone, ethanol and excess electricity is essential. These results also demonstrate that an increase in butanol yield further reduces the environmental impacts of the system. Current average butanol yield is about 36% less than the theoretical yield,26,34 thus, there is the opportunity to reduce environmental impacts of this system through increased butanol yield. If butanol concentration is increased by 5 wt % in the fermentation broth by increasing the product yield 14534


Article


Figure 6. Uncertainty associated with GHG emissions form butanol production with different pretreatment methods.

use and GHG emissions (SI Figures S1, S2, S9, and S10). Additionally, apart from ionic liquid pretreatment, uncertainty in the simultaneous saccharification, fermentation, and recovery, stillage utilization and ABE separation models were the major sources of uncertainty from the biochemical process (SI Figures S21−S26). In contrast, energy and emissions associated with ionic liquid production was overwhelmingly the key contributor to uncertainty from biochemical process for ionic liquid pretreatment (SI Figures S17 and S18). Dramatic process optimizations are required specifically for ionic liquid pretreatment methods to achieve GHG emissions below the emissions from gasoline. For instance, 99% IL recovery and butanol yield of 90% of the theoretical maximum (0.41g/g-sugar) hit the targeted GHG emissions of 55.8 gCO2eq/MJ (60% of the conventional gasoline) set by Renewable Fuel Standards program of the United States. Future research could focus on these areas, specifically achieving ∼99% ionic liquid-recovery or alternative protic ionic liquid30 with low ionic liquid loading rate or one-pot high gravity system29 with cholinium lysinate. Among the pretreatment processes evaluated, there was higher uncertainty associated with the biological and ionic liquid pretreatment methods. The lower uncertainty associated with sulfuric acid pretreatment methods (the most researched and well-characterized of the pretreatment methods) was due to less variability present in input parameters. The positively skewed probability distributions that are associated with the GHG emissions and net energy of the sulfuric acid, AFEX, and ionic liquid pretreatment methods indicate the higher probability of the extreme values of energy use and GHG emissions when compared to baseline values. On the other hand, negatively skewed probability distributions of energy use and GHG emissions associated with steam explosion and biological pretreatment methods indicate that these systems could be higher performing than the median as the uncertainties in the process parameters decrease in time with process development and optimization. Process Optimization and Policy Implications. Wellinformed policy and investment decisions should not be based on single point baseline analyses only. The sensitivity of the

process, therefore, identification and implementation of optimal pretreatment conditions are required to reduce environmental impact from the use of the enzyme. A process inhibitor-tolerant Clostridium strain(s) could improve the sustainability of steam explosion and sulfuric acid pretreatment processes by reducing the ammonia requirement for detoxification. Sustainable agricultural practices with minimal nitrogen application, an optimal biorefinery location, full utilization of truck carrying capacity, use of electric balers and trucks reduce the environmental impact from feedstock supply logistics. A small improvement in each lifecycle stage of butanol production system could greatly impact on the overall sustainability of the system, which is required to be addressed through future research. The sensitivity analysis results for lifecycle energy use were similar to those for GHG emissions. Uncertainties Associated with Life-Cycle Energy Use and GHG Emissions. Figures 5 and 6 depict the probability distributions of the overall lifecycle energy use and GHG emissions from the butanol production system with the different pretreatment methods considered in this study. The input probability distributions for each of these inputs and lifecycle stage is illustrated in SI Tables S1−S8. The resulting probability distributions of lifecycle energy use for the butanol production system have values at 95% certainty of 0.82, 0.75, 0.87, 3.7, and 1.1 MJ/MJ-butanol with the steam explosion, sulfuric acid, AFEX, ionic liquid, and biological pretreatment methods, respectively. Similar distributions were found for GHG emissions with the upper limit values of 33.3, 37.6, 44.1, 1888.3, and 77.8 gCO2e/MJ-butanol with the steam explosion, sulfuric acid, AFEX, ionic liquid and biological pretreatment methods, respectively. However, the mode value of these systems closely resembles the results from the baseline model by using baseline values (Table 2 and SI Table S10). The resulting uncertainty associated each stage of the entire production chain are presented in the SI Figures S1−S40. The extreme values of GHG emissions are less than the mean gasoline emissions of 93 g CO2e/MJ for steam explosion, sulfuric acid, AFEX, and biological pretreatment methods. Uncertainty in the nutrient replacement and transportation models were the major sources of uncertainty for both energy 14535


Article


(2) Biddy, M. J.; Scarlata, C.; Kinchin, C. Chemicals from Biomass: A Market Assessment of Bioproducts with Near-Term Potential; National Renewable Energy Laboratory (NREL): Golden, CO, 2016; https:// www.nrel.gov/docs/fy16osti/65509.pdf (accessed March 25, 2017). (3) Tao, L.; Tan, E. C. D.; Mccormick, R.; Zhang, M.; Aden, A.; He, X.; Zigler, B. T. Techno-Economic Analysis and Life-Cycle Assessment of Cellulosic Isobutanol and Comparison with Cellulosic Ethanol and n-Butanol. Biofuels, Bioprod. Biorefin. 2014, 8 (1), 30−48. (4) Green, E. M. Fermentative production of butanolthe industrial perspective. Curr. Opin. Biotechnol. 2011, 22 (3), 337−343. (5) Rahimi, A.; Ulbrich, A.; Coon, J. J.; Stahl, S. S. Formic-acidinduced depolymerization of oxidized lignin to aromatics. Nature 2014, 515 (7526), 249−252. (6) Monlau, F.; Barakat, A.; Trably, E.; Dumas, C.; Steyer, J.-P.; Carrère, H. Lignocellulosic materials into biohydrogen and biomethane: impact of structural features and pretreatment. Crit. Rev. Environ. Sci. Technol. 2013, 43 (3), 260−322. (7) Baral, N. R. Techno-Economic Analysis of Butanol Production through Acetone-Butanol-Ethanol Fermentation. PhD dissertation; The Ohio State University, 2016; https://etd.ohiolink.edu/pg_ 10?0::NO:10:P10_ACCESSION_NUM:osu1480501106426567 (accessed June 11, 2017). (8) Baral, N. R.; Shah, A. Comparative techno-economic analysis of steam explosion, dilute sulfuric acid, ammonia fiber explosion and biological pretreatments of corn stover. Bioresour. Technol. 2017, 232, 331−343. (9) Pereira, L. G.; Chagas, M. F.; Dias, M. O. S.; Cavalett, O.; Bonomi, A. Life cycle assessment of butanol production in sugarcane biorefineries in Brazil. J. Cleaner Prod. 2015, 96, 557−568. (10) Mullins, K. A.; Griffin, W. M.; Matthews, H. S. Policy implications of uncertainty in modeled life-cycle greenhouse gas emissions of biofuels. (Special Issue: Environmental Policy: Past, Present, and Future.). Environ. Sci. Technol. 2011, 45 (1), 132−138. (11) Brito, M.; Martins, F. Life cycle assessment of butanol production. Fuel 2017, 208, 476−482. (12) SuperPro Designer-User’s Guide; Intelligen. Inc.Scotch Plains, NJ, 2017. (13) 2016. http://www.gabi-software.com/america/index/ (accessed February 21, 2017). (14) National Renewable Energy Laboratory (NREL). U.S. Life Cycle Inventory Database; National Renewable Energy Laboratory (NREL): Golden, CO, 2011; http://www.nrel.gov/lci/ (accessed March 25, 2017). (15) Argon National Laboratory (ANL). GREET® Model; Argon National Laboratory: Lemont, IL, 2016; https://greet.es.anl.gov/ (accessed March 07, 2017). (16) Baral, N. R.; Quiroz-Arita, C.; Bradley, T. H. Uncertainties in Corn Stover Feedstock Supply Logistics Cost and Life-Cycle Greenhouse Gas Emissions for Butanol Production. Appl. Energy 2017, 208, 1343−1356. (17) Hess, J. R.; Wright, C. T.; Kenney, K. L.; Searcy, E. M. Commodity-Scale Production of an Infrastructure-Compatible Bulk Solid from Herbaceous Lignocellulosic Biomass; Idaho National Laboratory, 2009; https://bioenergy.inl.gov/Reports/ Uniform%20Format%20Bioenergy%20Feedstock.pdf (accessed January 11, 2018). (18) Hsu, D. D.; Inman, D.; Heath, G. A.; Wolfrum, E. J.; Mann, M. K.; Aden, A. Life cycle environmental impacts of selected U.S. ethanol production and use pathways in 2022. Environ. Sci. Technol. 2010, 44 (13), 5289−5297. (19) Morey, R. V.; Kaliyan, N.; Tiffany, D. G.; Schmidt, D. R. A Corn Stover Supply Logistics System. Appl. Eng. Agric. 2010, 26 (3), 455−461. (20) Kaliyan, N.; Morey, R. V.; Tiffany, D. G. Economic and Environmental Analysis for Corn Stover and Switchgrass Supply Logistics. BioEnergy Res. 2015, 8 (3), 1433−1448. (21) Baral, N. R.; Shah, A. Techno-economic analysis of cellulose dissolving ionic liquid pretreatment of lignocellulosic biomass for

outputs, and the degree of uncertainty in energy use, GHG emissions as evaluated in this study could help to make fully informed pretreatment policy and implementation decisions. Except for the sulfuric acid pretreatment, the other pretreatment methods evaluated in this study are yet to be commercially implemented for cellulosic biofuel production, and therefore have more risk and uncertainty associated with them. Of the pretreatment options considered, steam explosion and AFEX pretreatment methods are found to have lower GHG emissions (and potential to meet the Renewable Fuel Standard of the U.S.) and net energy consumption than other pretreatment methods, under the assumptions of the baseline analysis. These pretreatment methods have potential to be the near-term alternatives to the commercial sulfuric acid pretreatment, but the high degree of uncertainty associated with their environmental costs and benefits must be considered to evaluate these technologies. For many of the pretreatment methods considered here, feedstock cultivation and transportation, the rate of nitrogen application at the field, and the value of the coproducts (especially electricity) production are all major sources of sensitivity and uncertainty for this analysis. When these inputs can be optimized, many of these pretreatment methods can be used to realize GHG emissions and net energy reduction goals.

■

ASSOCIATED CONTENT

S Supporting Information *

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acs.est.8b05176. Data associated with feedstock supply logistics, and biochemical conversion process with different pretreatment methods. Additionally, the detailed uncertainty results for each stage of supply chain (PDF)

■

AUTHOR INFORMATION

Corresponding Author

*Phone: +1 (970) 491 3539; e-mail: thomas.bradley@ colostate.edu. ORCID

Nawa Raj Baral: 0000-0002-0942-9183 Notes

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS This work was mainly supported by funding from Department of Mechanical Engineering, Colorado State University. This work was conducted in part by the DOE Joint BioEnergy Institute (http:// www.jbei.org) supported by the U.S. Department of Energy, Office of Science, Office of Biological and Environmental Research, through contract DE-AC0205CH11231 between Lawrence Berkeley National Laboratory and the U.S. Department of Energy. The United States Government retains and the publisher, by accepting the article for publication, acknowledges that the United States Government retains a non-exclusive, paid-up, irrevocable, world-wide license to publish or reproduce the published form of this manuscript, or allow others to do so, for United States Government purposes.

■

REFERENCES

(1) Jones, D. T.; Woods, D. R. Acetone-butanol fermentation revisited. Microbiol. Rev. 1986, 50 (4), 484−524. 14536


Article

Environmental Science & Technology fermentable sugars production. Biofuels, Bioprod. Biorefin. 2016, 10 (1), 1−20. (22) Baral, N. R.; Shah, A. Microbial inhibitors: Formation and effects on acetone-butanol-ethanol fermentation of lignocellulosic biomass. Appl. Microbiol. Biotechnol. 2014, 98 (22).9151 (23) NREL. Lignocellulosic Biomass to Ethanol Process Design and Economics Utilizing Co-Current Dilute Acid Prehydrolysis and Enzymatic Hydrolysis for Corn Stover; National Renewable Energy Laboratory (NREL): Golden, CO, 2002; https://www.nrel.gov/docs/fy02osti/ 32438.pdf (accessed June 25, 2017). (24) NREL. Process Design and Economics for Biochemical Conversion of Lignocellulosic Biomass to Ethanol Process Design and Economics for Biochemical Conversion of Lignocellulosic Biomass to Ethanol; National Renewable Energy Laboratory (NREL): Golden, CO, 2011; https:// www.nrel.gov/docs/fy11osti/47764.pdf (accessed June 11, 2017). (25) Baral, N. R.; Shah, A. Techno-Economic Analysis of Cellulosic Butanol Production from Corn Stover through Acetone−Butanol− Ethanol Fermentation. Energy Fuels 2016, 30 (7), 5779−5790. (26) Mariano, A. P.; Qureshi, N.; Maciel Filho, R.; Ezeji, T. C. Assessment of in situ butanol recovery by vacuum during acetone butanol ethanol (ABE) fermentation. J. Chem. Technol. Biotechnol. 2012, 87 (3), 334−340. (27) Sun, J.; Shi, J.; Konda, N. M.; Campos, D.; Liu, D.; Nemser, S.; Shamshina, J.; Dutta, T.; Berton, P.; Gurau, G. Efficient dehydration and recovery of ionic liquid after lignocellulosic processing using pervaporation. Biotechnol. Biofuels 2017, 10 (1), 154. (28) Wan, C.; Li, Y. Microbial pretreatment of corn stover with Ceriporiopsis subvermispora for enzymatic hydrolysis and ethanol production. Bioresour. Technol. 2010, 101 (16), 6398−6403. (29) Neupane, B.; Konda, N. M.; Singh, S.; Simmons, B. A.; Scown, C. D. Life-Cycle Greenhouse Gas and Water Intensity of Cellulosic Biofuel Production Using Cholinium Lysinate Ionic Liquid Pretreatment. ACS Sustainable Chem. Eng. 2017, 5 (11), 10176−10185. (30) Brandt-Talbot, A.; Gschwend, F. J. V.; Fennell, P. S.; Lammens, T. M.; Tan, B.; Weale, J.; Hallett, J. P. An Economically Viable Ionic Liquid for the Fractionation of Lignocellulosic Biomass. Green Chem. 2017, 19 (13), 3078−3102. (31) Avery, G. M. A Life Cycle Assessment of Ionic Liquid Pretreatment for Lignocellulosic Biomass. MS thesis, The University of Toledo, 2016. Available at: https://etd.ohiolink.edu/!etd.send_file?accession= toledo1481273168926691&disposition=inline [Accessed June 15, 2017]. (32) Klein-Marcuschamer, D.; Simmons, B. A.; Blanch, H. W. Techno-economic analysis of a lignocellulosic ethanol biorefinery with ionic liquid pre-treatment. Biofuels, Bioprod. Biorefin. 2011, 5 (5), 562−569. (33) Perez-Pimienta, J. A.; Sathitsuksanoh, N.; Thompson, V. S.; Tran, K.; Ponce-Noyola, T.; Stavila, V.; Singh, S.; Simmons, B. A. Ternary ionic liquid−water pretreatment systems of an agave bagasse and municipal solid waste blend. Biotechnol. Biofuels 2017, 10 (1), 72. (34) Mariano, A. P.; Qureshi, N.; Filho, R. M.; Ezeji, T. C. Bioproduction of butanol in bioreactors: New insights from simultaneous in situ butanol recovery to eliminate product toxicity. Biotechnol. Bioeng. 2011, 108 (8), 1757−1765.

14537


Probabilistic Lifecycle Assessment of Butanol Production from Corn

Recommend Documents