Can Dispersed Biomass Processing Protect the ... - ACS Publications

Dec 21, 2012 - The analysis uses a spatial bioeconomic model that maximizes profit from ..... biorefinery (Top left) versus local biomass processing d...
0 downloads 0 Views 2MB Size
Download PDF
Article pubs.acs.org/est

Can Dispersed Biomass Processing Protect the Environment and Cover the Bottom Line for Biofuel? Aklesso Egbendewe-Mondzozo,*,† Scott M. Swinton,‡ Bryan D. Bals,§ and Bruce E. Dale§ †

Fondazione Eni Enrico Mattei (FEEM) & Euro-Mediterranean Center for Climate Change (CMCC), Corso Magenta, 63, 20123 Milan, Italy ‡ Department of Agricultural, Food, and Resource Economics & Great Lakes Bioenergy Research Center (GLBRC), Michigan State University, 202 Agriculture Hall, East Lansing, Michigan 48824, United States § Department of Chemical Engineering and Materials Science, Michigan State University, 3815 Technology Boulevard, Lansing, Michigan 48910, United States S Supporting Information *

ABSTRACT: This paper compares environmental and profitability outcomes for a centralized biorefinery for cellulosic ethanol that does all processing versus a biorefinery linked to a decentralized array of local depots that pretreat biomass into concentrated briquettes. The analysis uses a spatial bioeconomic model that maximizes profit from crop and energy products, subject to the requirement that the biorefinery must be operated at full capacity. The model draws upon biophysical crop input-output coefficients simulated with the Environmental Policy Integrated Climate (EPIC) model as well as market input and output prices, spatial transportation costs, ethanol yields from biomass, and biorefinery capital and operational costs. The model was applied to 82 cropping systems simulated across 37 subwatersheds in a 9-county region of southern Michigan in response to ethanol prices simulated to rise from $1.78 to $3.36 per gallon. Results show that the decentralized local biomass processing depots lead to lower profitability but better environmental performance, due to more reliance on perennial grasses than the centralized biorefinery. Simulated technological improvement that reduces the processing cost and increases the ethanol yield of switchgrass by 17% could cause a shift to more processing of switchgrass, with increased profitability and environmental benefits. year).9 The bulky nature of the cellulosic biomass feedstock and its uneven distribution across landscapes can result in high costs of storage and delivery of biomass feedstock to the biorefinery. A set of LBPDs (35−180 Gg/year) that pretreats and concentrates biomass before shipping it to a biorefinery plant for final processing into ethanol can potentially reduce logistical costs.7,10 Sustainable biofuel production entails both profitability for private producers and provision of environmental services for society at large. Profitability is a precondition for private farmers and biofuel processors to engage in the biofuel business, but society as a whole has a stake in the nonmarketed environmental outcomes of bioenergy production, as evidenced by laws like the U.S. Energy Independence and Security Act of 2007 that ties bioenergy mandates to life cycle emissions of greenhouse gases.11 Some recent studies have shown that biomass supply from annual crop residues (e.g., corn stover and wheat straw) as feedstock for biofuel production could impair water quality and increase soil erosion.5,12−14 In particular, biomass production from annual crops is associated with increased nutrient runoff (e.g., phosphorus (P) and nitrate

1. INTRODUCTION Cropping systems with perenniality, minimal tillage, low inputs, and high diversity offer the best prospects for ecologically sustainable production of energy biomass.1,2 Diverse perennial cropping systems not only reduce greenhouse gases emissions but also may provide habitat to support populations of beneficial insects that offer ecological benefits such as pollination and pest suppression.3,4 Centralized biorefineries could be a threat to diversified cropping systems, because the most profitable way to meet their feedstock demand will be to rely on the cheapest and most abundant biomass crop sources. Recent modeling has shown these to be crop residues from annual corn and wheat, supplemented at high biomass prices by monocropped perennial grasses. 5 Environmental policy offers one path to ensure sustainable cropping systems by balancing bioenergy market price drivers with environmental incentives, 6 but biomass processing technology may offer another avenue to sustainable feedstock production. Local biomass processing depots (LBPDs) have been proposed to address the logistical problems of a centralized biorefinery. 7 Yet LBPDs may also offer means to disperse bioenergy crop production and potentially lead to environmentally beneficial plant biodiversity. The original motivation for LBPDs was to moderate the cost of delivering and storing biomass. Centralized biorefinery designs range in capacity from medium (730−1400 Gg/year)8 to large (4700−7800 Gg/ © 2012 American Chemical Society

Received: Revised: Accepted: Published: 1695

March 27, December December December

2012 19, 2012 21, 2012 21, 2012

dx.doi.org/10.1021/es303829w | Environ. Sci. Technol. 2013, 47, 1695−1703

Environmental Science & Technology

Article

(NO3) losses) into streams.12By contrast, biomass supply from perennial energy crops (e.g., switchgrass, miscanthus, native prairies, and mixed grasses) tends to mitigate environmental impacts through reduction of greenhouse gas emissions and improvement of water quality.12,14 The presence of a decentralized configuration of LBPDs could influence farmer selection of annual or perennial bioenergy crops through its influence on transportation costs, which affect the biomass price received at the farm gate. In particular, if LBPDs increase the farm price received for biomass in some locations, then they could profitably induce improved environmental sustainability by making perennial energy biomass crops more attractive than they are for a centralized biorefinery. Whether and when a decentralized configuration of biomass processing could lead farmers to choose perennial biomass crops over annuals is an empirical question to be explored in this paper. The principal objective of this paper is to compare the effects on crop choice by a representative, profit maximizing firm that farms and also manages an ethanol biorefinery plant with two possible alternative spatial configurations: a) direct raw biomass supply to a centralized biorefinery or b) raw biomass first shipped to a system of local biomass preprocessing depots (LBPDs) that supply a densified biomass product to a central biorefinery. The profitability focus is primary, because production decisions are made by private business people. However, while profit maximization from crop choice is of private interest, environmental outcomes matter to society at large. Hence, for each biorefinery spatial configuration, the study will 1) evaluate biorefinery profitability based on farmlevel biomass production, as well as transport, pretreatment, and final processing costs, 2) simulate long-term environmental impacts in terms of soil erosion and nutrient runoff as well as greenhouse gas emissions, and 3) evaluate the impact of potential changes in biomass conversion technology on biorefinery profitability and environmental quality. To reach the objectives of this study, the following research questions will be addressed: a) What are the key parameters driving profitability of biorefinery spatial configuration? b) What are the land use changes and the environmental impacts associated with each of the biorefinery spatial configurations (nutrient runoff, greenhouse gas emissions, land use change, and soil erosion) c) How are biorefinery profitability and environmental impacts altered by technological change? The remainder of the paper is organized as follows. First, a spatially explicit bioeconomic model developed and applied to study biomass production and supply in southwest Michigan5 is extended to include biorefinery processing into cellulosic ethanol as well as the possibility of biomass densification in LBPDs prior to biorefinery conversion into ethanol. Second, the empirical data and the assumptions regarding the method of densification as well as the final processing of biomass are presented. Third, the results of the analysis related to the biorefinery spatial configurations and the corresponding environmental impacts and profitability are presented and discussed.

requirement that the ethanol biorefinery must operate at full capacity in order to match investment in fixed capital with availability of feedstock. The model selects cropping systems among a set of 82 alternatives that will maximize firm profit from the sale of grain, forage products, cellulosic ethanol, and coproduced electricity. The cropping systems simulated are defined in terms of four management practices: crop rotation, level of fertilization, tillage, and crop residue removal for energy biomass5 (also see the Supporting Information for details). The EPIC model15 is used to simulate crop yields and environmental indicators for each cropping system based on weather, topography, and soil data. The geographic region modeled is situated in southwest Michigan (counties of Allegan, Barry, Eaton, Van Buren, Kalamazoo, Calhoun, Cass, St. Joseph, and Branch) and divided into 37 watersheds, as defined by their 10digit hydrologic unit codes (HUC). The watersheds, in turn, are subdivided into two levels of soil quality to yield a total of 70 land units (note that four of the watersheds lacked the lower quality soil). “Good” quality soils are those in land capability classes (LCC) 1−4, as defined by USDA Natural Resources Conservation Service. “Poor” quality soils are those in LCC 5− 7, which have more restrictions to productivity. The two spatial biorefinery configurations represent two experimental treatments for comparison: a) a single centrally located biorefinery (in Kalamazoo city) that collects biomass and pretreats it before processing into ethanol and byproducts (Figure 1a) and b) multiple LBPDs that pretreat biomass to densify it before shipping to the central biorefinery in Kalamazoo for conversion into ethanol and byproducts (Figure 1b). The parametrized and calibrated model chooses which of the 82 cropping systems to practice on each of the 70 land units in order to maximize net returns from sales of crop products, ethanol, and cogenerated electricity, subject to the requirement that the biorefinery must operate at full capacity and subject to other resource constraints. The final model is a constrained quadratic optimization program that maximizes the joint profit of farm and ethanol production enterprises as follows: 70

82

3

Max xij ∑ ∑ [−cjxij −

∑ rol ljxij

i=1 j=1

l=1

15

+

∑ ps (1 − ⌀s)(ρs xij − δsxij2)] s=1 9

−

∑ TCh−(PTVC + PTFC + RCVC + RCFC) h=1

+ (RET + REL)

(1)

Subject to 82

∑ xij ≤ bi ,

∀ i = 1 to 70 (2)

j

70

82

∑ ∑ enjxij − Eno ≤ M ,

∀ n = 1 to 5 (3)

i=1 j=1

2. MATERIALS AND METHODS The empirical model is built to maximize profit for a representative multiproduct firm that manages crop land and refines cellulosic ethanol. Profits are maximized subject to the

70

82

∑ ∑ aijhxij(β + γyi + θzi) = TCh , i

1696

j

∀ h = 1 to 9 (4)

dx.doi.org/10.1021/es303829w | Environ. Sci. Technol. 2013, 47, 1695−1703

Environmental Science & Technology

Article

Table 1. Model Parameters and Variables Definitions parameters, sets, and variables Sets h i j k l n s Parameters aijs bi cj enj olj ps rl ⌀s ρs δs Eon q, d M φh ωh πh, μh Variables TCh Ψ xij

Figure 1. a. Southwest Michigan subregion with the centralized biorefinery spatial configuration. b. Southwest Michigan subregion LBPD spatial configuration. 9

70

PTFC PTVC RCFC RCVC RET REL

82

∑ ∑ ∑ aijh(1 − ⌀h)xij = Ψ h

i

j

9

70

82

(5)

∑ ∑ ∑ aijh(1 − ⌀h)xijφh = PTVC h=1

i

j

9

70

82

i

j

9

70

82

(7)

∑ ∑ ∑ aijh(1 − ⌀h)xijπhq = RET h=1

i

j

9

70

82

(8)

∑ ∑ ∑ aijh(1 − ⌀h)xijμh d = REL h=1

i

j

set of 9 biomass outputs studied in the model set of 70 subwatersheds with good or poor land quality set of 82 cropping systems simulated on each subwatershed set of 8 local biomass preprocessing depots (LBPDs) set of three fertilizer nutrients used in the cropping systems set of five environmental outputs of cropping systems set of 15 traditional (6) and biomass (9) crop products combined yield of crop s from land parcel i and cropping system j maximum quantity of cropland available in subwatershed i average cost of production for cropping system j value of environmental output n in cropping system j quantity per ha of fertilizer nutrient l used in cropping system j market price of crop s unit cost of fertilizer nutrient l storage loss coefficient for biomass products average base yield in the calibration of the output product s average linear yield decline with increasing land allocated to output product s quantity limit of environmental outputs allowed assumed ethanol prices (q) and electricity price (d) environmental constraints control parameter unit pretreatment variable cost for biomass type h unit ethanol conversion variable costs for biomass type h ethanol yield (πh) and electricity yield (μh) from biomass type h cost of transporting biomass product h to the demand point total quantity of all biomass produced in the region quantity of land in subwatershed i allocated to cropping system j pretreatment fixed costs pretreatment variable costs refinery conversion fixed costs refinery conversion variable costs revenues from ethanol sales revenues from net electricity sales

The objective function (1) contains six expressions. The first 82 expression (−Σ70 i Σj cjxij) represents the total variable production costs across all cropping systems and subwatersheds. The 82 3 second expression (−Σ70 i Σj Σl=1rloljxij) is the total cost of synthetic fertilizers across systems and subwatersheds. The 82 15 2 third expression (Σ70 i Σj Σs=1ps(1 − ⌀s)(ρsxij − δsxij)) is the total crop sales revenue from all cropping systems and subwatersheds adjusted for storage losses. The term (ρsxij − δsx2ij) defines the quadratic output levels obtained by multiplication of the calibrated marginal yield expression (ρs − δsxij) by the quantity of land xij allocated to the production of output s (details on model calibration in Egbendewe-Mondzozo et al.5). The fourth expression (Σ9h=1TCh) represents the total transport cost of each biomass type to the biorefinery plant. These four expressions calculate the gross margin of the representative farmer from

(6)

∑ ∑ ∑ aijh(1 − ⌀h)xijωh = RCVC h=1

definition

(9)

The mathematical sets, variables, and parameters used in the model are defined in Table 1. 1697

dx.doi.org/10.1021/es303829w | Environ. Sci. Technol. 2013, 47, 1695−1703

Environmental Science & Technology

Article

cereal and biomass sales. The fifth expression corresponds to the pretreatment variable costs (PTVC), biomass pretreatment fixed costs (PTFC), biorefinery conversion variable costs (RCVC), and biorefinery conversion fixed costs (RCFC). The last expression represents the revenues from ethanol sales (RET) and net electricity sales (REL). The final two expressions calculate the biorefinery’s profits and can be adapted to calculate profits for the LBPD configuration as well. Equation 2 expresses the set of land resource availability constraints. Along with the full capacity eq 5, these are the only systematically binding constraints in the model. The remaining equations are not binding constraints but are either accounting rows or identities. Equation 3 is a set of accounting rows of environmental output levels. The quantity M allows flexibility on whether the environmental constraint set is binding or not. For instance if M = 0 the environmental constraint set is binding and total environmental outputs must not exceed Eon. Setting M to a large number, M ≫ 0, is analogous to the case where environmental impacts are not binding. The setting in this paper corresponds to the no binding environmental constraints case where eq 3 is used only to account for cumulative environmental outcomes across all land units. Equation 4 calculates transport costs to the biorefinery. The term (β + γyi + θzi) is the transport cost of a metric ton (Mg) of biomass to the refinery site; with β being the cost of loading and unloading, γ is the cost per Mg per kilometer of hauling distance, and θ is the cost per hour of hauling time. The variables yi and zi are respectively the hauling distance and time from a parcel i to the refinery plant site. Equation 5 calculates the total biomass produced and imposes capacity constraints on the model. The biorefinery is designed such that full capacity is always met in order to minimize fixed capital investment costs. Equations 6 to 9 calculate the pretreatment variable costs, ethanol variable conversion costs, and the total ethanol and net electricity sales, as appropriate for each modeling scenario. In the presence of LBPDs, the transport costs in eq 4 add local transport to LBPD to become Σ8k=1Σ9h=1TCkh. This formula calculates the biomass transport from the farm points to the 8 LBPD locations plus the transport costs from each LBPD to the biorefinery. The pretreatment variable costs expression in eq 6 has to add costs for each of the 8 LBPDs. Finally, the pretreatment fixed costs in the objective function will be a summation of the fixed costs from each LBPD. After calibration of the bioeconomic optimization model to 2007−09 average crop market prices and land use, the model was run without ethanol production to generate predicted baseline environmental outcome levels corresponding to farming conditions in 2007−09.5 Then, holding all other parameters constant, the ethanol price was progressively raised from $1.78/gal to $3.36/gal (or $2.66/gal to $5.05/gal gasoline-gallon-equivalent, using the fact that 1.5 gallons of ethanol are needed to produce the energy equivalent of one gallon of gasoline). The range of ethanol prices was chosen to start at $1.78/gal, the minimum 2010 price16 and end at $3.36, a forward looking price that corresponds to gasoline just over $5.00/gal. The biomass processing capacity of the biorefinery is set to 700Gg/year for a medium size biorefinery of 2000Mg/ day. Given that energy biomass is not currently commercially produced in Michigan, the average biomass price at the biorefinery gate is calculated as the shadow price of the biorefinery capacity constraint expressed in eq 5. This value expresses the implied total cost (direct cost plus opportunity

cost) of producing biomass to meet the required biorefinery operating capacity. To test the effects of biorefinery configuration and processing technology, four scenarios are run on top of the baseline without biofuels. The scenarios are a) centralized biorefinery with most likely processing technology; b) decentralized LBPDs linked to a central refinery with most likely processing technology; c) centralized biorefinery with two levels of improved ethanol yields from switchgrass; and d) decentralized LBPDs linked to a central biorefinery with two levels of improved ethanol yields from switchgrass.

3. MODEL DATA 3.1. Biophysical Crop and Environmental Yields Data. The EPIC biophysical model15,17−21 is used to simulate average crop yield and environmental outcome parameters (soil erosion, phosphorus loss, nitrate loss, nitrous oxide emissions, and soil carbon loss) in southwest Michigan for a 24 year period (1986−2009). EPIC simulates 82 cropping systems defined by their crop rotations, tillage, fertilizer level, and biomass residue removal (details in Egbendewe-Mondzozo et al.5). The nitrous oxide emissions and soil carbon loss are used to calculate the total greenhouse gas emission in CO2 − carbon equivalents. The model includes grain and forage yields from six field crops plus biomass yields from seven cellulosic bioenergy crops and biomass residue yields from two field crops (a total of 9 biomass types). Field crops include corn grain, soybean, wheat, alfalfa, canola, and corn silage. Cellulosic energy crops simulated are switchgrass, miscanthus, native prairie cool season mix, native prairie warm season mix, grass mixes of five types and six types, and hybrid poplar. Crop residues include corn stover and wheat straw. Average crop yield assumptions are based on Egbendewe-Mondzozo et al.5 3.2. Crop Production Costs, Market Price, and Biomass Transport Cost Data. Crop production activity costs are obtained from Stein for 2009.22,23 Market prices for 2007−09 come from the National Agricultural Statistics Service (NASS).24 Geographic information system data are used to calculate transport costs based on travel time and distance from supply points to demand points with hauling distance and perhour hauling time cost drawn from Graham, English, and Noon.13 The model accounts for storage loss for all grassy biomass types (i.e., all types except poplar trees), using an 8.8% loss coefficient (⌀s in the model above). This level corresponds to dry matter losses for wrapped round bales stored at field edge for 6 months as reported in recent literature.25 Key parameters on input costs, output price, and transport costs and other fuller details on the crop production part of the model appear in Egbendewe-Mondzozo et al.5 The model does not include livestock production or the possibility of feeding byproducts of ethanol distillation to livestock. 3.3. Biomass Pretreatment, Conversion Yields, and Costs. After harvest, biomass is assumed to be processed yearround at LBPDs, which process 100 to 250 Mg/day of material.7 Biomass bales are milled and then pretreated using ammonia fiber expansion (AFEX) pretreatment (AFEX is a trademark of MBI International).10 After pretreatment, the biomass is partially dried in a drum dryer if a high severity pretreatment was used, and then all biomass is briquetted for shipment to a centralized biorefinery. These LBPDs purchase electricity and natural gas to produce steam and power for the AFEX process. At the biorefinery, the briquettes are saccharified, and the C5 and C6 sugars are fermented into 1698

dx.doi.org/10.1021/es303829w | Environ. Sci. Technol. 2013, 47, 1695−1703

Environmental Science & Technology

Article

Table 2. List of Properties for the Types of Biomass Considered in the Study biomass

glucan content (g/kg)

xylan content (g/kg)

pretreatment severity

ethanol yields (L/Mg)

electricity produceda (kWh/Mg)

variable costsb ($/Mg)

corn stover switchgrass miscanthus native prairies wheat straw grass mixes poplar

350 335 440 290 380 320 440

220 240 190 170 230 200 150

low high high low low low high

275.80 277.10 267.20 208.30 295.50 274.10 189.50

368.98 367.67 455.28 428.74 360.42 323.63 608.18

55.33 55.32 55.75 55.63 55.28 55.11 56.39

a

Electricity produced after combusting the nonfermented biomass and supplying all heat and power to all operations except pretreatment in the biorefinery. bVariable costs at the biorefinery, excluding pretreatment costs.

electricity. The biorefinery model is based on one developed by the NREL, which includes pretreatment, conversion, and purification of ethanol as well as steam and electricity coproduction from the unhydrolyzable residue.28 For this study, the costs associated with pretreatment were eliminated, and the fixed costs for all other processes were sized appropriately for the process conditions used in this study. The capital cost, which includes enzymatic hydrolysis and fermentation, ethanol separation, wastewater treatment, and electricity coproduction, was estimated annually as the total capital investment annuitized over the expected facility lifetime (20 years for the pretreatment and LBPD, 30 years for the refinery) assuming an annual real interest rate of 5%. Processing assumptions were 20% solid loading during saccharification,30 a total of 72 h residence time between saccharification and fermentation, 100% C6 sugar consumption, 80% xylose consumption,31 60% arabinose consumption, and an enzyme loading of 10 mg/g biomass32 at $3.60/kg enzyme.29 The variable operating costs of labor, maintenance, fly ash removal, and nutrients for fermentation were drawn from the NREL model. Total fixed and variable costs for the LBPDs, centralized biorefinery in the LBPD approach, and centralized biorefinery in the no-LBPD scenario are shown in Table 3.

ethanol. The remaining biomass is combusted to produce steam and electricity to provide heat and power to the biorefinery, with excess electricity exported to the power grid. The second spatial configuration scenario eliminates the LBPDs and moves the grinding and AFEX treatment to the centralized refinery (no briquetting or drying is performed in this operation), with all steam and power required for pretreatment provided via combustion of the noncarbohydrate portion of biomass. Both configurations are designed to process 700 Gg of biomass per year. Feedstocks react differently to pretreatment based on their cell wall structure and composition. To capture this variation, each feedstock in the model was assigned to a low severity (0.8:0.5:1.0 weight ratio of ammonia, water, and biomass) or high severity (1.5:0.8:1.0 ratio of ammonia, water, and biomass) pretreatment. The low severity pretreatment is for highly digestible material such as corn stover,26 wheat straw, and mixed grasses and requires much less energy input than the high severity pretreatment. The high severity is for highly recalcitrant biomass such as switchgrass,27 miscanthus,28 or native prairie, in which a high concentration of sugars is not attainable under low severity conditions. The amount of ethanol produced from each feedstock was estimated based on composition and response to pretreatment. Potential electricity production also varies with biomass type. The pretreatment severity, biomass composition, sugar yield, and gross electricity production of each type of biomass is shown in Table 2. In the model, the nonconverted fraction of biomass is transferred to a solid biomass boiler to produce high pressure steam, which is then transferred to a multistage turbine and generator to produce electricity.28 Net electricity revenue was estimated by calculating the gross electricity production and subtracting off steam use and electricity required in the biorefinery as determined via the National Renewable Energy Laboratory (NREL) techno-economic model.29 In the two spatial configurations, all steam and power requirements at the central biorefinery could be met by the combustion of the nonconverted fraction. However, the LBPDs purchased heat and power from the grid, as these biomass boilers are not installed at the LBPDs. Purchase price of electricity from the grid was assumed to be $0.068/kWh,10 while the selling price to the grid was set at $0.0572/kWh. To estimate the costs associated with ethanol production, process models of both the LBPDs10 and the biorefinery were developed. The LBPD biomass densification model sizes the major pieces of equipment based on expected incoming biomass to estimate the fixed costs associated with production. Variable costs include labor, maintenance, ammonia use (fixed at 22 g/kg biomass treated), and purchased natural gas and

Table 3. Variable and Capital Costs Associated with the Facilities Modeled types

a

LBPD LBPD pretreatment pretreatment biorefineryb

size (Mg/ day)

low severity variable costs ($/Mg)

high severity variable costs ($/Mg)

capital costs (million $/year)

100 250 550 2000 2000

42.73 30.09 30.40 31.70

49.60 36.96 37.94 39.24

0.79 1.77 1.50 5.25 10.09

a

The three types of facilities are 1) LBPDs, 2) pretreatment centers at a centralized biorefinery, and 3) the centralized biorefinery excluding pretreatment operations. b The biorefinery variable costs are determined by the biomass type and listed in Table 2.

4. MODEL SIMULATION RESULTS AND DISCUSSION The initial results show the biomass types supplied by farmers, the total profits earned from biomass production and conversion activities, the land use change, and the changes in environmental output levels compared to the baseline without biomass production. Ethanol yield parameters are subsequently modified to analyze the sensitivity of these initial results to technological change in biomass plant growth and conversion techniques. 1699

dx.doi.org/10.1021/es303829w | Environ. Sci. Technol. 2013, 47, 1695−1703

Environmental Science & Technology

Article

Figure 2. a. Biomass sources under centralized biorefinery (Top left) versus local biomass processing depots (bottom left). b. Environmental outputs change under centralized biorefinery (top right) versus local biomass processing depots (bottom right).

4.1. Initial Results. For the centralized biorefinery configuration, only low cost biomass types from annual crop residues (corn stover and wheat straw) are supplied as feedstock (Figure 2a). By contrast, in the LBPD configuration, diverse sources of feedstock are supplied, including not only annual crop residues (corn stover and wheat straw) but also perennial energy biomass such as grass mixes. The supply of perennial energy crops as feedstock under the LBPD configuration is explained by the fact that the LBPD production costs with LBPDs operate with minimum capacity levels, and additional feedstock from expensive biomass sources are needed to reach the minimum capacity in certain local depots. As a consequence of using more expensive sources of biomass plus the fixed costs needed for building these LBPDs, production costs with LBPDs are much higher than for the central biorefinery alone. Hence, the joint enterprise of biomass and ethanol production will be profitable under the LBPD configuration only if the ethanol price reaches or exceeds $2.30/gal (corresponding to a minimum biomass price level of $74/Mg). By contrast, with centralized ethanol production, the joint enterprise becomes profitable at ethanol price of $2.00/gal (for a minimum biomass price level of $44/Mg). As for land use, in the LBPD configuration about 2% of cropland (roughly 9,000 acres) is diverted to perennial energy biomass production. As a result of using more perennial crops, the dispersed LBPD configuration yields less environmental damage (nutrient runoff, greenhouse gas emissions, and soil erosion) than the centralized biorefinery configuration. Clearly, there is a trade-off between total farm and biorefinery enterprise profitability and environmental quality (Figure 2b). High biorefinery profitability implies use of the least cost biomass feedstock from crop residues of annual crops that cause more environmental damage than perennial crops. By contrast, moving to the LBPD configuration encourages the use of

more perennial crops that cause less environmental damage but lack revenue from a grain product and so require a higher biomass price to cover costs. The LBPD results suggest that variants of the LBPD configuration (like shifting capacity to areas with more corn and wheat residue availability) could increase profitability, but those same changes would result in more environmental damage. 4.2. Sensitivity to an Increase of Switchgrass Ethanol Yield. The initial model results in little production of perennial biomass crops, in part because they require a more severe pretreatment to achieve similar ethanol yields than the crop residues. In the face of active research globally to increase the yield of biofuel from perennial crops, it is relevant to consider technological change scenarios for lower cost processing via improvements in the ethanol yield from resistant forms of biomass. We specifically consider two scenarios for enhanced ethanol yield from switchgrass. A medium increase scenario raises the ethanol yield parameter for switchgrass from 277.1 to 298.6 L/Mg (an 8% increase) while also shifting its pretreatment requirements to low severity from high severity. The results show that the only major changes that would result from this medium improvement in ethanol yield consist of switchgrass replacing the grass mixes under the LBPD configuration. The small amount of switchgrass produced under the centralized biorefinery configuration is insufficient to induce much change in profitability and environmental outputs compared to the initial results. A more optimistic technological change scenario would cause a high increase in the switchgrass ethanol yield from 277.1 to 323.1/Mg liters (a 17% increase), while continuing to require only low severity pretreatment. If the ethanol yield from switchgrass increases by 17%, the centralized biorefinery would demand more switchgrass as feedstock and progressively 1700

dx.doi.org/10.1021/es303829w | Environ. Sci. Technol. 2013, 47, 1695−1703

Environmental Science & Technology

Article

Figure 3. a. Biomass sources under centralized biorefinery (top left) versus local biomass processing depots (bottom left) with high (17%) increase in switchgrass ethanol yields. b. Environmental outputs change under centralized biorefinery (top right) versus local biomass processing depots (bottom right) with high (17%) increase in switchgrass ethanol yields.

than centralized biorefining for cellulosic ethanol. However, with current technology, costs of ethanol production remain high. Especially for the spatially dispersed local biomass processing depots, costs of ethanol production exceed revenues at lower ethanol prices. Improvements in the ethanol conversion efficiency of more resistant forms of ligno-cellulosic biomass could trigger substantial shifts toward perennial feedstocks such as switchgrass, with attendant benefits in water quality and greenhouse gas emissions. Interestingly, however, such technological change leads to greater environmental gains (and profits) for the centralized biorefinery than for the decentralized local biomass processing depots. Clearly, both spatial configuration and technological gains in processing efficiency can have important repercussions for the environmental performance of biofuel production systems. While farm level greenhouse gas emissions might increase with residue collection, it is important to note that ethanol production from residues has been found to be carbon negative when fossil fuel displacement is accounted for in life cycle analysis.33 Further research will be needed to explore additional means to enhance profitability and mitigate the environmental consequences of increased production of cellulosic ethanol. One is to integrate into the biorefinery model the sale of ethanol production byproducts for animal feed, which has been found to reduce the threshold for biorefinery profitability.34 Such a change could shift the minimum ethanol price threshold for profitable biorefining. The second is to refine the spatial scale, to capture how placement of specific crops within a watershed can affect water quality35,36 Future extensions of this research should explore the potential profitably to achieve better environmental outcomes from biofuel production from the same set of resources by manipulating the locations both of

replace the low cost biomass from crop residues with switchgrass as the ethanol price increases (Figure 3a). In the LBPD configuration, the minimum capacity requirement at each LBPD produces a slower decline in the use of annual crop residues as biomass. The centralized biorefinery configuration would still be profitable at $2.00/gal ethanol price (as in the initial scenario) but at a higher minimum biomass price of $60/ Mg (since switchgrass remains an expensive source of biomass relative to annual crop residues). Even with the high increase in ethanol yield from switchgrass, in order for the LBPD configuration to become profitable ethanol price would have to reach $2.20/gal with biomass minimum biomass price of $74/Mg, as in the initial scenario. The scenario with a high increase in switchgrass ethanol yield causes a decline in land use for wheat cropping systems in favor of more perennial cropping systems. As ethanol price increases, land for perennials increases from 0% to 8% of total crop land (0 to 52,000 ha). As a consequence of perennial cropping systems adoption, the levels of environmental damage are reduced relative to the initial scenario results (Figure 3b). As ethanol price increases, the environmental results improve gradually as more switchgrass is used as feedstock in the biorefinery. Contrary to the initial scenario results, a large increase in ethanol yield from switchgrass causes environmental outputs to improve more under the centralized biorefinery configuration than under the LBPD configuration. At high ethanol prices (greater than $3/Gal), the environmental outputs improve to a point similar to the baseline level of no biomass production.

4. DISCUSSION This research establishes the potential for spatially dispersed biomass processing to generate better environmental outcomes 1701

dx.doi.org/10.1021/es303829w | Environ. Sci. Technol. 2013, 47, 1695−1703

Environmental Science & Technology

Article

(11) US Congress, Energy Independence and Security Act, H.R. 6. http://www.energy.senate.gov/public/index.cfm/files/serve?File_id= 1d4aca7e-3cea-4db1-a000-75d3a5673410. (12) Robertson, P. G.; Hamilton, S. K.; Del Grosso, J. S.; Parton, W. J. The biogeochemistry of bioenergy landscapes: carbon, nitrogen, and water considerations. Ecol. Appl. 2011, 21 (4), 1055−1067. (13) Graham, R. L.; English, B. C.; Noon, C. E. A geographical information system-based modeling system for evaluating the cost of delivered energy crop and feedstock. Biomass Bioenergy 2000, 18 (4), 309−329. (14) Love, J. B.; Nejadhashemi, P. A. Water quality impact assessment of large-scale biofuel expansion in agricultural regions of Michigan. Biomass Bioenergy 2011, 35 (5), 2200−2216. (15) Williams, J. R.; Jones, C. A.; Kinitry, J. R.; Spanel, D. A. The EPIC crop growth-model. Trans. ASAE 1989, 32 (2), 497−511. (16) Agricultural Marketing Resource Center Ethanol futures prices and basis by state. http://www.agmrc.org/renewable_energy/ethanol/ ethanol__prices_trends_and_markets.cfm (accessed December 19). (17) Izaurralde, R. C.; Williams, J. R.; McGill, W. B.; Rosenberg, N. J.; Jakas, M. C. Q. Simulating soil C dynamics with EPIC: model description and testing against long-term data. Ecol. Modell. 2006, 192 (3−4), 362−384. (18) Izaurralde, R. C.; Williams, J. R.; Post, W. M.; Thomson, A. M.; McGill, W. B.; Owens, L. B.; Lal, R. Long-term modeling of soil C erosion and sequestration at the small watershed scale. Clim. Change 2007, 80 (1−2), 73−90. (19) Jones, C. A.; Dyke, P. T.; Williams, J. R.; Kiniry, J. R.; Benson, V. W.; Griggs, R. H. EPIC: an operational model for evaluation of agricultural sustainability. Agric. Syst. 1991, 37 (4), 341−350. (20) Williams, J. R. The EPIC model. In Computer models of watershed hydrology; Singh, V. P., Ed.; Water Resources Publications: Highlands Ranch, 1995; pp 909−1000. (21) Zhang, X.; Izaurralde, R. C.; Manowitz, D.; West, T. O.; Post, W. M.; Thomson, A. M.; Bandaru, V. P.; Nichols, J.; Williams, J. R. An integrative modeling framework to evaluate the productivity and sustainability of biofuel crop production systems. Global Change Biol. Bioenergy 2010, 2 (5), 258−77. (22) Stein, D. 2009 crop enterprise budget. Michigan State University Extension. https://www.msu.edu/∼steind/ (accessed May 19). (23) Stein, D. 2009−2010 machine custom and work rate estimates. Michigan State University Extension. https://www.msu.edu/∼steind/ (accessed May 10). (24) USDA, National Agricultural Statistics. Field Crops Prices; Department of Agriculture: Washington, DC, 2010. (25) Brechbill, S. C.; Tyner, E. W.; Ileleji, K. E. The economics of biomass collection and transportation and its supply to Indiana cellulosic and electric utility facilities. BioEnergy Res. 2011, 4 (2), 141− 152. (26) Teymouri, F.; Laureano-Perez, L.; Alizadeh, H.; Dale, B. Optimization of the ammonia fiber explosion (AFEX) treatment parameters for enzymatic hydrolysis of corn stover. Bioresour. Technol. 2005, 96 (18), 2014−2018. (27) Garlock, R.; Balan, V.; Dale, B. E. Optimization of AFEX pretreatment conditions and enzyme mixtures to maximize sugar release from upland and lowland switchgrass. Bioresour. Technol. 2011, 104, 757−768. (28) Murnen, H.; Balan, V.; Chundawat, S.; Bals, B.; Sousa, L.; Dale, B. Optimization of Ammonia Fiber Expansion (AFEX) pretreatment and enzymatic hydrolysis of Miscanthus x giganteus to fermentable sugars. Biotechnol. Prog. 2007, 23 (4), 846−850. (29) Humbird, D.; Davis, R.; Tao, L.; Kinchin, C.; Hsu, D.; Aden, A.; Schoen, P.; Lukas, J.; Olthof, B.; Worley, M.; Sexton, D.; Dudgeon, D. Process design and economics for biochemical conversion of lignocellulosic biomass to ethanol; NREL/TP-5100-47764; National Renewable Energy Laboratory: 2011. (30) Bals, B. D.; Teymouri, F.; Campbell, T.; Jin, M.; Dale, B. E. Low temperature and long residence time AFEX pretreatment of corn stover. BioEnergy Res. 2012, 5, 372−379.

perennial crops within watersheds and of individual local biomass processing depots that service a centralized biorefinery.

■

ASSOCIATED CONTENT

S Supporting Information *

Table that explains the various cropping systems simulated in the paper. This material is available free of charge via the Internet at http://pubs.acs.org.

■

AUTHOR INFORMATION

Corresponding Author

*Phone: +39 025-203-6983. Fax: +39 025-203-6946. E-mail: [email protected]. Notes

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS This work was funded by the DOE Great Lakes Bioenergy Research Center (DOE BER Office of Science DE-FC0207ER64494). This research was conducted when the first author was a Visiting Assistant Professor at Michigan State University (MSU) and Researcher at the Great Lakes Bioenergy Research Center (GLBRC). For data and comments, the authors wish to thank Sarah AcMoody, Kurt Thelen, Dennis Stein, Eric Wittenberg, Robin Graham, Burton English, Charles Noon, and Randy Jackson.

■

REFERENCES

(1) Tilman, D.; Hill, J.; Lehman, C. Carbon-negative biofuels from low-input high-diversity grassland biomass. Science 2006, 314 (5805), 1598−1600. (2) Robertson, G. P.; Dale, V. H.; Doering, O. C.; Hamburg, S. P.; Melillo, J. M.; Wander, M. M.; Parton, W. J.; Adler, P. R.; Barney, J. N.; Cruse, R. M.; Duke, C. S.; Fearnside, P. M.; Follett, R. F.; Gibbs, H. K.; Goldemberg, J.; Mladenoff, D. J.; Ojima, D.; Palmer, M. W.; Sharpley, A.; Wallace, L.; Weathers, K. C.; Wiens, J. A.; Wilhelm, W. W. Sustainable biofuels redux. Science 2008, 322 (5898), 49−50. (3) Gardiner, M. A.; Tuell, J. K.; Isaacs, R.; Gibbs, J.; Ascher, S. J.; Landis, D. Implications of three biofuel crops for beneficial arthropods in agricultural landscapes. BioEnergy Res. 2010, 3 (1), 3−19. (4) Fletcher, J. R., Jr.; Robertson, A. B.; Evans, J.; Doran, J. P.; Alavalapati, R. J.; Schemske, W. D. Biodiversity conservation in the era of biofuels: risks and opportunities. Front. Ecol. Environ. 2011, 9 (3), 161−168. (5) Egbendewe-Mondzozo, A.; Swinton, S. M.; Izzauralde, C.; Manowitz, D. M.; Zhang, X. Biomass supply from alternative cellulosic crops and crop residues: A spatially explicit bioeconomic modeling approach. Biomass Bioenergy 2011a, 35 (11), 4636−4647. (6) Egbendewe-Mondzozo, A.; Swinton, S. M.; Izzauralde, R. C.; Manowitz, D. M.; Zhang, X. Maintaining environmental quality while expanding energy biomass production: policy simulations from Michigan, USA. Michigan State University: East Lansing, MI. 2011b. (7) Eranki, P. L.; Bals, B. D.; Dale, B. E. Advanced regional biomass processing depots: A key to the logistical challenges of the cellulosic biofuels industry. Biofuels, Bioprod. Biorefin. 2011, 5 (6), 621−630. (8) Aden, A.; Ruth, M.; Ibsen, K.; Jechura, J.; Neeves, K.; Sheehan, J.; Wallace, B.; Montague, L.; Slayton, A.; Lukas, J. Lignocellulosic biomass to ethanol process design and economics utilizing co-current dilute acid prehydrolysis and enzymatic hydrolysis for corn stover; NREL/TP-51032438; National Renewable Energy Laboratory, CO: 2002. (9) Wright, M.; Brown, R. C. Establishing the optimal sizes of different kinds of biorefineries. Biofuels, Bioprod. Biorefin. 2007, 1 (3), 191−200. (10) Bals, B. D.; Dale, B. E. Developing a model for assessing biomass processing technologies within a Local Biomass Processing Depot. Bioresour. Technol. 2011, 106, 161−169. 1702

dx.doi.org/10.1021/es303829w | Environ. Sci. Technol. 2013, 47, 1695−1703

Environmental Science & Technology

Article

(31) Jin, M.; Lau, M.; Balan, V.; Dale, B. Two-step SSCF to convert AFEX-treated switchgrass to ethanol using commercial enzymes and Saccharomyces cerevisiae 424A(LNH-ST). Bioresour. Technol. 2010, 101 (21), 8171−8178. (32) Gao, D.; Chundawat, S.; Krishnan, C.; Balan, V.; Dale, B. E. Mixture optimization of six core glycosyl hydrolases for maximizing saccharification of ammonia fiber expansion (AFEX) pretreated corn stover. Bioresour. Technol. 2010, 101 (8), 2770−2781. (33) Adler, P. R.; Del Grosso, J. S.; Parton, W. J. Life-cycle assessment of net greenhouse-gas flux for bioenergy cropping systems. Ecol. Appl. 2007, 17, 675−691. (34) Sendich, E. D.; Dale, B. E. Environmental and economic analysis of the fully integrated biorefinery. Global Change Biol. Bioenergy 2009, 1 (5), 331−345. (35) Gassman, P. W.; Reyes, M. R.; Green, C. H.; Arnold, J. G. The soil and water assessment toolhistorical development, applications, and future research directions. Trans. ASABE 2007, 50 (4), 1211− 1240. (36) Parish, E. S.; Hilliard, M. R.; Baskaran, L. M.; Dale, V. H.; Griffiths, N. A.; Mulholland, P. J.; Sorokine, A.; Thomas, N. A.; Downing, M. E.; Middleton, S. R. Multimetric spatial optimization of switchgrass plantings across a watershed. Biofuels, Bioprod. Biorefin. 2011, 6 (1), 58−72.

1703

dx.doi.org/10.1021/es303829w | Environ. Sci. Technol. 2013, 47, 1695−1703