Applied Energy xxx (xxxx) xxx–xxx
Contents lists available at ScienceDirect
Applied Energy journal homepage: www.elsevier.com/locate/apenergy
Economic and energetic analysis of biofuel supply chains Rex T.L. Ng, Christos T. Maravelias
⁎
Department of Chemical and Biological Engineering, University of Wisconsin-Madison, 1415 Engineering Drive, Madison, WI 53706, USA DOE Great Lakes Bioenergy Research Center, University of Wisconsin-Madison, 1415 Engineering Drive, Madison, WI 53706, USA
H I G H L I G H T S methods to calculate the cost and energy input of biofuel supply chains. • General supply chain configurations and transportation modes are studied. • Multiple impact of biorefinery capacity, biomass availability, and densification are studied. • The configuration, with depots preferred only at larger distances, is optimal. • Hybrid Insights can be used to constrain available options for biofuel supply chain design. •
A R T I C L E I N F O
A B S T R A C T
Keywords: Biorefinery Cellulosic ethanol Transportation Distributed biomass processing system
We show how different supply chain configurations affect the economic performance and energy efficiency of biofuel supply chains. Despite the additional costs and energy required for installing and operating regional depots, transportation costs and energy savings are observed when depots are combined with the appropriate transportation modes at longer distances. We introduce a hybrid configuration which combines various configurations and leads to better economic performance and higher energy efficiency. We further study the impact of various factors on the performance of the biofuel supply chain: distance between harvesting site and depot, biomass productivity, biorefinery size, and densification efficiency. We show that biomass should be shipped directly to biorefineries when the distance is small, while depots are preferred at larger distances. The hybrid configuration offers lower minimum ethanol selling price and energy input for larger biorefineries. Furthermore, we study how improvements in densification technologies can reduce transportation cost and energy consumption.
1. Introduction Cellulosic biomass is seen as a promising source for sustainable biofuels and bioenergy production. Different biofuel conversion platforms, e.g., biochemical [1], thermochemical [2,3], and catalytic [4,5], have been widely studied. Numerous techno-economic analyses [1,2,5] have identified biomass feedstock and handling costs as the primary biofuel cost drivers. To improve the handling and transportation efficiency of biomass, the concept of regional biomass processing depot [6,7] (referred to in this work as ‘depot’) has been introduced. At a depot, biomass is pretreated and/or densified into a stable and dense intermediate (e.g., briquette or pellet), with increased durability and reduced volume. The densified intermediate can be transported economically over long distances. Lamers et al. [8] categorized depots as standard and quality depots. In the former, biomass is dried and densified through different mechanical and thermal processing technologies, e.g.,
⁎
grinding, drying, torrefaction, or pelleting [9]. In the latter, biomass is pretreated prior to drying and densification. The pretreated densified biomass can bypass pretreatment technologies at the biorefinery. Different analyses have been conducted to compare biofuel supply chains (SCs) with and without depots. Argo et al. [10] and Muth et al. [11] concluded that larger biorefinery sizes with standard depots would be greatly preferred over a biofuel SC without depots. However, life cycle greenhouse gas emissions of the SC with depots are higher [10,11]. Similar conclusions were drawn from Kim and Dale [12] where large-scale biorefineries with quality depots were shown to lead to better economic performance, but most environmental impact categories are higher than those without depot [12]. Lamers et al. [13] found that cost and profitability risk reductions, as well as handling and storage efficiency improvement can be achieved with the installation of standard and quality depots. Lin et al. [14] showed that high-density pellets lead to higher transportation costs for short-distance
Corresponding author at: Department of Chemical and Biological Engineering, University of Wisconsin-Madison, 1415 Engineering Drive, Madison, WI 53706, USA. E-mail address:
[email protected] (C.T. Maravelias).
http://dx.doi.org/10.1016/j.apenergy.2017.08.161 Received 16 May 2017; Received in revised form 2 August 2017; Accepted 14 August 2017 0306-2619/ © 2017 Elsevier Ltd. All rights reserved.
Please cite this article as: Ng, R.T.L., Applied Energy (2017), http://dx.doi.org/10.1016/j.apenergy.2017.08.161
Applied Energy xxx (xxxx) xxx–xxx
R.T.L. Ng, C.T. Maravelias
feedstock/transportation costs ($ Mg−1) average feedstock/transportation costs ($ Mg−1) average feedstock/transportation costs for configuration n ($ Mg−1) C TFB total average feedstock cost in hybrid configuration ($ Mg1 ) Dj,k / Dj,l / Dk,l actual distance along arc j → k/j → l/k → l (km) average actual collection distance for biorefinery/depot Dl / Dk (km) Dn average actual collection distance for configuration n (km) EE energy input per gallon of ethanol (MJ gal−1) T/TR E energy input/transportation energy (MJ Mg−1) E T/TR average energy input/transportation energy (MJ Mg−1) T/TR average energy input/transportation energy for configEn uration n (MJ Mg−1) TB E total average energy input for hybrid configuration (MJ Mg−1) amount of biomass collected in configuration n (Mg Fn year−1) Q angle in the polar coordinate system (rad) r j,k / r j,l/ rk,l straight line distance along arc j → k/j → l/k → l (km) rk / rl maximum collection radius for depot k/biorefinery l (km) transition distance between configurations n and n″ (km) rn∗,n″ R radius in the polar coordinate system (km) ratio of biomass collected in configuration n over total xn biomass collected (%) Y annual biomass productivity (Mg km−2 year−1)
CFB/TR C FB/TR FB/TR Cn
Nomenclature Parameters operating days (day year-1) transportation and storage loss factor (%) annual corn stover yield (Mg km−2 year−1) fraction of corn stover collected (%) energy for harvesting and collection/depot processing/ biomass handling (MJ Mg−1 biomass) ∊B/P energy for transporting baled/densified biomass (MJ Mg−1 km−1) ζ ethanol yield (gal ethanol Mg−1 biomass) ratio of harvested corn area to the total land area (%) η θl / θk biorefinery/depot size (Mg day−1) percentage of participating cornfields within the collection ϑ radius (%) κ HC/DP/BH/FT cost for harvesting and collection/depot processing/ biomass handling/fixed transportation ($ Mg−1) B/P σ variable cost for transporting baled/ densified biomass ($ Mg−1 km−1) ω tortuosity factor
α β γ δ ε HC/DP/BH
Binary Parameter
χk
=1 if configuration with depot k is selected
Variables
CFE
feedstock cost per gallon of ethanol ($ gal−1)
biofuel SCs or the analysis of given configurations. There have been very limited analyses of biofuel SCs. Accordingly, the goal of this paper is to study the economic performance and energy efficiency of the SC under different configurations and transportation modes. The contributions of this paper are the following: (1) we develop systematic methods for the calculation of cost and energy inputs; (2) we introduce the concept of hybrid configuration, a combination of two or more configurations, which leads to better economics and higher energy efficiency; (3) we compare the economic performance and energy efficiency of various biofuel SC configurations; and (4) we investigate the impact of a number of parameters (e.g., biomass productivity, biorefinery size, and densification efficiency) on the efficiency of the SC. The analysis provides critical insights into the following questions: When should we ship biomass directly to the biorefinery? Where should we install a depot? Which transportation mode yields lower cost and higher energy efficiency? How should SC configurations be combined? The rest of this article is structured as follows: In Section 2, we introduce different SC configurations, and present the assumptions we adopt and calculations we perform. In Section 3, we present the results and discuss the key cost and energy efficiency drivers. We close in Section 4 with some concluding remarks. We use lowercase Greek letters for parameters; Latin letters for variables; lowercase Latin italic letters for subscripts; and uppercase Latin letters for superscript.
transportation, but lower overall costs if used for long-distance transportation. Kim and Dale [15] reported that large-scale biorefineries in the Midwest region of the United States, which are installed in high feedstock density areas, are economically competitive with SCs without depots. Similarly, de Jong et al. [16] found that depots are economically favorable only at large-scale forest-based biorefineries in Sweden. Recently, Gautam et al. [17] also reported that cost reductions and forest biomass quality improvement when a depot was added to the SC in Quebec, Canada. In terms of biofuels SC design, a number of approaches have been proposed (see [18–21] for reviews). Most of these approaches do not consider depots [22–25], while there are some that account for depot location and capacity [26–28]. Furthermore, Gonzales et al. [29] developed a geographic information system (GIS)-based heuristic to locate and size depots and biorefineries. Finally, there are a few studies that consider biofuel SC design, including depots, under uncertainty using Monte Carlo [30] and stochastic programming [31] approaches. Despite the research in the area, there is a number of open questions. First, previous analyses of SCs with depots considered a single configuration [10–17], where all biomass is assumed to be shipped to depots; i.e., no biomass is shipped directly to the biorefinery even when the harvesting sites are located nearby. However, it has been shown [27,28] that the economics of SCs can be improved through the combination of two configurations: (1) direct biomass shipments from harvesting sites (located nearby) to the biorefinery, and (2) biomass from the harvesting sites farther away from the biorefinery is first shipped and densified at the nearest depot and then shipped to the biorefinery. Second, most previous analyses were restricted to one transportation option (either truck or rail) for shipping densified biomass when depots are selected. However, multiple transportation modes should be considered since truck transportation of densified biomass may be essential in areas where no railway network is available. Finally, the focus of previous research has been on the design of
2. Methods 2.1. Configurations We define a configuration as a combination of depot selection and transportation modes. If a depot is selected, baled biomass is first sent to a depot for densification and densified biomass is then shipped to a biorefinery; else, baled biomass is shipped to a biorefinery directly. We 2
Applied Energy xxx (xxxx) xxx–xxx
R.T.L. Ng, C.T. Maravelias
2.2. Assumptions
consider two transportation modes: (1) Truck transportation, which is often used due to the extensive highway and state roadway network and its accessibility. (2) Rail transportation, which has the advantage of low energy consumption per unit mass of biomass and lower transportation cost at longer distances, given its relatively low variable transportation costs. We consider a total of seven configurations, which can be represented by a tuple: (transportation mode from harvesting site to depot, depot selection, transportation mode from harvesting site or depot to biorefinery) (Fig.1A), where “D”, “T”, and “R” denote depot, truck, and rail selection, respectively, while “-” denotes no depot or transportation mode. We further assume that the biorefinery is located next to a railroad line. For the configurations without depot, baled biomass (blue line) from the harvesting site is transported directly to the biorefinery by truck (-, -, T). If rail transportation is considered, baled biomass is shipped through (-, -, R) or (T, -, R) configurations. In the former, baled biomass is transported directly to the biorefinery if the harvesting site is located next to the rail station. In the latter, baled biomass is trucked to the nearest rail station and then shipped to the biorefinery. For the configurations with depots, densified biomass (red line) is transported by truck (-, D, T) or rail (-, D, R) when the harvesting site is co-located with the depot. In practice, the depot will receive biomass from more than one harvesting sites. Baled biomass is trucked to the nearest depot and densified biomass is shipped to the biorefinery either by truck (T, D, T) or rail (T, D, R). We define a hybrid configuration as a combination of configurations (the idea is illustrated in Fig. 1B). The transition distance, rn∗,n″ is the distance where configuration n″ becomes economically or energetically more efficient than configuration n , which is employed closer to the biorefinery. The transition includes changes in transportation mode or depot installation. We start with the configuration near the biorefinery. For example, in Fig. 1B, n = 1, 2, and 3 are used to denote (-, -, T), (T, D, T), and (T, D, R), respectively. In the orange circular area (-, -, T), baled biomass is trucked directly to the biorefinery. In the blue annular ∗ ∗ area (T, D, T) between r1,2 and r2,3 , baled biomass is sent to the depot and densified biomass is then trucked to the biorefinery. Depot should ∗ be installed at a distance larger than r2,3 (dashed-dot line) from the biorefinery at the center (red annular area) and densified biomass is transported by rail (T, D, R).
We introduce the following indices: harvesting site j, depot k, biorefinery l. In the base case, we consider corn stover as biomass feedstock and assume that the capacity of a depot is 200 metric tons per day [8] (Mg day−1). All costs are indexed to 2011 dollars, and the weight of biomass is in dry basis unless stated otherwise. We consider a standard depot where baled corn stover is dried and densified through a high moisture pelleting process [32], for which cost information is taken from Lamers et al. [8]. Note that different technologies used in the depot may have different processing costs and energy requirements. The bulk densities of baled and densified corn stover are assumed to be 120 and 560 kg m−3, with moisture contents of 20% and 10%, respectively. The material losses due to drying and densification are negligible [32]. The fixed mass loss factor for transportation and storage is 4.94% [33]. The capacity of the biorefinery is 2,000 Mg day−1 unless stated otherwise. Dilute acid pretreatment, separate hydrolysis and fermentation are employed at the biorefinery. Heat and power generation allows the biorefinery to be energetically self-sufficient while excess electricity can be sold to the grid. Process and cost information on the biorefinery are extracted from National Renewable Energy Laboratory report [1]. Note that the analysis can be repeated using the proposed methods for any feedstock and technology. We repeat the analysis for thermal decomposition with catalytic upgrading using oak as a feedstock [34]. Details and results are found in the Supplementary Material. We consider two transportation distances: (1) point-to-point, and (2) region-to-point. In the former, the transportation distance from a specific harvesting site/depot to the depot/biorefinery is used (Fig. 2A). In the latter, we calculate the average distance assuming that the biomass collection area is circular with the biorefinery or depot located at the center, and the harvesting sites uniformly distributed in the area (Fig. 2B). In a hybrid configuration, the average transportation distance from depots in the annular area to the biorefinery at the center is estimated (see Section S.1 in the Supplementary Material for more details). We assume that biomass productivity is on an annual basis, though productivity based on seasonal or monthly basis can also be considered. Note that the maximum collection radius may vary if seasonal productivity is considered. Biomass productivity of corn stover is assumed to be 8.5 Mg km−2 year−1 for the base case [12]. For further details regarding productivity calculations, see Kim and Dale [12].
Fig. 1. (A) Biofuel supply chain configurations with and without regional depot and alternate transportation modes. Blue and red lines represent the transportation of baled and densified biomass, respectively. “+” signs represent the rail station, or the depot is co-located with the harvesting site. (B) Hybrid configuration (-, -, T)/(T, D, T)/(T, D, R). (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)
3
Applied Energy xxx (xxxx) xxx–xxx
R.T.L. Ng, C.T. Maravelias
Fig. 2. Illustration of (A) point-to-point and (B) region-to-point distances.
determine the minimum ethanol selling price (MESP) in $ per gallon gasoline equivalent (GGE) in order to obtain a net present value of zero. We consider a 10% internal rate of return, 30 years of the biorefinery lifetime, and 35% corporate tax rate. We follow the assumption values of equity financing, depreciation, construction time, start-up time, etc., provided by National Renewable Energy Laboratory [1]. Next, we describe how we calculate the transportation cost, C TR in Eq. (1), in the two different cases:
2.3. Cost calculations The cost parameters and variables for the calculations are summarized in Fig. 3A. The harvesting and collection cost, κ HC , includes nutrient replacement, fuel use, labor and equipment for harvesting, binding, and stacking baled corn stover. The depot processing cost, κ DP , includes operating cost and annualized capital investment for drying and densifying corn stover [8]. The biomass handling cost, κ BH , includes size reduction (grinding and milling) prior to densification at the depot or pretreatment at the biorefinery [8]. The transportation cost, C TR , which consists of fixed and variable transportation costs, will be presented in the following sub-sections. The fixed transportation cost, κ FT , includes the equipment and labor needed to load/unload baled or densified biomass to/from truck or railcar [33,35,36]. The variable costs for transporting baled corn stover [33], σ B by truck and rail are 0.20 and 0.02 $ Mg−1 km−1, respectively, while the variable costs for shipping densified corn stover [35,36], σ P by truck and rail are 0.06 and 0.01 $ Mg−1 km−1, respectively. The calculations for truck transportation cost are based on round trip [33]. Table 1 summarizes the cost parameters for different configurations. The feedstock cost at the biorefinery gate, CFB is estimated as follows:
CFB = κ HC + κ BH + κ DPχk + C TR
2.3.1. Point-to-Point approach The transportation cost, C TR is determined based on the following equation:
C TR = κ FT + σ BDj,l (1−χk ) + σ Bχk Dj,k + σ Pχk Dk,l
where Dj,k / Dj,l / Dk,l is the actual distance along the arc between harvesting site j and depot k (denoted as j → k), j and biorefinery l (j → l), and k and l (k → l). When a depot is considered ( χk = 1), the second term on the RHS of Eq. (3) becomes zero. In contrast, the last two terms in the RHS of Eq. (3) become zero if a configuration without depot is considered ( χk = 0 ). Note that the actual distance D (Dj,k / Dj,l / Dk,l ) is calculated based on the straight line distance r (r j,k / r j,l/ rk,l along arcs j → k/j → l/k → l) (see Fig. 2A) as follows:
(1)
D = ω·r
where χk is the binary parameter for depot k ( χk = 1 if depot is included). The feedstock cost to produce one gallon of ethanol, CFE ($ gal−1) is given by:
CFE
CFB = ζ (1−β )
(3)
(4)
where ω is the tortuosity factor, which depends on the road infrastructure [37]. We assume ω = 1.3 for truck and ω = 1.79 for rail [33]. 2.3.2. Region-to-Point approach The maximum collection radius, r (rk for depot and rl for biorefinery) is first determined based on a radial geometric mean formula [12] (see Fig. 2B gray and green shaded circles):
(2)
where ζ is the ethanol yield (gal ethanol Mg-1 corn stover), and β is the fixed loss factor due to transportation and storage (%). The discounted cash flow rate of return analysis is adopted to
r=
θα Y π (1−β )
(5)
Fig. 3. (A) Cost and (B) energy parameters and variables. Solid and dashed arrows represent parameters and variables for the configurations with and without depots, respectively. Blue and red colors represent the transportation of baled and densified biomass, respectively. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)
4
Applied Energy xxx (xxxx) xxx–xxx
R.T.L. Ng, C.T. Maravelias
Table 1 Cost parameters normalized to $ Mg−1 baled biomass [8,33,35,36]. Configuration
(-, -, T)
(-, -, R)
(T, -, R)
(-, D, T)
(-, D, R)
(T, D, T)
(T, D, R) 40.0
Harvesting and Collection, κ HC
40.0
40.0
40.0
40.0
40.0
40.0
Biomass Handling, κ BH
17.7
17.7
17.7
17.7
17.7
17.7
17.7
Depot Processing, κ DP
–
–
–
18.1
18.1
18.1
18.1
Fixed Transportation, κ FT
6.9
49.4
6.9 + 49.4*
3.3
17.6
6.9 + 3.3*
6.9 + 17.6*
* Loading and unloading happen twice for (T, -, R), (T, D, T) and (T, D, R). TR
where θ is the facility size (θk for depot and θl for biorefinery) (Mg day−1), α is the operating days (day year−1), β is the fixed transportation and storage loss factor (%), and Y is the annual biomass productivity (Mg km−2 year−1), which is calculated based on the following equation:
Y = γ ·δ·η ·ϑ
considered in calculating Cn . For the example in Fig.1B, for n = 1 (no TR depot is installed): Cn = 1 = κ FT + σ BDn = 1. When depot installation is TR considered ( χk = 1), e.g., for n = 2: Cn = 2 = κ FT + σ BDk + σ PDn = 2 . The FB average feedstock cost for configuration n, Cn , can be calculated by TR TR replacing C in Eq. (1) with Cn . The total average feedstock cost C TFB for the hybrid configuration is given by:
(6) −2
−1
year ), δ is the where γ is the annual corn stover yield (Mg km fraction of corn stover collected (%), η is the ratio of harvested corn area to the total land area (%), and ϑ is the percentage of participating cornfields within the collection radius (%). In polar coordinates, if Q is the angle, and R is the radius, then for a circular area Q ∈ {0,2π } and R ∈ {0,r } . The average actual collection distance D to a biorefinery (Dl ) or depot (Dk ) which are located at the center of a circular area is given as:
C TFB =
xn =
Dn = ω
2π
rn,n″ RdRdQ 0
∗
=
∫0 ∫
Dn = ω
∗
2π
r
2π
n′,n ∗ rn,n″ r n∗′,n
∫0 ∫
The parameters and variables used for energy calculations are shown in Fig. 3B. The energy requirement for biomass harvesting and collection, ε HC , includes fuel use and nutrient replacement at the harvesting sites [38]. The energy for biomass handling, ε BH , includes grinding and milling at the depot or biorefinery [8]. The energy for depot processing, ε DP includes drying and densification of biomass (e.g., electricity and natural gas) [8]. Since the biorefinery is energetically self-sufficient, the process heat and electricity required at the biorefinery is not considered when calculating energy input. Based on the Greenhouse gases, Regulated Emissions, and Energy use in Transportation (GREET) model [39], energy consumptions of heavy-duty truck and rail are 16 MJ km−1 and 0.2 MJ Mg−1 km−1, respectively. The energy input for transporting baled biomass, ∊B by truck and rail, are 5.27 and 0.83 MJ Mg−1 km−1, respectively, while the energy input for transporting densified biomass, ∊P by truck and rail, are 1.13 and 0.18 MJ Mg−1 km−1, respectively. The calculation for truck transportation energy use is based on round trip [39]. Table 2 summarizes the energy parameters for the configurations with and without depots. We calculate the energy input per Mg of biomass, E T as follows:
2 ∗ ωrn,n″ 3
(9)
=
2 ω 3
(rn∗,n″ )2
2π [(rn∗,n″ )3 − (r n∗′,n )3]
=
RdRdQ rn∗′,n 2
+( ) + rn∗,n″ + rn∗′,n
3 2π [(rn∗,n″ )2 − ( r n∗′,n )2] 2
E T = ε HC + ε BH + ε DPχk + E TR
rn∗,n″·rn∗′,n (10)
Table 2 Energy parameters normalized to MJ Mg-1 baled biomass [8,38].
∗ ∗ ∗ ∗ 2 ∗ 2 (r2,3)2 + (r1,2)2 + r2,3·r1,2 D1 = ωr1,2 , D2 = ω ∗ ∗ 3 3 r2,3 + r1,2
TR
The average transportation cost, Cn
Configuration
is then written as:
Cn = κ FT + σ BDn (1−χk ) + σ Bχk Dk + σ Pχk Dn
(15)
where E TR is the transportation energy. The energy input to produce one gallon of ethanol, EE is given as:
For example, in Fig. 1B, Dn = 1 for the orange circular area and Dn = 2 for the blue annular area are:
TR
(14)
2.4. Energy calculations
For an annular area, the average actual distance, Dn between transition distances rn∗′,n and rn∗,n″ (see Fig. 2B blue shaded annulus) is calculated as follows:
∫0 ∫r ∗n,n″ R2dRdQ
(13)
(8)
∗
r
Fn
where x n is the ratio of the amount of biomass collected in the configuration n, Fn (Mg year−1) over the total biomass collected in the entire region (Mg year−1).
In Eq. (8), the average actual collection distance for depot, Dk is considered as biomass from different harvesting sites is transported to fulfil the depot feedstock demand. The average feedstock cost, C FB is calculated by replacing C TR in Eq. (1) with C TR . For a hybrid configuration, the shape of biomass collection area near the biorefinery is always a circle. The average distance, Dn , over the region, up to the transition distance rn∗,n″ (see orange shaded circle in Fig. 2B) is given as: 2π
Fn
∑
Fn = π·Y ·(1−β )·[(rn∗,n″ )2−(rn∗′,n )2]
The average cost for transporting baled biomass to a depot and shipping densified biomass to the biorefinery, C TR can then be calculated as follows:
∫0 ∫0 n,n″ R2dRdQ
(12)
n
(7)
C TR = κ FT + σ BDk + σ PDk,l
FB
x n Cn
n
2π r ∫0 ∫0 R2dRdQ
2 D = ω 2π r = ω·r 3 ∫0 ∫0 RdRdQ
∑
When χk = 0 , only the first two terms on the RHS of Eq. (11) are 5
Depot
1086.0
1086.0
Depot Processing, ε DP
–
882.3
Biomass Handling, ε BH
681.0
681.0
Harvesting and Collection,
(11)
No Depot
ε HC
Applied Energy xxx (xxxx) xxx–xxx
R.T.L. Ng, C.T. Maravelias
EE =
ET ζ (1−β )
viable as it has relatively high fixed transportation cost. While this analysis helps us understand the various trade-offs, all harvesting sites cannot be collocated with a depot or rail station (r j,k = 0 ); rather, depots are expected to receive biomass from different harvesting sites to meet their processing capacity. To understand the trade-offs in a more realistic setting, we study the feedstock costs by assuming r j,k = 35 km (Fig. 4B), which, as we will see later, is equal to the average distance (rk ) for a 200 Mg day−1 depot size if the productivity is 8.5 Mg km−2 year−1. Additional loading/unloading at the depot/rail station leads to higher feedstock costs for the (T, D, T), (T, D, R) and (T, -, R) configurations. The transition distance for depot installation is 155 km (intersection between orange and blue lines). Depot should be installed if r j,l > 155 km. Truck transportation of densified biomass ((T, D, T), blue line) has lower cost if r j,l ∈ [155, 260] km. Rail transportation of densified biomass ((T, D, R), red line) is the best mode for distances larger than 260 km. These transition distances can be used to remove arcs and thus reduce the size of large-scale spatially explicit biofuel SC optimization models. Specifically, we can use the proposed insights to develop a preprocessing algorithm that identifies and removes arcs that will never be selected in an optimal SC configuration. For example, given a set of potential biorefineries, we can exclude depots that are located less than 155 km from a biorefinery (short-dashed line in Fig. 4B), as well as the arcs from a depot to a biorefinery for truck and rail transportations if the harvesting site-biorefinery distances are less than 155 km (shortdashed line) and 260 km (dot-dashed line), respectively. From an energy point of view, rail transportation is always better than truck transportation. Fig. 5A displays energy input for various configurations if r j,k = 0 km. Rail shipment of baled biomass ((-, -, R), purple line) yields the lowest energy input if r j,l ≤ 760 km, due to lower rail transportation energy. However, this configuration is not economically viable as presented earlier. Due to the additional energy required at the depot, ((-, D, T), gray line) and ((-, D, R), green line) have higher energy input at zero distance. The depot installation with rail transportation (-, D, R) has lower energy input if r j,l > 135 km. If no railway network is available, the transition to (-, D, T) happens at r j,l = 165 km. When r j,k is fixed at 35 km (Fig. 5B), the transition distances from (-, -, T) to (T, D, R) and from (-, -, T) to (T, D, T) are shifted to 170 km and 199 km, respectively. We discuss the key cost and energy efficiency drivers in the next sub-sections. We also provide the results for a special case where no railway is available in the Supplementary Material (Section S.3).
(16)
For point-to-point approach, E TR is given by:
E TR = ∊B Dj,l (1−χk ) + ∊B χk Dj,k + ∊B χk Dk,l
(17)
The average transportation energy input for transporting baled biomass to a depot and shipping densified biomass to the biorefinery, E TR is calculated as follows:
E TR = ∊B Dk + ∊P Dk,l
(18)
The average energy input, E T is calculated by replacing E TR in Eq. (15) with E TR . For a hybrid configuration, the average transportation energy input for each configuration n is written as:
E nTR = ∊B Dn (1−χk ) + ∊B χk Dk + ∊P χk Dn
(19)
The average energy input for each configuration n, E nT in the hybrid configuration can be calculated by replacing E TR in Eq. (15) with E nTR (see Supplementary Material Section S.2 for example calculations). The total average energy input E TB for the hybrid configuration is given as:
E TB =
∑
x n E nT
n
(20)
where x n is the ratio of the amount of biomass collected, which can be calculated via Eq. (13). 3. Results and discussion We first study the feedstock cost, CFB /CFE as a function of point-topoint straight line distance, r j,l . The feedstock costs for the configurations where the depot/rail station is co-located with the harvesting site (the distance between the harvesting site and the depot/rail station, r j,k = 0 ) are illustrated in Fig. 4A. If r j,l ≤ 81 km, baled biomass should be trucked directly to the biorefinery ((-, -, T), orange line), which yields CFB ∈ [65, 86] $ Mg-1 or CFE ∈ [0.79, 1.05] $ gal-1. Both ((-, D, T), gray line) and ((-, D, R), green line) have high costs at zero distance due to the additional cost of installing and operating the depot. Depots help to obtain a lower feedstock cost if r j,l > 81 km. Truck transportation of densified biomass (-, D, T) is favorable if r j,l ∈ [81, 225] km, which yields CFB ∈ [86, 98] $ Mg−1. If r j,l > 225 km, (-, D, R) has the lowest cost (green line), due to a small variable transportation cost of rail. Rail shipment of baled biomass ((-, -, R), purple line) is not economically
Fig. 4. Feedstock cost as a function of straight line distance, r j,l if the distance between the harvesting site and the depot/rail station are (A) r j,k = 0 , and (B) r j,k = 35 km.
6
Applied Energy xxx (xxxx) xxx–xxx
R.T.L. Ng, C.T. Maravelias
Fig. 5. Energy input as a function of r j,l if (A) r j,k = 0 , and (B) r j,k = 35 km.
3.1. Harvesting site – depot distance
3.2. Biomass productivity
The distance between the harvesting site and the depot, r j,k has a significant influence on the feedstock cost. Fig. 6A illustrates the feedstock cost as a function of point-to-point distances r j,l and r j,k . At zero r j,k , if r j,l ∈ [81, 225] km, the depot is installed and densified biomass can be shipped by truck, (-, D, T); while densified biomass is transported by rail, (-, D, R), if r j,l > 225 km. The transition distance for depot installation starts at 120 km after considering additional loading/ unloading at the depot, and increases with r j,k . The transition distance for rail transportation of densified biomass begins at 225 km (dot-dashed line). A reference dark blue line is shown at r j,k = 35 km, which is equal to the average distance at the base case. The energy input for the best configurations as a function of r j,l and r j,k by excluding (-, -, R) is illustrated in Fig. 6B. The bold vertical purple line denotes (-, D, R) if r j,k = 0 . The transition distance for depot installation begins at 135 km (dot-dashed line) and increases with r j,k .
Biomass productivity, Y which depends on weather, tillage practice, etc., is highly variable. Productivity changes lead to changes in the average actual collection distance for the depot, Dk when the capacity of a depot is fixed. Fig. 7A illustrates the average feedstock cost, C FB along Dk and rk,l , based on the estimation of average transportation cost, C TR via Eq. (8). Dk (gray line with respect to the secondary axis) decreases with Y . Thus, the transition distance from (-, -, T) to (T, D, T) changes due to Dk . For example, if Y = 7.5 Mg km−2 year−1, Dk is 48 km. If the depot is located at rk,l = 150 km, densified biomass is trucked to the biorefinery and the feedstock cost is 108 $ Mg−1 (1.32 $ gal−1). If a depot is installed in an area with 2.5 km−2 year−1, Dk is 84 km. The feedstock cost at rk,l = 150 km increases to 115 $ Mg−1 (1.41 $ gal−1). Note that (T, D, T) and (T, D, R) are the optimal configurations if rk,l ∈ [120, 225] km and rk,l ≥ 225 km, respectively. Fig. 7B shows rail transportation (T, D, R) is preferable at long distances (≥ 135 km) from an energy point of view. The average energy input
Fig. 6. (A) Feedstock cost and (B) energy input for best configurations as a function of r j,l and r j,k . The dark blue line represents the average distance at the base case (Y = 8.5 Mg km−2 year−1 and 200 Mg day−1 depot size), which is equal to 35 km. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)
7
Applied Energy xxx (xxxx) xxx–xxx
R.T.L. Ng, C.T. Maravelias
Fig. 7. (A) Average feedstock cost and (B) average energy input for the best configurations as a function of distance and biomass productivity.
From an energy point of view, hybrid configuration (-, -, T)/(T, D, R) always yields lower average energy input if Y ≤ 7.5 Mg km−2 year−1, when a 2,000 Mg day−1 biorefinery is considered (Fig. 9A). Direct biomass shipment, (-, -, T), yields lower energy input if Y > 7.5 Mg km−2 year−1. The total average energy input for the case where no railway network is available, is shown at the top right of Fig. 9A. Hybrid configuration (-, -, T)/(T, D, T) has lower total average energy input if Y ≤ 4 Mg km−2 year−1. Fig. 9B shows the transition distance for different cases; e.g., if Y = 2.5 Mg km−2 year−1, the transition distance to (T, D, R) is 200 km, which yields an average energy input of 36.1 MJ gal−1 ethanol. If no railway network is available, the transition distance to (T, D, T) is 229 km. Note that no depot installation (-, -, T) is required if Y = 10 Mg km−2 year−1.
increases with rk,l , but decreases with Y . Note that rail transportation of densified biomass is always energetically favorable. Fig. 8A shows the total average feedstock cost, C TFB as a function of Y at a fixed biorefinery size (2,000 Mg day−1). If Y ≤ 6 Mg km−2 year−1, hybrid configuration (-, -, T)/(T, D, T) (blue line) has lower C TFB , while direct shipment ((-, -, T), orange line) always yields lower C TFB if Y > 6 Mg km-2 year-1. The target of US Department of Energy is to reach a fuel price of 3 $ GGE−1 by 2022 [40]. At this target, the feedstock cost should be 88 $ Mg−1 biomass (black line). This analysis implies that, if no other technology changes are implemented, the biorefinery should be installed in an area with average productivity of more than 13 Mg km−2 year−1 in order to achieve this target. Fig. 8B shows the transition distances between (-, -, T) and (T, D, T) for three values of Y . If Y = 2.5 Mg km−2 year−1, the maximum collection radius, rl , is 306 km and the transition distance is 184 km. Direct biomass shipment (-, -, T) is preferred if r j,l ≤ 184 km, whereas depots should be installed and densified biomass is trucked to the biorefinery if r j,l > 184 km. When Y increases to 5 Mg km−2 year−1, rl decreases to 217 km, and the transition distance decreases to 165 km. If Y = 15 Mg km−2 year−1 rl decreases to 125 km, which is less than the transition distance.
3.3. Biorefinery capacity Larger biorefinery facilities result in lower capital cost per gallon of ethanol ($ gal−1), but also higher transportation cost because biomass from a larger area has to be used. We calculate the total average feedstock cost, C TFB as a function of biorefinery capacity, θl at Y = 8.5 Mg km−2 year−1 (see Fig. 10A). The average actual collection distance,
Fig. 8. (A) Total average feedstock cost as a function of biomass productivity, and (B) transition distance at various productivities.
8
Applied Energy xxx (xxxx) xxx–xxx
R.T.L. Ng, C.T. Maravelias
Fig. 9. (A) Total average energy input as a function of biomass productivity, and (B) transition distance at various productivities.
Dk , for a 200 Mg day−1 depot is 45 km (=ω (1.3) × r (35 km)). The transition distances are 155 (short-dashed line) and 260 km (dot-dashed line), and the maximum collection radius increases with θl (black solid line with respect to the secondary axis). For small-scale biorefineries (θl ≤ 1,740 Mg day−1), biomass should be shipped directly to the biorefinery since (-, -, T) yields lower C TFB . Hybrid configuration (-, -, T)/(T, D, T) yields lower C TFB than direct shipment if θl ∈ (1,740, 4,920) Mg day−1. Hybrid configuration (-, -, T)/(T, D, T)/(T, D, R) is preferable if θl > 4,920 Mg day−1 as more biomass needs to be collected in a larger collection area. Rail transportation can potentially reduce the feedstock cost up to 9 $ Mg−1 for a 20,000 Mg day−1 biorefinery, as compared to truck transportation. The cost and revenue for different biorefinery capacities are illustrated in Fig. 10B. The minimum ethanol selling price (MESP) ranges from 3.67 to 4.64 $ GGE-1. The feedstock cost is a dominant cost; it accounts for 34–58% (1.58–2.11 $ GGE−1). The increase of cost is mainly due to the increase in transportation and depot installation costs. The production cost accounts for 24–30% (1.09 $ GGE-1) and the average income tax is about 2–6% (0.09–0.30 $ GGE-1). Note that the
cost related to capital investment decreases due to the economies of scale. The capital depreciation costs and average return on investment (ROI) account for 4–11% (0.16–0.52 $ GGE−1) and 11–29% (0.41–1.36 $ GGE−1), respectively. The credit for excess electricity is about 0.20 $ GGE-1. Note that the MESP continues to decline with biorefinery size, as the capital cost reduction exceeds the increased transportation costs at larger biorefinery sizes. From an energy point of view, hybrid configuration (-, -, T)/(T, D, R) (red line) has lower total average energy input if θl > 2,100 Mg day−1 (Fig. 11A). Fig. 11B shows the total average energy input based on the most economically attractive configuration as identified in Fig. 10A. The transition distances obtained from hybrid configuration (-, -, T)/(T, D, T)/(T, D, R) are used to calculate energy input. The total average energy inputs at θl ∈ [1,000, 1,740], [1,740, 4,920], and [4,920, 20,000] Mg day−1 are 28.1–30.2, 30.2–35.9, and 35.9–36.6 MJ gal−1, respectively.
3.4. Biorefinery capacity and biomass productivity In the previous sub-sections, we showed that both biomass
Fig. 10. (A) Total average feedstock cost as a function of biorefinery capacity. (B) Cost and revenue as a function of biorefinery capacity.
9
Applied Energy xxx (xxxx) xxx–xxx
R.T.L. Ng, C.T. Maravelias
Fig. 11. Total average energy input for the most (A) energetically and (B) economically attractive configurations as a function of biorefinery capacity.
axial pressure [41]. The transportation cost decreases with bulk density, but becomes less sensitive at higher densities [42]. Note that increasing the densification efficiency may reduce the transportation cost and energy, but it may require higher processing cost, κ DP , and energy input, ε DP . Thus, the change in transition distance depends on densification efficiency, κ DP , and ε DP . We define the densification factor (DF) as the ratio of bulk densities of densified biomass over baled biomass. For example, the DF is 4.7 for the base case (560 kg m−3/120 kg m−3). We first study the transition distance as a function of DF and κ DP , as shown in Fig. 13. When DF = 2 and κ DP = 10 $ Mg−1, the transition distances for hybrid configuration (-, -, T)/(T, D, T)/(T, D, R) are 177 km and 239 km (Fig. 13A). When κ DP increases to 30 $ Mg−1, hybrid configuration (-, -, T)/(T, D, R) is preferable beyond 296 km. For DF = 6 (Fig.13B), hybrid configuration (-, -, T)/(T, D, T)/(T, D, R) has the same transition distance between (T, D, T) and (T, D, R) (dashed-dot line), which is 268 km when we vary κ DP ∈ [10, 30]. However, the transition distance between (T, D, T) and (T, D, R) (short-dashed line) changes from 102 to 206 km when we vary κ DP ∈ [10, 30] $ Mg−1. Next, we consider a 10,000 Mg day−1 biorefinery, employing 200 Mg day−1 depots, installed in an area with Y = 8.5 Mg km−2 year−1. Based on the calculated transition distances, we summarize the MESP as a function of DF and κ DP (Fig. 14A). Hybrid configuration (-, -,
productivity and biorefinery capacity impact the cost and energy input. In this subsection, we study the effect of total average feedstock cost (Fig.12A). The total average feedstock cost decreases with Y , but increases with θl . At low productivity (e.g., Y = 5 Mg km−2 year−1), biomass should be trucked directly to the biorefinery if θl ≤ 1,170 Mg day-1. Hybrid configuration (-, -, T)/(T, D, T) yields lower costs if θl ∈ [1,170, 3,130] Mg day−1, whereas hybrid configuration (-, -, T)/(T, D, T)/(T, D, R) has the minimum cost values if θl > 3,130 Mg day−1. As Y increases (e.g., 25 Mg km−2 year−1), the transition capacities increase to 4,180 and 12,860 Mg day−1 for depot installation with truck and rail transportations, respectively. Fig. 12B shows the MESP as a function of Y and θl . Our analysis suggests that when the installation of small biorefineries is considered, potential depots can be excluded since direct shipment yields better economics, unless the biorefinery is located in a low productivity area.
3.5. Densification efficiency The densification efficiency at the depot can be improved by increasing the bulk density of biomass, which generally depends on material composition, particle shapes and size, particle size distribution, orientation of particles, specific density, moisture content and applied
Fig. 12. (A) Total average feedstock cost and (B) MESP as a function of biomass productivity and biorefinery capacity.
10
Applied Energy xxx (xxxx) xxx–xxx
R.T.L. Ng, C.T. Maravelias
Fig. 13. Feedstock cost as a function of straight line distance, r j,l for densification factors of (A) 2 and (B) 6.
T)/(T, D, R) is preferable if the depot has low DF and high κ DP . The black square indicates the MESP of the base case (3.81 $ GGE−1). To achieve the same, or lower MESP, a combination of DF and κ DP above and to the left of the dashed line should be used. For example, the maximum κ DP to achieve the same MESP at DF of 6 is 21.4 $ Mg−1 biomass. The total average energy input at various DF and ε DP is shown in Fig. 14B. Similarly, to achieve a total average energy input that is the same as or lower than that of the base case (35.2 MJ gal−1), a combination of DF and ε DP on the left of the dashed line is necessary.
combination of different SC configurations. We also studied how biomass productivity, biorefinery size, and densification efficiency affect the performance of biofuel SC. A SC without depots typically leads to better economics and higher energy efficiency for small-scale biorefineries, unless the biorefinery is constructed in a low productivity area. We also studied how improvements in densification efficiency can reduce transportation cost and energy consumption, while accounting for the increase in depot processing cost and energy. All calculations and analyses were performed assuming a single period (i.e., annual biomass availability) although biomass availability is seasonal. However, the analyses can be repeated using seasonal or even monthly availability data and then combined to obtain annual results. Similarly, the methods and analyses presented herein can be applied to other feedstocks or even biofuel SC with multiple feedstocks. The study of a multi-period, multi-feedstock SC will, in fact, be the topic of future research. In addition to allowing us to better understand the complex tradeoffs present in biofuels SCs, the insights offered in this study can also aid the formulation of large-scale optimization models for SC planning.
4. Concluding remarks In this paper, we studied the economic performance and energy efficiency of different biofuel SC configurations. We found that depot installation is not economically favorable close to the biorefinery, while rail transportation leads to better economics only at larger collection distances. From an energy point of view, rail transportation of densified biomass is always better than truck transportation. Thus, cost and energy reductions can be achieved through careful selection and
Fig. 14. (A) MESP and (B) total average energy input as a function of densification factor and depot processing cost or energy. Black square boxes indicate the MESP and the total average energy input of the base case. Dashed lines represent the combination of densification factor and depot processing cost or energy to achieve the same MESP or total average energy input. Solid lines represent the transition of configurations.
11
Applied Energy xxx (xxxx) xxx–xxx
R.T.L. Ng, C.T. Maravelias
biorefinery supply chain. Biofuels, Bioprod Biorefining 2015;9:648–60. [14] Lin T, Rodríguez LF, Davis S, Khanna M, Shastri Y, Grift T, et al. Biomass feedstock preprocessing and long-distance transportation logistics. GCB Bioenergy 2016;8:160–70. [15] Kim S, Dale BE. A distributed cellulosic biorefinery system in the US Midwest based on corn stover. Biofuels, Bioprod Biorefining 2016;10:819–32. [16] de Jong S, Hoefnagels R, Wetterlund E, Pettersson K, Faaij A, Junginger M. Cost optimization of biofuel production – The impact of scale, integration, transport and supply chain configurations. Appl Energy 2017;195:1055–70. [17] Gautam S, LeBel L, Carle M-A. Supply chain model to assess the feasibility of incorporating a terminal between forests and biorefineries. Appl Energy 2017;198:377–84. [18] Awudu I, Zhang J. Uncertainties and sustainability concepts in biofuel supply chain management: a review. Renew Sustain Energy Rev 2012;16:1359–68. [19] Sharma B, Ingalls RG, Jones CL, Khanchi A. Biomass supply chain design and analysis: Basis, overview, modeling, challenges, and future. Renew Sustain Energy Rev 2013;24:608–27. [20] Yue D, You F, Snyder SW. Biomass-to-bioenergy and biofuel supply chain optimization: Overview, key issues and challenges. Comput Chem Eng 2014;66:36–56. [21] Garcia DJ, You F. Supply chain design and optimization: Challenges and opportunities. Comput Chem Eng 2015;81:153–70. [22] Zamboni A, Shah N, Bezzo F. Spatially explicit static model for the strategic design of future bioethanol production systems. 1. Cost minimization. Energy & Fuels 2009;23:5121–33. [23] Akgul O, Shah N, Papageorgiou LG. Economic optimisation of a UK advanced biofuel supply chain. Biomass and Bioenergy 2012;41:57–72. [24] Alex Marvin W, Schmidt LD, Daoutidis P. Biorefinery location and technology selection through supply chain optimization. Ind Eng Chem Res 2013;52:3192–208. [25] Santibañez-Aguilar JE, González-Campos JB, Ponce-Ortega JM, Serna-González M, El-Halwagi MM. Optimal planning and site selection for distributed multiproduct biorefineries involving economic, environmental and social objectives. J Clean Prod 2014;65:270–94. [26] Čuček L, Martín M, Grossmann IE, Kravanja Z. Multi-period synthesis of optimally integrated biomass and bioenergy supply network. Comput Chem Eng 2014;66:57–70. [27] Ng RTL, Maravelias CT. Design of cellulosic ethanol supply chains with regional depots. Ind Eng Chem Res 2016;55:3420–32. [28] Ng RTL, Maravelias CT. Design of biofuel supply chains with variable regional depot and biorefinery locations. Renew Energy 2017;100:90–102. [29] Gonzales DS, Searcy SW. Biomass and Bioenergy GIS-based allocation of herbaceous biomass in biorefineries and depots. Biomass and Bioenergy 2017;97:1–10. [30] Hu H, Lin T, Wang S, Rodriguez LF. A cyberGIS approach to uncertainty and sensitivity analysis in biomass supply chain optimization. Appl Energy 2017;203:26–40. [31] Quddus MA, Ibne Hossain NU, Mohammad M, Jaradat RM, Roni MS. Sustainable network design for multi-purpose pellet processing depots under biomass supply uncertainty. Comput Ind Eng 2017;110:462–83. [32] Tumuluru JS, Cafferty KG, Kenney KL. Techno-economic analysis of conventional, high moisture pelletization and briquetting process. Am. Soc. Agric. Biol. Eng. Annu. Int. Meet. 2014, ASABE 2014, vol. 6, 2014, p. 4177–89. [33] Suh K, Suh S, Walseth B, Bae J, Barker R. Optimal corn stover logistics for biofuel production: a case in minnesota. Trans ASABE 2011;54:229–38. [34] Won W, Maravelias CT. Thermal fractionation and catalytic upgrading of lignocellulosic biomass to biofuels: Process synthesis and analysis. Renew Energy 2017;114:357–66. [35] Gonzales D, Searcy EM, Ekşioğlu SD. Cost analysis for high-volume and long-haul transportation of densified biomass feedstock. Transp Res Part A Policy Pract 2013;49:48–61. [36] Sokhansanj S, Kumar A, Turhollow A. Development and implementation of integrated biomass supply analysis and logistics model (IBSAL). Biomass and Bioenergy 2006;30:838–47. [37] Sultana A, Kumar A. Development of tortuosity factor for assessment of lignocellulosic biomass delivery cost to a biorefinery. Appl Energy 2014;119:288–95. [38] Morey RV, Kaliyan N, Tiffany DG, Schmidt DR. A corn stover supply logistics system. Appl Eng Agric 2010;26:455–61. [39] Argonne National Laboratory. Greenhouse gases, Regulated Emissions, and Energy use in Transportation (GREET) Model 2015. https://greet.es.anl.gov (accessed June 8, 2016). [40] U.S. Department of Energy. Bioenergy Technologies Office - Multi-Year Program Plan. US Dep Energy, Energy Effic Renew Energy 2016. http://energy.gov/sites/ prod/files/2016/07/f33/mypp_march2016.pdf (accessed October 8, 2016). [41] Lam PS, Sokhansanj S, Bi X, Mani S, Lim CJ, Womac AR, et al. Physical characterization of wet and dry wheat straw and switchgrass – bulk and specific density. ASABE Annu Int Meet 2008;300:23. [42] Sultana A, Kumar A. Optimal configuration and combination of multiple lignocellulosic biomass feedstocks delivery to a biorefinery. Bioresour Technol 2011;102:9947–56. [43] Villoria NB, Elliott J, Müller C, Shin J, Zhao L, Song C. Rapid aggregation of global gridded crop model outputs to facilitate cross-disciplinary analysis of climate change impacts in agriculture. Environ Model Softw 2016;75:193–201.
Specifically, the transition distance between configurations can be used to eliminate arcs in spatially explicit SC models (e.g., the direct arc from a harvesting site to a biorefinery can be eliminated if the distance is such that a configuration including a depot should be preferred), thereby allowing us to formulate detailed models based on spatially explicit biomass productivity data. The development of mixed-integer programming (MIP) planning models based on data from the Agricultural Model Intercomparison and Improvement Project (AgMIP) [43] combined with climate and historical cropland data will be the topic of future research. In summary, this study provides baseline results and a suite of tools for assessing the economics and energy efficiency of biofuel SCs. We hope that further studies will build upon this work to account for temporally and geographically nonuniform feedstock availability as well as different densification, pretreatment, and conversion technologies. Acknowledgements This work was funded by the DOE Great Lakes Bioenergy Research Center (DOE BER Office of Science DE-FC02-07ER64494). The authors would also like to acknowledge Professor Bruce Dale and Dr. Seungdo Kim, who provided helpful comments on this analysis. Appendix A. Supplementary material Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j.apenergy.2017.08.161. References [1] Humbird D, Davis R, Tao L, Kinchin C, Hsu D, Aden A, et al. Process Design and Economics for Biochemical Conversion of Lignocellulosic Biomass to Ethanol: Dilute-Acid Pretreatment and Enzymatic Hydrolysis of Corn Stover. Golden, CO: National Renewable Energy Laboratory (NREL); 2011. [2] Dutta A, Sahir A, Tan E, Humbird D, Snowden-Swan LJ, Meyer P, et al. Process Design and Economics for the Conversion of Lignocellulosic Biomass to Hydrocarbon Fuels - Thermochemical Research Pathways with In Situ and Ex Situ Upgrading of Fast Pyrolysis Vapors. Golden, CO: National Renewable Energy Laboratory (NREL); 2015. [3] Sikarwar VS, Zhao M, Clough P, Yao J, Zhong X, Memon MZ, et al. An overview of advances in biomass gasification. Energy Environ Sci 2016;9:2939–77. [4] Luterbacher JS, Rand JM, Alonso DM, Han J, Youngquist JT, Maravelias CT. Nonenzymatic sugar production from biomass using biomass-derived γ-valerolactone. Science 2014;343(80-):277–80. [5] Han J, Luterbacher JS, Alonso DM, Dumesic JA, Maravelias CT. A lignocellulosic ethanol strategy via nonenzymatic sugar production: process synthesis and analysis. Bioresour Technol 2015;182:258–66. [6] Eranki PL, Bals BD, Dale BE. Advanced regional biomass processing depots: a key to the logistical challenges of the cellulosic biofuel industry. Biofuels, Bioprod Biorefining 2011;5:621–30. [7] Bals BD, Dale BE. Developing a model for assessing biomass processing technologies within a local biomass processing depot. Bioresour Technol 2012;106:161–9. [8] Lamers P, Roni MS, Tumuluru JS, Jacobson JJ, Cafferty K, Hansen JK, et al. Technoeconomic analysis of decentralized biomass processing depots. Bioresour Technol 2015;194:205–13. [9] Chai L, Saffron CM. Comparing pelletization and torrefaction depots: optimization of depot capacity and biomass moisture to determine the minimum production cost. Appl Energy 2016;163:387–95. [10] Argo AM, Tan ECD, Inman D, Langholtz MH, Eaton LM, Jacobson JJ, et al. Investigation of biochemical biorefinery sizing and environmental sustainability impacts for conventional bale system and advanced uniform biomass logistics designs. Biofuels, Bioprod Biorefining 2013;7:282–302. [11] Muth DJ, Langholtz MH, Tan ECD, Jacobson JJ, Schwab A, Wu MM, et al. Investigation of thermochemical biorefinery sizing and environmental sustainability impacts for conventional supply system and distributed pre-processing supply system designs. Biofuels, Bioprod Biorefining 2014;8:545–67. [12] Kim S, Dale BE. Comparing alternative cellulosic biomass biorefining systems: Centralized versus distributed processing systems. Biomass and Bioenergy 2015;74:135–47. [13] Lamers P, Tan ECD, Searcy EM, Scarlata CJ, Cafferty KG, Jacobson JJ. Strategic supply system design – a holistic evaluation of operational and production cost for a
12