Cellulosic Biofuel Supply with Heterogeneous Biomass Suppliers: An Application to  Switchgrass-based Ethanol   Alicia Rosburg, John Miranowski, Keri Jacobs Working Paper No. 13015 August 2013   IOWA  STATE  UNIVERSITY   Department  of  Economics   Ames,  Iowa,  50011-­‐1070   Iowa  State  University  does  not  discriminate  on  the  basis  of  race,  color,  age,  religion,  national  origin,  sexual  orientation,   gender  identity,  genetic  information,  sex,  marital  status,  disability,  or  status  as  a  U.S.  veteran.  Inquiries  can  be  directed   to  the  Director  of  Equal  Opportunity  and  Compliance,  3280  Beardshear  Hall,  (515)  294-­‐7612. Cellulosic Biofuel Supply with Heterogeneous Biomass Suppliers: An Application to Switchgrass-based Ethanol Alicia Rosburg, John Miranowski, and Keri Jacobs Key Words biofuel, biomass, cellulosic ethanol, RFS2, switchgrass JEL codes Q16, Q11, Q42, Q41, Q48 Abstract The potential of biomass for alternative energy production has attracted considerable attention because of associated implications for energy security, food supply, and climate change. This paper considers the economic impacts of spatial variation and landowner behavior on potential biomass supply for U.S. cellulosic biofuel. We develop and apply a long-run biomass production through bioenergy conversion cost model that incorporates heterogeneity of biomass suppliers within and between local markets. An application to U.S. switchgrass-based cellulosic ethanol production suggests cost-minimizing biofuel production decisions, which include biorefinery size, biomass transportation distance, and price of biomass, vary significantly across locations. An aggregate switchgrass ethanol supply curve developed from local biofuel supply estimates is used to evaluate the potential for and cost of achieving cellulosic biofuel production targets such as the revised Renewable Fuels Standard. Empirical results suggest spatial variation in biomass production conditions plays an important role in the potential supply and distribution of U.S. cellulosic biofuel production. Cellulosic Biofuel Supply with Heterogeneous Biomass Suppliers: An Application to Switchgrass-based Ethanol1 Alicia Rosburg, John Miranowski, and Keri Jacobs 1. Introduction Unstable energy prices and concern about the environmental impacts of growing greenhouse-gas (GHG) emissions have increased interest in finding alternative sources of energy. The use of biomass, a renewable and potentially GHG-reducing energy source, has gained significant attention and political support in the United States. In addition to allocating federal funds to bioenergy research projects, the United States has imposed mandates and provided market incentives to stimulate bioenergy production and consumption. The revised U.S. Renewable Fuel Standard (RFS2) took effect July 2010 and mandates a minimum contribution from cellulosic biofuel – the form of bioenergy considered in this paper – to the U.S. transportation fuel mix through 2022. Several biomass-to-biofuel conversion methods exist, but the economics of cellulosic biofuel production has limited industry development. The first commercial scale cellulosic biorefinery isn’t expected to be operational until 2014.2 The amount of biomass that must be supplied for commercial scale production presents a significant challenge to industry development. A national biomass market does not exist and is unlikely to develop given the high costs of biomass transportation (Babcock, Marette, & Tregeur, 2011). As a result, cellulosic biorefineries will rely on local biomass markets for feedstock supply. The quantity and price at 1 This paper was prepared for and presented at the 16th ICABR Conference – 128th EAAE Seminar entitled “The Political Economy of the Bioeconomy: Biotechnology and Biofuel.” Ravello, Itally, June 24-27, 2012. Acknowledgements: This research was partially funded by the Biobased Industry Center (BIC). The authors would like to thank the Agricultural and Environmental Workshop participants at Iowa State University for valuable input on this paper. The authors are solely responsible for any remaining errors. 2 Three commercial-scale plants are under construction and expected to be operational in 2014: Abengoa’s 25 mgy plant in Hugoton, Kansas, POET-DSM Advanced Biofuels, LLC’s 25 mgy plant in Emmetsburg, Iowa, and DuPont Danisco Cellulosic Ethanol’s (DDCE) 27.5 mgy plant in Nevada, Iowa. 1 which biomass producers are willing and able to supply biomass to a biorefinery will vary both between and within local markets. Between markets, the amount of sustainable biomass production varies due to geographical and climate differences. Within local markets, potential suppliers differ in their perceived costs and benefits of biomass production even with relatively uniform production conditions.3 This paper evaluates the economic trade-offs faced by commercial-scale cellulosic biofuel production as a result of spatial variation and landowner heterogeneity in potential biomass supply. We begin with a theoretical long-run cost model, or supply model, from biomass production through bioenergy conversion that incorporates biomass supplier heterogeneity within and between local markets. A primary contribution of this paper is the treatment of local biomass supply within the theoretical model of cellulosic biofuel production cost. While previous literature has assumed the fraction of local landowners willing and able to participate in biomass supply is fixed and independent of the price of biomass, we incorporate a functional relationship between the rate of landowner participation and the price of biomass. The theoretical model is then applied to switchgrass-based ethanol production in the United States using biofuel processing costs, switchgrass production costs, and data on heterogeneity in the opportunity cost of potential biomass cropland. A non-linear mathematical programming model is used to determine the cost-minimizing production decisions – including biorefinery size, capture region distance, feedstock price, and average cost of cellulosic ethanol production – for each potential biorefinery location. The estimated local ethanol supplies are combined to generate an aggregate ethanol supply curve. The resulting supply curve is used to evaluate the economic trade-offs that exist as a result of spatial variation and landowner heterogeneity as well as the potential for and costs to meet the RFS2 cellulosic biofuel mandates. The paper is organized as follows. The economic trade-offs in cellulosic biofuel production not realized in petroleum-based transportation fuel and first-generation biofuel production (e.g., corn ethanol, soybean biodiesel) are discussed in the next section. Section 3 presents the theoretical model for cellulosic 3Bergtold, Fewell, & Williams (2011),Tyndall (2007), Tyndall, Berg, & Colletti (2011), Hipple & Duffy (2002), Wen, Ignosh, Parrish, Stowe, & Jones (2009), and Altman, Bergtold, Sanders, & Johnson (2011). 2 biofuel production cost with heterogeneous biomass suppliers. Section 4 describes the empirical specification and data used in the application of the theoretical model to U.S. switchgrass-based ethanol production. Results are presented in Section 5. Section 6 concludes. 2. Cost Structure for Cellulosic Biofuel Production The cellulosic biofuel industry has a different cost structure than petroleum-based transportation fuel or first-generation biofuel. Petroleum-based industries have relatively constant average feedstock costs with respect to plant size. Economies of scale in processing, up to a point, lead to large-scale petroleum-based refineries (Wright & Brown, 2007; Searcy & Flynn, 2009). By operating within local feedstock markets, biofuel producers face a trade-off between economies of scale in biofuel processing and diseconomies of scale in feedstock procurement. First-generation biorefineries use commoditized feedstocks with a market price (e.g., corn, soybeans). The increase in feedstock demand from a larger capacity first generation biorefinery is met by paying the market price for additional feedstock located farther from the biorefinery and paying a greater transportation cost. The trade-off between economies of scale in processing and diseconomies of transportation results in a cost-minimizing combination of feedstock transportation distance and biorefinery capacity. This cost-minimizing combination is independent of the market price of feedstock (Searcy & Flynn, 2009). By using a non-commoditized feedstock, the trade-offs are more complex for a potential cellulosic biorefinery. The biorefinery’s cost-minimizing decisions depend on the offered price of biomass. The fraction of land allocated into biomass production will be determined by the fraction of land for which the offered price of biomass covers all costs incurred from biomass production, including opportunity costs. We refer to this fraction as the “participation rate” in biomass supply. The relationship between the participation rate and price of biomass adds complexity to the cost structure of biomass procurement. To illustrate, a simplified example is outlined in Figure 1. Consider a potential biorefinery that offers biomass suppliers a price P per ton of biomass, and suppose the 0 participation rate within the local market is d at the biomass price P . Given d , let r denote the radius of 0 0 0 0 the circular capture region needed to satisfy feedstock demand if the biorefinery produces at capacity Q 0 3 gallons per year (Figure 1a). Now suppose the biorefinery would like to capitalize on economies of scale in biofuel processing by operating at capacity Q > Q . The increase in feedstock demand from the larger 1 0 capacity cellulosic biorefinery can be met in one of three ways. First, the biorefinery can maintain the offered price of biomass P and the participation rate d and satisfy the increase in feedstock demand by 0 0 traveling farther for additional feedstock. The larger capture region is depicted in Figure 1b by radius r > 2 r . Second, the biorefinery can maintain the size of the capture region (r ) but increase the participation 0 0 rate within the local market to d > d through an increase in the price offered for feedstock to P > P . 2 0 2 0 Third, the biorefinery could use a combination of both. This combination is illustrated in Figure 1b by radius r , price of biomass P , and participation rate d . Any of the three methods to satisfy feedstock 1 1 1 demand for a larger biorefinery increases average feedstock cost, resulting in diseconomies of feedstock procurement. This economic trade-off between economies of scale in cellulosic biofuel processing and diseconomies of scale in feedstock procurement leads to location-dependent cost-minimizing biofuel production decisions that not only include biorefinery size and biomass transportation distance but also the price of feedstock. [Figure 1] The theoretical model presented in the next section incorporates these economic trade-offs into a cellulosic biofuel long-run production cost model. Although to our knowledge, this is the first analysis to explicitly incorporate this relationship into a production cost model for cellulosic biofuel, previous literature has acknowledged the existence and importance of this relationship. Leboreiro & Hilaly (2011, p. 2713) note farmer participation rate in biomass supply is “a strong function of the economic incentive.” Yet, following previous literature4, Leboreiro & Hilaly go on to assume a fixed value and conduct sensitivity analysis. 4 See Cameron, Kumar, & Flynn (2007), Dornburg & Faaij (2001), Gan (2007), Gan & Smith (2010), Huang, Ramaswamy, Al-Dajani, Tschirner, & Cairncross (2009), Jenkins (1997), Kaylen, Van Dyne, Choi, & Blase (2000), Kaylen, Van Dyne, Kumar, Cameron, & Flynn (2003), Nguyen & Prince (1996), Searcy & Flynn (2009), Searcy & Flynn (2010), and Wright & Brown (2007). 4 3. Theoretical Framework: Long-run Cellulosic Biofuel Production Cost We consider production capacity choice of a potential cellulosic biorefinery prior to capital investment. The biorefinery is assumed to minimize the long-run average total cost of biofuel production (ATC) by choosing the production capacity and price of biomass to offer to local biomass suppliers, conditional on the biorefinery technology and local biomass supply conditions. A biorefinery’s average total cost is a function of its feedstock procurement costs (ATC ) and biofuel processing costs (ATC ). F P 3.1 Feedstock Procurement Cost The per ton cost of feedstock procurement includes the price paid to local biomass suppliers (P ), F storage cost (S), and transportation cost. The biofuel processor is assumed responsible for biomass transportation. The transportation cost per ton of feedstock is derived by multiplying the per mile per ton transportation cost (t) by the average hauling distance (D). Equation (1) calculates the per gallon cost of feedstock procurement (ATC ) by dividing total per ton feedstock costs by the gallons of biofuel produced F from each ton of feedstock, commonly referred to as the biofuel yield (Y ). O 1 𝐴𝑇𝐶 = ×[𝑃 +𝑆+𝑡×𝐷]. (1) 𝐹 𝐹 𝑌𝑂 The average hauling distance for a biorefinery operating with a capacity of Q gallons per year is derived using the equation from French (1960) for a circular capture area with a square road grid and uniform biomass density: 𝑄 𝐷 = 𝛾×√ (2) 𝑌𝑂×𝑌𝐵×𝑑 where Y is the biomass yield per acre, d is the fraction, or density, of land allocated to biomass B production within the region, and 𝛾 is a conversion factor.5 Holding all other variables constant, an increase in biorefinery capacity will increase the average 𝜕𝐷 hauling distance and average cost of feedstock transportation ( > 0). Conversely, an increase in the 𝜕𝑄 5 Alternative models exist for the transportation network, including those proposed by French (1960). Sensitivity analysis to alternative transportation arrangements had minimal impact on model results. 5 fraction of land allocated to biomass production will decrease the average biomass hauling distance and 𝜕𝐷 average cost of feedstock transportation holding all other variables constant ( < 0). Therefore, 𝜕𝑑 depending on the local biomass supply conditions, an increase in the fraction of land allocated to biomass supply could partially or fully offset the need to increase average transportation distance to meet feedstock demand for a larger biorefinery. Landowners respond to price incentives and the fraction of land allocated to biomass production will be non-decreasing in the price of biomass. This relationship is captured in the procurement cost model through a location-specific function 𝑑(𝑃 ), where 𝜕𝑑(𝑃𝐹) ≥ 0. 𝐹 𝜕𝑃𝐹 We refer to 𝑑(𝑃 ) as the local participation rate function. 𝐹 Incorporating equation (2) and the local participation rate function into equation (1) produces the following equation for the average cost of feedstock procurement (ATC ): F 1 𝑄 𝐴𝑇𝐶 (𝑄,𝑃 )= ×[𝑃 +𝑆+𝑡×𝛾×√ ]. (3) 𝐹 𝐹 𝐹 𝑌𝑂 𝑌𝑂×𝑌𝐵×𝑑(𝑃𝐹) 3.2 Biofuel Processing Cost Biofuel processing costs arise in converting biomass into cellulosic biofuel and depend on the biorefinery technology and capacity. There are per-gallon costs that depend on biorefinery capacity and exhibit economies of scale (𝐴𝑇𝐶 ) and there are per-gallon costs that are independent of biorefinery 𝑃,𝑄 capacity (𝐴𝑇𝐶 ). A power function is used to model biorefinery costs that exhibit economies of scale 𝑃,𝑁 (Brown, 2003) and assumes the following relationship between average cost at capacity Q and average 0 cost at capacity Q : 1 𝑘−1 Avg Cost =Avg Cost ×[𝑄1] . (4) 𝑄1 𝑄0 𝑄0 The scaling factor, 𝑘 ≥ 0, represents the rate at which total cost increases with capacity, or equivalently, 𝑘−1 represents the rate at which average cost changes with capacity. With economies of scale, k is strictly less than one. For a biorefinery with capacity Q, the power function for processing costs that exhibit economies of scale can be written as follows: 6 𝑘−1 𝑄 𝐴𝑇𝐶 = 𝐴𝑇𝐶 ×[ ] , (5) 𝑃,𝑄 𝑃,𝑄0 𝑄0 where 𝐴𝑇𝐶 and k are exogenous, known variables and 𝐴𝑇𝐶 represents per gallon costs for a 𝑃,𝑄0 𝑃,𝑄0 “baseline” biorefinery with capacity Q . Equation (5) and the biofuel processing costs independent of 0 biorefinery capacity together imply the following expression for the average total cost of processing biofuels (ATC ): P 𝑘−1 𝑄 𝐴𝑇𝐶 (𝑄)= 𝐴𝑇𝐶 +𝐴𝑇𝐶 ×[ ] . (6) 𝑃 𝑃,𝑁 𝑃,𝑄0 𝑄0 3.3 Biorefinery Objective Function Combining equations (3) and (6), the objective function for the cost-minimizing biorefinery can be written as follows: 𝑄 𝑘−1 1 𝑄 min 𝐴𝑇𝐶(𝑄,𝑃 )=min 𝐴𝑇𝐶 +𝐴𝑇𝐶 ×[ ] + ×[𝑃 +𝑆+𝑡×𝛾×√ ] (7) 𝑄,𝑃𝐹 𝐹 𝑄,𝑃𝐹 𝑃,𝑁 𝑃,𝑄0 𝑄0 𝑌𝑂 𝐹 𝑌𝑂×𝑌𝐵×𝑑(𝑃𝐹) where 𝜕𝑑(𝑃𝐹) ≥ 0 𝜕𝑃𝐹 𝑑(𝑃 )∈ [0,1] 𝐹 𝑄,𝑃 ≥ 0. 𝐹 The biorefinery objective function requires specification of the participation rate function, d(P ), and can F be solved using a non-linear mathematical programming model. 3.4 Fixed Biomass Density Approach: Use and Limitations The biorefinery’s problem has been simplified in previous literature by assuming the fraction of land allocated to biomass production and the price of biomass are fixed and independent of plant size. This approach assumes increased feedstock demand can only be met by traveling further to acquire additional biomass, leading to a cost-minimizing capacity and hauling distance that depend on an assumed fixed density value but are independent of the price of biomass. With an endogenous participation rate (equation 7), the biorefinery’s long-run cost can be depicted as a surface graph plotted over a range of capacities and local participation rates (Figure 2a). The solution to the biorefinery objective function is the capacity and local participation rate with 7 corresponding price of biomass at the minimum of the cost surface. A fixed participation rate approach used in previous literature is equivalent to selecting and evaluating a ‘slice’ from the biorefinery cost surface at a fixed participation rate (Figure 2b). The biorefinery objective function is simplified to a single variable problem for the minimum efficient capacity (Q). Unless the fixed participation rate and price of biomass are set exactly at the cost-minimizing values from the cost surface, the estimated minimum cost of biofuel production and minimum efficient capacity will differ between an endogenous and fixed value analysis. [Figure 2] The assumptions of a fixed biomass density and price provide a useful analytical simplification but at the sacrifice of an important economic relationship – potential biomass suppliers will respond to price incentives. Recent landowner surveys provide evidence that landowners are willing to allocate more land for biomass production as the price of biomass increases. Further, landowners are heterogeneous in the price at which they are willing to supply biomass even under relatively uniform production conditions (Bergtold, Fewell, & Williams, 2011; Jensen, et al., 2007; Tyndall, Berg, & Colletti, 2011; Qualls, Jensen, English, Larson, & Clark, 2011; Menard, Jensen, Qualls, English, & Clark, 2011). This leads to one of the basic hypotheses of this paper: heterogeneity between and within local biomass markets will create economic trade-offs with important impacts on the potential supply, distribution, and economics of cellulosic biofuel. The rationale underlying this hypothesis is that heterogeneity will create significant variation in the cost-minimizing production decisions across locations. To test this hypothesis, we apply the theoretical model to U.S. switchgrass-based ethanol production, relaxing the assumption of fixed biomass density and price. 8