A decision support system for biorefinery location & logistics Except where reference is made to the work of others, the work described in this thesis is my own or was done in collaboration with my advisory committee. This thesis does not include proprietary or classi ed information. Sujith Sukumaran Certi cate of Approval: Harry T. Cullinan Professor Chemical Engineering Kevin R. Gue, Chair Associate Professor Industrial and Systems Engineering Robert Bul n Technology Management Professor Industrial and Systems Engineering Burak Aksoy Research Fellow III Chemical Engineering George T. Flowers Dean Graduate School A decision support system for biorefinery location & logistics Sujith Sukumaran A Thesis Submitted to the Graduate Faculty of Auburn University in Partial Ful llment of the Requirements for the Degree of Master of Science Auburn, Alabama August 10, 2009 A decision support system for biorefinery location & logistics Sujith Sukumaran Permission is granted to Auburn University to make copies of this thesis at its discretion, upon the request of individuals or institutions and at their expense. The author reserves all publication rights. Signature of Author Date of Graduation iii Vita Sujith Sukumaran, son of Sukumaran and Geethabai, was born on May 26th, 1985, in Coimbatore, India. He graduated from Ramnagar Suburban Matriculation Higher Sec- ondary School in 2002. He attended Sri Krishna College of Engineering and Technology, a liated to Anna University, Coimbatore, where he earned his Bachelor of Engineering Degree in Mechanical Engineering in 2006. He then enrolled in the Graduate School at Auburn University, Department of Industrial and Systems Engineering, in January 2007. iv Thesis Abstract A decision support system for biorefinery location & logistics Sujith Sukumaran Master of Science, August 10, 2009 (B.E., Anna University, 2006) 87 Typed Pages Directed by Kevin Gue The use of forest biomass, a renewable resource, as a source of fuels or chemicals is hindered by logistics. Economic success of the bioenergy concepts and their products may also depend on the solution to the logistics problem. A decision support system (DSS) has been developed to identify locations for biore neries in the state of Alabama. The DSS is comprised of two models, a location and an economic model. In the location model, biore nery location and the catchment area for the biomass are identi ed in such a way that it will incur the least transportation cost. It also selects the type and number of equipment needed for the rst three stages of the supply chain such as loading, transport from forest site to road site and preprocessing. In the economic model, the DSS uses the cost outputs from the location model and cost inputs from the user to calculate the investment and the rate of return on investment. The user has initially to choose from a list of biomass conversion technologies which are to be used in the biore nery. The DSS supports decision makers in analyzing the biomass supply, estimating the pro tability of investments, and evaluating necessary investments in infrastructure and equipment for biore neries. v Acknowledgments Firstly, I o er my sincerest gratitude to my advisor, Dr. Kevin Gue, who has supported me throughout my thesis with his patience and knowledge whilst allowing me the room to work in my own way. I attribute the level of my Masters degree to his encouragement and e orts and without him this thesis, too, would not have been completed or written. I also wish to thank Dr. Burak Aksoy who has taught me all the basics of the biore nery process. His recommendations and suggestions from the preliminary to the concluding level have been invaluable for the development of the project and for software improvement. I would like to thank Dr. Harry Cullinan and Dr. Robert Bul n for their invaluable help during this study and for serving on my committee. Thanks also to Dr. Muehlenfeld of the School of Forestry and Wildlife Sciences for assistance in describing the practical aspects of biomass transportation. Finally, I would like to thank my family and friends for all their support and understanding without which I could have never completed this program. vi Style manual or journal used Latex: A Document Preparation System : User?s Guide and Reference Manual, by Leslie Lamport (together with the style known as \aums"). Computer software used to build this decision support system is Microsoft Excel 2007. The decision support system contains a written document, software and a user?s manual. To get a copy of the decision support system, contact Dr. Kevin Gue vii Table of Contents List of Figures x 1 Introduction 1 1.1 Background . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1 1.2 Biomass as a renewable energy . . . . . . . . . . . . . . . . . . . . . . . . . 1 1.3 Economics of a biore nery . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3 1.4 Biomass supply chain . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4 1.5 Research questions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6 1.6 Thesis organization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6 2 Facility Location Model 8 2.1 Data sources for the model . . . . . . . . . . . . . . . . . . . . . . . . . . . 9 2.1.1 Forest resources . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9 2.2 Assumptions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12 2.3 Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12 2.4 E ect of centroids on facility location . . . . . . . . . . . . . . . . . . . . . 21 3 Refinery technologies 22 3.1 Gasi cation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23 3.2 Lignocellulose to ethanol conversion re nery technologies . . . . . . . . . . 26 3.2.1 Simultaneous sacchari cation and fermentation . . . . . . . . . . . . 26 3.2.2 Dilute sulphuric acid hydrolysis and fermentation . . . . . . . . . . . 27 3.2.3 Integrated fast pyrolysis and fermentation . . . . . . . . . . . . . . . 28 3.3 Fast pyrolysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29 4 Economic analysis of refinery technologies 31 4.1 Assumptions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31 4.2 Capital Investment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32 4.3 Revenue Estimation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33 4.4 Operating or Production costs . . . . . . . . . . . . . . . . . . . . . . . . . 35 4.5 Economic Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37 5 Analysis 40 5.1 Impact of facility location on IRR . . . . . . . . . . . . . . . . . . . . . . . 40 5.2 Impact of product prices on IRR . . . . . . . . . . . . . . . . . . . . . . . . 45 5.3 Impact of diesel prices . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49 5.4 Impact of woody biomass buying cost . . . . . . . . . . . . . . . . . . . . . 50 viii 6 Decision Support System 52 6.1 Welcome screen . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52 6.2 Input screen . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53 6.3 Back-end screens . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54 6.3.1 Model screen . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55 6.3.2 Equipment selection screen . . . . . . . . . . . . . . . . . . . . . . . 56 6.3.3 Analysis Screen . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57 6.4 Result Screen . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57 7 Conclusions 59 Bibliography 65 Appendices 66 A Manual for using the decision support system 67 ix List of Figures 1.1 Summary of Energy Consumption . . . . . . . . . . . . . . . . . . . . . . . 3 2.1 Total forest residues available in Alabama (Muehlenfeld, 2003) . . . . . . . 10 2.2 Total mill residues available in Alabama (Muehlenfeld, 2003) . . . . . . . . 11 2.3 ten biomass rich counties in Alabama. . . . . . . . . . . . . . . . . . . . . . 15 2.4 Results for p = 1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16 2.5 Results for p = 2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17 2.6 Results for p = 3 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18 2.7 Results for p = 4 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19 2.8 Results for p = 5 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20 3.1 High pressure gasi er process ow chart (Craig and Mann, 1996) . . . . . . 25 3.2 FT fuels production process ow chart . . . . . . . . . . . . . . . . . . . . . 25 3.3 SSF process ow chart (So and Brown, 1999) . . . . . . . . . . . . . . . . . 27 3.4 Acid hydrolysis process ow chart (So and Brown, 1999) . . . . . . . . . . . 28 3.5 Integrated fast pyrolysis process ow chart (So and Brown, 1999) . . . . . . 29 3.6 Fast pyrolysis block ow diagram (Ringer et al., 2006) . . . . . . . . . . . . 30 4.1 Economic Model (Peters et al., 2003) . . . . . . . . . . . . . . . . . . . . . . 38 5.1 Impact of facility location on IRR . . . . . . . . . . . . . . . . . . . . . . . 41 5.2 Impact of facility location on transportation cost . . . . . . . . . . . . . . . 42 5.3 Impact of electricity price per MW on IRR . . . . . . . . . . . . . . . . . . 45 5.4 Impact of ethanol price per gal on IRR . . . . . . . . . . . . . . . . . . . . . 46 x 5.5 Impact of FT diesel price per gal on IRR . . . . . . . . . . . . . . . . . . . 47 5.6 Impact of Bio{Oil price per gal on IRR . . . . . . . . . . . . . . . . . . . . 48 5.7 Impact of Diesel price per gal on transportation cost . . . . . . . . . . . . . 49 5.8 Impact of Diesel price per gal on IRR . . . . . . . . . . . . . . . . . . . . . 50 5.9 Impact of Wood buying cost on IRR . . . . . . . . . . . . . . . . . . . . . . 51 6.1 Welcome screen snapshot . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52 6.2 Input screen snapshot . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54 6.3 Model screen snapshot . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55 6.4 Equipment selection screen snapshot . . . . . . . . . . . . . . . . . . . . . . 56 6.5 Sample economic analysis screen of bio-oil re nery . . . . . . . . . . . . . . 57 6.6 Result screen snapshot . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58 A.1 Welcome Screen Snapshot . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67 A.2 Input Screen Snapshot . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68 A.3 Result Screen Snapshot . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69 A.4 Model Screen Snapshot . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70 A.5 Decision Variables matrix Snapshot . . . . . . . . . . . . . . . . . . . . . . . 71 A.6 Constraints and Objective function Snapshot . . . . . . . . . . . . . . . . . 71 A.7 Model Screen Snapshot with premium solver dialog box . . . . . . . . . . . 72 A.8 Combinations screen snapshot . . . . . . . . . . . . . . . . . . . . . . . . . . 73 A.9 Data from FoRTS v5 snapshot . . . . . . . . . . . . . . . . . . . . . . . . . 73 A.10 Equipment selection screen snapshot . . . . . . . . . . . . . . . . . . . . . . 74 A.11 Sample economic analysis screen of bio-oil re nery Snapshot . . . . . . . . . 75 xi Chapter 1 Introduction 1.1 Background Global energy demand is expected to grow at 1.3% per year on average until 2030 (The outlook for energy, 2007). This increase will be due to economic and population growth. At the same time, signi cant energy e ciency gains will help us to balance overall demand increases. The United States economy depends on the continuous availability of low cost energy. This dependence on lower cost energy has assumed increased signi cance in the current economic and political environment. More than 85% of energy consumed in 2005 in the United States was from fossil fuels, namely coal, petroleum, and natural gas. Less than 7% of energy was from renewable resources, of which, alcohol fuels was a mere 0.4% (Department of Energy, 2006). In 2008, the United States consumed 20.68 million barrels of petroleum products per day (about 7.5 billion barrels per year). Roughly 58% (4.35 billion barrels) of petroleum consumed was imported, with about 13% ( 0.9 billion barrels) coming from Persian Gulf countries (Annual energy outlook, 2006). The widespread use of fossil fuels and the resulting release of greenhouse gasses have been blamed for global warming, increasing sea levels and changing climatic patterns. This has led to a renewed global interest in fuels from biological sources, primarily because they are usually net zero contributors to greenhouse gasses. 1.2 Biomass as a renewable energy Renewable energy is the energy generated from natural resources which are renewable, or naturally replenished, such as solar, wind, hydro and biomass. According to National Re- newable Energy Laboratory, biomass is de ned as any plant-derived organic matter which 1 includes herbaceous and woody energy crops, agricultural food and feed crops, agricultural crop wastes and residues, wood wastes and residues, aquatic plants, and other waste ma- terials including some municipal wastes. Also woody biomass is de ned as/: \the trees and woody plants, including limbs, tops, needles, leaves, and other woody parts, grown in a forest, woodland, or rangeland environment, that are the by-products of restoration and hazardous fuel reduction treatments" (U.S.Department of Agriculture, Department of Energy, and Department of the Interior, 2003). Biofuels are liquid, solid, or gaseous fuels derived from renewable biological sources and can be burned directly for thermal energy or converted to other high-value energy sources including ethanol, biodiesel, methanol, hydro- gen, or methane. There is a history of biomass being used as fuel for thousands of years and it is today a major fuel used worldwide. Biomass is also the only alternative way to obtain liquid fuels that are used today compared to other forms of renewable energy. Figure 1.1 shows the present state of renewable energy used in the United States, with renewable en- ergy representing only 6% of the total, and biomass representing a little above 2.5%. In an e ort to push forward greater utilization of renewable energy, the federal government through the Department of Energy has put forth benchmark biomass initiative goals for 2020 which are to have 5% of all power, 10% of all fuels, and 18% of all bioproducts being supplied by biomass and serving as replacements for what otherwise would be fossil fuel expenditures (DOE, 2002). 2 Figure 1.1: Summary of Energy Consumption Allen et al. (1998) has categorized biomass fuels into four groups: 1. Wood (such as forest fuel available after felling of trees, thinning of forests), 2. Crop residues, 3. Dedicated energy crops grown speci cally to be used as fuel, and 4. Urban wastes (human and animal excretement). There are re nery technologies to convert any type of biomass to energy. According to Allen et al. (1998), primary biomass fuels can be used directly or can be converted into secondary fuels such as liquid or gaseous fuels through the use of various technologies. In our research we have concentrated on woody biomass. 1.3 Economics of a biore nery The major constraint to the use of biomass fuels in the U.S. is economics| the cost of producing biofuels is higher than that of the wholesale price of fossil fuels. Biofuels 3 competitiveness depends on the wholesale price of the fossil fuels. Researchers are trying to develop an understanding of the economics of the biore nery processing because it is crucial in realizing eventual commercialization. Constructing and operating a new biore nery requires a commitment of large amount of money. The decision to make such a commitment is based upon many factors, one of which is the prices of the products obtained from the biore nery. Apart from product prices, other factors in uencing the decision making process are installation cost, operating costs, supply chain costs, market position, health, safety and environmental concerns. Biore neries require a large capital investment even before it can be put into operation. The various costs estimates associated with erecting and operating a biore nery are as follows: capital investment, raw material costs, supply chain costs, labor costs, overhead costs and general expenses. Revenue for the biore nery comes from sale of the products produced by the plant. These estimates become the data for evaluating the economical consequences of the project. Rate of return on investment is used as a measure of estimating pro tability in the economic analyzes. 1.4 Biomass supply chain The steps which control the biomass supply chain are as follows: 1. Harvesting the feedstock in the eld or forest. 2. Handling and transporting the biomass from the eld to a point where road transport vehicles can be used (First stage transport). 3. Storage of biomass; Biomass is harvested at speci c times of the year hence it is to be stored to ensure a year-round supply to the biore nery. 4. Preprocessing; This is generally done to improve the handling e ciency and quantity during transport. 4 5. Transportation; to transport the fuel from the collection site to the biore nery (Second stage transport). Among these, transportation has been identi ed as one of the largest cost contributors to the cost of biomass feedstock (Hess et al., 2007; Allen et al., 1998; Bhat et al., 1992; Kumar et al., 2004). The logistics of biomass fuel supply are complex and problematic, and logistics costs will have an important bearing on the total delivered cost of biomass (i.e. the total cumulative cost of biomass fuel at the point of delivery to a power station). For example, the economic competitiveness of cellulosic ethanol production is highly dependent on feedstock cost, which constitutes 35-50% of the total ethanol production cost, depending on various geographical factors and the types of systems used for harvesting, collecting, preprocessing, transporting, and handling the material (Hess et al., 2007). The logistics associated with moving the biomass from the land to the biore nery can make up 50-75% of those feedstock costs, however, logistical costs exceeding 25% of the total biomass value leave very little pro t margin for biomass producers and biore nery operators (Hess et al., 2007). Sokhansanj et al. (2006) provides a few factors which they believe are most important while designing a supply chain for biore nery: The maximum rate of biomass supply to biore nery, Form and bulk density of biomass, The distance biomass has to travel to reach to biore nery, and Transportation infrastructure available between the points of harvest and biore nery. Biomass has lower energy density and physical density than other fossil fuels. Fiedler et al. (2007) have classi ed the operations taking place during preprocessing. These operations in uence the attributes of the transported, stored and utilized biomass products. The low bulk density of most biomass fuels reduces the tonnage capacity of trucks, resulting in larger truck movements. Mani et al. (2006) has shown that the pelletizing operation increases the 5 bulk density of the biomass and also improves the ease of transportation. Each of these factors are responsible for high transportation and logistics costs in biomass supply. Because transportation is one of the major costs involved in the economics of biore- nery, we can say that it partially determines the feasibility of biore nery in the state. A reduction in transportation cost considerably increases the pro tability of the investment which, in turn, increases the rate of return on investment. Transportation cost can be minimized by optimizing the location of the biore nery. 1.5 Research questions The most important question we are trying to answer with this research is, \Can a biore nery be located pro tably in the state of Alabama?", other questions are: Where will the biore nery be located in the state? What is the region of supply to the biore nery? How much of feedstock is supplied by each of the counties in the supply region? What is the transportation cost incurred? How much capital investment is required for the biore nery? What is the estimated rate of return on investment? 1.6 Thesis organization The decision support system has two parts, a facility location model and a model for economic analysis. The facility location model is rst explained, followed by the economic analysis and then how they both are integrated in Excel. We also perform some experiments using the decision support system to examine the e ects on prices of various products and technologies. The use of these sensitivity analyzes, although restricted in scope, provides a framework for evaluating future development and identifying critical areas that need to 6 be addressed before commercial success can be assured. With a focused study, di erent component designs can be evaluated, the performance of the system responses can be ob- served and economic impact assessed. Though this thesis is devoted to the single state of Alabama, the framework and the concepts developed should be transferable to any other state of similar or lesser size and readapted for newer environment and management policies. 7 Chapter 2 Facility Location Model Facility location problems are a special case of optimization problems solved by oper- ation researchers. The objective of the problem is to locate a facility with \Minimal Cost" while satisfying all the constraints. There are many ways of classifying the location mod- els. Daskin (2008) has classi ed location modeling into four types: analytical, continuous, network, and discrete. Daskin (2008) further classi ed discrete location models into three broad areas: covering based models (set covering, maximum covering, p-center), median based model (p-median, xed charge) and other models. In covering based models, the objective is to cover critical distance or time and demands within them and to be served in order to count them as ?covered?. Examples include locating re stations, emergency vehicles bases, and so on. For a set covering model, the objective is to cover all the demand points with a minimum number of facilities. In the maximum covering model, the objective is to cover maximum demand with a xed number of facilities. In the p-center model, we seek the smallest possible coverage distance so that every node is covered. In median based models, the objective is to minimize the demand-weighted distance between the facility and the demand nodes. Such models are widely used in distribution planning. For the p-median model, the objective is to minimize the average distance between demands and the nearest p sites. An uncapacitated xed charge model is an extension of the p-median problem with capacities on facilities. Our problem is similar to the p-median problem. 8 2.1 Data sources for the model 2.1.1 Forest resources Alabama has rich forest resources and is home to many companies in the pulp and pa- per industries. The processing of forest resources in these industries generates a signi cant amount of biomass. These secondary forest residues constitute the major portion of the biomass that is in use today. Muehlenfeld (2003) reports the total forest resources available in the state of Alabama. We use this data in our model for determining the facility loca- tion and supply counties. Muehlenfeld (2003) categorizes the woody biomass available in the state of Alabama into three categories: standing woody biomass, forest residues, and manufacturing residues. Each category includes di erent types of wood, which have their own attributes and challenges. It is important to know the maximum volume of biomass available to us to determine the location of biore neries. In our decision support system we have incorporated all woody biomass available except for the standing woody biomass. Standing forest biomass inventory is de ned as the dry weight of all wood and bark above a one-foot stump in all live trees that are 1 inch or greater in diameter at breast height and located on commercial forest land (Muehlenfeld, 2003). This volume of biomass is generally not available for biore nery use because it is used for production in the pulp and paper industries. Hence, the other two categories (forest residues and manufacturing residues) are the actual fuel available for use in a biore nery. Forest residues Forest residues constitute both the logging residues and cull trees. Logging residues are the potential leftovers from harvesting operations, such as crowns, limbs and unused portions of growing stock trees. In Alabama, approximately 2.6 million oven-dry tons of logging residues can be recovered annually. Logging residues can be easily obtained using whole tree chipping operations. The operations produce \dirty chips" whose economics are well understood in practice (Muehlenfeld, 2003). Throughout our model, we assume that 9 the re nery uses feedstock in the form chips. Another source of forest residues is the cull, or rough trees, which are the trees that do not have any value other than their potential as biomass fuel. It is estimated that there are approximately 2.7 million dry tons available annually in Alabama. Figure 2.1 displays the distribution of forest residues in Alabama available annually. Figure 2.1: Total forest residues available in Alabama (Muehlenfeld, 2003) 10 Manufacturing residues Figure 2.2: Total mill residues available in Alabama (Muehlenfeld, 2003) Forest industries are categorized into two sectors, primary manufacturing and sec- ondary manufacturing. Primary manufacturing sectors use wood coming from the forest directly for operations. Secondary manufacturing sectors use products from the primary sector as raw materials and adds value to it. Both the sectors produce waste materials 11 which have potential for use as fuel in a biore nery. According to the Alabama forestry commission, 99% of the primary manufacturing residuals are used for other purposes like fuel, ber for pulp etc., but the availability of these residuals for the biore nery greatly de- pends on the economics. Even though secondary manufacturing residues are comparatively of lesser volume than primary residuals, it is estimated that Alabama produces slightly more than one-half million oven-dry tons per year. Figure 2.2 displays the distribution of manufacturing residues in Alabama available annually. 2.2 Assumptions We assume that the locations from which the feedstock are taken are the centroids of the particular counties. The distance matrix dij is obtained by calculating the round trip distance between centroids of counties. The actual road distances between the centroids were obtained using Google Earth. The model also assumes that the same cost is involved in transporting all types of feedstock which is not the case practically. The mode of transport for the second stage of transportation is assumed to be trucks. 2.3 Model We de ne the following notation: Indices i;j { Indices denoting locations { i is the re nery location, j is the counties of raw material procurement. Decision Variables xi{ Equals 1 if a factory is located in county i and 0 otherwise. yij{ Proportion of feedstock available in county j used for the re nery in county i. 12 Parameters 1. aj { Availability of forest residues in county j in dry tons. 2. bj { Availability of mill residues in county j in dry tons. 3. R { Requirement for feedstock in the re nery in dry tons. 4. p { Number of facilities to be located. 5. n { Number of counties (67 for Alabama). 6. Cf { Cost per ton for the rst three stages of the biomass supply chain in $/ton. 7. Cij { Transportation cost to move feedstock from county j to re nery in county i in $/ton-mile. 8. dij { Centroid distance between county j and county i in miles. 9. f { Percentage of forest residues in county that can be used as a feedstock for re nery. 10. m { Percentage of mill residues in county that can be used as a feedstock for re nery. Our objective is to Minimize nX i=1 nX j=1 [ faj + mbj]Cij dij yij Subject to nX j=1 [ faj + mbj]yij Rxi8i (2.1) nX i=1 xi = p8i (2.2) nX i=1 yij 18j (2.3) 0 yij 18i;j (2.4) xi 2 f0;1g8i (2.5) 13 The objective is to minimize the total variable cost of transportation. Constraint set 2.1 ensures feedstock requirements are met at the biore neries. Constraint set 2.2 requires the number of facilities to be located to be equal to p. Constraint set 2.3 ensures no county provides more than its available supply. The model developed here is a mixed integer problem which is complex to solve. The di culty arises from the fact that integer programming problems have many local optima and nding a global optimum to the problem requires one to prove that a particular solution dominates all feasible points. Premium solver 8.0 developed by Frontline systems is capable of solving mixed integer problems and is also compatible with Microsoft Excel. Hence, we build the decision support system in this widely used application, Microsoft Excel 2007, with addin Premium solver 8.0. We applied this model to the state of Alabama. We calculated the density of biomass available in each of the counties of the state. We obtained the ten richest counties in terms of density of biomass available for use in the biore nery and plotted them in a map as shown in Figure 2.3. As we can see in the Figure, the biomass rich counties are located in some counties in the center, in the southwest and central east of the state. Intuitively, we would predict the biore neries to be located somewhere close to these areas. For example, to nd out how the model behaves, we tried to locate a biore nery of the capacity of 550 tons/day. We also assumed that the percentage of forest and mill residues ( ?s) available from each of the counties to be 60%. The results of the model point to Chilton county as the optimal site for locating this biore nery. The results also indicate that the re nery uses 66% of the total biomass resources available from Chilton county, and none from surrounding counties. We can see from Figure 2.3 of the density plot that Chilton is among the top ten biomass rich counties in Alabama. The results from the model are plotted in Figure 2.4. 14 1 2 5 6 4 8 10 7 3 9 Figure 2.3: ten biomass rich counties in Alabama. With the same set of inputs, we increased the number of re neries to 2. The result of the model points to the optimal locations of Lamar and Russell counties. The re nery located in Lamar county uses 100% of its available biomass and also 24% of the biomass available from Fayette county. The re nery located in Russell county uses 96% of the total biomass available from itself. Both Lamar and Russell counties are among the top ten 15 Facility Location p = 1 Transp. Cost = $5,535,348 Figure 2.4: Results for p = 1 biomass dense counties in Alabama. Another important point to note in the result is, how the model tries to spread the facility locations over di erent parts of the state in order to make use of the biomass potential available in those areas rather than concentrating it to one part of the state. Figure 2.5 shows the results of the model. 16 Facility Location p = 2 Transp. Cost = $14,609,639 Figure 2.5: Results for p = 2 For p value equal to 3, the model chose Bibb, Lamar and Russell counties as opti- mal facility locations. The results follow the same pattern as the previous experiments. Figure 2.6 shows the optimal locations and the regions of supply for the re neries. 17 Facility Location p = 3 Transp. Cost = $30,419,970 Figure 2.6: Results for p = 3 For p value equal to 4, the optimal locations identi ed by the model are Butler, Lamar, Monroe and Russell counties. The re nery located in Lamar county uses 100% of its avail- able biomass and also 24% of the biomass available from Fayette county. The re nery located in Russell county uses total biomass available within it and also 91% of the biomass available from Lee county. The re nery located in Butler uses 100% of its available resource 18 and also gets the biomass supply from Crenshaw and Conecuh counties. 46% of the biomass supply from Conecuh county goes to the re nery located in Butler county and the rest goes to the re nery located in Monroe. The re nery located in Monroe county also gets 2% of the total biomass available from Clarke county. p = 4 Transp. Cost = $54,092,693 Facility Location Figure 2.7: Results for p = 4 For p value equal to 5, the model chose Lee, Butler, Monroe, Bibb and Lamar counties as optimal facility locations. The results follow the pattern as the previous experiments. 19 Figure 2.8 shows the optimal locations and the regions of supply for the re neries. The time taken to obtain an optimal solution by the model is in seconds for the values of p up to 3. It takes a couple of minutes to solve for values of p up to 5. The solution times are relatively quick and increase along with capacity and number of re neries. 0.10 0.26 0.61 0.95 p = 5 Transp. Cost = $86,603,780 Facility Location 0.07 Figure 2.8: Results for p = 5 20 2.4 E ect of centroids on facility location In the real world, the pick up points of the feedstock will be randomly located at di erent parts of the county. But in our model, we have assumed that all the feedstock is obtained from the centroid of the county. So, in order to justify the assumption that centroid to centroid distance does not have a signi cant impact in selecting the facility location, we performed some experiments using the distances. We created three sets of distances by using both the straight line distances (obtained using the great circle distance formula) and actual distances (obtained using Google Earth). They are as follows: Random distances were created by mutiplying distances with the expression 1+Rand(- 0.1,0.1) i.e randomdij = dij 1 + Rand( 0:1;0:1 ) Second set of distances(1:1dij) were obtained by increasing 10% of the distances. Third set of distances (0:9dij) were obtained by decreasing 10% of the distances. We ran the model for these three di erent scenarios at two di erent levels of requirement as well as two di erent levels of raw materials availability ( ). The results were the same in terms of facility location and the supply regions for all the three sets of distances. There were variations in the total cost, as we would expect. 21 Chapter 3 Refinery technologies Now, we move to the economic model of the decision support system. Conversion technologies are available to convert any type of biomass into fuels. Signi cant attention has been given by researchers to develop conversion technologies which are feasible, low cost and less complex to operate. Though most of the re nery technologies are not available on a commercial scale, a few of them have been erected on a smaller scale for research purposes. Using the literature, we have performed an economic analysis for some of the re nery technologies for which data were available. This chapter deals with those re nery technology types which were used in the decision support system. They are as follows: Gasi cation for power production, Gasi cation followed by FT synthesis, Simultaneous sacchari cation and fermentation (SSF), Dilute sulphuric acid hydrolysis and fermentation, Integrated fast pyrolysis and fermentation, and Fast pyrolysis. There are three capacity levels for each of the re nery technologies. The capacity of the biore nery (tons of biomass required per day), were obtained from the literature (Craig and Mann, 1996; Ringer et al., 2006; Hamelinck et al., 2004; So and Brown, 1999). Additional capacities were created to make the decision support system more exible. One is obtained by doubling the base capacity. The other capacity is two-thirds of the base capacity. 22 3.1 Gasi cation Gasi cation is a manufacturing process that converts carbon-containing materials, such as coal, biomass, or various wastes to a syngas which can then be used to produce elec- tric power, fuels and other valuable products (Gasi cation technologies council, 2008).The production of syngas is due to the partial combustion of biomass and takes place at temper- atures of about > 700 C (Craig and Mann, 1996). The reactor in which this process takes place is called a gasi er. The syngas comprises primarily of hydrogen (H2), carbon monox- ide (CO), traces of methane and non useful products like tar and dust. There are several types of gasi ers available for commercial use today. Gasi cation has a number of signi - cant economic bene ts as it converts low-value feedstocks to high value products, thereby increasing the use of available energy in the feedstocks while reducing disposal costs. The ability to produce a number of high-value products at the same time (polygeneration) helps a facility o set its capital and operating costs (Gasi cation technologies council, 2008). In addition, the principal gasi cation byproducts (sulfur and slag) are readily marketable pro- viding additional revenues to the plant. The principal issues with biomass gasi cation is using biomass syngas in a gas turbine. The technologies for biomass syngas cleanup is still evolving and the system used today is highly expensive (Antares Group, 2003). A Biomass Integrated Gasi cation Combined Cycle (BIGCC) power plant combines a gasi cation system with the \combined cycle" electric power system (consisting of one or more gas turbines integrated with a steam turbine). The basic elements of a BIGCC power plant include a biomass dryer (fueled by waste heat), a gasi er for converting the biomass into a combustible fuel gas, a gas cleanup system, a gas turbine-generator fueled by combustion of the biomass-derived gas, a heat recovery steam generator (HRSG) to raise steam from the hot exhaust of the gas turbine, and a steam turbine-generator to produce additional electricity (Larson et al., 2001). The BIGCC con guration achieves the highest thermal-to-electrical e ciency of any commercial power generation technology on the market today (Antares Group, 2003; Rajvanshi, 1986; Larson et al., 2001). 23 Gas-to-liquid (GTL) technologies convert hydrocarbon feedstock, such as biomass, nat- ural gas or coal, into a FT syncrude, which is processed further into a range of liquid hy- drocarbon products (Wilhelm et al., 2001). The GTL process can be broken into three distinct phases: generation of syngas, F-T synthesis, and upgrading (Sousa-Aguiar et al., 2005). Syngas generation is done with the help of gasifaction systems explained earlier. Fischer-Tropsch is a method of converting syngas into hydrocarbon products. The syngas from the gasi er is fed into a F-T reactor and is reacted in the presence of an iron or cobalt catalyst, which converts it into a para n wax that is then upgraded (hydrocracked) to make a variety of products. The range of products possible to create from F-T includes light hydrocarbons, methane, ethane, lique ed petroleum gas, gasoline, diesel, and waxes. In our decision support system we have incorporated both BIGCC con guration and the GTL process. Craig and Mann (1996) have analyzed BIGCC con guration to determine commercial potential of gasi cation systems. They performed the studies in di erent types of BIGCC systems as mentioned below: 1. High pressure, air-blown gasi cation with an aero-derivative gas turbine. 2. High pressure, air-blown gasi cation with an advanced utility turbine. 3. Low pressure indirectly heated gasi cation with an advanced utility turbine. 4. Low pressure, Air-blown gasi cation with an advanced utility turbine. In our decision support system, we have also incorporated circulated uidized bed (CFB) gasi ers. Hamelinck et al. (2004) has analyzed the production of FT diesel from biomass. In this process, gasi cation is followed by FT synthesis. Instead of air (N2,H2,Water) additional oxygen is added to the gasi er because gasi cation with oxygen o ers bene ts in downstream equipment size, compression energy and higher partial pressures for rele- vant components in Fischer-Tropsch (FT) diesel (Hamelinck et al., 2004). After pretreated, biomass is passed through a gasi er. Biomass is gasi ed to produce synthesis gas (biosyn- gas). The gas is then passed through a compressor for tar removal, and it is cleaned of other 24 Figure 3.1: High pressure gasi er process ow chart (Craig and Mann, 1996) impurities. The composition is then modi ed to t the speci cations for the FT synthesis in the FT reactor. The reactor produces the FT o -gas which is then recycled or combusted to produce electricity. The liquid FT products are treated to produces variety of fuels. The steps involved in the process are shown below. Figure 3.2: FT fuels production process ow chart 25 3.2 Lignocellulose to ethanol conversion re nery technologies Extensive research has been done on conversion of lignocellulosic materials to ethanol especially after the 1980 oil crisis (Du and Murray, 1996; Sun and Cheng, 2002; Estegh- lalian et al., 1997; Sivers and Zacchi, 1995). The processing of lignocellulose to ethanol consists of four major unit operations: pretreatment, hydrolysis, fermentation, and prod- uct separation/puri cation (Mosier et al., 2005). Biomass is a mixture of lignin, cellulose and hemicellulose. The purpose of pretreatment is to remove the lignin and hemicellulose, reduce cellulose crystallinity and increase the porosity of materials (McMillan, 1994). This is required to alter the structure of biomass to make cellulose accessible to enzymes. Hy- drolysis includes the processing steps that convert carbohydrate polymers into monomeric sugars. During hydrolysis, hemicellulose polymers releases it component sugars, which are fermented to ethanol by microorganisms. Ethanol is then recovered from the fermentation broth by distillation. There are three basic types of ethanol from cellulose processes { acid hydrolysis and fer- mentation, enzymatic hydrolysis and fermentation, and thermochemical followed/preceded by fermentation (Badger, 2002). In our decision support system, we have incorporated one variation of each type. They are as follows: dilute sulphuric acid hydrolysis and fermenta- tion, Simultaneous Sacchari cation and Fermentation (SSF) and Integrated fast pyrolysis and fermentation. Each type is explained in detail in the following sections. 3.2.1 Simultaneous sacchari cation and fermentation Cellulose hydrolysis carried out in the presence of fermentative microorganisms is re- ferred to as simultaneous sacchari cation and fermentation (SSF). So and Brown (1999) have considered an SSF process which has been built into the decision support system. In this process feedstock is fed into an acid prehydrolysis chamber where pretreatment takes place. It is pretreated with dilute sulphuric acid in this chamber to break lignin and cel- lulose. Then it is passed into the broth for hydrolysis and fermentation. Batch culture of 26 Trichoderma reesei is utilized for cellulase production (Hinman et al., 1992). During the hy- drolysis, three steps take place: adsorption of cellulase enzymes, biodegradation of cellulose to fermentable sugars and desorption of cellulase (Sun and Cheng, 2002). The fermentable sugars (pentose and xylose) generated by the hydrolysis are fermented using genetically engineered Escherichia coli (Hinman et al., 1992). After fermentation, ethanol is extracted by distillation. The processes involved are shown below in the owchart. Compared to Figure 3.3: SSF process ow chart (So and Brown, 1999) acid hydrolysis and fermentation, SSF has the following advantages: increased hydrolysis rate, lower enzyme requirement, higher product yields and shorter process time (Sun and Cheng, 2002). The disadvantages of SSF are incompatible temperature of hydrolysis and fermentation, ethanol tolerance to microbes and inhibition of enzymes by ethanol (Sun and Cheng, 2002). 3.2.2 Dilute sulphuric acid hydrolysis and fermentation There are two basic types of acid hydrolysis processes: dilute acid and concentrated acid (Badger, 2002). In our decision support system, we have used the process analyzed by Qureshi and Manderson (1995). In this process, heated dilute sulphuric acid at 180 C is used for the hydrolysis process on the feedstock. Dilute acid processes are conducted under high temperature and pressure, and have reaction times in the range of seconds or minutes, which facilitates continuous processing. During the hydrolysis process, hemicellulose is 27 broken down into its component sugars, pentose and hexose. Then these sugars are passed into the fermentation chamber where it is treated with microorganisms (strain of Candida shehatae) to ferment pentose and hexose to ethanol. After fermentation, ethanol is obtained by the process of distillation and membrane separation. By other auxiliary equipment, steam is also produced from the system which, in turn, is used to run a steam turbine to produce electricity. The electricity produced is used to run machines in the plant. Figure 3.4 explains the various steps in this process. Figure 3.4: Acid hydrolysis process ow chart (So and Brown, 1999) The advantage of acid hydrolysis and fermentation is that acids can serve both for pretreatment and hydrolysis. But the drawback of these processes are the cost of acids and the requirement to neutralize the acid after treatment to prevent production of inhibitory byproduct, furfural (Dale and Moelhman, 2000). 3.2.3 Integrated fast pyrolysis and fermentation So and Brown (1999) also analyzed an alternative approach to produce ethanol from lignocellulose by pyrolysis. They studied the Waterloo fast pyrolysis process developed at the University of Waterloo and Resource Transform International Ltd., in Ontario, Canada. The feedstock is fed into acid hydrolysis chamber for pretreatment. The biomass is treated with 5% sulphuric acid at about 80 90 C in the pretreatment process. After this step, a 28 part of the feedstock is passed through for fermentation. The other part of the feedstock is pyrolyzed in a pyrolyzer at 500 C and then it is passed for extraction of levoglucosan which, in turn, undergoes hydrolysis. After hydrolysis, it passed into the fermentation chamber. In the fermentation chamber, the hexose and pentose sugars from both the parts are fermented and converted into ethanol using two cultures of enzymes (Saccharomyces cerevisiae and Pichia stipitis). Ethanol is then removed by the process of distillation. Figure 3.5: Integrated fast pyrolysis process ow chart (So and Brown, 1999) 3.3 Fast pyrolysis There are lot of studies conducted on fast pyrolysis re nery types. We have used study conducted in NREL for the decision support system. Ringer et al. (2006) have studied the fast pyrolysis process used for the production of bio-oil. The process is composed of ve major processing areas: feed handling and drying, pyrolysis, char combustion, product recovery, and steam generation. In the feed handling section, the biomass is reduced in size to 1-5 mm and dried to 5%-10% moisture. It is then sent to pyrolysis where it is heated to 400 500 C in an oxygen-de cient atmosphere to degrade the biomass into a mix of gases, bio-oils, and char. Char is removed using high-e ciency cyclones and is combusted to fuel the pyrolysis 29 reaction. To maximize the yield of bio-oils, the reaction is rapidly quenched through heat exchange or direct liquid (e.g., water or recycled bio-oils) injection. The bio-oils are present in the gas stream as aerosols and require scrubbers and/or wet electrostatic precipitator for e cient capture. After cleaning, some of the clean pyrolysis gases are recycled to uidize the bed; and the remaining gases are combusted for process heat. Where feasible, heat is recovered from the pyrolysis gases to generate steam for electricity production. The gure below shows the processing areas. Figure 3.6: Fast pyrolysis block ow diagram (Ringer et al., 2006) We have incorporated all the above explained re nery technologies into our decision support system. Each re nery type has di erent products produced from them. Depending on the product the user is interested in, we have provided the user with the list of re nery technologies to choose from. We have performed economic analysis on each type of re nery technology which is explained in detail in the next chapter. 30 Chapter 4 Economic analysis of refinery technologies A plant construction will be undertaken only if it promises to be pro table. First, su cient capital must be promised for the project to bring up all aspects of the plant. It is essential for us to be aware of all the costs associated with operating the plant. In this chapter, investment and plant operation costs involved in all the re nery technologies are explained, as well as cash ow and gross and net pro ts. 4.1 Assumptions Data for this decision support system were obtained from published journals (Craig and Mann, 1996; Ringer et al., 2006; Hamelinck et al., 2004; So and Brown, 1999). Because each reference used di erent sets of assumptions, a common set of assumptions were developed so that the system could be built more robust and future technologies can be added with ease. Some of the general assumptions are tabulated in the following table. More speci c assumptions concerning costs and revenues are explained in their appropriate sections. 31 Economic Assumptions 2008 USD Technical lifetime = 25 years Economic lifetime = 20 years Interest rate = 10% Modi ed Accelerated Cost Recovery System (MACRS) depreciation Federal and State income tax = 40% Debt to Equity ratio = 0.5 Equity invested in two stages (50% each year) Plant operation hours = 8000 per year 330 operating days / year No Production in the rst two years of plant construction 50% production in the third year All the given production output is on the basis of 90% plant capacity except for the fermentation processes. Table 4.1: Assumptions for Economic Analysis 4.2 Capital Investment The rst major component in an economic analysis is the capital investment. Cap- ital investment is the total amount of money needed to supply the necessary plant and manufacturing facilities plus the amount of money required as working capital for opera- tion of the facilities. Capital costs for all the re nery technologies were estimated using combination of capacity, equipment based cost estimates, contingencies and fees. In the case of gasi ers and fast pyrolysis, the factored estimation method (percentage of delivered equipment cost) is used (Craig and Mann, 1996; Ringer et al., 2006). This method requires determination of the delivered-equipment cost. The other items included in the investment are then estimated as percentages of the total delivered-equipment cost. Capital costs for the lignocellulose to ethanol product re neries were assumed using the order-of-magnitude method. This method related the capital investment of a new plant to the capital invest- ment of a similar previously constructed plant by an exponential power ratio. This power has been found to average between 0.6 and 0.7 for many re neries (Peters et al., 2003). Table 4.2 shows the capital investment required for some of the plant used in the decision support system. 32 Re nery Capacity Capital Type (tons/day) Investment Integrated fast pyrolysis 800 $91,233,660 SSF 812 $84,622,525 Acid Hydrolysis 796 $88,589,206 High Pressure gasi er 683 $138,914,559 CFB gasi er 1848 $260,670,731 Fast pyrolysis 550 $51,016,723 Table 4.2: Capital Investment values for re nery types (Craig and Mann, 1996; Ringer et al., 2006; Hamelinck et al., 2004; So and Brown, 1999) 4.3 Revenue Estimation The second component of the analysis is the estimation of revenue. The revenue is generated from the sale of product, or products, by the plant. The total annual revenue from product sales is the sum of the unit price of each product multiplied by its rate of sales. A plant is designed for production of a major product, such as ethanol, or bio-oil, but along with it, additional secondary products are also produced like electricity. Rate of production of these secondary products is determined by the chemistry and operating characteristics of the re nery technology. Though these secondary products may not be huge in volume, they are still used and generate revenues for the re nery. These are also taken into account and added along with the total revenue. The other important thing to note in the economic analysis is that we assume no production for the rst two years of operation and 50% production in the third year because, during the start-up period, production rates are very low, the length of the start-up period is usually unknown as well as the year of start-up. Product prices are best established by market studies and historical data available. In our decision support system, we have obtained the product prices from Energy Infor- mation Administration (2009). The historical data is available in the Energy Information Administration (2009) regarding the prices of ethanol, diesel and electricity. Bio-oil and FT diesel are not available as commodities in the fuel market. Hence, the prices of bio-oil 33 and FT diesel are not available and has to be estimated. Bio-oil could be directly used as a replacement fuel for other types of fuels such as #2 and #6 fuel oil or even natural gas. The di erence between these fuels is the energy value (heating value) of the fuels. We use this energy di erence to estimate the prices. For example, the heating value of #6 heating oil is 153,000 btu/gal and for bio-oil it is 75,000 btu/gal which means #6 heating oil has 2.042 times more energy than bio-oil. Therefore to replace one gallon of #6 heating oil approximately 2.042 gallons of bio-oil is required to obtain the equivalent energy value. Then at equivalent energy value the bio-oil price will follow the same wholesale price of replacement fuel trend. So, we used the #6 heating oil selling price and divided it by 2.042 to obtain the bio-oil selling price. Similarly, we obtained the price of FT diesel as it is used as a replacement fuel for diesel. The heating value of diesel is 130,500 btu/gal whereas that of FT diesel is 124,675 btu/gal. Diesel has 1.047 times more energy than FT diesel. Hence, we obtain the price of FT diesel by dividing the current price of diesel by 1.047. In the decision support system, the user can enter the di erent values of product unit prices and evaluate the impact of the prices on the revenue and pro ts by doing sensitivity analysis. Another major source of revenue to biore nery is government incentives. Tax incentives and grants are made available to the energy industry as a result of public policy to promote clean technologies. These have been primarily promoted as a method of creating energy independence, reducing the pollutant levels associated with fossil fuels, or both. Signi cant government incentives are available for the biore nery operators to encourage entrepreneurs and companies to take up renewable energy projects. Initially such incentives are needed for the new biore nery markets to emerge. Muehlenfeld (2003) has identi ed that the investment returns of biore neries can be hugely a ected by these incentives. These incen- tives are considered to be revenue and are added to the cash ow. Listed below are some of the regulations which are used in the decision support system for some of the re nery technologies. 1. Section 45(a) of Energy Policy Act of 2005 \ Section 45(a) provides that the renewable electricity production credit for a taxable 34 year is 1.5 cents (adjusted for in ation) for each kilowatt hour of electricity that the taxpayer (1) produces from quali ed energy resources at a quali ed facility during the 10-year period beginning on the date the facility was originally placed in service." This regulation is applicable to the gasi er re neries which have their output as elec- tricity. 2. Energy Policy Act of 2005 (H.R. 6) \Small ethanol producer (Section 1345-1347): expands the de nition of a small ethanol producer to include plants of up to 60 million gallons per year capacity; and creates a production incentive of 10 cents per gallon on the rst 15 million gallons of ethanol produced each year." This regulation is applicable for the lignocellulose to ethanol conversion re neries. The newly introduced \Farm Bill 2008" has provided further tax credits to the biore- nery operators which we have not considered in our economic analyzes because the bill was passed in the later part of 2008. This bill provides ethanol producers with a tax credit of $1.01 per gallon of ethanol produced. 4.4 Operating or Production costs The third component of an economic analysis is operating costs. All expenses directly connected with the manufacturing operation or the physical equipment of a re nery are included in operating costs. These costs are divided into two classi cations for easy under- standing: 1. Fixed costs 2. Variable production costs Fixed costs are expenses which are independent of production rate. Expenditures for depreciation, property taxes, insurance, nancing (loan interest), and rent are usually classi ed as xed costs. These costs, except for depreciation, tend to change due to in ation. 35 Because depreciation is on a schedule established by tax regulations, it may di er from year to year, but it is not a ected by in ation. In our economic analysis, we have used the Modi ed Accelerated Cost Recovery System (MACRS) to calculate depreciation each year. We also assume debt to asset ratio of 0.5. Each year?s loan interest is paid along with some part of the principal. We have assumed the repayment period to be 25 years at 10% interest for the loan. All the other xed cost values were obtained from the literature of the respective technologies (Craig and Mann, 1996; Ringer et al., 2006; Hamelinck et al., 2004; So and Brown, 1999). The other cost which occurs in the beginning is the startup cost. It is assumed to be 10% of the total investment and is spent in the rst year. Variable production costs include expenses directly associated with the manufacturing operation. This type of cost involves expenditures for raw materials (including transporta- tion, unloading, etc.), direct operating labor, supervisory and clerical labor directly applied to the manufacturing operation, utilities, plant maintenance and repairs, operating supplies, laboratory supplies, royalties, catalysts and solvents. These costs are incurred for the most part only when the plant is operating, hence, the term variable costs. Careful consideration was taken to develop the variable costs in our decision support system. Expenditure on the raw materials were analyzed in detail. Depending on the biomass requirement of the biore nery, we can calculate the cost of biomass to be pur- chased annually. The rst three stages of the biomass are harvesting, rst stage transport and preprocessing. The United States Forest Service research unit has built a residue truck- ing model to help truck operators to calculate costs involved while transporting biomass (Forest Residues Trucking Model, 2005). The cost for the rst three stages of the biomass supply for the model were predicted using the data they have used in their FoRTS model. FoRTS model also provides the list of equipment needed for the various operations like loaders, containers, vans etc. Using this cost data, biore nery raw material requirement, and equipment information, an optimization model was created to choose the equipment in such a way that least cost per ton is spent on the rst three stages. It creates a set of various combinations of equipment needed for the operations and speci c requirements 36 of the re nery. We have a facility location model to calculate the variable transportation cost based on the requirement of re nery. The other variable costs like labor, royalties and catalysts were provided by their respective literature of the technologies (Craig and Mann, 1996; Ringer et al., 2006; Hamelinck et al., 2004; So and Brown, 1999). In ation e ects are calculated depending on the current in ation rate entered by the user. The costs are then recalculated for the subsequent years automatically. 4.5 Economic Model Figure 4.1 shows the economic model of a biore nery. The total capital investment required for the plant is assumed to occur as a lump sum in the rst three years from the start of construction. Cash ows into the re nery as dollars of income (Si) from the sales of products while the annual costs for operating the re nery, such as raw material cost and labor cost but not including depreciation, are shown as out ow costs (Co). These cash ows for income and operating expenses represents rate of ow in dollars per year. The di erence between the income and the operating costs (Si Co) is the gross pro t before depreciation charge. Depreciation (d) is subtracted as a cost before income tax charges are calculated and is reduced. The resulting gross pro t (Si Co d) is taxable. The income tax to be paid depends on the tax rate. In our model, we have assumed the tax rate ( ) to be 40% including the state taxes. The remainder after the income taxes are paid ((Si Co d)(1 )) is the net pro t and this is returned to the capital reservoir. When the depreciation charge (d) is added to the net pro t, this makes up the total cash ow. The total generated cash ow returned to the reservoir on an annual basis is Aj = (Sij Coj)(1 ) + dj where Aj is the cash ow from the project in the year j in dollars, Sij is the sales rate in the year j in dollars, Coj is the cost of operation in the year j in dollars, dj is the depreciation 37 charge for the year j in dollars and is the income tax rate. Figure 4.1: Economic Model (Peters et al., 2003) Using the same methodology, the cash ow for the subsequent years of the re nery is obtained. After calculating the cash ow for the whole life period, net present value of the re nery is calculated in order to take into account the time value of the money. Then the internal rate of return on the investment (IRR) is calculated. It is a rate used to measure 38 the investment worth or pro tability. It is de ned as the break-even interest rate, i , which equates the net present value of a project?s cash out ows to the net present value of its cash in ows. It is mathematically expressed as follows, NPV (i ) = A0(1 + i )0 + A1(1 + i )1 + A2(1 + i )2 + + AN(1 + i )N = 0 where N is the project life span. In the case of the biore nery we have assumed N to be 25 years. IRR is very important to investors to know whether they are investing in a pro table venture or not. Using this model, we have calculated the value of IRR for all the re nery types as well as di erent re nery capacities in each type. 39 Chapter 5 Analysis 5.1 Impact of facility location on IRR The Decision Support System (DSS) gives the user the optimum facility location, trans- portation cost and the supply counties; but it fails to capture some important aspects of the decision process such as real estate costs, quality of life considerations, and so on. To investigate the sensitivity of the IRR calculations on the location of the plant, we used the model to locate each of the six re nery types in every county. For this experiment, we chose the average capacity of six re nery technology types. They are integrated fast pyrolysis and fermentation (IFP) with a capacity of 800 tons/day, Simultaneous Sacchari cation and Fer- mentation (SSF) with a capacity of 812 tons/day, dilute acid hydrolysis and fermentation (ACH) type of re nery with a capacity of 796 tons/day, oxygen fed Circulated Fluidized Bed gasi er (CFB) with a capacity of 1848 tons/day, fast pyrolysis (FP) with a capacity of 567 tons/day and high pressure gasi er (HPG) with a capacity of 1467 tons/day. The product prices of diesel, ethanol and electricity for the experiments were obtained by taking the average of the prices for the past 5 years. The prices of bio-oil and F-T diesel are calculated using the averages of the corresponding fuels as mentioned in Chapter 4. Thus, the prices obtained are diesel price as $3 per gallon, ethanol price as $1.80 per gallon, F-T diesel price as $2.90 per gallon, bio-oil price as $0.91 per gallon and electricity price as $70 per MW. We also assumed the percentage of forest and mill residues ( ?s) available from each of the counties to be 60% and the in ation rate to be 3.1% for the analysis. We obtained the IRR values and transportation costs as results from the model for each of the counties. We then plotted the IRR of each county as shown in Figure 5.1. We did the same for the transportation costs and plotted them as shown in the Figure 5.2. We 40 Figure 5.1: Impact of facility location on IRR can see from the IRR plot that the di erence in IRR between the rst and last county is approximately 3%. This shows that the location of the biore nery does not signi cantly in uence the IRR. But considering the large amount of money invested in the biore nery, even 3% increase in IRR results in large savings. We can also clearly see from both the plots that certain counties are more favorable for locating a re nery than others. An IRR value of 10% is pro table but may not be su cient for an investment like this since the technology is considered as a high risk business. For a high risk business, marginal acceptance rate of return for investors is 24%-32% (Peters et al., 2003). From Figure 5.1, we can see that this criteria rules out locating all the six types of re neries at their average capacities. But the SSF re nery has an IRR value above 17% which could attract some investors as it falls within the range (16%-24%) of IRR for medium risk business. Also, increasing the capacity of the SSF re nery from 812 tons/day to 1624 tons/day increases the IRR value above 24% for the same set of inputs. This is important because we can now locate a biore nery of this type and capacity in Alabama based on logistics| the goal of this research. 41 Figure 5.2: Impact of facility location on transportation cost Using the output from the model, we also ranked the counties based on IRR for each re nery type. We obtained the average rank of each county for all the six re nery types and the results are shown in Table 5.1. We have sorted the results in the table based on the average rank of each county from the experiments. From the table, we can also see that based on the average rank, Lamar county is the best location in the state of Alabama for the six re nery types. The values in the ?number of times in top 10 column? of the table also reiterates the fact that, the counties with richest biomass in the state are the frequently chosen counties for re nery location. As such, the table also provides information regarding alternate locations available for each re nery type. 42 County Name IFP SSF ACH CFB FP HPG Average Rank No of times in top 10 Lamar 1 1 1 3 3 2 1.8 6 Lee 3 4 3 2 4 3 3.2 6 Russell 4 3 4 1 7 1 3.3 6 Bibb 2 2 2 5 6 4 3.5 6 Butler 6 6 6 4 12 5 6.5 5 Winston 7 7 7 8 5 9 7.2 6 Chilton 5 5 5 13 9 11 8.0 4 Talladega 8 8 8 10 11 7 8.7 5 Chambers 10 10 10 9 8 8 9.2 6 Marshall 9 9 9 16 1 17 10.2 4 Geneva 11 11 11 18 2 20 12.2 1 Escambia 13 13 13 7 31 10 14.5 2 Walker 14 14 14 19 13 19 15.5 Fayette 12 12 12 26 10 25 16.2 1 Crenshaw 15 15 15 25 14 18 17.0 Lawrence 22 21 22 14 15 16 18.3 Monroe 20 19 20 6 43 6 19.0 2 Pike 17 18 17 20 19 23 19.0 Clay 18 17 18 21 20 22 19.3 Pickens 16 16 16 23 25 24 20.0 Conecuh 26 25 26 12 21 15 20.8 Greene 19 20 19 27 22 21 21.3 Montgomery 25 26 25 24 17 26 23.8 Marion 23 22 23 30 28 29 25.8 Jackson 30 30 30 11 45 14 26.7 Bullock 31 31 31 36 16 33 29.7 Sumter 29 29 29 31 37 27 30.3 Barbour 33 33 33 28 26 30 30.5 Randolph 27 27 27 42 23 38 30.7 Macon 24 24 24 44 24 46 31.0 Mobile 36 34 37 15 52 13 31.2 Etowah 32 32 32 32 32 31 31.8 Perry 28 28 28 38 33 39 32.3 Table 5.1: Ranking of counties based on IRR for di erent re nery technologies 43 County IFP SSF ACH CFB FP HPG Avg. Rank No. of times in top 10 Colbert 21 23 21 60 18 55 33.0 Marengo 34 35 34 29 39 32 33.8 Clarke 38 36 39 22 54 28 36.2 Tuscaloosa 44 44 44 17 56 12 36.2 Wilcox 39 39 38 37 27 42 37.0 Cullman 42 42 42 33 30 37 37.7 DeKalb 37 38 36 45 40 34 38.3 Henry 40 40 40 43 34 40 39.5 St. Clair 41 41 41 39 36 41 39.8 Morgan 35 37 35 49 44 43 40.5 Autauga 46 46 46 34 38 36 41.0 Franklin 43 43 43 54 29 50 43.7 Dallas 48 48 48 40 41 44 44.8 Tallapoosa 50 50 50 35 51 35 45.2 Choctaw 51 51 51 48 35 51 47.8 Calhoun 47 47 47 53 47 52 48.8 Blount 53 53 53 41 53 45 49.7 Dale 45 45 45 59 48 56 49.7 Hale 49 49 49 58 42 57 50.7 Co ee 52 52 52 56 50 49 51.8 Je erson 54 54 54 46 58 48 52.3 Lowndes 55 55 55 47 49 53 52.3 Washington 56 56 56 50 46 58 53.7 Baldwin 59 59 59 51 61 47 56.0 Coosa 58 58 58 55 57 54 56.7 Cleburne 57 57 57 62 55 62 58.3 Covington 62 61 62 52 65 59 60.2 Madison 60 60 60 57 64 60 60.2 Cherokee 61 62 61 64 60 64 62.0 Elmore 63 63 63 63 59 63 62.3 Shelby 64 64 64 61 62 61 62.7 Limestone 65 65 65 65 66 65 65.2 Houston 66 66 66 66 63 66 65.5 Lauderdale 67 67 67 67 67 67 67.0 Table 5.2: Ranking of counties based on IRR for di erent re nery technologies contd. 44 5.2 Impact of product prices on IRR How does IRR change with product prices? To nd out, we used the model to locate one re nery of each type of product with the average capacity and in ation rate at 3.1%. Certain costs were kept constant in the following analysis on product prices such as woody biomass buying cost to be $30 and diesel price as $3 per gallon for the trucks. We also assumed the percentage of forest and mill residues ( ?s) available from each of the counties to be 60%. All the inputs were entered on a dry-ton basis. For example, we chose a high pressure gasi er type of re nery with a capacity of 683 tons/day. The product from this type of re nery is electricity. A graph was drawn by plotting price per MW against IRR as shown in Figure 5.3. In 2008, the average wholesale price of electricity per MW was $83 (Energy Information Administration, 2009). At that price, IRR for the re nery is 15.09% making this type of re nery a potential candidate to be located in the state. But the prices dropped signi cantly in 2009 and the current price is $45 per MW. We can see from the graph that operating this type of re nery at this price incurs a loss. Operating the re nery becomes pro table only after the prices crosses $52 per MW. Also, the pro ts obtained from this type of re nery are low compared to other re nery types in the state. 1% 6% 11% 16% 21% 26% $40 $45 $50 $55 $60 $65 $70 $75 $80 $85 $90 $95 $100 $105 $110 Electricity Price ($/MW) IRR Figure 5.3: Impact of electricity price per MW on IRR 45 To study the impact of changes in ethanol prices per gallon, we performed experiments to locate an SSF type of biore nery with a capacity of 812 tons/day. We have used the data obtained from So and Brown (1999) to build the DSS. So and Brown (1999) assumed that 93 gallons of ethanol can be produced from a dry-ton of biomass. This value is slightly higher compared to the re neries in practice. Due to processing and practical constraints, the ethanol obtained from one dry ton of biomass is usually 70-80 gallons. So, we varied the prices for all the three values of ethanol output and obtained the results. The results were plotted in a graph as shown in Figure 5.4. Historically, the prices of ethanol per gallon in the past years have been hovering between $1.80 and $2.50 (Energy Information Administration, 2009). When the price of ethanol touches $2.50, the re neries with ethanol yield of 93 gallons and 80 gallons per dry ton has IRR value more than 24% and for 70 gallons per dry ton the IRR value is 20.58%. Hence at that price, these re neries become attractive to the investors as it falls within the acceptance range for high risk businesses. The current price of ethanol per gallon in 2009 is $1.50, the IRR at this price is negative for 70 gallon per dry ton output and less than 10% for the other values of ethanol output making it less attractive for investors. 2% 7% 12% 17% 22% 27% 32% 37% 42% 47% 52% 57% 62% $1.00 $1.25 $1.50 $1.75 $2.00 $2.25 $2.50 $2.75 $3.00 $3.25 $3.50 $3.75 $4.00 Ethanol Price ($/gal) IRR 93 gal/dry ton 80 gal/dry ton 70 gal/dry ton Figure 5.4: Impact of ethanol price per gal on IRR 46 We can estimate the prices of FT diesel from the prices of diesel by equating their energy values as explained in Chapter 4. The prices of gasoline and diesel are highly variable around the world. For example, in July 2008 the price of diesel touched $4.70 per gallon (Estimated FT diesel price $4.53 per gallon) but the current price is just $2.10 per gallon (Estimated FT diesel price $2.02 per gallon). Hence, there are signi cant changes in the price and this a ects the IRR of the re nery. Therefore, to investigate the sensitivity of IRR on FT diesel prices, we used the model to locate a CFB gasi er type of re nery with a capacity of 1848 tons/day. The results were plotted on a graph. As expected, IRR was found to increase with increase in FT diesel prices. From the graph, we can see that IRR becomes positive between $1.60 and $1.65. Any value below $1.60 results in signi cant losses for the re nery operator. An increase in the price by $1 causes the IRR to increase by more 10%. Figure 5.5 shows how sensitive IRR is to FT diesel prices. -5% 0% 5% 10% 15% 20% 25% 30% 35% 40% $1.50 $2.00 $2.50 $3.00 $3.50 $4.00 $4.50 $5.00 F-T diesel Price($/gal) IRR Figure 5.5: Impact of FT diesel price per gal on IRR Bio-oil can be directly used as a replacement to #2 and #6 oil and can be upgraded to replace diesel and gasoline fuels. In order to study the impact on bio-oil prices, we used the model to locate a fast pyrolysis biore nery which has a capacity of 550 tons/day. The rest 47 of the inputs were kept constant and the results were obtained. Like the FT diesel prices, the price of bio-oil is estimated from the price of replacement fuels as explained in Chapter 4. The price of the bio-oil follows the same trend as the replacement fuels. Average price of #6 heated oil per gallon in the past ten years is $1.30 (Energy Information Administration, 2009) and the corresponding estimated price of bio-oil is $0.64 per gallon. Operating the re nery at this price incurs loss as shown in the graph. The re nery becomes pro table only after the prices cross $0.80 per gallon (Estimated #6 heated oil price is $1.64 per gallon). From the graph, we can also see that even a 15 cent rise in bio-oil prices increases IRR by approximately 10%. Also, the bio-oil obtained is a low value product and requires further upgrading, if they are to be used as a replacement for diesel and gasoline fuels (Huber et al., 2006). The upgrading process will increase the costs involved and may reduce the values of IRR. 10% 15% 20% 25% 30% 35% 40% 45% 50% 55% 60% $0.50 $0.65 $0.80 $0.95 $1.10 $1.25 $1.40 $1.55 $1.70 $1.85 $2.00 Bio-Oil Price ($/gal) IRR Figure 5.6: Impact of Bio{Oil price per gal on IRR 48 5.3 Impact of diesel prices As we have assumed the mode of transportation to be trucks, the diesel price has an impact on the transportation cost which, in turn, a ects facility location and IRR. In order to analyze the sensitivity of IRR, we used the same setup that was used in determining the impact of the facility location (Section 6.1) but varied the price of diesel. The diesel price a ects the transportation cost of the second stage of transport ($/ton-mile), from road side to biore nery, which is used in calculating variable transportation cost. We ran the model for all the six re nery types and obtained the values of variable transportation cost and IRR as results. The transportation cost increases with the increase in biomass requirement of the biore nery because of the increased amount of truck distances and movements. This is clearly shown in the Figure 5.7 by the di erence in transportation costs between the CFB gasi er (1848 tons/day) and the rest of the re nery types which have comparatively less requirement (< 850 tons/day). $5.00 $6.00 $7.00 $8.00 $9.00 $10.00 $11.00 $12.00 $13.00 $14.00 $15.00 $16.00 $17.00 $18.00 $19.00 $20.00 $21.00 $22.00 $2.00 $2.50 $3.00 $3.50 $4.00 $4.50 $5.00 M i l l i o n s Diesel Price ($/gal) Transportation Cost IFP SSF ACH CFB FP HPG Figure 5.7: Impact of Diesel price per gal on transportation cost 49 Figure 5.8 helps the user to understand how diesel price per gallon for trucks a ects the IRR. The diesel prices have no signi cant impact on the IRR as shown in the gure. This may be due to the fact that the variable transportation cost involved is small compared to large amount of capital invested in the plant. 8.00% 9.00% 10.00% 11.00% 12.00% 13.00% 14.00% 15.00% 16.00% 17.00% 18.00% 19.00% 20.00% $2.00 $2.50 $3.00 $3.50 $4.00 $4.50 $5.00 Diesel Price ($/gal) IRR IFP SSF ACH CFB FP HPG Figure 5.8: Impact of Diesel price per gal on IRR 5.4 Impact of woody biomass buying cost Increase in biomass buying cost increases the operating cost which, in turn, decreases the pro t. To study the impact on biomass buying cost, we used the same setup that was used in determining the impact of the facility location (Section 6.1) but varied the price of woody biomass buying cost. From Figure 5.9, we can see that the biomass buying cost is also one of the factors which a ects the IRR. Currently, woody biomass cost is approximately $30 per ton (Muehlenfeld, 2003). A reduction in price by $10 per ton from the current price increases IRR signi cantly making some of the re neries attractive to investors as it crosses the 16% mark used by some investors. Woody biomass prices have generally not shown the volatility of fossil fuel prices, but they do move up and down based on industrial conditions. 50 However, the price of woody biomass fuels have been in uenced by the prices for fossil fuels (Muehlenfeld, 2003). The value of woody biomass cost could further increase due to the growing markets for alternative uses of woody biomass residues (Muehlenfeld, 2003). 0.00% 2.50% 5.00% 7.50% 10.00% 12.50% 15.00% 17.50% 20.00% 22.50% $20 $25 $30 $35 $40 $45 $50 Woody Biomass buying cost ($/ton) IRR IFP SSF ACH CFB FP HPG Figure 5.9: Impact of Wood buying cost on IRR 51 Chapter 6 Decision Support System In the previous chapters, all the concepts behind the decision support system were explained. In this chapter, the actual user interface for the decision support system is discussed. The system was built using Microsoft Excel 2007 and the Premium Solver addin built by Frontline Systems. In this chapter, each screen in the system is explained in detail. 6.1 Welcome screen This is the initial screen which will be displayed to the user once the application is opened. In this screen, the title of the decision support system and its use is explained for the user. It also contains the set of instructions that the user has to follow to obtain the needed results. It contains the start button which the user has to click to initiate the process. Figure 6.1: Welcome screen snapshot 52 6.2 Input screen When the user clicks on the start button the input screen is displayed. The input screen is divided into three parts. They are, General information Raw material information Product information In the general information section, the user enters the number of re neries that he wants to locate in the state of Alabama. Then the user chooses the type of re nery technology which he is planning to use for the re nery from the drop down list. All the re nery technologies along with the capacity is displayed in the list. Each technology has three di erent capacities of the demand they require per day. Depending on the user, he can choose the size of re nery (small, medium, large) depending on the capacities. The other inputs required in this section are the current price of diesel per gallon and the current in ation rate. The user can perform sensitivity analysis to study the e ects of in ation and diesel prices on the pro tability by changing these values. In the raw material information section, the user has to rst select the type of data whether it is in wet tons or dry tons. The user has to then enter the percentage of availability of forest and mill residues ( f, m) to be used for biore nery. The user also enters the cost of obtaining the biomass. The user must also be aware of the moisture content of biomass when it is bought and should be entered into the system. Moisture content a ects the cost calculations because it limits the transportation capacity of the trucks. The other costs, wood procurement cost for the rst three stages of the biomass supply chain and transportation cost per ton-mile, in this section are automatically calculated based on rest of the data which has been entered into the system. The whole decision support system is built based on dry ton basis, so if the user enters the data in wet tons the system automatically converts them into dry ton basis for further calculations. 53 In the product information section, the user has to enter the selling price of the products per unit. This section gives the user the exibility to study the impact of product prices on the pro t. Depending on the user?s choice of re nery technology in the general information section, the user has to enter the product price per unit in this section. The user doesn?t have to enter all the other product information if the technology they have chosen doesn?t produce them. After entering all the required information, the user clicks the solve button. The program now enters into solving mode and results will be displayed once the problem is solved. Figure 6.2: Input screen snapshot 6.3 Back-end screens So far, all the screens which the user will be able to see were explained. The rest of the screens are hidden from the user. In the back end i.e. in these screens, the facility location optimization and economic analysis are done. We will see the important screens one by one. 54 6.3.1 Model screen In this sheet, the facility location model is built. The data that are already built into the sheet are listed below: Distances between counties Availability of forest residues Availability of mill residues Based on the input, the rest of the parameters required for the model are obtained. De- pending on the re nery type selected in the input sheet, the requirement will be lled in the model sheet automatically. Also, the availability matrices are calculated based on the percentages of forest and mill residues entered in the input. The cost matrix is then gener- ated with these matrices and the costs obtained from the input sheet. The constraints and the objective function are already built into this sheet and will recalculate automatically based on the matrices generated. When the solve button is clicked, all the new data cal- culations are performed and then the premium solver application is called by the program. The solver then solves the facility location optimization based on the data. The optimized transportation cost, location and the supply counties are thus obtained. Figure 6.3: Model screen snapshot 55 6.3.2 Equipment selection screen The rst three stages of the biomass supply chain resources optimization is done in this sheet. They include the in-woods loading, the rst stage transportation and the preprocess- ing resources optimization. Depending on the re nery type chosen by the user, the annual requirement of biomass is passed onto this sheet. Based on the requirement, the equipment used for the three stages are chosen from the pool of equipment. Data regarding all the equipment for the processes are obtained from the residue trucking model (FoRTS) built by US Forest Service research unit. Capacities, life, investment, costs involved and other miscellaneous costs are all obtained from their model and used in this program. Using this data, cost per ton mile for each equipment can be calculated. With the requirement known, all possible combinations of resources needed to satisfy the requirement is listed out and costs for each combination is calculated. Then the optimal set of equipment is selected from the list so as to minimize the cost associated with operating the machinery. All the above processes are initiated and run in a single module which gives the optimal set of equipment as output. Figure 6.4: Equipment selection screen snapshot 56 6.3.3 Analysis Screen The sheets in which economic analysis is done were grouped together to be called analysis screens. When the user chooses the re nery type and enters the cost data in the input sheet, they are transferred to the respective re nery technology sheet. Then the facility location model calculates the transportation cost which is also transferred to the sheet. The sheet already contains the rest of the costs involved which had been obtained from their respective literature (Craig and Mann, 1996; Ringer et al., 2006; Hamelinck et al., 2004; So and Brown, 1999). Cash ows are generated and rate of return of the investment is calculated. Figure 6.5: Sample economic analysis screen of bio-oil re nery 6.4 Result Screen The results from the analysis are displayed in three steps in this sheet. First, the results of the economic analysis are displayed which are investment cost, optimized transportation cost and the rate of return of the investment. In the second step, the optimized combination 57 of equipment needed for the rst three stages of the supply chain are listed. Along with the list of equipment, the number of equipment required for the hauling operation are also displayed. Finally, the facility location matrix is displayed. In the matrix, the optimized facility locations are displayed on the rows and the supply counties from which raw materials are obtained is displayed in the columns. The matrix is lled with proportion of raw materials obtained from each of the counties for the respective facility locations. Figure 6.6: Result screen snapshot 58 Chapter 7 Conclusions The goal of this research has been to build a decision support system to answer the question, \Can a biore nery be located pro tably in the state of Alabama?". The answer to this question is \Yes" for the processes we investigated under certain conditions like product prices, feedstock price and economies of scale. Optimizing the biore nery location by minimizing transportation cost also helps biore neries become more competitive. The DSS we have developed also provides information on location in order to minimize the transportation cost of a re nery. It also provides information on the amount of capital investment required for the project, revenues generated from the re nery, costs incurred in operating it and the rate of return on investment in the re nery. Though the decision support system gives the user the optimum facility location, cost and the supply counties; it fails to capture some important aspects of the decision making process such as real estate costs, quality of life considerations, and so on. Through our research we have identi ed that the biore nery locations have less impact (less than 3%) on the values of IRR than one might expect. But considering the large amount of money involved in the business, even small variations in IRR results in signi cant savings. We have also provided insights on the viable locations and the regions of supply for the re neries. We also found that the optimal locations of the biore neries are close to the biomass rich areas which would reduce the travel distances and in turn, the transportation cost. With the help of the model, we were able to identify the optimal re nery technology for each of the counties. We were also able to identify the list of counties which were best for each re nery type. The selling price of the products obtained from the biore nery is one of the major factors which makes a biore nery feasible or not. Through our research, we have identi ed 59 that product prices signi cantly a ect the IRR, which in turn, a ects the feasibility. The model helps to determine this feasibility and also captures the sensitivity of pro tability to product prices. For example, an increase in FT diesel price per gallon by $1 increases the IRR by 10%. Likewise, any drop in FT diesel price below $1.60 per gallon makes IRR negative, incurring losses. Similarly for bio-oil a small increase in price by 10 cents increases the IRR by 10%. It also helps us identify the threshold values of the product prices which makes the biore nery pro table. For example, a gasi er type of re nery with a capacity of 683 tons/day which has electricity as output will be pro table only if the price of electricity is greater than $52 per MW. Also, for re neries which have ethanol (yield of 93 gallons per dry-ton) as output the value is $1.25 per gallon. Similarly, a drop in the price of woody biomass buying cost by $10 makes most of the re nery types attractive to potential investors. The presented decision support system could help in strategic decision-making about locating and operating biore neries in the state of Alabama or with modi cations, in another state. It is particularly advantageous within the early stages of planning a biore nery plant. The user can: analyze the initial situation of the decision problem related to biore nery operation determine the optimal biore nery location which incurs the least transportation cost choose supply counties according to biomass availability and planning options estimate the pro tability of investments based on realistic assumptions regarding sup- ply chain costs adjust biore nery technology available and types of biomass resources estimate which investments in equipment are necessary and pro table plan capacities for processing, handling and transport. evaluate threshold values for product prices 60 react to di erent developments in the market. This is the rst such decision support system to be built for forest biomass resources to the best of our knowledge. Due to the exibility of the system, the planner can take current conditions into account each time they use the tool. The process of using this system leads to greater understanding of the processes and relationships. However, the accuracy and signi cance of the results depend on the available data and on the correctness of underlying estimations and assumptions. 61 Bibliography Allen, J., Browne, M., Hunter, A., Boyd, J., and Palmer, H. (1998). Logistics management and costs of biomass fuel supply. International Journal of Physical distribution & logistics, 28(6):463{477. Annual energy outlook, 2006 (2006). Annual energy outlook 2006. Technical report, Energy Information Administration. Antares Group, I. (2003). Assessment of power production at rural utilities using forest thinnings and commercially available biomass power technologies. Technical report, Pre- pared for the U.S. Department of Agriculture, U.S. Department of Energy, and National Renewable Energy Laboratory. Badger, P. (2002). Ethanol from cellulose: a general review. In Janick, J. and Whipkey, A., editors, Trends in new crops and new uses. ASHS Press. Bhat, M., English, B., and Ojo, M. (1992). Regional costs of transporting biomass feedstocks. Liquid Fuels From Renewable Resources; proceeding of an Alternate Energy Conference, 14-15 December 1992, Nashville, Tennessee, pages 50{57. Craig, K. R. and Mann, M. K. (1996). Cost and Performance Analysis of Biomass-Based Integrated Gasi cation Combined-Cycle (BIGCC) Power Systems. Technical report, Na- tional Renewable Energy Laboratory. Dale, M. C. and Moelhman, M. (2000). Enzymatic simultaneous sacchari cation and fer- mentation (ssf) of biomass to ethanol in a pilot 130 liter multistage continuous reactor separator. Ninth Biennial BioEnergy Conference, Newyork. 62 Daskin, M. S. (2008). What you should know about location modeling. Naval Research Logistics, 55(4):283{294. Du , S. J. B. and Murray, W. D. (1996). Bioconversion of forest products industry waste cellulosics to fuel ethanol: A review. Bioresource Technology, 55(1):1 { 33. Energy Information Administration (2009). Technical report. Esteghlalian, A., Hashimoto, A. G., Fenske, J. J., and Penner, M. H. (1997). Modeling and optimization of the dilute-sulfuric-acid pretreatment of corn stover, poplar and switch- grass. Bioresource Technology, 59(2-3):129 { 136. Fiedler, P., Lange, M., and Schultze, M. (2007). Supply Logistics for the industrialized use of Biomass{ Principles and Planning Approach. IEEE. Forest Residues Trucking Model (2005). Gasi cation technologies council (2008). Gas cation: Rede ning clean energy. Technical report, Gasi cation technologies council. Hamelinck, C. N., Faaij, P., Uil, H., and Boerrigter, H. (2004). Production of FT trans- portation fuels from biomass; technical options, process analysis and optimisation, and development potential. Energy, pages 1743{1771. Hess, J., Wright, C., and Kenney, K. (2007). Cellulosic biomass feedstocks and logistics for ethanol production. Biofuels, Bioproducts and Biore ning, 1:181{190. Hinman, N., Schell, D., Riley, J., Power, A., and Walter, P. (1992). Preliminary estimate of the cost of ethanol production for ssf technology. Applied Biochemistry and Biotechnology, 34-35:639{649. Huber, G. W., Iborra, S., and Corma, A. (2006). Synthesis of transportation fuels from biomass; chemistry, catalysts, and engineering. Chemical Reviews, 106(9):4044{4098. 63 Kumar, A., Cameron, J., and Flynn, P. (2004). Pipeline Transport of Biomass. Applied Biochemistry and Biochemistry, (27-40). Larson, E. D., Williams, R. H., and Leal, M. R. L. (2001). A review of biomass integrated- gasi er/gas turbine combined cycle technology and its application in sugarcane industries, with an analysis for cuba. Energy for Sustainable Development, 5(1). Mani, S., Tabil, L. G., and Sokhansanj, S. (2006). E ects of compressive force, particle size and moisture content on mechanical properties of biomass pellets from grasses. Biomass and Bioenergy, 30(7):648 { 654. McMillan, J. (1994). Pretreatment of lignocellulosic biomass. In Himmel, M., Baker, J., and Overend, R., editors, Enzymatic conversion of biomass for fuels production. American Chemical Society. Mosier, N., Wyman, C., Dale, B., Elander, R., Lee, Y., Holtzapple, M., and Ladisch, M. (2005). Features of promising technologies for pretreatment of lignocellulosic biomass. Bioresource Technology, 96(6):673 { 686. Muehlenfeld, K. J. (2003). Assessment of Wood-Based Syngas Potential for Use in Com- bined Cycle Power Plants in Alabama: A Guide For Economic Development Opportuni- ties. Technical report, Auburn University. Peters, M. S., Timmerhaus, K. D., and West, R. E. (2003). Plant design and economics for chemical engineers. McGraw-Hill Higher Education. Qureshi, N. and Manderson, G. (1995). Bioconversion of renewable resources into ethanol: An economic evaluation of selected hydrolysis, fermentation, and membrane technologies. Energy Sources, 17:241{265. Rajvanshi, A. K. (1986). Alternative Energy in Agriculture. CRC Press. 64 Ringer, M., Putsche, V., and Scahill, J. (2006). Large-Scale Pyrolysis Oil Production: A Technology Assessment and Economic Analysis. Technical report, National Renewable Energy Laboratory. Sivers, M. and Zacchi, G. (1995). A techno-economical comparison of three processes for the production of ethanol from pine. Bioresource Technology, 51(1):43 { 52. So, K. and Brown, R. (1999). Economic Analysis of Selected Lignocellulose-to-Ethanol Conversion Technologies. Applied Biochemistry and Biotechnology, 77-79:633{640. Sokhansanj, S., Kumar, A., and Turhollow, A. (2006). Development and implementation of integrated biomass supply analysis and logistics model (IBSAL). Biomass and Bioenergy, pages 838{847. Sousa-Aguiar, E. F., Appel, L. G., and Mota, C. (2005). Natural gas chemical transforma- tions: The path to re ning in the future. Catalysis Today, 101(1):3 { 7. Gas-to-Liquid Technology. Sun, Y. and Cheng, J. (2002). Hydrolysis of lignocellulosic materials for ethanol production: a review. Bioresource Technology, 83(1):1 { 11. The outlook for energy, 2007 (2007). The outlook for energy. Technical report, ExxonMobil. Wilhelm, D. J., Simbeck, D. R., Karp, A. D., and Dickenson, R. L. (2001). Syngas pro- duction for gas-to-liquids applications: technologies, issues and outlook. Fuel Processing Technology, 71(1-3):139 { 148. 65 Appendices 66 Appendix A Manual for using the decision support system This manual gives users and developers an understanding of how to use the decision support system (DSS) e ectively to make decisions and strategies. The manual is structured in such a way that each screen is explained with a screenshot of it from the DSS. The important points to be noted by the user or developer are highlighted in the screenshot. In some shots, explanations are also provided for easy understanding. Welcome Screen This is the initial screen which will be displayed to the user once the application is opened. In this screen, the title of the decision support system and its use have been explained for the user. It also contains the set of instructions that the user has to follow to obtain the needed results. Figure A.1: Welcome Screen Snapshot 67 Input Screen When the user clicks on the start button the input screen is displayed. The input screen is divided into three parts. Data collection has to be performed before using the DSS. The necessary data are: Re nery information Feedstock cost information Product cost information All the data entered in this screen are duplicated in other sheets, as required. Figure A.2: Input Screen Snapshot Re nery information Data to be entered in this section are: Number of re neries to be located, Re nery technology for the biore nery (selected from a list), and Diesel price per gallon used in the trucks. The choice of re nery technology is linked to a cell in the input sheet. Each re nery type and its capacity are assigned a number. Depending on the re nery technology chosen respective number for that technology is displayed in the cell. This number is used in the Visual Basic code, which triggers the respective values for that technology to be used in the forthcoming calculation. 68 Feedstock cost information Select the data in wet-tons or dry-tons Percentage availability of forest and mill residues Biomass buying cost Moisture content Product cost information Depending on the re nery technology, selling price per unit is entered in this section. Result Screen The results from the analysis are displayed in three steps in this sheet. First, the results of the economic analysis are displayed which are investment cost, optimized transportation cost and the rate of return of the investment. After this, in the second step the optimized combination of equipment needed for the rst three stages of the supply chain are listed. Along with the list of equipment, the number of equipment required for the hauling opera- tion is also displayed. Finally, the facility location matrix is displayed. In the matrix, the optimized facility locations are displayed on the rows, and the supply counties from which raw materials are obtained is displayed in the columns. The matrix is lled with propor- tion of raw materials obtained from each counties for the respective facility locations. An Figure A.3: Result Screen Snapshot Alabama map is provided on a separate sheet to help the user to relate the results in a graphical format. A button has been provided in the result screen to help the user start a new project. 69 Back-end Screens The DSS consists of two models; the facility location model and the economic analysis model. The facility location model is built into a single sheet, i.e. model screen. The economic analysis spans over many sheets. Each sheet representing a re nery capacity and corresponding re nery technology. Model Screen In this sheet, the facility location model is built. Various screenshots from di erent parts of the model screen are shown. The data built into the sheet are listed below. Distances between counties Availability of forest residues Availability of mill residues Based on the input, the rest of the parameters required for the model are obtained. Depending on the re nery type selected in the input sheet, the requirement will be lled in the model sheet based on the VB code. Also, the availability matrices are calculated based on the percentages of forest and mill residues entered in the input. Figure A.4: Model Screen Snapshot The cost matrix is then generated with these matrices and the costs obtained from the input sheet. The constraints and the objective function are already built into this sheet and will recalculate automatically based on the matrices generated. 70 Figure A.5: Decision Variables matrix Snapshot Figure A.6: Constraints and Objective function Snapshot When the solve button is clicked, all the new data calculations are performed and then the premium solver application is called by the program. The solver then solves the facility 71 location optimization based on the data. The optimized transportation cost, location and the supply counties are obtained. Figure A.7: Model Screen Snapshot with premium solver dialog box In the screen shot, the premium solver dialog box with all the constraints entered are shown. The DSS is capable of helping the user to decide the optimal re nery technology for a county of the user?s choice. In order to incorporate this request from the user, changes have to made in the premium solver dialog box in the constraints. In this case, decision variable is only y and for the value of x, 1 must be entered by the user in the cell corresponding the county of the user?s choice rest of the counties x-value will be 0. Then rerun the premium solver. This will give the transportation cost. Trying it out for all the re nery technologies will help the user to obtain the optimal technology for that particular county. 72 Equipment selection screen The rst three stages of biomass supply chain, in-woods loading, rst stage transporta- tion and preprocessing resources optimization are done in this sheet. Depending on the re nery type chosen by the user, the annual requirement of biomass is passed onto this sheet. Based on the requirement, equipment used for the three stages are chosen from a pool of equipment. Figure A.8: Combinations screen snapshot Data regarding all the equipment for the processes are obtained from the residue truck- ing model (FoRTS) built by the US Forest Service research unit. Capacities, life, investment, costs involved and other miscellaneous costs are all obtained from their model and used in this program. Figure A.9: Data from FoRTS v5 snapshot 73 Using this data, the cost per ton mile for each equipment can be calculated. With the requirement known, the equipment needed for each stage is calculated easily. The next step is to choose the optimal set of equipment to be used for the re nery. This can be achieved by running through all possible combinations of equipment in the three stages and choosing the set which yields the least cost for the rst three stages of supply chain. Figure A.10: Equipment selection screen snapshot Analysis Screen The sheets in which the economic analysis are done were grouped together to be called analysis screens. When the user chooses the re nery type and enters the cost data in the input sheet, they are transferred to the respective re nery technology sheet. Then the facility location model calculates the transportation cost which is also transferred to the sheet. The sheet already contains the rest of the costs involved, which were obtained from their respective literatures. Cash ows are generated and the rate of return of the investment is calculated. 74 Figure A.11: Sample economic analysis screen of bio-oil re nery Snapshot Things to know regarding the code In order to add a re nery technology, the steps to be used are as follows: 1. Collect all the required data for the technology. 2. Using the template and assumptions of the previous technologies, perform economic analysis for the new technology. 3. Attach the sheet to the decision support system. 4. Add it to the list of re nery technologies in the input page. 5. Make changes in the code in 3 areas by adding an extra case statement. Those are for example, (a) In the requirement section of the code, Case 27 Sheets(\Model").Activate Range(\B1").Value = 330 * Worksheets(\unkeconanlysissheetname").Range("B9").Value * nfacilities 75 (b) In the result display section of the code, Case 26 Worksheets(\Result").Range("C2").Value =\Unknown technology (Capacity" Worksheets(\Result").Range("C4").Value = Worksheets(\unkeconanalysissheetname") .Range("B2").Value Worksheets(\Result").Range("D5").Value = Worksheets(\unkeconanalysissheetname") .Range("B65").Value (c) Hide the unknown analysis sheet. To add new equipment in the rst three stages of the supply chain, the steps to be used are as follows, 1. Collect the required data for the equipment and add it to the ?combinations? sheet as well as the ?data? sheet. 2. Add it to a list on the optimal sheet in the appropriate column based on the stages of the supply chain. 3. Calculate the costs involved. 4. Make changes to the code by adding cases similar to the procedure followed for adding a re nery technology. Most of the code is written to display results except for a few areas where they a ect the model. Those areas have been pointed out in the above cases. The code also has comments written side by side to help the developer understand its purpose and also provide an option for the developer to tweak it to obtain new properties, or results. 76