Hardwood Biomass to Gasoline, Diesel, and Jet Fuel - ACS Publications

Article pubs.acs.org/EF

Hardwood Biomass to Gasoline, Diesel, and Jet Fuel: 1. Process Synthesis and Global Optimization of a Thermochemical Refinery Richard C. Baliban,† Josephine A. Elia,† Christodoulos A. Floudas,*,† Barri Gurau,‡ Michael B. Weingarten,‡ and Stephen D. Klotz‡ †

Department of Chemical and Biological Engineering, Princeton University, Princeton, New Jersey 08544, United States Lockheed Martin Mission Systems and Sensors (MS2), 199 Borton Landing Road, Moorestown, New Jersey 08057, United States

‡

S Supporting Information *

ABSTRACT: A process synthesis framework is introduced for the conversion of hardwood biomass to liquid (BTL) transportation fuels. A process superstructure is postulated that considers multiple thermochemical pathways for the production of gasoline, diesel, and jet fuel from a synthesis gas intermediate. The hardwood is dried and gasified to generate the synthesis gas, which is converted to hydrocarbons via Fischer−Tropsch or methanol synthesis. Six different types of Fischer−Tropsch units and two methanol conversion pathways are analyzed to determine the topology for liquid fuel production that minimizes the overall system cost. Several upgrading technologies, namely, ZSM-5 catalytic conversion, oligomerization, hydrocracking, isomerization, alkylation, and hydrotreating, are capable of outputting fuels that meet all necessary physical property standards. The costs associated with utility production and wastewater treatment are directly included within the process synthesis framework using a simultaneous heat, power, and water integration. The solution quality of the optimal topology is mathematically guaranteed to be within a small fraction of the best possible value through the use of piecewise linear underestimation of nonlinear terms and a rigorous global optimization branch-and-bound strategy. A total of 12 case studies are investigated to determine the effect of refinery capacity and liquid fuel composition on the overall system cost, the BTL refinery topological design, the process material/energy balances, and the lifecycle greenhouse gas emissions.

1. INTRODUCTION In 2011, the United States consumed 18 835 thousand barrels per day (kBD) of petroleum-derived products, 13 223 kBD of which were required for the transportation sector.1 The majority of the transportation demand was attributed to gasoline (8565 kBD), diesel (2779 kBD), and jet fuel (1425 kBD), which combine to represent 96.6% of the national sector requirement. Over the next 2 decades, the transportation sector demand is expected to rise 9% to 14 410 kBD, while the net imports of petroleum are projected to decline approximately 0.8% per year over that period.2 The U.S. Energy Information Administration (EIA) has identified “non-petroleum”-derived sources as the means for satisfying the supply demand gap, with a particular focus on biomass. In general, liquid fuels that can be derived from domestic carbon-based feedstocks will reduce the dependence upon crude imports. Consequently, they provide significant means for increasing national security through enhanced energy independence. A recent review has highlighted the process design alternatives that can produce gasoline, diesel, and jet fuel using any one or a combination of coal, biomass, or natural gas feedstocks.3 Biomass-based processes are generally focused on corn-based ethanol and soybean-based diesel, but the use of these feedstocks have led to concerns regarding the impact on the price and availability of these feedstocks as sources of food or feed.4 Lignocellulosic plant sources (e.g., corn stover or forest residue) could be a more considerable source of biofuels in the future, although an increase in crop production will be required to generate an appropriate amount of sustainable residue for fuels production.5−8 © 2013 American Chemical Society

One of the major sources of biomass that is available throughout the nation is forest residues. These residues consist of (1) fuelwood harvested for current residential/commercial heating, (2) primary and secondary mill residues, (3) pulping liquors from paper manufacture, and (4) municipal solid waste sources. The United States Department of Energy (U.S. DOE) has estimated that approximately 129 million dry tons per year (MDTY) of forest residues may be harvested sustainably and a total of 226 MDTY may be available by 2030.5 Residues that are derived from timberland removal are generally less expensive on a dry basis than biomass that is categorized as agricultural residues or perennial crops. A key factor in this pricing scheme comes from the higher moisture content of the forest residues, which can represent up to 50% or more of the as-received weight of the feedstock.5 A significant portion of the biomass moisture will have to be dried off before the biomass can be processed efficiently in gasification systems. Hardwood and softwood residues from fuelwood or mill activities can be an important source of biomass feedstock because the infrastructure for harvesting and transportation is currently in place. These feedstocks would also make excellent candidates for gasification systems because consistency in the feedstock composition can be achieved. However, the use of land as a Special Issue: Accelerating Fossil Energy Technology Development through Integrated Computation and Experiment Received: December 6, 2012 Revised: January 30, 2013 Published: January 31, 2013 4302

dx.doi.org/10.1021/ef302003f | Energy Fuels 2013, 27, 4302−4324

Energy & Fuels

Article

neously analyze many distinct process designs and identify the optimal topology.24−27,89−94 The strategy considers a superstructure of energy processes that can convert the same feedstock to fuel products, and the selected conversion route is determined via an optimization model with an economic and/ or environmental objective function that is solved through deterministic global optimization approaches. You et al.95−99 proposed optimization models that simultaneously integrated the design (i.e., technology selection) with the supply chain of a network of hydrocarbon biorefineries. The biorefinery includes biomass to fuel conversions via gasification systems, followed by Fischer−Tropsch (FT) conversions, and fast pyrolysis, followed by hydroprocessing steps, such as hydrotreating and hydrocracking. You and Wang95 proposed a multi-objective, multi-period mixed-integer linear programming (MILP) model to simultaneously maximize the economic performance and minimize the environmental impacts of a biorefinery. You et al.96 also incorporated the social impacts of a biorefinery that took into account the number of jobs created and the improvement to the regional economy. Thus, the multiobjective MILP model includes economic, environmental, and social impact objective functions. Gebreslassie et al.97 considered the optimal hydrocarbon biorefinery supply chains under supply and demand uncertainties using conditional valueat-risk and downside risk. The objective function minimizes the annualized cost and the financial risk of the supply chain. The last two papers focus on the optimal design and synthesis of the hydrocarbon biorefineries. Wang et al.98 focused on the gasification pathway and proposed a multi-objective mixedinteger nonlinear programming (MINLP) model, where the nonlinearities are due to stream splittings and the capital cost functions, and Gebreslassie et al.99 focused on the fast pyrolysis, hydrotreating, and hydrocracking systems. 1.1. Problem Definition. The goal of this paper is to investigate and propose, through an optimization-based process synthesis framework, a thermochemical-based refinery that is capable of converting hardwood biomass (45 wt % moisture) to the liquid fuels gasoline, diesel, and kerosene. No fossil-based feedstocks, including coal or natural gas, will be input to the refinery. Additionally, no other forms of biomass feedstock beyond hardwood residues will be used in the refinery. Electricity that is provided for the refinery will be extracted from a local grid, and no renewable sources, including solar or wind, will be used. CO2 supercritical compression and sequestration will not be considered in the study. Four target barrel per calendar day capacities will be examined that have target values of 0.8, 1, 2.5, and 10 thousand barrels per day (kBD). For each capacity, one case study will be performed to output gasoline, diesel, and jet fuel that are in volumetric ratios commensurate with the 2010 United States demand. A second set of case studies will be designed to maximize the diesel production at each capacity level by setting a lower bound of 75 vol % on the total diesel output with respect to the total C5+ hydrocarbon output. A final set of four case studies will attempt to maximize the jet fuel output by setting the lower bound of jet fuel to be 75 vol % on the C5+ product at each capacity level. To achieve the above goal, an optimization-based process synthesis strategy is proposed that is capable of simultaneously analyzing thousands of distinct process designs of a hardwood BTL refinery. This work is based on recent successes in process synthesis that have been developed for any coal, biomass, or natural gas feed,24−27,91,92 specifically switchgrass and natural gas93 or specifically natural gas.94 A process superstructure of

source of biomass for liquid fuels production must be analyzed in context with the other potential land requirements, including fiber demand and the preservation of natural habitats. Ultimately, removal of forest residues for biofuel production will have to be investigated using a holistic framework that recognizes the connectivity of multiple processes that occur on both the forest and the ecosystem, including soil carbon management, erosion mitigation, nutrient management, water/ air quality, and global fiber production.8 Nevertheless, it is possible to responsibly develop forest biomass feedstocks in a sustainable manner to generate a significant national supply without deforestation.5 Biomass is an advantageous feedstock because a strong environmental benefit can be achieved during the lifecycle of a biomass to liquid (BTL) refinery. Liquid fuels derived from coal or natural gas use a nonrenewable resource, but these feedstocks possess distinct advantages, including the low delivered cost of coal ($2.0−$2.5/MMBtu)2 relative to natural gas ($4.8−$5.8/MMBtu) 9 or biomass ($4.0−$9.0/ MMBtu)8,10,11 and the high carbon conversion efficiency of natural gas. Hybrid refineries that can use both a fossil-based feedstock and biomass can take advantage of the strengths of each feed and produce refineries that are economically or environmentally superior to single feedstock refineries. The three major hybrid refinery classifications that use biomass are coal/biomass to liquids, 11−18 biomass/natural gas to liquids,19−23 and coal/biomass/natural gas to liquids.24−27 Although hybrid-feedstock refineries can take advantage of the benefits of using lower cost fossil-based feedstocks, it may not always be practical to use multiple feedstocks within a single refinery. In fact, the distributed network of biomass feedstock locations will contain several points where infrastructure for coal or natural gas delivery is limited. Therefore, the potential of forest resources as a sustainable feedstock will depend critically upon the economic development of the refinery used to produce the liquid fuels. The larger capital investment cost for “second-generation” biofuels facilities is a strong motivating factor for the development of new processes that can use existing technology to refine the fuels at lower costs. The general strategy for BTL refinery development is the generation of a process design with a fixed topology (i.e., the combination of process units and streams) that is subsequently analyzed using simulation software packages to find the heat and mass balances throughout the refinery. An economic analysis can then be carried out to determine the appropriate financial metrics, including total investment cost or net present value.10−12,17,18,28−88 The inherent limitation in the above approach is the uncertainty whether the design being considered is optimal in the context of all possible alternative designs of biomass conversion to the liquid fuels of interest. Given the numerous combinations of unit operations, recycle streams, utility systems, and wastewater systems, the determination of the optimal topology represents a major challenge. A brute force approach that aims at enumerating all designs would require a significant amount of computational time and manpower to analyze all process designs. The effort required to conduct such an analysis scales with the number of designs under consideration and becomes intractable beyond some upper threshold of possibilities. Moreover, this strategy offers no guarantee that the “best” design is superior to novel process topologies that were not considered in the comparative analysis. To address the aforementioned issue, an optimization-based process synthesis strategy has been introduced to simulta4303

dx.doi.org/10.1021/ef302003f | Energy Fuels 2013, 27, 4302−4324

Energy & Fuels

Article

authors24−27 for more detailed information of the units. The novel components in this paper are highlighted in the subsections below. 2.1. Hardwood Biomass Handling and Gasification. The BTL refinery is assumed to input hardwood biomass, which has a representative composition shown in Table 1 that is derived from

alternatives is constructed that considers multiple pathways for converting hardwood biomass to the liquid transportation fuels gasoline, diesel, and jet fuel. A large-scale MINLP model is proposed to rigorously describe the input−output relationships of each unit within the refinery superstructure. The simultaneous consideration of each process unit is incorporated into the large-scale model, which is solved to global optimality using a rigorous deterministic branch-and-bound global optimization approach.26,100−103 Key units within the refinery, including hardwood biomass gasification, synthesis gas cleaning, hydrocarbon generation, and multiple refining technologies, are described. A series of heat engines24,89,90 is incorporated into the process synthesis framework using a pinch-point methodology104 to ensure the simultaneous consideration of the utility plant. Similarly, a detailed wastewater network based on a superstructure approach23,105−107 is incorporated to minimize the freshwater usage and the wastewater contaminants. The process synthesis framework will be used to examine (i) hardwood biomass gasification with/without recycle synthesis gas, (ii) synthesis gas conversion via FT synthesis or methanol synthesis, (iii) hydrocarbon upgrading via ZSM-5 zeolite catalysis, olefin oligomerization, or carbon number fractionation and subsequent treatment, and (iv) methanol conversion via methanol-to-gasoline (MTG) or methanol-to-olefins (MTO). The key products from the hardwood BTL refinery will be gasoline, diesel, and jet fuel with allowable byproducts of liquefied petroleum gas (LPG) and electricity. The quantitative trade-offs associated with key metrics for the BTL refinery are illustrated using 12 case studies, which are selected to demonstrate the capability of the proposed process synthesis framework.

Table 1. Feedstock Proximate and Ultimate Analyses for Hardwood108 proximate analysis (wt %, db) moisture (ar)

ash

45

VMa

heating values (kJ/kg) FCb

2.14 N/A N/A ultimate analysis (wt %, db)

HHVc

LHVd

19130

17842

C

H

N

Cl

S

O

50.19

5.9

0.32

0

0.03

41.42

VM = volatile matter. bFC = fixed carbon. cHHV = higher heating value. dLHV = lower heating value. a

the Energy Research Centre of the Netherlands (ECN) Phyllis database.108 The hardwood is assumed to be delivered to the refinery as woodchips, which must be further processed prior to entry to the hardwood gasifier. The woodchips are screened to remove particles with sizes greater than 2 in. and then sent to a grinder for further size reduction.109 The process flow diagram for the generation of synthesis gas is detailed in Figure 1. The hardwood biomass moisture content is reduced to 20 wt % through a preliminary drying step10 before the feedstock can be injected to the gasifier. Any heat necessary for the dryer is provided by flue gas generated through combustion within the refinery. The flue gas leaves the dryer at 110°C and 1.05 bar and is passed through an air cyclone and a baghouse filter to remove any particulates that are present.109 The heated hardwood leaves the dryer at 105°C and 1.05 bar and is transferred to a high-pressure gasifier (30 bar) using compressed CO2 (10 wt %) and a lockhopper. The effluent of the hardwood gasifier is a mixture of synthesis gas, C1−C2 hydrocarbons, acid gases (e.g., NH3 and H2S), tar, char, and ash.24,89 The solid ash and char are separated from the vapor phase using cyclones and are recycled back to the gasifier. It is assumed that the recycle of char will effectively provide a 100% conversion of the carbon in the hardwood to vapor species, while the ash is removed from the gasifier as slag. The composition of the gasifier effluent is a function of the hardwood composition, gasifier temperature, and oxidizer flow rate, and a detailed mathematical model has been formulated to determine the composition.24−27,89 Gasification is facilitated by purified oxygen and steam, which are preheated to 800°C before entering the gasifier unit. The oxygen equivalence ratio (ER) is defined as the mass ratio of input oxygen divided by the total amount of oxygen necessary for complete combustion of the hardwood. The steam/hardwood biomass ratio (SBR) is defined as the mass ratio of input steam divided by the bonedry weight of input hardwood. For this study, the ER is varied between 0.2 and 0.4 and the SBR is varied between 0.2 and 1.5 to allow for flexibility of gasifier operation within the parameter space that was used to construct the gasifier model.89 The parameters associated with eqs (1)−(10) are listed in Table 2 and were calculated using a parameter estimation optimization model that minimized the difference between several experimental gasifier outlets and that predicted the gasifier operating equations.89 All of the parameters are valid for a given gasifier unit u that is selected from one of two different types, which are based on the input of solid feedstock only (BGS) or a combination of solid and vapor feedstocks (BRGS). Equations (1)− (10) have been written using the former gasifier (BGS) for reference, and similar constraints are enforced for the latter gasifier with recycle. Operation of the hardwood gasifier assumes that the effluent is in equilibrium with respect to the water-gas shift reaction (eq 1). In eq (1), NBGS,BC1,s indicates the molar flow of species s from the gasifier unit (BGS) to the first cyclone (BC1) and KWGS BGS represents the equilibrium constant for the water-gas shift reaction. The equilibrium

2. HARDWOOD BTL PROCESS SUPERSTRUCTURE: CONCEPTUAL DESIGN AND MATHEMATICAL MODELING This section will outline the conceptual design and mathematical modeling of the key sections of the hardwood BTL refinery.24−27,89,91,92 The complete mathematical model, all relevant nomenclature, and a complete set of process flow diagrams are provided as Supporting Information. All relevant thermodynamic information for the hardwood refinery was computed in Aspen Plus, version 7.3 using the Peng−Robinson equation of state with Boston− Mathias α function. The superstructure is primarily divided into several sections, namely, (i) syngas generation from biomass, (ii) syngas cleaning section, (iii) hydrocarbon production section, (iv) hydrocarbon upgrading section, (v) light gas recycle section, (vi) hydrogen and oxygen production, and (vii) heat, power, and water integration. Section (i) converts hardwood biomass into syngas via gasifiers that may or may not include gas recycle. The purpose of the gas recycle is to take advantage of the high temperature of the biomass gasifier for the endothermic reverse water-gas shift reaction, so that input CO2 can be shifted to CO for further fuel production. The syngas cleaning section removes acid gases and nitrogenous components as well as some CO2 from the stream prior to the hydrocarbon production section. The hydrocarbon production section contains two major routes, the first via FT systems, where different temperatures, catalysts, and levels of wax production are considered and second via methanol intermediate and subsequent conversion to gasoline or olefins. The upgrading section has two options, namely, upgrading via ZSM-5 zeolite catalysis, olefin oligomerization, or fractionations. The light gases are channeled to an autothermal reactor (for syngas reforming), fuel combustor to supply process heats, or an optional gas turbine for electricity generation. The flue gases of the combuster and gas turbine units pass through a CO2 separation unit before being vented to the atmosphere, and CO2 is recycled within the process. The readers are referred to previous publications by the 4304

dx.doi.org/10.1021/ef302003f | Energy Fuels 2013, 27, 4302−4324

Energy & Fuels

Article

Figure 1. Hardwood biomass gasification flowsheet. Hardwood is dried to 20 wt % moisture and then transferred to the gasifier system via a lockhopper. The gasifiers will operate with either a solid fuel (hardwood) or a combination of solid and recycle gases as fuel. Residual ash and char that are generated within the gasifiers are separated via the cyclones and recycled to the gasifiers. The raw syngas is then transferred to a tar cracker to remove most of the tar species in the vapor phase. (2) also includes all hydrocarbons that are input to the gasifiers from additional source units u′ that are located throughout the refinery. Given the initial amount of C1−C2 hydrocarbons that are present after ), eq (3) denotes the amount of hydrocarbon leaving pyrolysis (NS,Calc s the gasifier using a conversion fraction (cf HC BGS,s) for each species s.

Table 2. Parameters Used for Operation of the Hardwood Biomass Gasifiera parameter

value

parameter

value

mf u cf HC u,CH4

0.10 0.6667

nf u a1u,char

0.9801 5.002 × 10−3

cf HC u,C2H2

0.6667

a2u,char (°C−1)

0.2710

cf HC u,C2H4

0.6667

a1u,N2

−0.9437

cf HC u,C2H6

0.6667

a2u,N2 (°C−1)

1.250 × 10−3

sru,(NO/N2O)

1.276 × 10−1

sf u,ash

0.99

−2

sru,(HCN/NH3)

3.590 × 10

hlu

0.01

sru,(COS/H2S)

1.0000

pyrHC CH4,hardwood

0.0678

pyrCHC2H2,hardwood

0.0002

pyrCHC2H4,hardwood

0.0206

pyrCHC2H6,hardwood

0.0009

KWGS u

KWGS (900°C) u

0.9137

KWGS (1000°C) u

(800°C)

WGS NBGS,BC1,CONBGS,BC1,H2O − KBGS NBGS,BC1,CO2NBGS,BC1,H2 = 0

(1)

∑

NsS,Calc − pyr HC NS − s , s BLK,BGS, sB B

NuS′ ,BGS, s = 0

(u ′ ,BGS) ∈ UC

∀ s ∈ SHC

(2)

HC NBGS,BC1, s − cfBGS, N S,Calc = 0 s s

∀ s ∈ SHC

(3)

The nitrogen present in the feedstock will vaporize as N2, NH3, HCN, NO, and N2O. The species ratio (sru) of NO/N2O is determined using eq (4), and the ratio of NH3/HCN is determined using eq (5). The amount of nitrogen present in N2 and NH3 is set using a nitrogen fraction (nf u) parameter, as shown in eq (6). Finally, the fraction of nitrogen as NH3 is calculated using eq (7).

1.0815 0.7849

a All of the parameters are valid for a given gasifier unit u that is selected from one of two different types, which are based on the input of solid feedstock only (BGS) or a combination of solid and vapor feedstocks (BRGS).

constant in eq (1) was extracted from Aspen Plus, version 7.3. The index s will represent the set of all species that are present in the hardwood gasifier, including syngas components (H2, CO, and CO2), water, hydrocarbons (CH4, C2H2, C2H4, and C2H6), nitrogen species (N2, NH3, HCN, NO, and N2O), sulfur species (H2S and COS), char, or ash. The equilibrium constant is a function of the operating temperature of the gasifier, which is selected from a discrete set of alternatives by the optimization framework. The C1−C2 hydrocarbon species (SHC) are assumed to be present in concentrations above their equilibrium values;89 therefore, a restriction on the extent of reaction of each hydrocarbon species is enforced using eq (3). The C1−C2 hydrocarbons are generated during pyrolysis of the hardwood species, which is based on the constraint in eq (2). The species index sB in eq (2) represents the hardwood biomass input to the gasifier, and the parameter pyrHC represents the molar amount of hydrocarbon species s generated per molar amount of hardwood species sB. Note that eq

NBGS,BC1,NO − srBGS,(NO/N2O)NBGS,BC1,N2O = 0

(4)

NBGS,BC1,HCN − srBGS,(HCN/NH3)NBGS,BC1,NH3 = 0

(5)

NBGS,BC1,NH3 + 2NBGS,BC1,N2 − nfBGS

∑ (BGS,BC1, s) ∈ S

UF

S NBGS,BC1, sAR s ,N

(6)

=0 1 2 NBGS,BC1,NH3 − (aBGS,N + aBGS,N T )(NBGS,BC1,NH3 2 2 BGS

+ 2NBGS,BC1,N2) = 0

(7)

The sulfur will be completely transformed into COS and H2. The species ratio (sru) for sulfur will determine the relative amounts of the two sulfur species (eq 8). 4305

dx.doi.org/10.1021/ef302003f | Energy Fuels 2013, 27, 4302−4324

Energy & Fuels

Article

Figure 2. Synthesis gas (syngas) handling flowsheet. Syngas may be passed over a forward/reverse water-gas shift reactor to alter the H2 to CO/CO2 ratio prior to FT synthesis or methanol synthesis. The syngas is then cooled, flashed to remove water, and directed to a dual-capture methanol-based unit for CO2 and H2 removal. The H2S gases are then directed to a Claus plant for recovery of the sulfur. The captured CO2 may be vented, sequestered, or recycled back to process units.

NBGS,BC1,COS − srBGS,(COS/H2S)NBGS,BC1,H2S = 0

gasifier. The high temperature of the unit will facilitate the reverse water-gas shift equilibrium of the synthesis gas effluent, although the concentration of the hydrocarbons in the effluent will be far above the equilibrium values. Using the reverse water-gas shift reaction, CO2 may be consumed within the gasifier unit by reaction with H2 that is present within the gasifier. Therefore, any CO2 that is generated by the process can be recycled to the gasifier along with H2 that is produced from pressure-swing adsorption or electrolysis of water. An example of the syngas composition from the biomass gasifier ranges between 19 and 27% of H2, CO, CO2, and H2O, 5−6% of CH4, less than 3% of C2 compounds, and small percentages of nitrogen- and sulfur-containing compounds. The effluent of the gasifier is passed through a catalytic tar reformer (825°C), which will reform (i) tar species to CO and H2, (ii) NH3 to N2 and H2, and (iii) C1−C2 hydrocarbons to CO and H2. The current bench-scale performance of a tar reformer from the National Renewable Energy Laboratory (NREL) has a conversion of 80% of CH4, 99.6% of tars, 99% of C2H6, 90% of C2H4, and 90% of NH3.109 Equipment is currently being installed for pilot-scale demonstration of the tar reformer performing over a continuous period of time.110 The steam that is present in the gasifier effluent is sufficiently high to reform the syngas without the need of additional input steam.110 Heat for the tar reformer is provided by a circulating catalyst between the tar reformer and a catalyst regenerator to remove the coke deposits on the catalyst surface. The level of coke deposited on the catalyst is insufficient to provide the heat needed for the endothermic reforming reactions; therefore, additional fuel gases are passed through the regenerator.110 The fuel gases contain a high degree of methane and unreacted syngas. The syngas exiting the tar reformer is cooled to 185°C and passed to the cleaning section (see Figure 2). 2.2. Synthesis Gas Cleaning. The syngas exiting the tar cracker is directed to a cleaning section, which is shown in Figure 2. The syngas must be cooled to 185°C and sent to a scrubbing system (SCRUB) to remove any residual tar, particulates, and NH3 from the gas phase. The wastewater generated from wet scrubbing is sent to a biological digestor for treatment of the organic contaminants, and the scrubbed

(8)

The char entrained in gasifier vapor is a function of the input hardwood flow rate and the gasifier temperature and uses two known coefficients (aBGS,char), as described by eq (9). The ash that is removed from the gasifier is based on a split fraction (sf BGS,ash), as modeled by eq (10). S 1 2 S NBGS,BC1,char − (aBGS,char + aBGS,char TBGS)NBLK,BGS, sB = 0

S NBGS,out − sfBGS,ash ash,ash

∑

(9)

NuS′ ,BGS,ash = 0

(u ′ ,BGS) ∈ UC

(10)

The heat loss in the gasifier is modeled to be a fraction (hl) of the input dry lower heating value (LHV) of the hardwood (eq 11). L S Q BGS + hlBGS MWsBNBLK,BGS, sB LHVs = 0

(11)

The logical selection of one hardwood gasifier temperature is set using eq (12). The set UBGS defines the set of possible operating temperatures from 800 to 1000°C. This temperature will directly affect the extent of the water-gas shift reaction, the residual char, and the ratio of N2/NH3. The remaining vapor species compositions will be indirectly affected by the temperature change through the constraints above. Equality is enforced in eq (12) because there must be at least one operating temperature in the hardwood BTL process. Although the operating temperature of the gasifier may be above 800°C, the syngas effluent from the unit will always be output at a temperature of 800°C to ensure that a majority of the ash is removed from the gasifier as slag.

∑ u ∈ UBGS

yu − 1 = 0

(12)

The gasifier will use steam for gasification of hardwood/char, reforming of C1−C2 hydrocarbons, and reforming of tar species. High-purity oxygen is input to the gasifier to provide the heat needed for the reforming reactions and to help crack the tar species in the 4306

dx.doi.org/10.1021/ef302003f | Energy Fuels 2013, 27, 4302−4324

Energy & Fuels

Article

Figure 3. FT synthesis flowsheet. Clean syngas is converted to hydrocarbons over cobalt- or iron-based catalysts operating at either low or high temperature. The residue/wax produced from FT synthesis is directed to a hydrocracker, and the vapor-phase C3−C22 hydrocarbons are sent for further upgrading. syngas is directed to a dual-capture Rectisol system. Prior to entry into the scrubber unit, the synthesis gas may be partially passed over a water-gas shift reactor that can operate in the presence of sulfur species. The reactor operates at 28 bar and a temperature between 400 and 600°C and is used to either (i) increase the H2/CO ratio of the syngas via the forward water-gas shift reaction or (ii) decrease the CO2 concentration of the syngas via the reverse water-gas shift reaction. The capability to use the reverse water-gas shift concentration largely depends upon the inlet concentration of CO2, which generally must be high (>0.5) to be implemented effectively. The CO2 concentration may be increased by recycling CO2 from the dual-capture unit in Figure 2 or the post-combustion capture unit in Figure 10. The reverse water-gas shift reaction must be facilitated by the addition of hydrogen from pressure-swing adsorption, electrolysis of water, or reformed gases from the autothermal reactor. Oxygen is provided from air separation or electrolysis to provide the heat necessary for the reverse water-gas shift reaction. If the reactor implements the forward watergas shift reaction, then CO2 will be generated, which must be subsequently captured from the syngas. The exothermic heat of reaction is removed via steam generation. The decision to incorporate the dedicated water-gas shift unit into the hardwood BTL refinery is dependent upon the H2/(CO + CO2) ratio necessary for syngas conversion, the amount of H2 that can be economically extracted via pressure-swing adsorption, the capital/operational costs associated with H2 generated from electrolysis, and the lifecycle emissions requirement of the refinery. The Rectisol unit is used to co-remove H2S and CO2 from the syngas11 and provide a clean gas stream that is ready for FT synthesis or methanol synthesis. The Rectisol unit must be used to remove the sulfur species to prevent poisoning of the catalysts used in the synthesis reactors. The unit operates at 24 bar,11 and all of the input streams enter at 24 bar. The output gas exits at 21 bar, and CO2 exits in two places at 1.01 and 3 bar. The capture unit will remove 100% of H2S and 90% of CO2 from the input gases.11 A total of 3 mol of CO2 are entrained in the sulfur-rich stream for every mole of H2S that is recovered. This sulfur-rich gas stream exits at 1.01 bar and is

compressed to 1.1 bar prior to entering the Claus system to recover solid sulfur. The per-pass conversion of H2S and SO2 to solid sulfur is approximately 95%;111 therefore, the tail gas from the Claus process is hydrolyzed to form H2S and recycled back to the Rectisol unit. This allows for 100% overall recovery of the sulfur from the hardwood BTL refinery. Pure CO2 leaves the Rectisol unit at 25°C and 1.5 bar and may either be (i) vented from the refinery, (ii) compressed to 31 bar for recycle to refinery units, or (iii) compressed to 150 bar for sequestration. Multi-stage units with intercooling will be used to control the temperature rise during compression and ensure that no single compressor outlet is above 200°C. 2.3. FT Hydrocarbon Production. The clean syngas from Figure 2 may be directed to the FT synthesis section shown in Figure 3. The FT units operate at 20 bar and use either an iron- or cobalt-based catalyst.24,25,27 The iron-based units will operate using either low or high temperature and facilitate equilibrium via the water-gas shift reaction. Thus, the iron-based FT units have the potential to consume CO2 within the reactor using H2 and produce CO for the FT reaction. The value of the equilibrium constant within the FT temperature range requires that the ratio of CO2/(CO + CO2) is above a critical threshold of about 0.75, which is possible through recycle of process CO2 to the inlet of the FT units. If the CO2/(CO + CO2) ratio is below the critical threshold, then CO will likely be converted to CO2 within the units. The amount of inlet hydrogen with respect to both CO and CO2 will dictate where the critical threshold will lie and, therefore, plays an important role in FT synthesis. Low-temperature iron-based units have been successfully operated with inlet H2/CO ratios between 0.5 and 1112−114 and take advantage of in situ forward water-gas shift to increase the H2/CO ratio to near 1.7−2.0. These units require substantially less hydrogen than processes that have an input H2/CO closer to 2, but approximately 50% of the input CO is converted to CO2. To prevent such a large increase in the outlet CO2 concentration, the inlet ratio of H2/(CO + CO2) can be set to ensure that CO2 can be used as a carbon source via the reverse water-gas shift reaction. The Ribblett ratio114,115 is defined such that H2/(2CO + 3CO2) is 4307

dx.doi.org/10.1021/ef302003f | Energy Fuels 2013, 27, 4302−4324

Energy & Fuels

Article

Figure 4. FT hydrocarbon upgrading flowsheet. Vapor-phase FT effluent may be cooled and scrubbed to remove the aqueous phase and several oxygenates before being directed to a hydrocarbon recovery column. Alternatively, the FT effluent (including oxygenate) may be converted to gasoline-range hydrocarbons via a ZSM-5 zeolite catalyst. The ZSM-5 reactor effluent may be further separated into gasoline and LPG for resale.

Figure 5. FT distillate upgrading flowsheet. The water-lean FT hydrocarbons are split into distillate, naphtha, and light gas fractions, which can be subsequently upgraded. The straight-run distillate may be hydrotreated and output as diesel or kerosene. Alternatively, the distillate may be hydrocracked along with FT wax/residue to form smaller, paraffinic molecules. high conversion rates of CO and CO2 are achieved in the hardwood BTL refinery. To examine the effects of the H2/(CO + CO2) ratio, the synthesis gas entering the iron-based FT units will be handled in one of two ways. One low-temperature (240°C) unit and one high-

approximately equal to 1 and is useful because the effluent composition of unreacted syngas from a FT unit will maintain roughly the same value as the inlet. Therefore, the internal or external gas loop designs for FT synthesis can be theoretically designed such that very 4308

dx.doi.org/10.1021/ef302003f | Energy Fuels 2013, 27, 4302−4324

Energy & Fuels

Article

Figure 6. FT naphtha upgrading flowsheet.

Figure 7. FT C3−C5 gases upgrading flowsheet. temperature (320°C) unit will require an inlet Ribblett ratio that is equal to 1 and will facilitate the reverse water-gas shift reaction as the CO2 inlet concentration increases. The other two units will operate at 267°C and have an effluent composition that is based on two previous U.S. DOE reports.27,112,113 These two units will have a H2/CO inlet

ratio between 0.5 and 0.7 and will ensure that the H2/CO ratio in the effluent is equal to 1.7 from forward water-gas shift conversion. Hydrogen may be recycled to any of the FT units to shift either the H2/CO ratio or the H2/CO2 ratio to the appropriate level. Steam may 4309

dx.doi.org/10.1021/ef302003f | Energy Fuels 2013, 27, 4302−4324

Energy & Fuels

Article

be alternatively used as a feed for the two iron-based forward water-gas shift FT units to shift the H2/CO ratio in situ. The FT reactors may also operate using a cobalt-based catalyst. These FT units will be ideal for achieving high per-pass conversion of CO to FT liquids because cobalt-based catalysts will not facilitate the water-gas shift reaction. The extent of catalyst oxidative degradation because of high water partial pressure is a contentious topic, and current data imply that the effect of catalyst stability based on the presence of water is not clear.114 Although cobalt oxidation is reversible, increased time on stream will result in higher levels of methane formation and lower levels of C5+ liquids from the FT unit.114 For this study, the CO per-pass conversion was set to 60% to avoid catalyst oxidation, although conversion levels may approach 80% if catalyst stability can be achieved.114 Note that both high- and lowtemperature cobalt-based FT systems are considered in this analysis. To date, only the low-temperature FT unit has been commercially available, but this study investigates the possibility of a hightemperature system using an α value that is consistent with hightemperature FT operation (e.g., α = 0.72). The two streams exiting the iron or cobalt FT units will be a waxy liquid phase and a vapor phase containing a range of hydrocarbons. The wax will be directed to a hydrocracker (WHC), while the vapor phase is split for further processing. 2.4. FT Hydrocarbon Upgrading. The vapor-phase effluent from FT synthesis will contain a mixture of C1−C30+ hydrocarbons, water, and some oxygenated species. Figure 4 details the process flowsheet used to process this effluent stream. The stream will be split and can pass through a series of treatment units designed to cool the stream and knock out the water and oxygenates for treatment.24,89 Initially, the water-soluble oxygenates are stripped from the stream. The stream is then passed to a three-phase separator to remove the aqueous phase from the residual vapor and any hydrocarbon liquid. Any oxygenates that are present in the vapor phase may be removed using an additional separation unit. The water-lean FT hydrocarbons are then sent to a hydrocarbon recovery column for fractionation and further processing. The oxygenates and water removed from the stream are mixed and sent to the biological digestor for wastewater treatment. The following sections will detail the individual sections of the FT upgrading. Note that all of the appropriate utility requirements for all units can be extracted from literature sources112,116−118 to ensure the appropriate operational costs for the refinery. 2.4.1. ZSM-5 Catalytic Upgrading. The FT hydrocarbons may also be passed over a ZSM-5 catalytic reactor operating at 408°C and 16 bar112 to be converted into mostly gasoline-range hydrocarbons and some distillate.112,113 The ZSM-5 unit will be able to convert the oxygenates to additional hydrocarbons; therefore, no separate processing of the oxygenates will be required for the aqueous effluent. The raw product from FT ZSM-5 is fractionated to separate the water and distillate from the gasoline product. The water is mixed with other wastewater knockouts, and the distillate is hydrotreated to form a diesel product. The raw ZSM-5 hydrocarbon product is sent to a LPG/gasoline separation section for further processing. 2.4.2. Distillate Upgrading. The water-lean FT hydrocarbons leaving the oxygenate scrubber in Figure 4 are passed to a hydrocarbon recovery column, as shown in Figure 5. The hydrocarbons are split into C3−C5 gases, naphtha, kerosene, distillate, offgas, and wastewater streams.89,116 Figures 5−7 detail the upgrading units associated with converting the C2−C22 hydrocarbons into the final gasoline, diesel, and kerosene products. The distillate contains C11−C22 hydrocarbons and may be split to either a hydrocracker or a hydrotreater. The hydrocracker will also input the residue/wax from the FT units and has an effluent composition that is based on distribution of the carbon atoms into C1−C4 offgas, C5−C6 gases, naphtha, C11−C14 distillate, C15−C22 distillate, and C23+ residual hydrocarbons.24,116,117 The C23+ residual hydrocarbons are recycled back to the hydrocracker, and the two distillate streams are split for kerosene production, diesel production, or recycle to the hydrocracker. The extent of hydrocracking of the two distillate streams is dependent upon the fuel requirements of the hardwood BTL refinery. No additional hydrotreating of the hydrocracker distillate is necessary because the

hydrocarbons are assumed to be saturated on the basis of the addition of hydrogen to the hydrocracker.114 The C10 hydrocarbons are directed to additional upgrading units. The straight-run C11−C22 distillate from the hydrocarbon recovery column may also be hydrotreated to saturate the olefinic compounds and remove oxygen heteroatoms. The hydrotreated distillate may be split to diesel and kerosene fractions or may be output completely as diesel fuel. C1−C4 offgas from the hydrotreater is directed to saturated gas recovery, and the sour water generated from input oxygenates is sent to wastewater treatment. Heavy C9−C12 aromatics may also be hydrotreated to add to the pool of output kerosene. For both hydrotreaters, enough hydrogen is input to the units to fully saturate the double bonds present in the input species.24,116,117 2.4.3. Naphtha Upgrading. The naphtha upgrading section of the refinery is detailed in Figure 6. The C6−C10 naphtha from the FT reactor is split to one of two major upgrading units. The naphtha reformer will produce an aromatic-rich hydrocarbon effluent from a naphtha feedstock and is highly useful for providing a reformate blendstock for gasoline production. An alternative means of aromatization based on metal-promoted acidic ZSM-5 catalysis can also be used. Although these two units both serve the purpose of producing aromatic hydrocarbons from the FT effluent, the product compositions will be very different. Prior to entry into the naphtha reformer, the feed is passed through a hydrotreater to saturate the double bonds and remove oxygen heteroatoms. The offgas is directed to a saturated gas plant, while the C7−C10 hydrocarbons are directed to reforming. The C5−C6 hydrocarbons may be directed to a C5/C6 isomerizer or may be split into two separate streams. The C6 hydrocarbons will be directed to the naphtha reformer or to C6 isomerization. The C5 hydrocarbons will be processed in Figure 7. Note that the use of separate C5 and C6 isomerization units as opposed to a combined C5/C6 isomerization unit is a decision to be taken by the process synthesis model. Although a single C5/C6 isomerization unit will require less capital cost than two separate units, the per-pass isomerization of the n-alkanes can limit the octane number increase of the hydrocarbons. The overall conversion of linear hydrocarbons may be increased by separation and recycle of the nalkanes. The complexity of unit operations is relatively low for single carbon number effluents and will increase for C5/C6 feeds. Additionally, it may be difficult to incorporate all of the isomerized C5/C6 hydrocarbons in gasoline because this material comprises about 50% of the straight-run FT naphtha from high-temperature units and 40% of the straight-run FT naphtha from low-temperature units. Isomerization of the C5 hydrocarbons and refining of the C6 hydrocarbons may allow for additional flexibility within the hardwood BTL refinery to produce gasoline that meets all appropriate specifications.114 After hydrotreating, the naphtha may be passed through a reforming unit. Conventional naphtha reforming is based on the UOP platforming process that uses a Pt/Al2O3-based catalyst to convert low octane heavy naphtha into high octane gasoline and simultaneously produce H2 for refinery use. The rate of conversion of hydrocarbons in the reforming process increases with the carbon number, and the process performs very well for C8−C10 hydrocarbons, while it is inefficient for C6 or C7 hydrocarbons (specifically n-alkanes). Although the C6 n-alkanes may be isomerized into a higher octane number branched hydrocarbon, this is not as readily achieved for C7 nalkanes. Furthermore, the liquid yield loss from the reformer is affected by both the octane number of the hydrogenated straight-run naphtha and the cycloalkane content of the feed.114 FT naphtha is considered to be lean because of the low naphthalene and aromatic contents and is difficult to convert into a high octane reformate without significant yield loss. The RZ platforming process from UOP118 employs a monofunctional Pt/L-zeolite reforming catalyst, which demonstrates high selectivity for aromatization when using an alkane-rich feed that contained C6 and C7 n-alkanes. Direct dehydrocyclation produces Cn aromatics from Cn alkanes in the Pt/L unit; therefore, the carbon number distribution of the aromatic products can be controlled from the carbon number of the naphtha feed.114 Because of the presence of 4310

dx.doi.org/10.1021/ef302003f | Energy Fuels 2013, 27, 4302−4324

Energy & Fuels

Article

Figure 8. Methanol synthesis and upgrading flowsheet. some hydrogenolysis, the selectivity of light aromatics will be higher than that expected from the feed distribution. Thus, there will be benzene in the aromatic product even without the presence of C6 hydrocarbons in the feed. Appropriate unit operations must be included in the refinery to reduce the benzene concentration to appropriate levels for gasoline. The Pt/L reforming unit is modeled assuming an 80% conversion of the C6−C7 paraffins, 100% conversion of the C8+ paraffins, and 100% conversion of the olefins.114 The distribution of carbon in the effluent of the RZ platforming process is based on the idea that Cn aromatics and olefins will mostly form Cn alkanes.114 Thus, the yield of Cn aromatics from the Cn feed is assumed to be 80%. The decomposition of some Cn species will yield higher levels of benzene;114 therefore, it is assumed that 10% of the carbon in the Cn species will be output as benzene. The balance of the carbon for each species is distributed to C1−C4 linear paraffins with a distribution similar to that of a Pt/Al2O3 unit.116 The effluent from the Pt/L-zeolite reforming unit is mixed with the aromatic effluent from the M/Z-ZSM-5 unit for separation of the C7+ aromatics. These aromatic species are split such that the C7/C8 fraction is output as gasoline, the C11−C15 fraction is output as jet fuel, and the C9−C10 fraction may be output as either gasoline or jet fuel. A hydrogen-rich fuel gas containing C1−C2 hydrocarbons is split from the aromatic effluent and directed to either the pressure-swing adsorption unit for H2 generation or the inlet of the syngas conversion units to increase the H2/CO ratio. The C3−C4 gases contain mostly saturated hydrocarbons and are directed to the saturated gas plant for further processing. 2.4.4. C3−C5 Upgrading. The final section of the FT upgrading is shown in Figure 7. The straight-run FT C5 gases may be split to either an alkylation process, an oligomerization process, or hydrotreating for isomerization (see Figure 6). The C3−C5 alkylation unit will use the olefinic C3−C5 gases from the FT synthesis to convert isobutane into a high octane number blendstock. The level of isobutane present in straight-run FT synthesis is typically much lower than that necessary to convert all olefinic species; therefore, additional isobutane is generated through C4 isomerization. n-Butanes are recovered from various offgas streams in the saturated gas plant and are mixed with input n-butanes in the C4 isomerization unit. The saturated gas plant will also extract

propane that can be sold as byproduct LPG or converted via aromatic alkylation. The C1−C2 fuel gases from the saturated gas plant are recycled for light gas handling (see Figure 10). Modeling of the C3−C5 alkylation unit and the C4 isomerization unit is based on a carbonnumber decomposition scheme, while the saturated gas plant uses known recovery factors for the C 3 and C 4 hydrocarbon species.24,116,117 The straight-run C5 FT gases may be split to a solid phosphoric acid (SPA)-catalyzed aromatic alkylation unit. This unit will react C3−C5 olefins with benzene or toluene to produce heavier aromatics that can be used within jet fuel blends. This unit provides a critical pathway for reduction of refinery benzene while simultaneously increasing the jet fuel yield from either high-temperature or low-temperature FT. The SPA unit assumes that the C3−C5 olefins will completely react to form either heavy aromatics or C6+ olefins. The input concentration of benzene and toluene to olefins will be 4:1 to maximize the output concentration of aromatics.114 Recycle of unreacted benzene and toluene to the SPA reactor will ensure that these aromatics are completely reacted. The SPA unit assumes that 95% of the olefins reacts with the C6−C7 aromatics to form heavier species, while the remaining 5% will form heavier C6−C10 olefins. The effluent from the SPA aromatic alkylation unit will be separated to recover the C9+ aromatic for jet fuel production. The aromatics are partially hydrotreated to meet the required specifications for jet fuel.114 The SPA naphtha fraction is comprised of the alkenes produced from the reaction of two C3−C5 olefins in the alkylation unit and is recycled back to the naphtha upgrading section for further processing (see Figure 6). The C3−C5 alkanes will be passed to a metal-promoted HZSM-5 catalytic reactor for aromatization. A portion of the propane may be distilled from this feed stream to provide byproduct LPG. Using a Ga/H-ZSM-5 catalytic process known as the Cyclar process, the C3 and C4 alkanes may be readily converted to aromatics using dehydrogenation and subsequent aromatization. Furthermore, the naphtha present in the refinery (i.e., straight-run FT naphtha, hydrocracked naphtha, and SPA naphtha) may also be converted readily to aromatics within the ZSM-5 unit. The liquid yield loss associated with aromatics production from naphtha typically prevents H-ZSM-5 catalytic technology from being widely used within standard 4311

dx.doi.org/10.1021/ef302003f | Energy Fuels 2013, 27, 4302−4324

Energy & Fuels

Article

Figure 9. LPG/gasoline separation flowsheet. refining processes. However, the potential to reduce the LPG within the refinery and produce additional liquid fuels makes metal-promoted H-ZSM-5 reforming an advantageous process. If the Cyclar process is used within the refinery, the process synthesis model can determine whether the loss of liquid yield from a naphtha feedstock is economically superior to the addition of a separate reforming unit for naphtha. 2.5. Methanol Synthesis. The syngas entering the methanol synthesis reactor must be sulfur-free to prevent catalyst poisoning.109 The reactor is shown in Figure 8 and will operate at 300°C and 50 bar. The syngas leaving the Rectisol unit will be compressed to 51 bar prior to entering the synthesis reactor. Equilibrium conditions are assumed to exist within the reactor between H2, CO, CO2, H2O, and CH3OH for the water-gas shift reaction (eq 14) and the methanol synthesis reaction (eq 13).109

CO + 2H 2 ↔ CH3OH

(13)

CO2 + H 2 ↔ CO + H 2O

(14)

that are tolerant to higher concentrations of water can eliminate the need for downstream purification. In this study, the ZSM-5-based reactors for methanol conversion are assumed to tolerate crude methanol containing up to 50 wt % water; therefore, no additional purification of methanol is necessary. These downstream processing units will produce approximately 50 wt % water for each unit mass of methanol reacted; therefore, high levels of water in the crude methanol input are not anticipated to be a concern.119 The methanol synthesis effluent is cooled to 35°C to flash out most of the entrained vapor at 48 bar. A total of 95 vol % of the methanol input to the flash unit is assumed to be recovered in the liquid phase along with water and some CO. The crude methanol exiting the flash unit is heated to 300°C to form a gas, which is then passed over a turbine to recover some of the electricity needed to recycle syngas. The turbine effluent at 5 bar is cooled to 60°C and then directed to a degasser distillation column to remove 100% of the CO2 and other entrained gases while recovering 99.9 wt % of the methanol. The entrained gases will be recycled to the process as fuel gas. The vapor exiting the flash unit will be split such that 95% of the stream is compressed to 51 bar and recycled to the methanol synthesis unit. The balance of the stream is purged as fuel gas to limit the build-up of inert species. 2.6. Methanol Conversion. Methanol produced in the synthesis reactor may be split to either the MTG or the MTO process. Methanol will be catalytically converted to gasoline-range hydrocarbons in the MTG process using a fluidized-bed reactor and a ZSM5 catalyst. The effluent of the MTG is described in Table 3.4.2 of a Mobil study120 and in process flow diagram P850-1402 of a NREL study.109 The composition of the MTG reactor in this paper was taken from the NREL report because of the high level of component detail provided by NREL. The MTG unit operates adiabatically at a temperature of 400°C and 12.8 bar, and the methanol feed will be pumped to 14.5 bar and heated to 330°C for input to the reactor. The input methanol is completely converted to 44 wt % water and 56 wt % crude hydrocarbons, of which 2 wt % will be light gas, 19 wt % will be C3−C4 gases, and 19 wt % will be C5+ gasoline.109 The crude hydrocarbons will ultimately be separated into finished fuel products, of which 82 wt % will be gasoline, 10 wt % will be LPG, and the balance will be recycle gases.109,120

This equilibrium between the five species will have important consequences on the downstream processing of the crude methanol and the hardwood BTL refinery in general. The syngas exiting the Rectisol unit will have a low composition of CO2, which will result in minimal formation of both CO2 and H2O in the methanol synthesis effluent. If methanol must be purified prior to conversion in subsequent units or if it is to be sold as a byproduct, then this is beneficial because of the reduced separation costs necessary with distillation. The per-pass conversion of CO to methanol can approach 45−50% depending upon the concentration of inert species, including CH4 and N2 that are present in the synthesis reactor. If the inlet concentration of CO2 were to increase, the reverse water-gas shift reaction will begin to be more prevalent and will generate additional output water through the reaction of CO2. The per-pass conversion of inlet carbon (CO + CO2) to methanol will also decrease. However, if H2, CO, and CO2 are inlet to the synthesis reactor such that the Ribblett ratio is equal to 1, then unreacted CO and CO2 can be recycled back to the unit to increase the per-pass conversion of CO + CO2 to high levels (e.g., >90%). The presence of water in the outlet methanol stream will increase the costs necessary for methanol purification, but the use of methanol conversion units 4312

dx.doi.org/10.1021/ef302003f | Energy Fuels 2013, 27, 4302−4324

Energy & Fuels

Article

Figure 10. Light gas handling flowsheet. Light gases are generated from (i) external recycle of FT synthesis, (ii) external recycle of methanol synthesis, and (iii) upgrading unit offgas. The gases may be split to an autothermal reformer to generate additional syngas, a fuel combustor to provide process heating, or a gas turbine to provide process electricity. The flue gas from the gas turbine and the fuel combustor may be directed to a post-combustion capture unit to recover CO2. to the process. The second knockout unit (HCKO2) will separate out CO2 from the recovered light gases via a one-stage Rectisol unit (CO2SEP) for sequestration or recycle back to additional process units. The CO2-lean light gases will be recycled back to the process. A deethanizer is used to remove any C1−C2 hydrocarbons, and the light gases exiting this column are sent to an absorber, where a lean oil recycle is used to strip the C3+ species. The liquid bottoms from the absorber are then refluxed back to the deethanizer, while the lights from the absorber are recycled as fuel gas. The bottoms from the deethanizer are sent to a stabilizer column to remove the C3−C4 hydrocarbons for alkylation to produce iso-octane and a LPG byproduct. The stabilizer bottoms are sent to a splitter column to recover the lean oil for recycle in the absorber column. Light and heavy gasoline fractions are recovered from the splitter. The distillation units are modeled within the process synthesis model as splitter units with a split fraction determined by process flow diagrams P850-A1501 and P850-A1502 from a NREL study.109 Low-pressure steam and cooling water necessary for all units are also derived from this study.109 The alkylate was modeled as isobutane,109 and the alkylation unit was modeled using a species balance, where the key species, butene, was completely converted to isobutane. Butene is used as the limiting species in this reaction because it is generally present in a far smaller concentration than isobutane. 2.8. Light Gas Handling. Light gases are generated from the BTL refinery as either unreacted synthesis gas from the FT reactor, unreacted synthesis gas from the methanol reactor, or saturated hydrocarbon fuel gases that leave the product upgrading. These gases will consist of H2, CO, CO2, H2O, C1−C2 hydrocarbons, and inert species (e.g., N2 and Ar). The source of the light gases will dictate the concentrations of each species, and the light gases from FT synthesis and methanol synthesis may have large fractions of H2 and CO that could be directly recycled to the synthesis units to increase the perpass conversion. This “internal” gas loop design helps to minimize the capital cost of the BTL refinery by reducing the process units necessary to transfer the unreacted syngas to the synthesis reactor inlet. In an internal gas loop, a compressor or heat exchanger are the

Methanol entering the MTO process is heated to 400°C and then enters the MTO fluidized-bed reactor operating at 482°C and a pressure of 1 bar. Low-pressure steam is used to control the exothermic heat of reaction from the MTO unit. Methanol is completely converted to a hydrocarbon product containing 1.4 wt % CH4, 6.5 wt % C2−C4 paraffins, 56.4 wt % C2−C4 olefins, and 35.7 wt % C5−C11 gasoline.119 The MTO unit is modeled mathematically using an atom balance and a typical composition seen in the literature.119 The MTO product is fractionated (MTO-F) to separate the light gases, olefins, and gasoline fractions. The MTO-F unit is assumed to operate as a separator unit, where 100% of the C1−C3 paraffins is recycled back to the refinery, 100% of the C4 paraffins and 100% of the olefins are directed to the Mobil olefins-to-gasoline/ distillate (MOGD) unit, 100% of the gasoline is combined with the remainder of the gasoline generated in the process, and 100% of the water generated in the MTO unit is sent for wastewater treatment. The MOGD process will take in the olefins generated from the MTO unit. Using a fixed-bed reactor, the olefins will be converted into gasoline and distillate over a ZSM-5 catalyst in a product ratio ranging from 0.12 to >100. In this study, the ratio was selected to be 0.12 to maximize the production of diesel. The MOGD unit operates at 400°C and 1 bar and will use low-pressure steam generation to remove the exothermic heat of reaction within the unit. The MOGD effluent will contain 82% distillate, 15% gasoline, and 3% light gases119 and will be fractionated (MTOD-F) to remove diesel and kerosene cuts from the gasoline and light gases. The MTODF unit will be modeled as a separator unit, where 100% of the C11−C13 species is directed to the kerosene cut and 100% of the C14+ species is directed to the diesel cut. 2.7. LPG/Gasoline Separation. The C3−C10 hydrocarbons produced by ZSM-5 upgrading of the FT hydrocarbons, the MTG unit, or the MOGD process must be sent to the LPG/gasoline separation flowsheet depicted in Figure 9. Each hydrocarbon stream is split and sent to a hydrocarbon knockout unit (35°C and 10 bar) for light gas removal via vapor−liquid equilibrium. The first knockout unit (HCKO1) will not incorporate additional CO2 separation; therefore, the CO2-rich light gases recovered from HCKO1 will be recycled back 4313

dx.doi.org/10.1021/ef302003f | Energy Fuels 2013, 27, 4302−4324

Energy & Fuels

Article

Table 3. Hardwood BTL Refinery Upgrading Unit Reference Capacities, Costs (2011 $), and Scaling Factors description hardwood biomass handling hardwood gasification, tar cracking, and gas cooling water-gas shift unit Rectisol unit Claus plant FT unit hydrocarbon recovery column methanol synthesis methanol degasser MTG unit MTO unit distillate hydrotreater kerosene hydrotreater naphtha hydrotreater wax hydrocracker naphtha reformer C5−C6 isomerizer C4 isomerizer C3−C5 alkylation unit Cyclar process SPA alkylation oligomerization saturated gas plant FT ZSM-5 reactor olefins-to-gasoline/diesel unit CO2 separation unit deethanizer absorber column stabilizer column splitter column HF alkylation unit LPG/alkylate splitter autothermal reformer pressure-swing adsorption air separation unit air compressor oxygen compressor electrolyzer gas turbine steam turbine sour stripper biological digestor reverse osmosis cooling tower

Co (MM$)

So

Smax

Hardwood Biomass Conversion 4.65 17.9 30.6 55.22 17.9 33.3 Synthesis Gas Handling/Clean Up 3.75 150 250 58.30 54.9 192.0 27.60 1.59 10.0 Hydrocarbon Production 12.26 23.79 60.0 0.65 1.82 25.20 8.22 35.647 3.82 11.169 5.80 10.60 3.48 10.60 Hydrocarbon Upgrading 2.25 0.36 81.90 2.25 0.36 81.90 0.68 0.26 81.90 8.42 1.13 72.45 63.53 27.27 94.50 0.86 0.15 31.50 9.50 6.21 52.29 12.64 115.16 16.42 8.99 0.61 7.83 4.23 4.93 10.60 3.48 10.60 5.39 8.54 0.58 5.13 0.91 0.96 1.03 4.57 1.01 3.96 8.99 0.61 1.06 0.61 10.26 12.2 35.0 Hydrogen/Oxygen Production 7.96 0.29 27.6 21.3 41.7 6.03 10 30 8.07 10 20 500 1 Heat and Power Integration 81.59 266 334 66.29 136 500 Wastewater Treatment 3.992 11.52 4.752 115.74 0.317 4.63 4.055 4530.30

only units necessary. This is contrary to an “external” gas loop, which would use one or more separation or reaction process units on the unreacted syngas prior to entry to the synthesis unit. The key drawback of the internal gas loop design is the build-up of inert species in the recycle loop, which could make operational costs prohibitively high if not properly addressed. The external gas loop design is capable of reducing or eliminating the inert species and provides a topological alternative for the unreacted syngas in the BTL refinery. The fraction of unreacted syngas that is directed to either the internal or external gas loop is a design consideration that must be optimized by the

scale basis

sf

kg/s kg/s

as-received hardwood dry hardwood

0.77 0.67

11 10

kg/s kg/s kg/s

feed feed gas recovered sulfur

0.67 0.63 0.67

110 11 121

kg/s kg/s kg/s kg/s kg/s kg/s

feed feed feed feed feed feed

0.72 0.70 0.65 0.70 0.65 0.65

112 and 113 116 110 110 110 and 120 120

kg/s kg/s kg/s kg/s kg/s kg/s kg/s kg/s kg/s kg/s kg/s kg/s kg/s kg/s kg/s kg/s kg/s kg/s kg/s kg/s kg/s

feed feed feed feed feed feed feed feed feed feed feed feed feed feed feed feed feed feed feed feed natural gas feed

0.60 0.60 0.65 0.55 0.60 0.62 0.60 0.60 0.67 0.65 0.60 0.65 0.65 0.62 0.68 0.68 0.68 0.68 0.65 0.68 0.67

116 116 116 116 118 116 116 116 123 110 116 112 120 120 110 110 110 110 110 110 116

kmol/s kg/s MW MW kW

purge gas O2 electricity electricity electricity

0.65 0.5 0.67 0.67 0.9

10 10 10 10 10

MW MW

electricity electricity

0.75 0.67

10 10

kg/s kg/s kg/s kg/s

feed feed feed feed

0.53 0.71 0.85 0.78

121 122 122 121

units

reference

and 120 and 113

and and and and and and

120 120 120 120 120 120

process synthesis model. Note that the saturated hydrocarbon light gas from the upgrading section will likely contain lower percentages of H2 and CO is likely to be optimally treated using an external gas loop design (e.g., autothermal reforming). Therefore, these light gases are directly recycled to the external gas loop. The external gas loop design is based on one of three major processing units, which are detailed in Figure 10. The first unit, the autothermal reformer, is designed to input steam and oxygen along with the light gases to generate an output rich in H2 and CO. This syngas can be directly recycled to the dual-capture acid-gas recovery 4314

dx.doi.org/10.1021/ef302003f | Energy Fuels 2013, 27, 4302−4324

Energy & Fuels

Article

unit to extract CO2 prior to delivering additional syngas to the FT synthesis or methanol synthesis. The light gases in the external gas loop may also be directed to a fuel combustor to provide hightemperature heat for the BTL refinery or to a gas turbine to generate electricity for use within the refinery or for byproduct resale. The effluent from these two units will be cooled to 35°C, passed through a water knockout unit, and may be directed to a low-pressure amine absorption CO2 recovery unit121 or simply vented to the atmosphere. CO2 that is recovered in the absorption unit may be recycled to the refinery to be consumed via the reverse water-gas shift reaction. 2.9. Hydrogen/Oxygen Production. The hydrogen requirement for the refinery will be met through pressure-swing adsorption or electrolysis of water. If electrolysis is used, then the generated oxygen will be used to satisfy the purified oxygen requirement. If not, then cryogenic air separation will be used. These three technologies are outlined in Supplementary Figure S11 of the Supporting Information. 2.10. Wastewater Treatment. Supplementary Figures S12 and S13 of the Supporting Information detail the complete wastewater treatment network that is incorporated into the hardwood BTL refinery. The wastewater from multiple process units, including acidgas wastewater knockout, hydrocarbon production, hydrocarbon upgrading, and post-combustion knockout, will be directed to either a sour stripper or a biological digestor to remove H2, NH3, or oxygenates that will be in the waste streams. The sulfur-rich effluent from the biological digestor and sour stripper are directed to the Claus combustor to recover and remove the sulfur while simultaneously providing additional heat for steam production. The blowdown from the cooling tower and the boilers is processed in a reverse osmosis unit to remove suspended and dissolved solids prior to wastewater discharge. The output water from the network includes any process water to the electrolyzers, refinery steam, or discharged wastewater. The discharge must meet all appropriate standards for contaminant species.25 2.11. Unit Costs. The total direct cost (TDC) for the hardwood BTL refinery units is calculated using estimates from several literature sources10,11,109,111−113,116,118,120,122,123 using the cost parameters in Table 3 and eq (15)

TDC = (1 + BOP)Co

Sr sf So

fuels production and includes contributions from the feedstocks cost (costF), the electricity cost (costEl), the CO2 sequestration cost (costSeq), and the levelized unit investment cost (costU). Because there is no sequestration option considered in this study, the value of costSeq will be zero. Each of the terms in eq (18) is normalized to the total volume of products produced (prod). Note that other normalization factors (e.g., total volume of gasoline equivalent and total energy of products) and other objective functions (e.g., maximizing the net present value) can be easily incorporated into the model framework. MIN +

cost uU (18)

The process synthesis model with simultaneous heat, power, and water integration represents a large-scale non-convex MINLP model that was solved to global optimality using a branch-and-bound global optimization framework.26 The large-scale model has 21 binary variables, 14 128 continuous variables, 14 012 constraints, and 321 non-convex terms. At each node in the branch-and-bound tree, a mixed-integer linear relaxation of the mathematical model is solved using CPLEX,125 and then the node is branched to create two children nodes. The solution pool feature of CPLEX is used during the solution of the relaxed model to generate a set of distinct points (150 for the root node and 10 for all other nodes), each of which is used as a candidate starting point to solve the original model. For each starting point, the current binary variable values are fixed and the resulting non-linear optimization (NLP) is minimized using CONOPT.126 If the solution to the NLP is less than the current upper bound, then the upper bound is replaced with the NLP solution value. At each step, all nodes that have a lower bound that is within an ε tolerance of the current upper bound [(LBnode/UB) ≥ 1 − ε] are eliminated from the tree. For a more complete coverage of branch-and-bound algorithms, the reader is directed to the textbooks by Floudas100,127 and reviews of global optimization methods.128−130

3. COMPUTATIONAL STUDIES A total of 12 case studies were performed to demonstrate the capability of the proposed process synthesis model using an average representation of hardwood residues (Table 1). The global optimization framework was terminated if all nodes in the branch-and-bound tree were processed or if 100 central processing unit (CPU) hours had passed.26 The effect of scale on the hardwood BTL refinery was examined through four representative capacities of 0.8 kBD, 1 kBD, 2.5 kBD, and 10 kBD. The gasoline, diesel, and kerosene compositions output from the refinery were selected to either (a) represent the 2010 U.S. demand (i.e., 67 vol % gasoline, 22 vol % diesel, and 11 vol % kerosene),9 (b) maximize the diesel production (i.e., ≥75 vol %), or (c) maximize the kerosene production (i.e., ≥75 vol %). The case studies are labeled as N-C, where N represents the type of product composition (i.e., R, 2010 U.S. ratios; D, maximum diesel; and K, maximum kerosene) and C represents the capacity in kBD. For example, the K-1 label represents the 1 kBD capacity refinery with a kerosene production that is at least 75 vol % of the total amount of C5+ hydrocarbon products. The lifecycle greenhouse gas (GHG) emissions from the refinery will be measured for each case study and then compared to the emissions of current fossil-fuel-based processes. The GHG emissions avoided from petroleum-based liquid fuels (91.6 kg of CO2 equiv/GJLHV) and natural gas combined cycle plant electricity (101.3 kg of CO2 equiv/GJ) will also be determined. If electricity is input to the refinery, then the associated GHG emissions with electricity production are added to the lifecycle GHG emissions for the refinery. If electricity is output from the

(15)

Kreutz et al.11 calculates a LCCR value of 14.38%/year and an IDCF of 7.6%. Thus, a multiplier of 15.41%/year is used to convert the TPC into a CC rate. Assuming an operating capacity (CAP) of 330 days/ year and operation/maintenance (OM) costs equal to 5% of the TPC, the total levelized cost (costU) associated with a unit is given by eq (17).

⎛ LCCR × IDCF OM ⎟⎞⎛ TPCu ⎞ ⎜ + ⎟ ⎜ ⎝ CAP 365 ⎠⎝ prod ⎠

∑ u ∈ UInv

where Co is the installed unit cost, So is the base capacity, Sr is the actual capacity, sf is the cost scaling factor, and BOP is the balance of plant percentage (site preparation, utility plants, etc.).11 The BOP is estimated to be 20% of the total installed unit cost. All capital cost numbers are converted to 2011 dollars using the Chemical Engineering Plant Cost Index.124 The TPC for each unit is calculated as the sum of the total direct capital (TDC) plus the indirect costs (ICs). The ICs include engineering, startup, spares, royalties, and contingencies and are estimated to 32% of the TDC. The TPC for each unit must be converted to a levelized cost to compare to the variable feedstock and operational costs for the process. Using an Electric Power Research Institute (EPRI) Technical Assessment Guide (TAG) methodology,11 the capital charges (CCs) for the refinery are calculated by multiplying the levelized capital charge rate (LCCR) and the interest during construction factor (IDCF) by the total overnight capital (eq 16). (16) CC = LCCR × IDCF × TPC

cost uU =

costFBio + cost FH2O + costFBut + costEl + costSeq

(17)

2.12. Objective Function. The objective function for the model is given by eq (18). The summation represents the total cost of liquid 4315

dx.doi.org/10.1021/ef302003f | Energy Fuels 2013, 27, 4302−4324

Energy & Fuels

Article

(x) the split of the clean synthesis gas to the various process operations. Table 5 demonstrates that these topological decisions are strongly dependent upon both the operating capacity of the refinery and the choice of liquid fuel composition output from the refinery. For all 12 case studies, a raw synthesis gas will be produced from the hardwood biomass via gasification. The gasifier may input only hardwood as a feedstock or a combination of hardwood and recycle gases. The benefit of feeding recycle gases to the gasifier is based on the ability to consume CO2 via the reverse water-gas shift reaction in the high-temperature gasifier. It is noteworthy that all 12 case studies do not use a recycle gas feed and only input solid hardwood to the gasifier. The H2/CO ratio exiting the gasifier will be less than that required for FT synthesis or methanol synthesis. A forward water-gas shift reaction downstream of the gasifier will be required to bring the ratio to approximately 2:1. This reaction will generate additional process CO2 that must be vented because it cannot be economically recycled back to the hardwood BTL refinery for consumption via the reverse water-gas shift reaction, because this would effectively decrease the H2/CO ratio in the gasifier effluent. A non-carbon-based source of H2 could be used as an input to the refinery to both increase the H2/CO ratio and decrease the refinery CO2 via the reverse water-gas shift reaction. However, typical processes, including electrolysis of water, have prohibitively high capital and operating costs that prevent this option from being viable in the refinery. Three possible temperature options were considered for the gasifier (900, 1000, or 1100°C). The case studies that output fuels consistent with the U.S. demand ratios have a gasifier operating at 900°C for the 0.8, 1, and 2.5 kBD capacities and 1000°C for the 10 kBD capacity. The maximum diesel cases used a 1000°C gasifier for all capacities, and the maximum kerosene cases used a 1100°C gasifier for all capacities. These case studies show an important trade-off in the utility and capital costs associated with the gasifier and the other process units in the refinery. The selection of a lower operating temperature will require less oxygen, produce less waste heat from syngas cooling, and may have higher levels of CO2 because of the higher value of the forward water-gas-shift equilibrium constant. Higher operating temperatures will need more oxygen, but the ability to reduce CO2 in the effluent of the gasifier can potentially increase the overall carbon conversion efficiency of the refinery. A dedicated water-gas

refinery, the avoided GHG emissions are subtracted from the total lifecycle GHG emissions. The cost parameters used for the refinery are listed in Table 4. The costs for feedstocks (i.e., hardwood biomass, freshwater, Table 4. Cost Parameters (2011 $) for the Hardwood BTL Refinery item hardwood residues butanes electricity

cost

item

cost

$70/dry metric ton

freshwater

$0.50/metric ton

$1.84/gallon $0.07/kWh

propanes

$1.78/gallon

and butanes) include all costs associated with delivery to the plant gate. The products (i.e., electricity and propane) are assumed to be sold from the plant gate and do not include the costs expected for transport to the end consumer. CO2 sequestration was not considered in the 12 case studies. Once the global optimization algorithm has completed, the resulting process topology provides (i) the operating conditions and working fluid flow rates of the heat engines, (ii) the amount of electricity produced by the engines, (iii) the amount of cooling water needed for the engines, and (v) the location of the pinch points denoting the distinct sub-networks. Given this information, the minimum number of heat-exchanger matches necessary to meet specifications (i)−(iv) are calculated as previously described.24,25,127,131 Upon solution of the minimum matches model, the heat-exchanger topology with the minimum annualized cost can be found using the superstructure methodology.90,127,131 The investment cost of the heat exchangers is added to the investment cost calculated within the process synthesis model to obtain the final investment cost for the superstructure. 3.1. Optimal Process Topologies. Table 5 shows the major topological decisions that are selected by the process synthesis framework for each of the 12 case studies. Key decisions feature (i) the type of hardwood gasifier, (ii) the operating temperature of the hardwood gasifier, (iii) the selection of a water-gas shift unit, (iv) the operating temperature of the water-gas shift unit, (v) the selection of synthesis gas conversion technologies, (vi) the type of upgrading for FT synthesis, (viii) the selection of an autothermal reformer for synthesis gas production, (ix) the operating temperature of the advanced test reactor (ATR), and

Table 5. Topological Information for the Optimal Solutions for the 12 Case Studiesa topological design

R-0.8

R-1

R-2.5

hardwood conversion gasifier temperature WGS/RGS temperature min wax FT nominal wax FT FT upgrading MTG usage MTOD usage GT usage

S 900 450

S 900 450

S 900 450

Y Y

Y Y

Y Y

R-10 S 1000 450 ir-LTFT fraction Y

D-0.8

D-1

D-2.5

D-10

S 1000 450

S 1000 450

S 1000 450

S 1000 450

Y Y

Y Y

Y Y

K-0.8

K-1

K-2.5

K-10

S 1100 450

S 1100 450

S 1100 450

S 1100 450

co-LTFT fraction

co-LTFT fraction

co-LTFT fraction

co-LTFT fraction

Y Y

a Hardwood biomass conversion (hardwood conversion) is gasification with a solid (S) or solid/vapor (S/V)-fueled system. The temperature (°C) of the hardwood gasification is selected along with the operating temperature of the water-gas shift unit (WGS), if used. The presence of a gas turbine (GT) is noted using yes (Y) or no (N). The FT units will operate at low temperature (LT) or high temperature (HT) with a cobalt (Co) or iron (Ir) catalyst. The FT vapor effluent will be upgraded using fractionation into distillate and naphtha (fraction) or ZSM-5 catalytic conversion. The use of MTG and MTO/MOGD is noted using yes (Y) or no (blank space).

4316

dx.doi.org/10.1021/ef302003f | Energy Fuels 2013, 27, 4302−4324

Energy & Fuels

Article

Figure 11. First process flow diagram for case study K-1.

topological switch for the highest capacity refinery. The case studies that look at the smallest three capacities (i.e., 0.8, 1, and 2.5 kBD) will use methanol synthesis, and then methanol will be split between the MTG and MTO processes. Diesel and jet fuel will be produced from the MOGD process, while gasoline is produced from the MTG process. At a capacity of 10 kBD, the MTO/MOGD processes are no longer used to generate diesel and jet fuel. In this study, a portion of the clean syngas is split to the iron-based FT unit operating with a Ribblett ratio114,115 of 1. Similar to the cobalt-based FT unit, the ironbased low-temperature FT unit will produce a large quantity of wax and distillate that can be used to produce diesel and jet fuel. Methanol synthesis is still used within the study as an intermediate before entering the MTG process to create gasoline. The methane- and ethane-rich gases generated from the process upgrading units are completely directed to the fuel combustor to provide a fraction of the necessary heating requirement for reducing the hardwood moisture. Additionally, a portion of the clean syngas (less than 10%) is directed to the fuel combustor to provide the balance of the hardwood dryer heating requirement. The energy required to reduce the hardwood moisture from 45 to 20 wt % prevents the use of an autothermal reactor or a gas turbine from being selected within the refinery. The combustion of all recycle C1 and C2 hydrocarbon-rich light gases is still insufficient for satisfaction of the full hardwood dryer heating requirement. Therefore, any diversion of the recycle gas to the autothermal reactor to produce synthesis gas will also increase the quantity of synthesis gas required for combustion. Although diversion of the recycle gas to the gas turbine can increase the electricity produced by the refinery, it is not economical to do so given the effect of increased synthesis gas combustion on the cost of the refinery.

shift is used in all case studies to increase the H2/CO ratio for syngas conversion. The unit operates at 450°C for all case studies. The unit operates isothermally, and all excess heat is used to generate low-pressure steam for use throughout the refinery. The hydrocarbon generation and hydrocarbon upgrading units that are used in the refinery are strongly dependent upon the composition of the liquid fuels. For the case studies that maximize diesel production, methanol synthesis is used to first generate a methanol intermediate from the synthesis gas. Methanol is converted to hydrocarbons via the MTO process, and the olefins are subsequently upgraded to mostly distillaterange hydrocarbons via the MOGD process. In each of the four maximum diesel case studies, a total of 75 vol % diesel was produced and the remaining amount of C5+ hydrocarbons was output as gasoline. A small quantity of LPG is output from the system and sold as a byproduct. The case studies that maximized the kerosene output used the cobalt-based low-temperature FT units to generate a raw hydrocarbon product that is rich in long carbon chain waxes. A combination of wax/distillate hydrocracking, SPA oligomerization, M/H-ZSM-5 aromatic alkylation, and RZ naphtha platforming is used to create output products that have 75 vol % kerosene and 25 vol % gasoline. The gasoline is an aromatic-rich product that must be blended to meet all physical property specifications. Through C5 isomerization, the gasoline produced from the refinery can meet all specifications but the kerosene output would be reduced to 60 vol %.114 For the maximum diesel and maximum kerosene case studies, the hydrocarbon production and upgrading topology was constant for a given output fuel requirement, regardless of the selection of the refinery capacity. The four case studies that output fuels equal to the U.S. demand ratios show a defined 4317

dx.doi.org/10.1021/ef302003f | Energy Fuels 2013, 27, 4302−4324

Energy & Fuels

Article

Figure 12. Second process flow diagram for case study K-1.

Table 6. Overall Cost Results for the 12 Case Studiesa ($/GJ of products)

R-0.8

R-1

R-2.5

R-10

D-0.8

D-1

D-2.5

D-10

K-0.8

K-1

K-2.5

K-10

hardwood butane water investment OM electricity LPG total ($/GJ) BEOP ($/bbl) lower bound ($/GJ) gap (%)

5.59 0.01 13.77 3.23 0.35 −0.24 22.70 116.60 21.48 5.37

5.40 0.01 12.67 2.97 0.34 −0.24 21.14 107.69 20.00 5.40

5.28 0.01 9.89 2.32 0.33 −0.25 17.58 87.42 16.47 6.33

5.30 0.01 6.37 1.49 0.35 −0.25 13.27 62.84 12.49 5.91

5.55 0.01 13.02 3.05 0.50 −0.20 21.93 112.21 20.54 6.36

5.38 0.01 11.54 2.71 0.50 −0.19 19.94 100.86 18.69 6.29

5.36 0.01 9.70 2.28 0.51 −0.21 17.65 87.79 16.55 6.24

5.34 0.01 6.54 1.53 0.49 −0.20 13.71 65.35 12.99 5.27

5.74 0.01 13.76 3.23 0.50 23.25 119.71 21.79 6.26

5.33 0.01 12.87 3.02 0.48 21.71 110.96 20.20 6.97

5.32 0.01 9.78 2.30 0.50 17.90 89.24 16.79 6.22

5.67 0.01 6.45 1.51 0.48 14.13 67.72 13.15 6.87

a

The contribution to the total costs (in $/GJ) comes from hardwood biomass, natural gas, butanes, water, investment, OM, and electricity. LPG is sold as byproducts (negative value). The overall costs are reported on a $/GJ and $/bbl basis, along with the lower bound values in $/GJ and the optimality gap between the reported solution and the lower bound.

As an illustrative example, the process flow diagram for the K-1 case study is shown in Figures 11 and 12. Hardwood biomass is dried and lockhopped into the gasifier using captured CO2. Oxygen from an air separation unit and steam are input to the gasifier to produce a raw syngas. The syngas is passed through a tar cracker and cooled before having the H2/ CO ratio shifted to 2:1 in a forward water-gas shift unit. The syngas is scrubbed to remove NH3 and residual tar in a wastewater stream and then passed over a dual-capture Rectisol system. The sulfur-rich gases exiting the Rectisol unit are sent to a Claus plant to recover sulfur, and CO2 is mostly vented from the system. The clean gas is passed to Figure 12, where it enters a cobalt-based low-temperature FT synthesis reactor. The wax and all C10+ vapor-phase hydrocarbons are directed to a hydrocracker to generate naphtha and jet fuel. Recycle of all hydrocarbons above the appropriate boiling range for jet fuel (i.e., ≥360°C) ensures that the production of jet fuel from the hydrocracker is maximized. The cracked naphtha is mixed with (a) straight-run C6−C9 naphtha from the FT unit and (b) LPG/naphtha from the SPA alkylation oligomerization unit in

the M/H-ZSM-5 aromatization reactor to produce (i) the aromatic component of the jet fuel, (ii) an aromatic-rich naphtha, and (iii) a H2-rich fuel gas. Benzene is distilled from the aromatic naphtha and combined with the olefinic C3−C5 from the FT vapor phase to produce a cyclic hydrocarbon jet fuel after hydrotreating. Hydrogen is provided for the hydrocracker and hydrotreater using pressure-swing adsorption. The C1−C2 offgas from the upgrading units is sent to a fuel combustor to provide the heat for hardwood drying. 3.2. Overall Costs of Liquid Fuels. The overall cost of liquid fuel production is based on the sum of the individual cost components for the hardwood BTL refinery. These components include purchase of the hardwood and freshwater feedstocks, the capital charges associated with investment cost, the OM costs, and the cost of electricity. The sum of these costs indicates the total cost of the refinery and can be partially offset by the sale of byproduct LPG. Table 6 shows each of these contributions to the total cost and expresses all results in $/GJ of liquid product LHV. Using the refiner’s margin for gasoline, diesel, and jet fuel,11,24 the total cost in $/GJ may be 4318

dx.doi.org/10.1021/ef302003f | Energy Fuels 2013, 27, 4302−4324

Energy & Fuels

Article

geographical variability in biomass availabilities and prices as well as the transportation costs for each delivery will be addressed explicitly in part 2 of this study, where a nationwide supply chain network for the BTL systems will be reported. A parametric analysis on the hardwood purchase price was performed using values of $50, $70, and $90/dry metric ton. The resulting values of the BEOP for these three purchase prices are plotted in Figure 13 for the three 2.5 kBD refineries.

converted into a price of crude oil at which value the refineries would be economically competitive with petroleum-based processes. This break-even oil price (BEOP) is expressed in $/barrel (bbl) of oil and is also listed in Table 6. The final lower bound on the total cost (in $/GJ) that was calculated by the global optimization framework is listed in the table along with the corresponding optimality gap that is calculated by dividing the lower bound by the total cost and subtracting from unity. The global optimization framework was able to close the optimality gap to between 5% and 7% for all case studies. The BEOP ranges from $112 to $120/bbl for a 0.8 kBD plant, from $101 to $111/bbl for a 1 kBD plant, from $87 to $89/bbl for a 2.5 kBD plant, and from $63 to $68/bbl for a 10 kBD plant. The costs associated with the refinery capital investment (i.e., capital charges and OM) contribute the largest fraction to the overall cost for each capacity level. At refinery capacities of 0.8 and 1 kBD, the investment and OM charges represent 71−75% of the total cost of the system. As the refinery capacity increases, this percentage decreases to 67− 70% for the 2.5 kBD refineries and 56−59% for the 10 kBD refineries. This result is due to a decrease in the absolute cost associated with investment or OM at increased capacity and is explicitly a function of the expected economy of scale achieved from 0.8 to 10 kBD. The other major cost component, the hardwood purchase price, maintains a roughly constant absolute cost value between $5.28 and $5.74/GJ across all capacity levels. Therefore, the relative contribution of the hardwood will begin to increase as the refinery capacity increases. Moreover, because the levelized cost values (i.e., the total cost amount divided by the amount of fuels produced at each capacity in GJ of LHV) of hardwood, electricity, water, or LPG (if output) are not affected by capacity, the benefit of economy of scale is a strong motivating factor for the cost decrease from 0.8 to 10 kBD. No butane input is required in all case studies. The decrease in the investment and OM charges is very pronounced at these low capacity levels because the major sections within the refinery will require only one or two units to operate. Thus, there is a large benefit for increasing the size of these units to their maximum operational capacity to secure the cost benefit. However, as the refinery capacity begins to grow beyond 5−10 kBD, most of the units within the refinery will not be able to handle the working capacity using only one processing unit. The introduction of two or more units (e.g., two or more gasifiers) begins to erode the benefits of economy of scale. Although multiple processing units can share some auxiliary equipment, the scaling factor for such units will begin to approach 0.911 as opposed to the lower values of 0.5−0.75 estimated in Table 3. The contribution of the hardwood purchase price to the BEOP for a particular refinery will range from 25% to 40% depending upon the capacity and liquid fuel composition that is selected. Note that other parameters, such as the biomass composition and the capital cost of process units, also introduce uncertainties in the BEOP. In this paper, it is assumed that hardwood at 45% moisture level can be supplied and that the degree of uncertainty associated with the capital cost estimations is similar to standard cost estimation for power plants and energy systems.111 Addressing the process synthesis problem under uncertainty will be the subject of future work. A nominal cost of $70/dry metric ton was assumed for hardwood, although local fluctuations in total supply, ease of production, and available transportation infrastructure can have an effect on the delivered cost of this feedstock.132 The

Figure 13. Parametric analysis of hardwood biomass purchase price on 2.5 kBD case studies.

The BEOP ranges from $79 to $81/bbl for a $50/dry metric ton hardwood price, between $87 and $89/bbl for $70/dry metric ton hardwood, and between $96 and $98/bbl for $90/ dry metric ton hardwood. A clear trend from the analysis shows that an increase or decrease in hardwood purchase price changes the BEOP for the three liquid fuel compositions in a similar fashion. This is expected because the absolute value of the hardwood price was similar for the different liquid fuel compositions. The U.S. ratios and maximum diesel case studies show a BEOP that is almost equivalent, while the maximum kerosene case study had a BEOP that was consistently $2/bbl higher. Similar results can be obtained from the three other capacities (i.e., 0.8, 1, and 10 kBD) because the hardwood contribution to the overall cost maintained the same trends as the 2.5 kBD capacity. 3.3. Investment Costs. The total plant cost (TPC) for a hardwood BTL refinery included the installed cost of all major pieces of equipment along with the balance of plant items and indirect costs. The TPC may be converted into a “total overnight capital” by adding the inventory capital, financing costs, pre-production costs, and other owner’s costs, which typically range from 10% to 15% of the TPC.11,111 The “total as-spent capital” figure, which incorporates the interest on the debt/equity during construction, can be calculated using the real discount rate and the construction time. This information has been accounted for when determining the levelized capital charge rate used in Table 6. For the seven major sections of the refinery, the contribution to the TPC (in 2011 MM$) is shown in Table 7. These sections include syngas generation, syngas cleaning, hydrocarbon generation, hydrocarbon upgrading, heat and power integration, wastewater treatment, and hydrogen/ oxygen production. Each section includes the cost of several process units that are listed above in Table 3. For each case 4319

dx.doi.org/10.1021/ef302003f | Energy Fuels 2013, 27, 4302−4324

Energy & Fuels

Article

Table 7. Breakdown of the Investment Costs for the 12 Case Studiesa refinery section syngas generation syngas cleaning hydrocarbon production hydrocarbon upgrading hydrogen/oxygen production heat and power integration wastewater treatment total (MM$) total ($/bpd)

R-0.8

R-1

R-2.5

R-10

D-0.8

D-1

D-2.5

D-10

K-0.8

K-1

K-2.5

K-10

37 27 22 10 16

44 31 25 11 19

86 61 47 23 37

216 156 119 61 101

36 25 20 10 17

40 33 21 6 18

121 60 39 8 23

354 147 111 16 77

37 28 21 10 17

44 33 24 12 20

83 62 46 23 37

218 164 122 59 98

15

16

33

83

12

16

29

84

14

17

32

85

8 155 154558

16 302 120741

42 777 77696

7 127 158841

7 141 140791

15 296 118402

8 797 79745

7 134 167963

8 157 157080

16 298 119328

41 787 78727

7 134 167999

a

The major sections of the plant include the syngas generation section, syngas cleaning, hydrocarbon production, hydrocarbon upgrading, hydrogen/ oxygen production, heat and power integration, and wastewater treatment blocks. The values are reported in MM$ and normalized with the amount of fuels produced ($/bpd).

Table 8. Overall Input/Output Balances for the 12 Case Studiesa R-0.8

R-1

R-2.5

hardwood (dt/h) water (kBD) gasoline (kBD) diesel (kBD) kerosene (kBD) LPG (kBD) vented CO2 (ton/h)

16.67 0.53 0.54 0.17 0.09 0.02 11.11

20.14 0.46 0.67 0.22 0.11 0.01 12.20

49.23 1.15 1.68 0.54 0.28 0.06 28.67

hardwood gasoline diesel kerosene LPG electricity efficiency (%)

94 34 12 6 1 1 56.68

113 43 15 8 1 1 58.09

277 107 38 19 3 3 60.03

hardwood gasoline diesel kerosene LPG vented CO2 percent conversion (%)

1.77 0.62 0.23 0.11 0.01 0.84 55.12

2.14 0.78 0.28 0.14 0.01 0.93 56.68

5.22 1.95 0.71 0.35 0.04 2.17 58.38

hardwood gasoline diesel kerosene LPG vented CO2 LGHG GHGAF GHGAE GHGI

−5.48 2.28 0.83 0.41 0.05 3.09 1.18 4.93 −0.10 0.24

−6.62 2.85 1.04 0.51 0.03 3.39 1.20 6.10 −0.12 0.20

−16.19 7.13 2.61 1.28 0.15 7.96 2.94 15.41 −0.29 0.19

R-10

D-0.8

D-1

D-2.5

Material Balances 197.71 16.58 20.09 50.00 4.60 0.37 0.46 1.15 6.72 0.20 0.25 0.63 2.15 0.60 0.75 1.88 1.13 0.15 0.01 0.01 0.04 116.47 9.54 11.38 28.18 Energy Balances (MW) 1114 93 113 282 428 13 16 40 153 43 53 133 78 9 1 1 2 12 1 2 4 59.39 58.93 60.76 61.04 Carbon Accounting (kg/s) 20.97 1.76 2.13 5.30 7.78 0.23 0.29 0.72 2.85 0.79 0.99 2.48 1.40 0.11 0.01 0.01 0.03 8.83 0.72 0.86 2.14 57.90 58.88 59.50 59.71 Lifecycle Analysis (kg of CO2 equiv/s) −65.02 −5.45 −6.61 −16.44 28.52 0.85 1.06 2.65 10.43 2.91 3.63 9.08 5.12 0.42 0.04 0.05 0.12 32.35 2.65 3.16 7.83 11.82 0.95 1.25 3.11 61.25 5.08 6.35 15.86 −1.20 −0.14 −0.17 −0.43 0.20 0.19 0.20 0.20

D-10

K-0.8

K-1

K-2.5

K-10

199.33 4.60 2.50 7.50 0.15 111.69

17.14 0.37 0.20 0.60 11.10

19.90 0.46 0.25 0.75 11.75

49.60 1.15 0.63 1.88 29.16

211.72 4.60 2.50 7.50 135.30

1123 159 533 9 17 61.26

97 13 42 1 55.43

112 16 52 2 59.65

279 40 130 4 59.78

1193 159 519 16 56.11

21.14 2.90 9.91 0.13 8.47 59.95

1.82 0.23 0.74 0.84 53.70

2.11 0.29 0.93 0.89 57.80

5.26 0.72 2.33 2.21 57.98

22.46 2.90 9.30 10.26 54.33

−65.55 10.61 36.32 0.49 31.03 12.39 63.45 −1.67 0.20

−5.64 0.85 2.73 3.08 1.02 4.97 −0.14 0.21

−6.54 1.06 3.41 3.26 1.19 6.21 −0.16 0.20

−16.31 2.65 8.52 8.10 2.97 15.54 −0.43 0.20

−69.63 10.61 34.09 37.58 12.66 62.15 −1.64 0.21

a

The inputs to the BTL refinery are hardwood biomass, electricity, and water, while the outputs include gasoline, diesel, kerosene, LPG, and vented CO2. Material balances are represented in the listed units; energy balances are based on LHV and are shown in MW; carbon accounting is shown in kg/s; and lifecycle GHG emissions are shown in kg of CO2 equiv/s. Process efficiency is calculated by dividing the energy sum of the outputs by the sum of the inputs. Carbon conversion is calculated by dividing the sum of the carbon leaving in liquid product by the sum input to the process. The total GHG emissions (LGHG) are reported along with the emissions avoided from liquid production (GHGAF), the emissions attributed by electricity use (GHGAE), and the emissions index (GHGI).

4320

dx.doi.org/10.1021/ef302003f | Energy Fuels 2013, 27, 4302−4324

Energy & Fuels

Article

balances, carbon accounting, and lifecycle emissions. For the energy balances, the key input to the process is the hardwood feedstock, which represents approximately 98−99% of the total. The balance of the energy is supplied by electricity, which is required by the process units. For all of the case studies, waste heat is used to generate electricity on-site using steam turbines, but this electricity requirement is insufficient to satisfy the total demand of the refinery. This factor is directly attributed to the additional heat that is required to dry the hardwood, which restricts the heat that can be delivered to the steam cycle. All 12 case studies report refineries that have a efficiency between 55.4% and 61.3%. The efficiency of the case studies varies depending upon the composition of the liquid products. The four maximum diesel case studies have higher efficiencies for any capacity than the corresponding other two fuel compositions. The U.S. ratio case studies typically provided the next highest efficiency, although the K-1 case study was slightly higher than the R-1 case study (59.6% versus 58.1%). Note that this result demonstrates an important trade-off between efficiency and economics for the lower process capacities. Because each case study is being optimized to minimize the overall cost, a trade-off to reduce the cost for a slight reduction in process efficiency would be favored in these systems. The overall carbon balance (in kg/s) is shown in Table 8 for the six major points in the refineries, where carbon is input or output from the system. The trends evident from the material balances are consistent with the associated carbon balances. The flow of carbon associated with each product is consistent with the volumetric flow rate of each product because the percentage of carbon in the liquid products is roughly similar. The total amount of carbon output in the LPG, gasoline, diesel, and kerosene products is therefore approximately constant in each case study. The carbon conversion efficiency is around 54−60% for all case studies, which are slightly higher than those reported in previous BTL studies.10,109 The high levels of internal synthesis gas recycle in the hardwood BTL refineries to the methanol synthesis and FT synthesis reactors play a critical role in improving the overall conversion efficiency, although this does cause a decrease in the net electricity produced (or increase in net consumed) because of the increase of compression costs. However, the refinery designs that target this higher level of carbon conversion ultimately lead to reduced overall costs for the entire plant. Table 8 also lists the major sources of lifecycle GHG emissions in the refinery. No GHG emissions target was set for the refineries because these processes will inherently have reduced GHG emissions with respect to fossil-fueled processes. The total lifecycle GHG emissions (LGHG) for each case study are defined as the sum of the total emissions from each stage of the process. The GHG emission rates (in kg of CO2 equiv/s) include (a) acquisition and transportation of the hardwood feed, (b) transportation and use of the gasoline, diesel, kerosene, and LPG, (c) venting of any process emissions, and (d) atmospheric sequestration of CO2 during growth of hardwood that occurs because of photosynthesis. The GREET model for well-to-wheel emissions133 is used to calculate the GHG emissions for feedstock acquisition and transportation in part (a) and product transportation in part (b). Transportation distances of 50 miles are assumed for feedstocks, and transportation distances of 100 miles are assumed for products. The final liquid products are assumed to be completely combusted to generate CO2 that is released into

study, the cost to produce an acid-gas-free syngas from the hardwood biomass constitutes 48−51% of the overall cost of the refinery. This includes the costs associated with syngas generation (e.g., hardwood gasification and drying), along with the costs of syngas cleaning (e.g., tar cracking and acid-gas removal). The third major cost component is the hydrocarbon production, which includes FT synthesis or methanol synthesis and the appropriate costs of internal recycle of the synthesis gas. The heat and power integration section involves all of the process heat exchangers and the steam turbines and contributes approximately 10−12% to the overall cost. A similar percentage is attributed to the hydrogen and oxygen production. The wastewater treatment is associated with about 5% of the cost. The value of TPC ranges from $127 to $134 MM for 0.8 kBD plants, from $141 to $157 MM for 1 kBD plants, from $296 to $302 MM for 2.5 kBD plants, and from $777 to $797 MM for 10 kBD plants. The TPC is divided by the total plant capacity and expressed as a value in $/barrel per day ($/bpd). This normalized value helps to illustrate the economy of scale associated with the refineries and ranges from $157 000 to $168 000/bpd for 0.8 kBD plants, from $141 000 to $157 000/ bpd for 1 kBD plants, from $118 000 to $121 000/bpd for 2.5 kBD plants, and from $78 000 to $80 000/bpd for 10 kBD plants. The clear economy of scale between 0.8 and 1 kBD is emphasized with these normalized values because a 25% increase in the plant capacity results in an 8−11% decrease in the cost required per unit capacity. This benefit continues for 2.5 and 10 kBD but is not as pronounced as it would be for low-capacity levels. The case studies that maximize the diesel production generally have the lowest capital investment, although the 10 kBD case study is higher than both the U.S. ratio and the maximum kerosene study. 3.4. Overall Process Input/Output Results. The material and energy balances for all case studies are displayed in Table 8. The units for the material balances are dry metric tons per hour (dt/h) for the hardwood and kBD for the liquid feedstocks and products (i.e., butanes, LPG, water, gasoline, diesel, and jet fuel). All components for the energy balances are expressed in MW and are based on the LHV for all feedstocks and products associated with the hardwood BTL refinery. The efficiency of the processes are based on the sum of the LHV of the products divided by the LHV of the feedstocks. Note that electricity is input for all 12 case studies; therefore, this energy is added to the denominator of the efficiency fraction. The carbon accounting and lifecycle GHG emissions for the case studies are also included in Table 8. The total input carbon from hardwood is listed, along with the quantity of carbon that ends up as a liquid product (i.e., gasoline, diesel, jet fuel, and LPG) or is vented. The total GHG emissions (in kg of CO2 equiv/s) are listed for each major input and output of the refinery. Purchased butanes were not used in any of the case studies; therefore, this feedstock amount was not included in Table 8. The material balances in Table 8 show that the key inputs and outputs for the refineries are relatively similar for a fixed capacity. That is, for any target capacity, the amount of input hardwood, total C5+ liquid products, and vented CO2 are all within a 10% range for all three liquid product compositions. The differences between the values for a given capacity can be directly attributed to either (i) the additional liquid products that are produced (i.e., LPG or butanes) or (ii) the topological differences in the hydrocarbon production and upgrading sections that were outlined earlier. This result from the material balances also provides consistency in the results of the energy 4321

dx.doi.org/10.1021/ef302003f | Energy Fuels 2013, 27, 4302−4324

Energy & Fuels

Article

capacity level are attributed to the configuration of the topology that is necessary to produce the particular liquid products. A parametric analysis on the hardwood purchase price indicates that there exists a threshold price of hardwood above which the refinery will no longer be economically competitive with crude oil refining. This threshold level for hardwood purchase is dependent upon the desired refinery capacity and will decrease as the capacity decreases. If crude oil was priced around $105/ bbl, then the 1 kBD refineries would be economically competitive with a hardwood purchase price of $70/dry metric ton. A reduction in the hardwood purchase price to $50/dry metric ton will make the hardwood refineries competitive at crude prices between $95 and $100/bbl. For the 10 kBD refineries, the process synthesis framework demonstrates the economic viability at a crude price of $65/bbl for hardwood purchase prices at $70/dry metic ton. If this purchase price was raised to $90/dry metric ton, the refineries would be competitive at crude priced between $75 and $82/bbl. The implementation of environmental legislation that penalizes GHG emissions will improve the economics of the refinery for all capacity levels. Given the EIA estimates for crude oil over the next 2 decades, it seems clear that hardwood BTL refineries at or above 2.5 kBD can be economically viable throughout the nation. Moreover, these hardwood BTL refineries use existing technology that has been demonstrated on a variety of scales, including pilot-plant level and full-scale commercial level.

the atmosphere, and the amount of atmospheric CO2 that is removed during photosynthesis in part (d) is based on the carbon content of the hardwood. The total GHG emissions avoided from liquid fuels (GHGAF) are equivalent to the total energy of fuels produced times a typical petroleum-based emissions level of 91.6 kg of CO2 equiv/GJLHV.11 The emissions attributed to electricity input (GHGAE) are equivalent to the energy input by electricity times a typical natural-gas-based emissions level of 101.3 kg of CO2 equiv/ GJ.111 Note that a negative value for GHGAE implies that the electricity is input to the refinery, and because the steam generation and wastewater cleaning are incorporated in the heat, power, and water integration approach in the process, no additional emissions are incorporated. The GHG emissions index (GHGI) for the refinery represents the division of LGHG by the sum of GHGAF and GHGAE. Values that are less than 1 are representative of processes that have superior lifecycle emissions to fossil-based processes. The GHGI ranges between 0.19 and 0.24 for all 12 case studies and indicates that the lifecycle emissions are reduced between 76% and 81% from petroleum-based processes. This is a substantial reduction from petroleum-based transportation fuels and was achieved without the inclusion of environmental legislation (e.g., a carbon tax) in the economic assumptions. Although such legislation will clearly benefit the BEOP associated with the refineries, it is not necessary for some of the larger scale refineries (2.5 and 10 kBD) to be competitive at crude prices above $87/bbl. The results of Table 8 show that the major points of GHG emissions are from liquid fuel consumption and from process venting. The tailpipe emissions from liquid fuel use contribute approximately 50−60% to the overall process emissions, while the process CO2 venting contributes about 35−45%. The balance of the lifecycle emissions is due to acquisition and transportation of the feedstocks and products. The use of CO2 supercritical compression and sequestration could virtually eliminate the process emissions and ultimately make the refineries net CO2 negative processes. However, this increased system cost would not be justified without the imposition of a economic penalty for plant emissions.

■

ASSOCIATED CONTENT

■

AUTHOR INFORMATION

S Supporting Information *

Complete process flowsheet and mathematical model. This material is available free of charge via the Internet at http:// pubs.acs.org.

Corresponding Author

*Telephone: (609) 258-4595. Fax: (609) 258-0211. E-mail: fl[email protected]. Notes

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS The authors acknowledge financial support from the National Science Foundation (NSF EFRI-0937706 and NSF CBET1158849) and support from the Lockheed Martin Corporation.

4. CONCLUSION A rigorous optimization-based framework for the process synthesis and simultaneous heat, power, and water integration of a hardwood BTL refinery was proposed in this study. Multiple existing technologies including hardwood gasification, acid-gas cleanup, FT synthesis, methanol synthesis, MTG, MTO, and hydrocarbon upgrading units were combined into a superstructure of possibilities. A large-scale nonlinear mixedinteger optimization model was developed to model all of the units in the refinery and was solved to a 5−7% optimality gap using a rigorous deterministic global optimization branch-andbound scheme. A total of 12 case studies that focused on four target capacities (i.e., 0.8, 1, 2.5, and 10 kBD) and three product compositions (i.e., U.S. demand ratios, maximum diesel, and maximum kerosene) were used to demonstrate the capability of the process synthesis framework and determine the process design that has the lowest overall cost. The price of crude oil for which the hardwood BTL refineries will be competitive ranges from $112 to $120/bbl for a 0.8 kBD plant, from $101 to $111/bbl for a 1 kBD plant, from $87 to $89/bbl for a 2.5 kBD plant, and from $63 to $68/bbl for a 10 kBD plant. Key differences between the refineries at a given

■

REFERENCES

(1) U.S. Energy Information Administration (EIA). Monthly Energy ReviewAugust 2012; EIA: Washington, D.C., 2012; DOE-EIA0035(2012/08), http://www.eia.gov/totalenergy/data/monthly/ archive/00351208.pdf. (2) U.S. Energy Information Administration (EIA). Annual Energy Outlook 2012 with Projections to 2035; EIA: Washington, D.C., 2011; DOE/EIA-0383(2012), http://www.eia.gov/forecasts/aeo/pdf/ 0383(2012).pdf. (3) Floudas, C. A.; Elia, J. A.; Baliban, R. C. Comput. Chem. Eng. 2012, 41, 24−51. (4) Lynd, L. R.; Larson, E.; Greene, N.; Laser, M.; Sheehan, J.; Dale, B. E.; McLaughlin, S.; Wang, M. Biofuels, Bioprod. Biorefin. 2009, 3, 113−123. (5) U.S. Department of Energy (DOE). Biomass as Feedstock for a Bioenergy and Bioproducts Industry: The Technical Feasibility of a BillionTon Annual Supply; DOE: Washington, D.C., 2005; DOE/GO102005-2135, http://www1.eere.energy.gov/biomass/publications. html. 4322

dx.doi.org/10.1021/ef302003f | Energy Fuels 2013, 27, 4302−4324

Energy & Fuels

Article

(6) National Research Council (NRC). Water Implications of Biofuels Production in the United States; NRC: Washington, D.C., 2008; http:// www.nap.edu/catalog.php?record_id=12039. (7) de Fraiture, C.; Giordano, M.; Liao, Y. Water Policy 2008, 10, 67−81. (8) National Academy of Sciences, National Academy of Engineering, and National Research Council (NRC). Liquid Transportation Fuels from Coal and Biomass: Technological Status, Costs, and Environmental Issues; United States Environmental Protection Agency (U.S. EPA): Washington, D.C., 2009. (9) U.S. Energy Information Administration (EIA). Annual Energy Outlook 2011 with Projections to 2035; EIA: Washington, D.C., 2011; DOE/EIA-0383(2011), http://www.eia.gov/forecasts/archive/aeo11/ pdf/0383(2011).pdf. (10) Larson, E. D.; Jin, H.; Celik, F. E. Biofuels, Bioprod. Biorefin. 2009, 3, 174−194. (11) Kreutz, T. G.; Larson, E. D.; Liu, G.; Williams, R. H. Fischer− Tropsch fuels from coal and biomass. Proceedings of the 25th International Pittsburg Coal Conference; Pittsburgh, PA, Sept 29−Oct 2, 2008. (12) Larson, E. D.; Fiorese, G.; Liu, G.; Williams, R. H.; Kreutz, T. G.; Consonni, S. Energy Environ. Sci. 2010, 3, 28−42. (13) Chen, Y.; Adams, T. A., II; Barton, P. I. Ind. Eng. Chem. Res. 2011, 50, 5099−5113. (14) Chen, Y.; Adams, T. A., II; Barton, P. I. Ind. Eng. Chem. Res. 2011, 50, 4553−4566. (15) Liu, P.; Whitaker, A.; Pistikopoulos, E. N.; Li, Z. Comput. Chem. Eng. 2011, 35, 1359−1373. (16) Warren, A.; El-Halwagi, M. Fuel Proc. Technol. 1996, 49, 157− 166. (17) Liu, G.; Larson, E. D.; Williams, R. H.; Kreutz, T. G.; Guo, X. Energy Fuels 2011, 25, 415−437. (18) Williams, R. H.; Liu, G.; Kreutz, T. G.; Larson, E. D. Energy Procedia 2011, 4, 1843−1850. (19) Liu, G.; Williams, R. H.; Larson, E. D.; Kreutz, T. G. Energy Procedia 2011, 4, 1989−1996. (20) Borgwardt, R. H. Biomass Bioenergy 1997, 12, 333−345. (21) Li, H.; Hong, H.; Jin, H.; Cai, R. Appl. Energy 2010, 87, 2846− 2853. (22) Dong, Y.; Steinberg, M. Int. J. Hydrogen Energy 1997, 22, 971− 977. (23) Grossmann, I. E.; Martín, M. Chin. J. Chem. Eng. 2010, 18, 914− 922. (24) Baliban, R. C.; Elia, J. A.; Floudas, C. A. Comput. Chem. Eng. 2011, 35, 1647−1690. (25) Baliban, R. C.; Elia, J. A.; Floudas, C. A. Comput. Chem. Eng. 2012, 37, 297−327. (26) Baliban, R. C.; Elia, J. A.; Misener, R.; Floudas, C. A. Comput. Chem. Eng. 2012, 42, 64−86. (27) Baliban, R. C.; Elia, J. A.; Weekman, C. A.; Floudas, V. W. Comput. Chem. Eng. 2012, 47, 29−56. (28) Laser, M.; Jin, H.; Jayawardhana, K.; Dale, B. E.; Lynd, L. R. Biofuels, Bioprod. Biorefin. 2009, 3, 231−246. (29) Tock, L.; Gassner, M.; Maréchal, F. Biomass Bioenergy 2010, 34, 1838−1854. (30) Hamelinck, C.; Faaij, A.; Uil, H.; Boerrigter, H. Energy 2004, 29, 1743. (31) Perales, A. L. V.; Valle, C. R.; Ollero, P.; Barea, A. G. Energy 2011, 36, 4097−4108. (32) Tijmensen, M. J. A.; Faaij, A. P. C.; Hamelinck, C. N.; Hardeveld, M. R. M. Biomass Bioenergy 2002, 23, 129−152. (33) Bridgwater, A. V.; Double, J. M. Fuel 1991, 70, 1209−1224. (34) Swanson, R. M.; Platon, A.; Satrio, J. A.; Brown, R. C. Fuel 2010, 89, S11−S19. (35) Clausen, L. R.; Elmegaard, B.; Houbak, N. Energy 2010, 35, 4831−4842. (36) Sharma, P.; Sarker, B. R.; Romagnoli, J. A. Comput. Chem. Eng. 2011, 35, 1767−1781.

(37) Henrich, E.; Dahmen, N.; Dinjus, E. Biofuels, Bioprod. Biorefin. 2009, 3, 28−41. (38) Sunde, K.; Brekke, A.; Solberg, B. Energies 2011, 4, 845−877. (39) Bao, B.; Ng, D. K. S.; Tay, D. H. S.; Gutiérrez, A. J.; El-Halwagi, M. M. Comput. Chem. Eng. 2011, 35, 1374−1383. (40) Tay, D. H. S.; Ng, D. K. S.; Sammons, N. E.; Eden, M. R. Ind. Eng. Chem. Res. 2011, 50, 1652−1665. (41) Giarola, S.; Zamboni, A.; Bezzo, F. Comput. Chem. Eng. 2011, 35, 1782−1797. (42) Bai, Y.; Hwang, T.; Kang, S.; Ouyang, Y. Transp. Res., Part B 2011, 45, 162−175. (43) Bowling, I. M.; Ortega, J. M. P.; El-Halwagi, M. M. Ind. Eng. Chem. Res. 2011, 50, 6276−6286. (44) Parker, N.; Tittmann, P.; Hart, Q.; Nelson, R.; Skog, K.; Schmidt, A.; Gray, E.; Jenkins, B. Biomass Bioenergy 2010, 34, 1597− 1607. (45) Huang, Y.; Chen, C. W.; Fan, Y. Transp. Res., Part E 2010, 46, 820−830. (46) Aksoy, B.; Cullinan, H. T.; Sammons, N. E., Jr.; Eden, M. R. Environ. Prog. 2008, 27, 515−523. (47) Williams, R. H.; Larson, E. D.; Liu, G.; Kreutz, T. G. Energy Procedia 2009, 1, 4379−4386. (48) Kumabe, K.; Fujimoto, S.; Yanagida, T.; Ogata, M.; Fukuda, T.; Yabe, A.; Minowa, T. Fuel 2008, 87, 1422−1427. (49) Ju, F.; Chen, H.; Ding, X.; Yang, H.; Wang, X.; Zhang, S.; Dai, Z. Biotechnol. Adv. 2009, 27, 599−605. (50) Hamelinck, C. N.; Faaij, A. P. C. J. Power Sources 2002, 111, 1− 22. (51) Clausen, L. R.; Houbak, N.; Elmegaard, B. Energy 2010, 35, 2338−2347. (52) Kim, H.; Han, K.; Yoon, E. S. J. Chem. Eng. Jpn. 2010, 43, 671− 681. (53) He, J.; Zhang, W. Appl. Energy 2011, 88, 1224−1232. (54) Ng, K. S.; Sadhukkan, J. Biomass Bioenergy 2011, 35, 1153− 1169. (55) Xiao, J.; Shen, L.; Zhang, Y.; Gu, J. Ind. Eng. Chem. Res. 2009, 48, 9999−10007. (56) Erturk, M. Renewable Sustainable Energy Rev. 2011, 15, 2766− 2771. (57) Vliet, O.; Faaij, A.; Turkenburg, W. Energy Convers. Manage. 2009, 50, 855−876. (58) Dal-Mas, M.; Giarola, S.; Zamboni, A.; Bezzo, F. Biomass Bioenergy 2011, 35, 2059−2071. (59) Amigun, B.; Gorgens, J.; Knoetze, H. Energy Policy 2010, 38, 312−322. (60) Leduc, S.; Schwab, D.; Dotzauer, E.; Schmid, E.; Obersteiner, M. Int. J. Energy Res. 2008, 32, 1080−1091. (61) Leduc, S.; Schmid, E.; Obersteiner, M.; Riahi, K. Biomass Bioenergy 2009, 33, 745−751. (62) Leduc, S.; Lundgren, J.; Franklin, O.; Dotzauer, E. Appl. Energy 2010, 87, 68−75. (63) Mignard, D.; Pritchard, C. Chem. Eng. Res. Des. 2008, 86, 473− 487. (64) Williams, R. H.; Larson, E. D.; Katofsky, R. E.; Chen, J. Energy Sustainable Dev. 1995, 1, 18−34. (65) Sarkar, S.; Kumar, A.; Sultana, A. Energy 2011, 36, 6251−6262. (66) Kou, N.; Zhao, F. Biomass Bioenergy 2011, 35, 608−616. (67) Cucek, L.; Martin, M.; Grossmann, I. E.; Kravanja, Z. Comput. Chem. Eng. 2011, 35, 1547−1557. (68) Marvin, W. A.; Schmidt, L. D.; Benjaafar, S.; Tiffany, D. G.; Daoutidis, P. Chem. Eng. Sci. 2011, 67, 68−79. (69) Terrados, J.; Almonacid, G.; Higueras, P. P. Renewable Sustainable Energy Rev. 2009, 13, 2022−2030. (70) Akgul, O.; Zamboni, A.; Bezzo, F.; Shah, N.; Papageorgiou, L. G. Ind. Eng. Chem. Res. 2011, 50, 4927−4938. (71) van Dyken, S.; Bakken, B. H.; Skjelbred, H. I. Energy 2010, 35, 1338−1350. (72) Rentizelas, A. A.; Tatsiopoulos, I. P. Int. J. Prod. Econ. 2010, 123, 196−209. 4323

dx.doi.org/10.1021/ef302003f | Energy Fuels 2013, 27, 4302−4324

Energy & Fuels

Article

(73) Rentizelas, A. A.; Tatsiopoulos, I. P.; Tolis, A. Biomass Bioenergy 2009, 33, 223−233. (74) Eksioglu, S. D.; Acharya, A.; Leightley, L. E.; Arora, S. Comp. Ind. Eng. 2009, 57, 1342−1352. (75) Eksioglu, S. D.; Li, S.; Zhang, S.; Sokhansanj, S.; Petrolia, D. Transp. Res. Rec. 2010, 2191, 144−151. (76) Zamboni, A.; Shah, N.; Bezzo, F. Energy Fuels 2009, 23, 5121− 5133. (77) Dunnett, A. J.; Adjiman, C. S.; Shah, N. Biotechnol. Biofuels 2008, 1, 13−40. (78) Marti, B. V.; Gonzalez, E. F. Renewable Energy 2010, 35, 2136− 2142. (79) Panichelli, L.; Gnansounou, E. Biomass Bioenergy 2008, 32, 289−300. (80) Zhang, F.; Johnson, D. M.; Sutherland, J. W. Biomass Bioenergy 2011, 35, 3951−3961. (81) An, H.; Wilhelm, W. E.; Searcy, S. W. Bioresour. Technol. 2011, 102, 7860−7870. (82) Gan, J.; Smith, C. T. Biomass Bioenergy 2011, 35, 3350−3359. (83) Leduc, S.; Starfelt, F.; Dotzauer, E.; Kindermann, G.; McCallum, I.; Obersteiner, M.; Lundgren, J. Energy 2010, 35, 2709−2716. (84) Hacatoglu, K.; McLellan, P. J.; Layzell, D. B. Bioresour. Technol. 2011, 102, 1087−1094. (85) Cherubini, F. Renewable Energy 2010, 35, 1565−1573. (86) Kocoloski, M.; Griffin, W. M.; Matthews, H. S. Energy Policy 2011, 39, 47−56. (87) Cucek, L.; Lam, H. L.; Klemes, J. J.; Varbanov, P. S.; Kravanja, Z. Clean Technol. Environ. Policy 2010, 12, 635−645. (88) Zwart, R. W. R.; Boerrigter, H. Energy Fuels 2005, 19, 591−597. (89) Baliban, R. C.; Elia, J. A.; Floudas, C. A. Ind. Eng. Chem. Res. 2010, 49, 7343−7370. (90) Elia, J. A.; Baliban, R. C.; Floudas, C. A. Ind. Eng. Chem. Res. 2010, 49, 7371−7388. (91) Baliban, R. C.; Elia, J. A.; Floudas, C. A. AIChE J. 2013, 59, 505−531. (92) Baliban, R. C.; Elia, J. A.; Floudas, C. A. Energy Environ. Sci. 2013, 6, 267−287. (93) Martin, M.; Grossmann, I. E. Ind. Eng. Chem. Res. 2011, 50, 13485−13499. (94) Ellepola, J.; Thijssen, N.; Grievink, J.; Baak, G.; Avhale, A.; van Schijndel, J. Comput. Chem. Eng. 2012, 42, 2−14. (95) You, F.; Wang, B. Ind. Eng. Chem. Res. 2011, 50, 10102−10127. (96) You, F.; Tao, L.; Graziano, D. J.; Snyder, S. W. AIChE J. 2012, 58, 1157−1180. (97) Gebreslassie, B. H.; Yao, Y.; You, F. AIChE J. 2012, 58, 2155− 2179. (98) Wang, B.; Gebreslassie, B. H.; You, F. Comput. Chem. Eng. 2013, 52, 55−76. (99) Gebreslassie, B. H.; Slivinsky, M.; Wang, B.; You, F. Comput. Chem. Eng. 2013, 50, 71−91. (100) Floudas, C. A. Deterministic Global Optimization: Theory, Methods and Applications; Kluwer Academic Publishers: Dordrecht, The Netherlands, 2000. (101) Misener, R.; Thompson, J. P.; Floudas, C. A. Oper. Res. 2011, 35, 876−892. (102) Misener, R.; Floudas, C. A. Math. Program., Ser. B 2012, DOI: 10.1007/s10107-012-0555-6. (103) Misener, R.; Floudas, C. A. J. Global Optim. 2012, DOI: 10.1007/s10898-012-9874-7. (104) Duran, M. A.; Grossmann, I. E. AIChE J. 1986, 32, 123−138. (105) Karuppiah, R.; Grossmann, I. E. Comput. Chem. Eng. 2006, 30, 650−673. (106) Ahmetovic, E.; Grossmann, I. E. Ind. Eng. Chem. Res. 2010, 49, 7972−7982. (107) Ahmetovic, E.; Grossmann, I. E. AIChE J. 2010, 57, 434−457. (108) van der Drift, A.; van Doorn, J. Analysis of Biomass Data in ECN Database Phyllis; Energy Research Centre of the Netherlands (ECN): Petten, The Netherlands, 2002; http://www.ecn.nl/phyllis/.

(109) National Renewable Energy Laboratory (NREL). Gasoline from Wood via Integrated Gasification, Synthesis, and Methanol-to-Gasoline Technologies; NREL: Golden, CO, 2011; U.S. DOE Contract DEAC36-08GO28308. (110) National Renewable Energy Laboratory (NREL). Process Design and Economics for Conversion of Lignocellulosic Biomass to Ethanol: Thermochemical Pathway by Indirect Gasification and Mixed Alcohol Synthesis; NREL: Golden, CO, 2011; U.S. DOE Contract DEAC36-08GO28308. (111) National Energy Technology Laboratory (NETL). Cost and Performance Baseline for Fossil Energy Plants. Volume 1: Bituminous Coal and Natural Gas to Electricity Final Report; NETL: Pittsburgh, PA, 2007; DOE/NETL-2007/1281, http://www.netl.doe.gov/energyanalyses/baseline_studies.html. (112) Mobil Research and Development Corporation. Slurry Fischer−Tropsch/Mobil Two Stage Process of Converting Syngas to High Octane Gasoline; Mobil Research and Development Corporation: Paulsboro, NJ, 1983; U.S. DOE Contract DE-AC22-80PC30022. (113) Mobil Research and Development Corporation. Two-Stage Process For Conversion of Synthesis Gas to High Quality Transporation Fuels; Mobil Research and Development Corporation: Paulsboro, NJ, 1985; U.S. DOE Contract DE-AC22-83PC60019. (114) de Klerk, A. Fischer−Tropsch Refining; Wiley-VCH Verlag and Co. KGaA: Weinheim, Germany, 2011. (115) Steynberg, A. R.; Dry, M. E. Stud. Surf. Sci. Catal. 2004, 152. (116) Bechtel Corporation. Aspen Process Flowsheet Simulation Model of a Battelle Biomass-Based Gasification, Fischer−Tropsch Liquefaction and Combined-Cycle Power Plant; Bechtel Corporation: San Francisco, CA, 1998; DE-AC22-93PC91029, http://www.fischer-tropsch.org/. (117) Bechtel Corporation. Baseline Design/Economics for Advanced Fischer−Tropsch Technology; Bechtel Corporation: San Francisco, CA, 1992; DE-AC22-91PC90027. (118) Meyers, R. A. Handbook of Petroleum Refining Processes; McGraw-Hill: New York, 2003. (119) Tabak, S. A.; Yurchak, S. Catal. Today 1990, 6, 307−327. (120) Mobil Research and Development Corporation. Research Guidance Studies To Assess Gasoline from Coal by Methanol-to-Gasoline and Sasol-Type Fischer−Tropsch Technologies; Mobil Research and Development Corporation: Paulsboro, NJ, 1978; U.S. DOE Contract EF-77-C-O1-2447. (121) National Energy Technology Laboratory (NETL). Assessment of Hydrogen Production with CO2 Capture Volume 1: Baseline State-ofthe-Art Plants; NETL: Pittsburgh, PA, 2010; DOE/NETL-2010/1434. (122) Balmer, P.; Mattsson, B. Water Sci. Technol. 1994, 30, 7−15. (123) Gregor, J. H.; Gosling, C. D.; Fullerton, H. E. Upgrading Fischer−Tropsch LPG with the Cyclar Process; UOP, Inc.: Des Plaines, IL, 1989; DE-AC22-86PC90014. (124) Access Intelligence. Chemical Engineering Plant Cost Index; Access Intelligence: New York, 2012; http://www.che.com/pci/. (125) IBM Corporation. CPLEX, ILOG CPLEX C++ API 12.1 Referece Manual; IBM Corporation: Armonk, NY, 2009. (126) Drud, A. Math. Program. 1985, 31, 153−191. (127) Floudas, C. A. Nonlinear and Mixed-Integer Optimization; Oxford University Press: New York, 1995. (128) Floudas, C. A.; Pardalos, P. M. J. Global Optim. 1995, 7, 113. (129) Floudas, C. A.; Akrotirianakis, I. G.; Caratzoulas, S.; Meyer, C. A.; Kallrath, J. Comput. Chem. Eng. 2005, 29, 1185−1202. (130) Floudas, C. A.; Gounaris, C. E. J. Global Optim. 2009, 45, 3− 38. (131) Floudas, C. A.; Ciric, A. R.; Grossmann, I. E. AIChE J. 1986, 32, 276−290. (132) United States Forest Service. Timber Product Output Reports; U.S. Forest Service: Washington, D.C., 2008; http://srsfia2.fs.fed.us/ php/tpo_2009/tpo_rpa_int1.phps. (133) Argonne National Laboratory, GREET 1.8b, The Greenhouse Gases, Regulated Emisssions, and Energy Use in Transportation (GREET) Model; Argonne National Laboratory: Argonne, IL, 2007.

4324

dx.doi.org/10.1021/ef302003f | Energy Fuels 2013, 27, 4302−4324