Techno-economic analysis of liquid fuel production from woody biomass via hydrothermal liquefaction (HTL) and upgrading

Applied Energy 129 (2014) 384–394

Contents lists available at ScienceDirect

Applied Energy journal homepage: www.elsevier.com/locate/apenergy

Techno-economic analysis of liquid fuel production from woody biomass via hydrothermal liquefaction (HTL) and upgrading Yunhua Zhu a,⇑, Mary J. Biddy b, Susanne B. Jones a, Douglas C. Elliott a, Andrew J. Schmidt a a b

Pacific Northwest National Laboratory, Richland, WA 99354, USA National Renewable Energy Laboratory, Golden, CO 80401, USA

h i g h l i g h t s  Bench-scale hydrothermal liquefaction (HTL) and hydrotreating tests were conducted.  A techno-economic analysis was conducted for the HTL and upgrading systems.  A state-of-technology case was evaluated based on the best available test data.  A goal case was evaluated considering potential process improvements.  Sensitivity analyses were conducted for alternative configuration and selected parameters.

a r t i c l e

i n f o

Article history: Received 17 September 2013 Received in revised form 11 February 2014 Accepted 21 March 2014 Available online 14 June 2014 Keywords: Techno-economic analysis Biomass Hydrothermal liquefaction Upgrading

a b s t r a c t Techno-economic analysis (TEA) was implemented to evaluate the feasibility of developing a commercial large-scale woody biomass HTL and upgrading plant. In this system, woody biomass at 2000 dry metric ton/day was assumed to be converted to bio-oil via HTL and further upgraded to produce liquid fuel. Two cases were evaluated: a state-of-technology (SOT) case with HTL experimental testing results underpinning the major design basis and a goal case considering future improvements for a commercial plant with mature technologies. Process simulation and cost analysis were conducted. The annual production rate for the final hydrocarbon product was estimated to be 42.9 and 69.9 million gallon gasoline-equivalent (GGE) for the SOT and goal cases, respectively. The minimum fuel selling price (MFSP) was estimated to be $4.44/GGE for the SOT case and $2.52/GGE for the goal case. For advancing from the SOT to the goal case, the assumption of reducing the organics loss to the water phase led to the largest reduction in the production cost. Alternative configuration of small scale distributed HTL plants was evaluated. Sensitivity analysis identified key factors affecting the goal case and its cost uncertainties resulting from the assumed uncertainties in selected parameters. Ó 2013 Elsevier Ltd. All rights reserved.

1. Introduction Biomass is an important domestic resource with the potential to make a significant impact on domestic fuel supplies. Biomass can be converted to liquid fuels and chemicals via a number of thermochemical approaches (e.g., gasification and liquefaction). In contrast to gasification, direct liquefaction features a simple and direct conversion of biomass to liquid fuel and relatively high liquid-fuel yields [1,2]. Typical direct liquefaction processes include fast pyrolysis and high-pressure hydrothermal liquefaction (HTL) [3,4]. The HTL bio-oil has an oxygen content of 10–20 wt%, which is much lower than that of the pyrolysis bio-oil, which is ⇑ Corresponding author. Tel./fax: +1 703 327 9987. E-mail address: [email protected] (Y. Zhu). http://dx.doi.org/10.1016/j.apenergy.2014.03.053 0306-2619/Ó 2013 Elsevier Ltd. All rights reserved.

about 40 wt% [5]. The HTL bio-oil has a heating value of about 35 MJ/kg, which is also higher than the heat value of 16–19 MJ/ kg for the pyrolysis bio-oil. The heating value of the HTL bio-oil is more comparable to the heating value of 40–45 MJ/kg for conventional petroleum fuels [2,5]. The concept of biomass liquefaction in hot water to produce liquid oil was originally developed in the 1920s and used alkali as a buffering agent. In the 1970s, a biomass HTL process was developed at the Pittsburgh Energy Research Center (PERC), and a pilot plant based on this process was demonstrated at the Albany Biomass Liquefaction Experimental Facility at Albany, Oregon, at a scale of 100 kg/h [1] with bio-oil recycled and a reducing gas. During the same period, the Lawrence Berkeley Laboratory (LBL) developed an HTL process, at an equivalent scale, with an acid hydroloysis pretreatment and water as the reaction media [6,7].

385

Y. Zhu et al. / Applied Energy 129 (2014) 384–394

In 1982, Shell Laboratory in the Netherlands began the research and development (R&D) of the HydroThermal UpgradingÒ (HTU) process [1,8–12]. Major technical features of these processes are provided in Table 1. In general, HTL reactions occur at temperatures from 250 to 380 °C, at pressures between 5 and 30 MPa, and with a residence time between 5 and 60 min [13,14]. HTL of biomass under alkaline or neutral conditions has been widely investigated [15] and reactions involve dehydration, deoxygenation, and decarboxylation. Compressed hot water has enhanced solvent properties that facilitate the formation of liquid–oil products from biomass [10,11]. Biomass is dissolved and liquefied in this process, and the major products are bio-oil, gas, and water with dissolved organics. The key in biomass HTL is oxygen removal; about 85% of the oxygen in biomass can be removed as CO2 and water [11]. The oxygen content of bio-oil can be as low as 10 wt%, and thus has a higher caloric value than the original biomass. The bio-oil produced from the HTL process can be upgraded to a conventional hydrocarbon fuel by near complete oxygen removal and molecular weight reduction via hydrotreating and hydrocracking [1,16]. Compared to biomass gasification and pyrolysis, HTL uses wet biomass as feedstock and thus avoids the energy consumption for biomass drying. In addition, hot compressed water stays in the liquid phase in HTL, eliminating the energy penalty for water vaporization present in gasification and pyrolysis. Biomass gasification and pyrolysis systems are commercially available [17,18], while biomass HTL has only been demonstrated at pilot scale. Hydrotreating to remove nitrogen and sulfur in heavy oil is a common and well established refinery process [19]. However, oxygen removal from HTL bio-oil by hydrotreating has not been demonstrated at commercial scale, and hydrocracking to remove heavy compounds in the bio-oil has not been demonstrated at experimental scale. Since HTL technology has not been commercialized yet, extensive techno-economic analyses (TEAs) are required to provide guidance for decision making about commercial development. A number of TEAs have been conducted for gasification and fast pyrolysis-based thermochemical biomass conversion to fuels technologies [20–27]. However, only limited TEAs have been performed for biomass HTL-based system evaluation [11,28–30]. Further, little cost analysis has been conducted for HTL bio-oil upgrading processes. An important reason is lack of detailed and consistent technical information to support a systematic analysis and evaluation. Pacific Northwest National Laboratory (PNNL)

under the National Advanced Biofuels Consortium (NABC) sponsorship conducted a series bench scale HTL and upgrading tests for woody biomass. The HTL process testing conducted at PNNL differs from other HTL processes developed previously: no reducing gas is used and water, versus recycled bio-oil, serves as the reaction medium. The above experimental work provided information for a techno-economic down selection process for ‘‘drop-in’’ biofuels pathways. The purpose is to select those strategies that showed the most promise to rapidly advance to commercialization. In this study, TEA was implemented to evaluate the feasibility of developing a large-scale woody biomass based HTL and upgrading system. Experimental information provided a detailed and consistent design basis for this analysis. Two cases are investigated: the state-of-technology (SOT) case (based on bench scale testing results) and the goal case (based on optimal assumptions for the product yields and process design). The purpose of this study is to provide preliminary evaluation of a large-scale HTL and upgrading system, identify potential improvements effects on process economics, identify key factors affecting the cost, and estimate the uncertainty in the production cost. 2. Materials and methods To implement TEA for a large-scale system, a detailed design basis that represents the battery limits of the system being assessed must be developed first [31]. In this study, the inside battery limits (ISBL) for a commercial scale stand-alone HTL and upgrading plant is assumed to include HTL, upgrading (hydrotreating and hydrocracking), and the hydrogen plant. In this study, the operation conditions and performance of the HTL and hydrotreating processes are mainly based on the experimental results. For other processes such as hydrocracking and hydrogen generation, their design specifications are based on literature sources. With design basis developed, process simulation can be developed. With detailed mass and energy balance information obtained from the simulation results, process economics can be estimated. 2.1. Process overview Fig. 1 shows a simplified block diagram of the biomass HTL and upgrading system. The overall model consists of four major processes: feedstock preparation, HTL, upgrading, and the hydrogen plant. Woody biomass feedstock is ground to fine particles and

Table 1 Historical development of hydrothermal liquefaction processes on biomass. Source: Refs. [1,8–12]. Process

PERC

Lawrence Berkley Laboratory Process

Dutch Shell HTU

Testing and development Scale-tested, dry biomass Pretreatment

1970s and 1980s 18 kg/h Drying and grinding (500 l) w/bio-oil 7.5%

1970s and 1980s 10 kg/h Acid pre-hydrolysis, neutralization, wet grinding (500 l)

1980s–2011 10 kg/h Thermal softening (250 °C, several minutes) 10.5%

wt% biomass in slurry Process medium Catalysts, wt% dry wood basis Reducing gas Temperature/pressure Space velocity Mass yield to bio-oil Mass yield to solids Carbon yield to bio-oil Oxygen content in bio-oil Melting point of bio-oil Bio-oil viscosity Bio-oil upgrading by hydrotreating Reactor and tested feedstock

Recycled bio-oil + water Na2CO3, 10% 60% CO/40% H2 350 °C/20 MPa 1–3 h1 53% 1% 77% 11–15% n/a 120 cSt@99 °C 72,000 cSt@52 °C Yes Plug-flow reactor, 570 h test, Douglas Fir (TR12)

18% (total solid) 12% (suspended solid) Water/no recycle Na2CO3, 8% 60% CO/40% H2 330–340 °C/20 MPa 0.15–1.3 h1 22–30% 0.50% 35% 13% n/a 70 cSt@99 °C

Water/no recycle None None 350 °C/18 MPa 1–3 h1 38% n/a 50% 13.5% wt% 80 °C n/a

Yes

No

Continuous stirred-tank reactor (CSTR), 500 h test with Douglas Fir (TR7), conduct at residence time of 7 h

200 h+, at 10 kg/h (dry basis) test was conducted with onion waste in 2004

386

Y. Zhu et al. / Applied Energy 129 (2014) 384–394

Natural gas Hydrogen plant

Offgas

Steam from steam cycle

Solid wastes Offgas

H2

Biomass

Feedstock pretreatment

Upgrading (hydrotreating only or hydrotreating & hydrocracking)

Hydrothermal liquefaction (HTL)

Liquid fuels

Recycled water Wastewater

Wastewater

Fig. 1. Simplified block diagram of a biomass HTL and upgrading system.

softened by direct injection of hot recycle water from the HTL process to form a biomass–water slurry. The slurry is pumped to the HTL reaction pressure level and preheated by the hot effluent from the HTL reactor. Hot biomass slurry is converted in the HTL reactor. Solid waste, mostly inorganic solids, such as ash, is separated from the hot HTL effluent as solid wastes. The filtered effluent is cooled and separated into a gas phase and then two liquid phases: bio-oil and an aqueous phase with predominately water and some dissolved organics. The bio-oil is sent to the upgrading process. Most of the aqueous phase is recycled to the feedstock preparation step while the remaining portion is purged to wastewater treatment. The gas phase is sent to the hydrogen plant to generate hydrogen for the upgrading process. The crude bio-oil is upgraded by hydrotreating only or hydrotreating and hydrocracking to remove most of oxygen in the bio-oil. The upgraded hydrocarbon product is stabilized by cooling and distillation to produce liquid fuels consisting of gasoline, diesel, and heavy oil fractions based on their boiling point ranges. Due to the limited scope of the experimental work, the hydrocracking step has not yet been demonstrated. Therefore, the SOT case assumes that only hydrotreating is used for upgrading and the goal case assumes that the heavy oil is further hydrocracked. 2.2. Major processes The design details of each major process in the woody biomass HTL and upgrading system are described in the following sections. 2.2.1. Biomass preparation Finely ground woody biomass with moisture content of about 50 wt% at a feed rate of 2000 dry metric ton/day (83,300 kg/h) is assumed to be used as the feedstock for HTL processing. The feedstock is mixed with recycled water from the HTL process to form a biomass–water slurry with 15 wt% dry biomass content and pumped to about 0.6 MPa. 2.2.2. Hydrothermal liquefaction The HTL process design is based on the best available HTL experimental tests results. The detailed experimental information for HTL and upgrading processes was described in Appendix A and Schmidt et al. [32]. The major design specifications for this process simulation are listed in Table 2. The biomass–water slurry is pumped to 20.4 MPa and preheated to 327 °C by the hot liquid effluent from the HTL reactor. Ongoing work with pump manufacturers suggests that large-scale biomass–water slurry pumping, though unverified in field tests, is feasible [33]. The preheated slurry is sent to the HTL reactor operating at 20.3 MPa and 355 °C. For this large-scale system, the HTL reactor is assumed to be plug flow type because plug flow reactors are more economical

at the given design scale than the continuous stirred-tank reactors (CSTRs) used in most experimental tests. The reactor has a shelland-tube design with slurry in the tube and a heat-transfer hot fluid system on the shell side. A fired heater is used to provide heat to the heat-transfer fluid to maintain isothermal conditions in the reactor. Biomass slurry is converted to oil, water, gas, and solid compounds in the reactor. The HTL reactor is simulated by specifying the mass yield distribution for each compound in the products. Experimental testing results from the gas chromatography–mass spectrometry (GC–MS) of the HTL oil phase and high-performance liquid chromatography (HPLC) for the aqueous phase were used to determine the major compounds used in the model to simulate the HTL product. The yield distribution of each compound specified for the HTL reactor model is determined by matching the testing results for the elemental distribution and mass yield of each phase, including oil, gas aqueous, and solid. Minor adjustments to the measured yields were made to facilitate closure of the mass balance. The boiling point curve and density estimated for the simulated oil by the model are compared with the simulated distillation (SimDis) and density results of actual bio-oil product obtained in the testing to verify that key features of the actual bio-oil is appropriately represented by the simulation. The hot effluent from the HTL reactor is sent to a filter to separate fine particles from the hot liquid. The filtered solids, composed of mostly ash, are assumed to be disposed of as solid wastes. The liquid effluent from the filter is cooled by the incoming biomass– water slurry to 148°C. The cooled stream is depressurized to 0.2 MPa and separated into gas, aqueous, and oil phases. The gaseous compounds include hydrogen (H2), carbon dioxide (CO2), methane (CH4), and other hydrocarbons. The gaseous product is assumed to be used to produce hydrogen by steam reforming in a hydrogen plant. The aqueous phase contains about 2 wt% carbon in the form of dissolved alcohols, acids, and other oxygenates. Approximately 20% of the aqueous phase is purged, and the remainder is recycled back to the biomass feed slurry. The split fraction of recycled water is adjusted to meet the design specification of dry biomass wt% in the biomass–water slurry. The purged aqueous phase is cooled and sent to a wastewater treatment process. Anaerobic digestion (AD) is assumed to be used to convert the aqueous phase organics to gas rich in CH4 and CO2. A portion of this offgas is sent to the hydrogen plant as fuel gas to generate heat for the reforming process. The remainder is sent to a boiler in the steam cycle to generate steam for process use. The oil phase, consisting mainly of oxygenates and hydrocarbons, is sent to the downstream upgrading process. 2.2.3. Upgrading For the SOT case, hydrotreating only is assumed for the upgrading of HTL bio-oil. The hydrotreating process involves contacting

Y. Zhu et al. / Applied Energy 129 (2014) 384–394 Table 2 Major inputs and assumptions for the biomass HTL and upgrading system. Cases

SOT

Goal

Biomass feed rate, metric ton/day, dry basis Dry biomass wt% in biomass–water slurry

2000

2000

15

15

355 20.3 99.9 4

336 16.6 99.9 4

29.4 17.8 49.7 3.0

40.5 17.8 38.6 3.0

88.3 0.9 1.8 9.0

88.3 0.9 1.8 9.0

67.2 32.8

72.9 27.1 Single-stage fixed bed 165 13.5 0.18

H2 consumption, g H2/g dry bio-oil

Two-stage fixed bed 165 13.5 0.54 (stage 1) 0.18 (stage 2) 0.033

Products distribution, wt% Deoxygenated oil Water Gas

81.7 15.7 2.5

79.0 18.5 2.5

Hydrothermal liquefaction Temperature, °C Pressure, MPa Biomass conversion, % LHSV, h1 Yields, kg/100 kg dry wood Bio-oil Gas Water (with dissolved organics) Solid wastes Gas compositions, wt% CO2 H2 CH4 Other hydrocarbons Water compositions, wt% H2O Dissolved organics Hydrotreating Temperature, °C (inlet) Pressure, MPa (inlet) LHSV, h1

Deoxygenated oil distillation streams, wt% Light hydrocarbons (C4) 0.8 Gasoline (C5–C10) 41.3 Diesel (C10 – 400 °C boiling 39.6 point) Heavy oil (boiling point >400 °C) 18.4 Hydrocracking Temperature, °C (inlet) Pressure, MPa (inlet) LHSV, h1

n/a n/a n/a

0.033

0.8 41.3 39.6 18.4 376 10.5 1

the bio-oil with hydrogen under elevated pressure and high temperature to remove oxygen and reduce the molecular weight of the bio-oil via hydrodeoxygenation reactions [1]. Dual stage processing, where mild hydrotreating is followed by more severe hydrotreating, has been found to overcome the reactivity of fast pyrolysis oil [34] and was employed in the experimental tests. Therefore, a two-stage hydrotreating process is assumed in the model. The operating conditions and product yields for the hydrotreating reaction simulation are based on the experimental results. The major design specifications for the hydrotreating process are listed in Table 2. Other design conditions for hydrogen compression, hydrogen recycle, and products separation in this process are based on the previously developed design case [24]. In this process, the HTL bio-oil is pumped to about 14 MPa and combined with compressed hydrogen from a hydrogen plant. The hydrogen sent to this process is adjusted to meet the goal of H2 at the hydrotreating reactor inlet to be twice of its consumption. The mixture is preheated by the hydrotreating reactor effluent and sent to the hydrotreating reactor system. Deoxygenated oil, water, and gaseous compounds are generated from the bio-oil upgrading. Based on the testing results, the HTL bio-oil is almost completely converted to hydrocarbons, mainly C4–C17, with some

387

C1–C3 and a small fraction of heavier components. The simulation of the hydrotreating reaction is by specifying the yield distributions of each product compounds based on experimental results, including the GC–MS and SimDis results of actual deoxygenated oil product, feed and product elemental balance, and mass yields for each phase. The effluent from the hydrotreating reactors is initially cooled by preheating the reactor inlet stream. The effluent is further cooled by other process streams, then air, and finally trim-cooled with cooling water. The cooled effluent is separated into product oil, wastewater, and offgas streams. The offgas from the hydrotreaters is sent to a pressure swing adsorption (PSA) system and 80% of hydrogen in the feed is assumed be recovered. The recovered hydrogen is recycled back to the reactors. The PSA tail gas stream, which is rich in hydrogen and light hydrocarbons, is sent to the hydrogen plant for hydrogen production. The hydrotreated oil is stabilized by removing butane and lighter components in a lights-removal column. This column also serves to adjust the initial boiling point (BP) of the gasoline fraction. The overhead gas containing light organics is sent to the hydrogen plant for hydrogen production. The stable oil is further separated into gasoline and diesel-range fuels, and into a heavy fraction based on their boiling points. For the goal case, both hydrotreating and hydrocracking are assumed to be used for bio-oil upgrading. Hydrocracking is a catalytic cracking process. The purpose is to increase the yield of fuel products rich in hydrogen (i.e., diesel) from heavy hydrocarbons. By adding hydrogen, the heavy oil from the hydrotreating product separation process is decomposed and rearranges to produce a mixture of liquids spanning the gasoline and diesel range and some light gases. The product stream is cooled and separated to form condensate and gas phases. The hydrogen rich offgas is partly recycled to the hydrocracking reactor. The purged offgas is sent to the hydrogen plant for hydrogen production. The condensate is sent to a separation column and distilled into gasoline and diesel-range blend stocks. These products are blended with the gasoline and diesel products from the hydrotreating process. The major specifications for the hydrocracking process are provided in Table 2. The design and simulation of the hydrocracking process is based on Jones et al. [24] and Haynes et al. [35].

2.2.4. Hydrogen plant A conventional steam reforming process is assumed to be used in the hydrogen plant. Offgas, from the HTL and upgrading areas, is assumed to be compressed to 3.1 MPa by a multiple-stage air-intercooling compressor. The compressed offgas and makeup natural gas are used to generate the hydrogen required by the upgrading hydrodeoxygenation reactors. Superheated steam at 4.5 MPa and 371 °C is mixed with the natural gas and process offgas and sent to the steam reformer. The steam reformer is assumed to be operated at temperature of 850 °C and inlet pressure of 3.1 MPa. The steam reformer effluent is cooled by heating process streams and boiler feed water. The cooled effluent is sent to a water–gas shift reactor to further increase the hydrogen production. After condensing out the water, the hydrogen is purified by using a PSA unit. A recovery rate of 80% and a product purity of 99.9% are assumed, which are specified based on the literature [36] and the simulated inlet gas conditions. Offgas from the PSA is recycled to the reformer burners. Offgas from the treatment of the aqueous wastewater generated in the HTL process is split and part of it is used as fuel for the reformer burner. Another part of it is burned as fuel in the fired heaters for the HTL and upgrading areas. The remainder is sent to the boiler in the steam cycle for steam generation. The simulation of the steam reforming process is based on SRI International [37] and Meyers [38].

388

Y. Zhu et al. / Applied Energy 129 (2014) 384–394

2.2.5. Steam cycle Saturated steam at 4.6 MPa is generated by recuperating heat from the steam reformer process hot effluents, the upgrading process hot effluents, and the steam-cycle boiler hot flue gas. The boiler flue gas is also used to generate superheated steam at 4.5 MPa and 370 °C. Part of the superheated steam is used in the steam reformer as a reactant, and the remainder is used as driver steam for some of the rotating equipment. Purchased electricity supplies the remainder of the power demand. 2.3. SOT and goal cases The major process inputs and assumptions in the SOT and goal cases are listed in Table 2. The design basis of the SOT case is developed based on actual data obtained in current experimental testing. The goal case represents the nth-of-a-kind (NOAK) plant design, which assumes that future improvements required for commercializing the process at the given scale is completed and technologies applied in the system are mature. The major improvements for the goal case include: (1) lower HTL reactor pressure; (2) less organics loss to the water phase; (3) adding a hydrocracking process for heavy oil treatment; and (4) using a single reactor hydrotreating process. For the HTL process, the goal case assumes lower HTL operating pressure with the same biomass conversion efficiency and yield for each compound as the SOT case. This assumption is based on the current HTL operating pressure range reported in literature [13] and the potential improvement is cost reduction because of lower pressure requirement for equipment, including pumps, heat exchangers, and HTL reactor. For the oil–water phase separation, the goal case assumes less organics loss to the water phase. This assumption is based on potential improvements in the oil–water phase separation for an NOAK plant. Therefore, the bio-oil yield in the goal case is higher than in the SOT case, and the water phase yield is lower because of less dissolved organics. The yields for gas and solid wastes are the same as in the SOT case. The bio-oil in the goal case has a slightly higher oxygen content than in the SOT case because some high oxygen content phenols originally dissolved in the water phase in the SOT case are now assumed to be in the bio-oil phase in the goal case. For the upgrading process of the goal case, a single-stage hydrotreating step with the same conservative operating conditions and compounds yield as in the SOT case is assumed (as shown in Table 2). This assumption can reduce the equipment cost of the hydrotreating process in the goal case because less equipment is used. The goal case single-stage hydrotreating design is justified by the likely prospect that HTL bio-oil does not need the first stage mild condition hydrotreating, which is typically used for hydrotreating of pyrolysis bio-oil [24], because the HTL bio-oil has much lower oxygen content than the fast pyrolysis one [2]. The bio-oil oxygen content of the goal case is higher than in the SOT case, which in turn causes more water generation during hydrodeoxygenation. This accounts for the slightly lower deoxygenated oil production in the goal case than in the SOT case. Both cases produce the same gas yield. The SOT case does not include a hydrocracking unit for heavy oil treatment as in the goal case because hydrocracking was not tested in this work. The heavy oil in the SOT case is assumed to be a by-product for the cost analysis. The goal case assumes a hydrocracker for further converting the heavy fraction, which helps to increase the yield of final fuel-grade products. This assumption needs to be demonstrated by testings. Ongoing research and development (R&D) is underway for the improvement of the upgrading process. A number of tests have been conducted to explore less severe operating conditions, higher space velocities, and better catalysts. Therefore, with R&D efforts,

more upgrading technology improvements are anticipated, and the assumptions for the goal case in this study have great potential to be realized in commercial plants in the future. 2.4. Process simulation approach With design basis developed, a process flowsheet for the biomass HTL and upgrading system was built in Advanced System for Process Engineering Plus (Aspen PlusÒ), a deterministic steadystate chemical process simulator [39]. Unit operations blocks in Aspen PlusÒ are used to simulate the actual pieces of equipment in the system. These blocks are connected by using material, heat, or work streams to simulate fluid flows and energy transfer between unit blocks. Input assumptions are specified for unit blocks and feed streams based on information from the design basis. Final product yields, energy efficiency, and other performance of the system are estimated based on the modeling results. 3. Economic analysis The capital costs of standard equipment, such as pumps, compressors, and heat exchangers, were estimated using Aspen Process Economic Analyzer. The HTL reactor cost was estimated by assuming a shell-and-tube type heat exchanger configuration for the reactor and the reactor size was estimated based on inlet slurry flow rate and space velocity. Since this study is a preliminary analysis for the purpose of economic feasibility evaluation, the estimation of HTL reactor cost in this study is sufficiently robust to meet this purpose. More detailed sensitivity evaluations are planned and a contract has been placed with an engineering firm to confirm HTL reactor cost assumptions. In addition, a missing equipment contingency factor was specified, as shown in Table 3, in order to account for the uncertainty in the equipment cost estimation resulting from lack of technical data and the gap between a preliminary analysis and a more detailed engineering analysis. For the high-pressure biomass slurry pumps in the HTL process, cost estimates were confirmed by vendor budgetary quotations. Non-standard equipment costs (e.g., the two-stage hydrotreating system and hydrocracking reactor) were estimated by scaling base equipment costs from literature references [37,40] based on the appropriate metric (e.g., flow and duty) and applying an appropriate scaling factor:

 Costnew ¼ Costbase

Capacitynew Capacitybase

scaling factor ð1Þ

The installed equipment costs were estimated based on the purchased equipment costs and their installation factors. The total capital investment was factored from installed equipment costs

Table 3 Economic analysis assumptions. Assumption

Value

Missing equipment contingency factor Installation factor

10% of total other equipment cost

Indirect cost factor, % Equity financing, % Stream factor Cost year for analysis Internal rate of return, % Plant life, yr Working capital, % Maintenance and overhead, % Maintenance materials, % Local taxes and insurance, %

2.47 of total purchased equipment cost (TPEC) 62% of TIC 100% 90% 2007 10% 20 5% of fixed capital investment (FCI) 95% of labor and supervision 2% of FCI 2% of FCI

389

Y. Zhu et al. / Applied Energy 129 (2014) 384–394

according to heuristics given in Dutta et al. [27]. All costs are reported in 2007 U.S. dollars to maintain consistency with other TEA work in NABC. Variable operating costs were calculated based on the consumption of raw materials, chemicals, wastes, and utilities from the simulation results and their unit prices from literature sources or vendor quotations [37,41,42]. The feedstock plant-gate price was assumed to be $70.00/dry metric ton in 2007 U.S. dollars, which was converted from $66.12/dry short ton in 2010 U.S. dollars (estimate from Catchlight Energy LLC) [43]. The feedstock price assumption does not include the procurement overhead cost associated with the feedstock. The fixed operating costs mainly include the maintenance and labor costs. The final production cost was calculated as minimum fuel selling price (MFSP) in a Microsoft Excel spreadsheet. A discounted cash flow rate of return analysis method was used [27]. MFSP is the selling price of the fuel that makes the net present value of the process equal to zero, with a specified discounted cash flow rate of return over the overall plant life (i.e., 20 years). A stream factor of 90% was assumed for both SOT and goal cases [27]. Table 3 gives the economic parameters used to calculate the MFSP. In addition, the gasoline-equivalent price at $/GGE (gallon gasoline-equivalent) for the final product was calculated to compare the higher heating value (HHV) of fuel and generic commercial gasoline and the MFSP of the fuel product:

MFSP ð$=GGEÞ ¼

MFSP of final product  gasoline HHV Final product HHV

ð2Þ

This method provided a consistent comparison of prices for final fuel product with different heating values. 4. Results and discussion 4.1. Performance results Table 4 lists major performance results in the SOT and goal cases for the biomass HTL and upgrading system. Although the same biomass feed flow rate is assumed in both cases, the goal case consumes more natural gas because it produces more bio-oil and has a hydrocracker, leading to higher hydrogen requirements, and thus higher natural gas consumption. In addition, lower organic loss to the water phase leads to less offgas generated from the HTL wastewater treatment for use in the hydrogen plant. The final hydrocarbon production rate of the goal case is about 63% higher than in the SOT case. In the goal case, fewer organics are assumed to be lost to the water phase, thus increasing the bio-oil production rate. In addition, the heavy end of the hydrotreated bio-oil is treated as by-product in the SOT case, while in the goal case it is hydrocracked into additional gasoline and diesel-range products. The final product yield is also calculated on a GGE basis, which provides a common basis for yield comparison between the two cases with products having different heating content. The water usage per unit of product in the goal case is lower than in the SOT case. This stems from the higher goal case product yield and lower cooling demand. Steam condensing associated with rotating equipment drivers is the largest cooling-water consumer. The lower level of dissolved organics in the goal case water phase leads to less offgas generated from wastewater treatment, which leads to less fuel gas available for steam generation. In addition, the higher bio-oil yield requires more hydrogen and thus more steam to the hydrogen plant, leaving less steam available for electricity generation. Therefore, the goal case has a lower stream flow rate to the steam condensers than the SOT case and thus a lower cooling-water demand. The goal case also generates less electricity because of less steam available for steam drivers.

Therefore, the overall electricity requirement in the goal case is higher than in the SOT case. However, because of higher product yields, the electricity requirement per unit of product in the goal case is lower than in the SOT case. Carbon efficiencies for the final product and by-products are calculated. Carbon efficiency is calculated by dividing the sum of the carbon molar flow rates for each carbon-containing component in the products by the carbon molar flow rates of biomass and other carbon-containing raw materials. Fig. 2 demonstrates the overall carbon balance of the major processes in the SOT case. The 99.7% of total carbon input to the system is from the biomass feedstock and 0.3% of the total carbon input is from the natural gas used for hydrogen generation. About 39.9% of the carbon partitions to the aqueous phase. This is only slightly lower than the carbon in the bio-oil. Therefore, reducing carbon or organics lost to the aqueous phase is important for increasing carbon efficiency. With an assumption of a 30% improvement in the carbon recovery to the bio-oil phase for the goal case, the carbon partitioning to the aqueous product decreases to 24.2% and the bio-oil carbon partitioning increases to 59.1% based on the total carbon input from the biomass and natural gas. The addition of hydrocracker to further treat the heavy fraction from the hydrotreating process increases the carbon reporting to gasoline and diesel products. Therefore the

Table 4 Performance results for the biomass HTL and upgrading system. Case

SOT

Goal

Raw materials Biomass feedstock flowrate, dry metric ton/day Natural gas flowrate, kg/h

2000 165

2000 1420

Overall process yields Hydrocarbon (final product) production rate, L/h Hydrocarbon production rate, million GGE/yr Hydrocarbon yields, L/kg dry feedstock Heavy oil (by-product) production rate, L/h Heavy oil yield, L/kg dry feedstock

18,560 42.9 0.22 3358 0.04

30,240 69.9 0.36 0 0

Hydrogen consumption Hydrogen feed to hydrotreater, kg/kg bio-oil Hydrogen feed to hydrocracker, kg/kg bio-oil

0.041 0

0.038 0.006

Water usage Cooling water makeup, L/L product Boiler feed water makeup, L/L product Total water usage, L/L product

9.44 0.96 10.4

3.96 0.66 4.62

Electricity usage Electricity consumption, MWe Electricity generation, MWe Net electricity requirement, kWh/L product

29.4 16.2 0.71

26.8 11.0 0.52

Carbon efficiency Overall carbon efficiency, % Ca Hydrocarbon, % C based on biomass only Hydrocarbon, % C based on biomass + natural gas

43.7 35.1 35.0

56.4 57.8 56.4

Energy efficiency Energy input Biomass feedstock, GJ/h, HHV basis Natural gas, GJ/h, HHV basis

1627 9.2

1627 79.4

Energy output Hydrocarbon, GJ/h, HHV basis Heavy oil, GJ/h, HHV basis

714.0 162.3

1163 0.0

Energy efficiency for hydrocarbon onlyb, % HHV basis Overall energy efficiencyc, % HHV basis

42.4 52.0

65.9 65.9

a The overall carbon efficiency is calculated as the total carbon moles in final liquid fuel and by-product heavy oil divided by the total carbon input in the biomass feedstock and natural gas. b Energy efficiency is calculated as the total energy output divided by total energy inputs. The total energy inputs include energy in biomass, natural gas, and required electricity. c Overall energy efficiency is for all fuel products, including both final liquid fuel product and by-product heavy oil.

390

Y. Zhu et al. / Applied Energy 129 (2014) 384–394

Wastewater 9.1% C

Natural gas 0.3% C

Gas 30.8% C

Anaerobic digestion

Aqueous product 39.9% C

Flue gas 44.6% C

Hydrogen plant

Gas 11.8% C

Gas 1.7% C

Solids 2.8% C

Separation

Hydrotreator

Separation

Filter

HTL

Biomass 99.7% C

Feed pretreat

H2

Bio-oil 45.3% C

Final product 35.0% C By-product 8.6% C

Aqueous 0.0% C

Fig. 2. Carbon balance of the SOT case.

goal case has higher carbon efficiency than the SOT case. As moving from the SOT case to goal case, the final product carbon fraction increases from 35.0% to 56.4%. The carbon efficiency of the goal case is comparable to the estimated carbon efficiency of a biomass fast pyrolysis and upgrading system [24]. The energy content of biomass, natural gas, and fuel products are calculated based on their flow rates and HHV. The goal case efficiency is 14% higher than the SOT case. The primary reason is the lower organics loss to the water phase and thus higher product yields. 4.2. Cost results The major cost results are listed in Table 5. The total installed cost (TIC) in the goal case is about 8.6% lower than in the SOT case. The HTL step and the bio-oil upgrading by hydrotreating only or hydrotreating and hydrocracking represent the two most expensive processes in the overall system. The sum of the installed cost for these two processes is about 61% and 49% of the TIC in the SOT and goal case, respectively. Therefore, the cost reduction of these two steps is important for reducing the total capital cost. The lower operating pressure and temperature of the HTL reactor system in the goal case (Table 2) leads to about 10% reduction in the installed cost compared to the SOT case. Compared to other major biomass conversion technologies, such as pyrolysis or indirectly-heated gasification, the HTL technology is more expensive for the capital cost [24,27]. The major reason is the much higher operating pressure and the more expensive shell-and-tube design for the HTL reactors compared to a pyrolyzer or gasifier. Commercial or large-scale HTL plants have not been built. Uncertainties exist in the design and cost estimation of the HTL processing step as a result of lack of knowledge and field data. Developing pilot- or demonstration-scale plants will provide more detailed design information and thus reduce the possibility of an overestimate or underestimate of cost. The cost of the hydrotreating step in the goal case is about 30% lower than in the SOT case. The single-stage hydrotreating configuration assumed in the goal case reduces the equipment cost of the process compared to the two-stage hydrotreating configuration assumed in the SOT case. Compared to the SOT case, the goal case adds a hydrocracking step and thus has extra cost for this process. The total upgrading cost, including both hydrotreating and hydrocracking, of the goal case is about 7% lower than that of the SOT case. The goal case assumes lower organics loss to the water phase and thus higher bio-oil production as well as the addition of the hydrocracking step. Therefore, more hydrogen is required in the goal case and thus the hydrogen plant cost in the goal case is about

Table 5 Cost results for the biomass HTL and upgrading system. Case (2007 US dollars)

SOT

Goal

Installed costs, $ million Biomass conditioning HTL reactor system Upgrading (hydrotreating) Upgrading (hydrocracking) Hydrogen plant Utilities Missing equipment

27.8 88.7 95.9 n/a 23.8 37.3 27.4

27.8 80.0 66.7 22.1 31.1 22.3 25.0

Total installed cost (TIC) Indirect costs Fixed capital investment (FCI) Total capital investment (TCI)

301 187 488 512

275 171 446 468

Operating cost, $ million/yr Variable operating cost Feedstock Natural gas Catalysts and chemicals Waste disposal Electricity and other utilities Co-product credits Fixed costs

46.0 0.26 5.48 25.0 7.48 -7.97 23.9

46.0 2.28 5.57 13.8 8.74 0 22.2

Capital depreciation Average income tax Average return on investment

24.4 16.3 45.4

22.3 14.9 40.2

MFSP, $/L product MFSP, $/GGE product

1.29 4.44

0.74 2.52

31% higher than in the SOT case. The lower organics loss to the water phase also leads to less offgas generation from the wastewater treatment and thus less steam generation from available offgas, which, in turn, leads to a higher cost for the utilities section of the goal case. In the goal case, the lower cost for the HTL and hydrotreating steps more than offsets the cost increases from the addition of the hydrocracking step and the larger hydrogen plant. Comparing the cost reductions in the HTL step to that in the hydrotreating step, the effect of reducing operating pressure and temperature of the HTL step is not as significant as replacing the two reactors in series design with a single reactor in the hydrotreating process. Further, the SOT and goal cases have differences in their operating costs. The higher hydrogen demand and lower offgas available for hydrogen generation in the goal case require a greater usage of natural gas than in the SOT case. The cost of catalysts and chemicals in the goal case is slightly higher than in the SOT case because the higher bio-oil production leads to higher upgrading catalyst

391

Y. Zhu et al. / Applied Energy 129 (2014) 384–394

$5.00 $4.50 $4.00

MFSP, $/GGE

consumption. More than 90% of the waste disposal cost results from the wastewater treatment with the remainder from the HTL reactor solid waste disposal. The waste-disposal cost in the goal case is only about 50% of those of the SOT case because the goal case assumes less organics loss to the aqueous phase, which reduces the wastewater treatment cost. Electricity and other utility costs in the SOT case are lower than in the goal case, mainly due to higher electricity generation, and thus lower net power consumption, in the SOT case. Heavy compounds in the goal case are treated by the hydrocracking step to form final products; thus, no co-product credit is taken in the goal case. The annual total variable cost in the goal case is slightly higher than in the SOT case. For the variable cost, the lower wastewater treatment costs in the goal case are offset by higher natural gas and utility costs and the lack of co-product credits. Without considering the feedstock cost, the waste disposal cost represents the biggest fraction of the total variable cost for both cases, which are 33% and 18% in the SOT and goal cases, respectively. Therefore, to some degree, reducing the wastewater treatment cost is important for reducing production costs. To reduce the wastewater treatment cost, either the dissolved organics in the water phase must be reduced or a lower cost wastewater treatment process should be developed. The overall production cost comparison shows that the MFSP in the goal case is about 43% lower than in the SOT case, mainly resulting from the lower capital cost and higher final products yield. The MFSP in unit of $/gallon is $4.89/gallon (converted from the value in $/L), which is higher than its gasoline-equivalent value of $4.44/GGE. The final product is a blend of diesel and gasoline range fuels, which has a heating value about 10% relatively higher than the conventional gasoline. From 2007 to 2012, the U.S. total gasoline wholesale price ranges from $1.77 to 2.93/gallon [44]. Therefore, the liquid fuel produced via the current HTL and upgrading technologies are still not competitive with the conventional petroleum-based gasoline product. However, with the future improvements assumed in the goal case, the liquid fuel production from biomass HTL and upgrading becomes economically attractive. The impacts of major improvements for moving from the SOT to the goal case are investigated and shown in Fig. 3. Impacts of these improvements are quantified by the changes in MFSP. The reduction of the organics loss to the water phase leads to the biggest reduction in the production cost, which is $1.22/GGE or 27% of the MFSP of the SOT case. Reduced organics loss to water increases the bio-oil yield and reduces the wastewater treatment cost. The bio-oil yield increase directly leads to the increase in the final products yield. This increase more than offsets the cost of additional hydrotreating and hydrogen generation. To meet the goal of reducing organics loss to the water phase, additional research is ongoing to improve the three-phase separation performance. In this study, the equipment cost of the three-phase separation process is assumed to be unchanged. With improved separation performance, the possible trade-offs between good separation and high equipment cost need further investigation to determine the optimum approach for this critical unit operation. Adding a hydrocracker step in the goal case leads to the second biggest cost reduction for the system, which is about 8% of the MFSP of the SOT case. The heavy oil comprises about 18 wt% of the stable oil from the hydrotreating process (as seen in Table 2). Adding a hydrocracker eliminates the credit for selling heavy fuel oil as by-product of approximately $8 million per year in the SOT case. It also leads to an increase of $22.1 million in the installed cost. However, adding a hydrocracker increases the final product yields. The final production cost reduction by adding a hydrocracker demonstrates that the benefits of the final products yield increase exceed the disadvantages of the capital cost increase and by-product credit elimination.

$3.50 $3.00 $2.50 $2.00 $1.50 $1.00 $0.50 $0.00 SOT

Lower HTL Reactor Pressure

Single-step Reduce organics loss to hydrotreating water phase

Add hydrocracker for heavier components

Goal

Fig. 3. Cost effect of improvements of moving from SOT to goal case.

Lowering the HTL reactor operating pressure has moderate benefits, which is an approximate 4% decrease in the MFSP compared to the SOT case. Using a single-stage hydrotreating reactor also reduces the MFSP by about 4%. These results are based on the assumptions that lowering the HTL operating pressure and using a single-stage hydrotreating reactor do not cause changes in the product yield or quality. This study does not consider the effects of any government subsidies on the production cost estimation and feasibility evaluation. If there are government incentives, such as mandatory blending requirements, financial support for project development, and tax credit, for developing commercial scale biomass-based HTL and upgrading plants, it will promote the development of HTL technologies and increase the market competitiveness of the liquid fuel production from this technology. 4.3. Sensitivity analysis The goal case results demonstrated a promising future for woody biomass conversion to liquid fuel in a commercial plant with mature HTL and upgrading technologies. The goal case is selected for the sensitivity analysis. In this section, an alternative configuration of small scale distributed HTL plant only is investigated. The effects of selected parameters on the production cost are also investigated. The uncertainty in the production cost resulting from the uncertainty in parameters is investigated. 4.4. Distributed HTL plant The analysis in the previous section provides estimates for a stand-alone plant, which includes both HTL and upgrading units in a single system. An alternative configuration is small scale distributed HTL plants combined with a centralized upgrading unit, such as upgrading by an existing refinery unit. Compared to a stand-alone plant, a distributed HTL plant eliminates the need of the upgrading units and a hydrogen generation plant. The bio-oil is the final product of the HTL plant. To evaluate the feasibility of developing small scale distributed HTL plants with a centralized upgrading unit system, the production cost of a different scale distributed HTL plant is investigated and the results are shown Fig. 4. With the biomass feed rate varying between 10 and 2000 dry metric ton/day, the MFSP of bio-oil ranges from $1.87 to $7.76/GGE. The MFSP of a distributed HTL plant with a biomass feed rate at 2000 dry metric ton/day has a MFSP of about $1.87/GGE, which is about 26% less than that of the goal case with the same biomass feed rate, which is $2.52/GGE. This reduction mainly results from

392

Y. Zhu et al. / Applied Energy 129 (2014) 384–394

8.0

MFSP, $/GGE

7.0 6.0 5.0 4.0 3.0 2.0 1.0 0

500

1000

1500

2000

Woody biomass feed rate, dry metric ton/day Fig. 4. Effects of plant scale on the bio-oil MFSP of distributed HTL plant.

the decrease in capital cost, which decreases about 44% compared to the stand-alone plant. Compared to the 2012 average gasoline wholesale price of $2.93/gallon [44], the size of the distributed HTL plant should be larger than 200 dry metric ton/day and thus the produced bio-oil has a competitive price compared to the conventional petroleum-based fuel product. Considering smaller HTL plants should have lower biomass feedstock price due to lower transportation cost and higher feedstock availability, the minimum distributed HTL plant size can be 150 dry metric ton/day when the biomass feedstock price decreases about 50% compared to the baseline value. 4.5. Identification of key factors Sensitivity analysis for selected parameters is implemented to demonstrate their effects on the production cost. The sensitivity analysis results are shown in Fig. 5. The change in the final product yield has the most significant impact on the final production cost. A 20% decrease in the yield causes a 25% increase in the MFSP. The assumption of the product yield decrease is based on the potential uncertainty in the HTL conversion efficiency and the organics loss to the water phase. The MFSP is also strongly dependent upon the upgrading equipment cost and feedstock price. The design basis of the upgrading process is the bench scale testing information. To account for the cost risks when scaling up to a commercial plant, an upper value of 100% increase in the baseline cost for the upgrading process, which is $36 million, is assumed. This leads to approximately 36% in the total capital investment and 19% increase in the MFSP. For the feedstock price, large scale plants require high availability of woody biomass and usually high transportation cost. These factors will lead to an increase in the feedstock price. For every 100% increase in feedstock price, the MFSP increases about 26%.

The liquid hourly space velocity (LHSV) of the HTL reaction has a moderate cost impact. If the HTL reaction is greater than zero order, a plug flow reactor would have a much shorter residence time. Therefore, a 100% increase from the baseline LHSV value of 4–8 h1 is assumed. This change leads to a 7% decrease in the total capital investment, which causes nearly a 4% decrease in the MFSP. Decreasing the LHSV by 50% (4–2 h1) causes the total capital investment to increase by 13%, and the MSFP to increase by 7%. Therefore, further increasing the LHSV only has limited benefits towards reducing the HTL system cost, assuming that the bio-oil yield does not change with the higher LHSV. Little or no economic benefit may be possible if higher LHSV leads to lower bio-oil yields. Another parameter with moderate impact is the missing equipment contingency factor. This factor is used to quantify the expected increase in the capital cost when applying advanced technologies that are currently not commonly practiced in a commercial scale plant. Because of lack of knowledge and experience for a commercial scale biomass HTL and upgrading plant, there are uncertainties in this factor. An upper value of 30% of this factor is assumed and the total capital investment increases by 18% from the baseline value. The corresponding value for the MFSP is $2.76/GGE. The uncertainty of the missing equipment contingency factor can be reduced when field data from large-scale biomass HTL and upgrading plants are available and detailed engineering analysis has been conducted. The variations in wastewater treatment and biomass feeding equipment cost have similar effects as the missing equipment contingency factor. For the lower and upper values of wastewater treatment operating cost (ranging from 50% to 50% of the baseline value), the MFSP changes ±4% relative to its baseline value. Other parameters, including the natural gas price and catalysts cost, show limited impacts on the production cost. 4.6. Uncertainty in production cost The previous analysis provides deterministic (single point) estimates for the production cost. It does not reflect inherent uncertainties in cost estimates resulting from a lack of field data and experience for advanced technologies. In addition, the above sensitivity analysis only reflects the effect of single parameter in the production cost. In this section, uncertainties of the selected model parameters for the above sensitivity analysis are characterized by probability distributions based on expert judgments about their potential variations in commercialization. In Table 6, the deterministic (or baseline) values and probability distributions for each uncertain input parameter are provided. The baseline value is assumed to be the most likely value in the distributions. Latin Hypercube Sampling (LHS) method is adopted in this study [45].

Product yield (0.36: 0.29 L/kg dry biomass) HTL LHSV (8: 4: 2 v/v/h) Upgrading equipment cost (35.96: 71.91 MM$) Biomass feeding equipment cost (9.87: 19.74 MM$) Missing equipment contingency factor (0.1: 0.3) Catalysts cost (2.79: 5.57: 8.36 MM$/yr) Natural gas (0.15: 0.20: 0.46 $/kg) Wastewater treatment (6.41: 12.83: 19.24 MM$/yr) Feedstock price ($60: $70: $95/dry tonne) -0.2

-0.1

0

0.1

0.2

0.3

0.4

0.5

Change to MFSP, $/GGE Fig. 5. Effects of parameters variation on the MFSP of the goal case.

0.6

0.7

393

Y. Zhu et al. / Applied Energy 129 (2014) 384–394 Table 6 Uncertainties in Selected Input Parameters. Parameters

Deterministic value

Probability distributionsa

Feedstock price, $/tonne Wastewater treatment, million$/yr Natural Gas, $/kg Catalyst cost, million$/yr Missing equipment contingency factor Biomass feeding equipment cost, million$ Upgrading equipment cost HTL LHSV, v/v/h Product yield, L/kg dry biomass

70 12.83 0.20 5.57 0.1 9.87 35.96 4 0.36

Triangular; 60–95 (70) Normal; 6.42–19.25 Triangular; 0.15–0.46 (0.20) Normal; 2.79–8.36 Triangular; 0.1–0.3 (0.1) Triangular; 9.87–19.73 (9.87) Triangular; 35.96–71.91 (35.96) Triangular; 2–8 (4) Triangular; 0.29–0.36 (0.36)

Cumulative Probability

a

For triangular distributions, the mode value is given in parentheses. For normal distributions, the 99.8% probability range is provided.

1 0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0 2.00

Deterministic Uncertainty Current gasoline price

2.50

3.00

3.50

4.00

MFSP, $/GGE Fig. 6. Comparison of deterministic and uncertainty estimate for the MFSP of the goal case.

Samples of the selected parameters are produced based on their probability distributions. A sample size of 100 is selected and the samples are input to the cost calculation model. After each run of the cost model, the results of MFSP are collected. The uncertainty in the MFSP is represented by using summary statistics and cumulative probability curve. The uncertainty analysis results for the MFSP of the HTL and upgrading plant is depicted in Fig. 6. There is only about 1.5% probability that the MFSP is lower than the deterministic analysis result of $2.52/GGE. The 95% probability range for the MFSP is $2.64–$3.70/GGE or +5% to +47% relative to the deterministic result. The mean value of the MFSP is $3.09/GGE and the standard deviation is $0.29/GGE. Therefore, for the goal case design, considering the uncertainties in advanced technologies, the deterministic estimate normally underestimates the production cost of the plant. With the assumed uncertainties in the selected parameters, the production cost of a commercial woody biomass HTL and upgrading plant will be normally higher than its deterministic estimate and the high end value is about 47% higher than the deterministic result. Compared to the current gasoline wholesale price, there is about 30% probability that the production cost of the HTL and upgrading plant is lower. 5. Conclusions This study investigated the technical and economic feasibility for producing liquid fuel from biomass conversion by the HTL and hydroprocessing technologies at commercial scales. The TEA results provided key estimates about the performance and cost for developing a large-scale biomass HTL and upgrading system. The SOT case used best available experimental tests results as the major design basis for TEA. This method enables a reliable connection between the current innovative technologies and their future applications in a commercial system. The HTL and upgrading technologies efficiently convert woody biomass to liquid fuels, but with a high production cost. The cost results of the SOT case demonstrate that liquid fuel production cost based the current

HTL technology is not competitive with the conventional petroleum-based gasoline. Among the assumed future improvements for the goal case, reducing organics loss to the water phase leads to the most significant cost reduction resulting from higher final products yields and lower wastewater treatment costs. Alternative configuration of small scale distributed HTL plants is evaluated and plant scale significantly affects the final production cost of bio-oil. Selected factors, including the key improvement of reduced organics loss assumed for the goal case, are evaluated. Key factors affecting the production cost include the product yield, upgrading equipment cost, and feedstock price. The identification of key factors provided insights for the priority for future R&D work. Uncertainty analysis provides a more objective estimation than the deterministic analysis for the risks of commercialization of advanced technologies. Although the goal case technology is promising for future liquid fuel production via woody biomass HTL, there are still financial risks resulting from lack of knowledge and experience in advanced technologies. With additional R&D, more information can be obtained for this technology and uncertainties in cost estimation can be reduced. Acknowledgments This work was supported by the National Advanced Biofuels Consortium, which is funded by the U.S. Department of Energy’s Bioenergy Technologies Office with recovery act funds. PNNL work was conducted under U.S. Department of Energy contract DE-AC05-76RL01830. We would like to thank Michael Talmadge at National Renewable Energy Laboratory for HTL reactor cost estimations used in this report. Appendix A. Experimental tests Bench-scale continuous flow experimental tests were conducted to convert woody biomass to liquid oil via HTL and catalytic hydrotreating. Biomass feedstock (pine forest product residual, including bark) was ground to fine particles (with 99% < 0.4 mm) in an immersion mill and mixed with by-product water from previous runs to form slurry compatible with the syringe pumps used in the experimental system. The solids content in the slurry for various experimental runs ranged from 8 to 15 wt%. The slurry was pumped to the HTL reactor, which was a 1 L CSTR. Initial tests were conducted with the CSTR operated at LHSVs ranging from 1 to 2 h1. Subsequent HTL tests conducted with a tubular reactor at a higher LHSV of about 4 h1 produced bio-oil of roughly similar quality. Feedstock was processed continuously for 5–10 h with feed rates of about 1600 g/h. HTL reactions were run at approximately 350 °C and 20 MPa. Sodium carbonate (Na2CO3) was used as the buffering agent at 1 wt% of total feed slurry. The purpose is to reduce acid catalyzed decomposition of carbohydrates, phenol aldehyde cross linking reactions and favor more useful aldol condensation reactions. During HTL, if the pH of the slurry drops below

394

Y. Zhu et al. / Applied Energy 129 (2014) 384–394

4 pH units, the dehydration, polymer formation, and ultimately char formation, are favored. Therefore, a buffering agent is employed to inhibit the formation of high molecular weight compounds and solid wastes and favor the bio-oil production. The hot product from the HTL reactor was sent to a high pressure backwash type filter to separate solids (e.g., unreacted biomass, ash, and char) from the gas and liquid phases. The gas and liquids were separated and collected for composition and physical property analysis. The HTL bio-oil was upgraded by hydrotreating to deoxygenate and reduce the boiling point. Hydrotreating tests were conducted in a two-stage 440 ml continuous fixed-bed reactor system charged with presulfided CoMo/F-Al2O3 catalyst. Bio-oil was treated continuously at feed rates of 65 g/h and 76 g/h for about 26 h (total). The experimental LHSVs used for this study were approximately 0.54 and 0.18 h1 for stages 1 and 2, respectively. The initial HTL and upgraded bio-oils were analyzed. Results included elemental analysis, density, distillation curves, and qualitative gas chromatography–mass spectrometry data. A distillation curve from analysis of the upgraded bio-oil, along with a quality control (QC) diesel sample shows that the BP of the upgraded bio-oil from woody biomass ranges from 110 to 410 °C. About 50% of the upgraded bio-oil is in the QC diesel range. References [1] Elliott DC. Historical developments in hydroprocessing bio-oils. Energy Fuels 2007;21:1792–815. [2] Xu C, Lad N. Production of heavy oils with high caloric values by direct liquefaction of woody biomass in sub/near-critical water. Energy Fuels 2008;22(1):635–42. [3] Hammerschmidt A, Boukis N, Hauer E, Galla U, Dinjus E, Hitzmann B, et al. Catalytic conversion of waste biomass by hydrothermal treatment. Fuel 2011;90:555–62. [4] Oasmaa A, Kuoppala E. Fast pyrolysis of forestry residue. 3. Storage stability of liquid fuel. Energy Fuels 2003;17(4):1075–84. [5] Demirbas A. Competitive liquid biofuels from biomass. Appl Energy 2011;88(1):17–28. [6] Elliott DC. In: Brown Robert C, editor. Hydrothermal processing, thermochemical processing of biomass: conversion into fuels, chemicals and power. John Wily & Sons; 2011 [chapter 7]. [7] Rust International Corporation. An investigation of liquefaction of wood at the biomass liquefaction facility Albany Oregon, PNL-5114, Pacific Northwest Laboratory, Richland Washington, April 1982. [8] Berend RH, Zeevalkink JA, Goudriaan F, Naber JE. Results of the first long duration run of the HTU pilot plant at TNO-MEP. In: Biomass for energy, industry, and climate protection: proceedings of 2nd world conference held in Rome, Italy, May 10–14, 2004. [9] Goudriaan F, Naber JE. HTU process design and development: innovation involves many disciplines. In: Bridgwater AV, Boocock DGB, editors. Science in thermal and chemical biomass conversion. Newbury Berks (UK): CPL Press; 2006. p. 1069–81. [10] Goudriaan F, Naber JE. HTUÒ diesel from wet waste streams. In: Symposium new biofuels, Berlin, German, May 2008. [11] Goudriaan F, van de Beld B, Boerefijn FR, Bos GM, Naber JE, van der Wal S, et al. Thermal efficiency of the HTU process for biomass liquefaction. In: Bridgwater AV, editor. Thermochemical biomass conversion. Oxford (England): Blackwell Science, Ltd.; 2001. p. 1312–25. [12] Goudriaan F, Peferoen DGR. Liquid fuels from biomass via a hydrothermal process. Chem Eng Sci 1990;45(9):2729–34. [13] Akhtar J, Amin NAS. A review on process conditions for optimum bio-oil yield in hydrothermal liquefaction of biomass. Renew Sustain Energy Rev 2011;15(3):1615–24. [14] Mørup AJ, Christensen PR, Aarup DF, Dithmer L, Mamakhel A, Glasius M, et al. Hydrothermal liquefaction of dried distillers grains with solubles: a reaction temperature study. Energy Fuels 2012;26:5944–53. [15] Yin S, Tan Z. Hydrothermal liquefaction of cellulose to bio-oil under acidic, neutral and alkaline conditions. Appl Energy 2012;92:234–9. [16] Feng W, van der Kooi HJ, de Swaan Arons J. Biomass conversions in subcritical and supercritical water: driving force, phase equilibria, and thermodynamic analysis. Chem Eng Process 2004;43:1459–67. [17] Bolhàr-Nordenkampf M, Hofbauer H. Gasification demonstration plants in Austria, IV. In: International Slovak biomass forum, Bratislava, Slovakia, February 9–10, 2004. [18] Ensyn. Ensyn’s commercial merchant facility for fuels production. [last accessed in May 2012]. [19] Robinson PR, Dolbear GE. Hydrotreating and hydrocracking: fundamentals. In: Hsu CS, Robinson PR, editors. Practical advances in petroleum processing, vol. 1. New York: Springer Science + Business Media Inc.; 2006 [chapter 7].

[20] Hamelinck CN, Faaij APC. Future prospects for production of methanol and hydrogen from biomass. J Power Sources 2002;111:1–22. [21] Tijmensen MJA, Faaija APC, Hamelinck CN, van Hardeveld MRM. Exploration of the possibilities for production of Fischer Tropsch liquids and power via biomass gasification. Biomass Bioenergy 2002;23:129–52. [22] Zwart WR, Boerrigter H, Deurwaarder EP, van der Meijden CM, van Paasen SVB. Production of synthetic natural gas (SNG) from biomass – development and operation of an integrated Bio-SNG system, ECN-E–06-018, Energy Research Centre of the Netherlands (ECN), Petten, The Netherlands, November 2006. [23] Phillips SD. Technoeconomic analysis of a lignocellulosic biomass indirect gasification process to make ethanol via mixed alcohols synthesis. Ind Eng Chem Res 2007;46:8887–97. [24] Jones SB, Holladay JE, Valkenburg C, Stevens DJ, Walton CW, Kinchin C, et al. Production of gasoline and diesel from biomass via fast pyrolysis, hydrotreating and hydrocracking: a design case. PNNL-18284, Pacific Northwest National Laboratory, Richland, WA, February 2009. [25] Jones SB, Zhu Y. Techno-economic analysis for the conversion of lignocellulosic biomass to gasoline via methanol-to-gasoline (MTG) process, PNNL-18481, Pacific Northwest National Laboratory, Richland, WA, August 2011. [26] Zhu Y, Gerber MA, Jones SB, Stevens DJ. Analysis of the effects of compositional and configurational assumptions on product costs for thermochemical conversion of lignocellulosic biomass to mixed alcohols – FY 2007 progress report, PNNL-17949, revision 1. Pacific Northwest National Laboratory, Richland, WA, February 2009. [27] Dutta A, Talmadge M, Hensley J, Worley M, Dudgeon D, Barton D, et al. Process design and economics for conversion of lignocellulosic biomass to ethanol – thermochemical pathway by indirect gasification and mixed alcohol synthesis. NREL/TP-5100-51400, National Renewable Energy Laboratory, Golden, Colorado, May 2011. [28] Beckman D, Elliott D, Gevert B, Hornell C, Kjellstom B, Ostman A, et al. Technoeconomic assessment of selected biomass liquefaction processes. Technical Research Center of Finland (VTT) report #697, Espoo, Finland, 1990. [29] Kampman BE, den Boer LC, Croezen HJ. Biofuels under development – an analysis of currently available and future biofuels, and a comparison with biomass application in other sectors. Report 05 4894 11, CE Delft, Delft, The Netherlands, May 2005. [30] Kerssen M, Berends RH. Life cycle analysis of the HTU process. Dutch research organisation for environment, energy and process innovation (TNO-MEP), Apeldoorn, The Netherlands; 2004. [31] Frey HC, Zhu Y. Techno-economic analysis of combined cycle systems. In: Rao Ashok D, editor. Combined cycle systems for near-zero emission power generation. Cambridge (UK): Woodhead Publishing, May 2012 [chapter 11]. [32] Schmidt AJ, Lindstrom TH, Lupton FS, Talmadge MS, Zhu Y. Mid stage 2 report on the hydrothermal liquefaction strategy for the NABC Leadership Team. PNNL-21768 (limited distribution), Pacific Northwest National Laboratory, Richland, WA, September 2012. [33] Berglin EJ, Enderlin CW, Schmidt AJ. Review and assessment of commercial vendors/options for feeding and pumping biomass slurries for hydrothermal liquefaction, PNNL-21981, Pacific Northwest National Laboratory, Richland, Washington, November 2012. [34] Solantuasta Y. Co-processing of upgraded bio-liquids in standard refinery units – BIOCOUP. In: European conference on biorefinery research proceedings. European Commission, Helsinki, Finland, October 19–20, 2006. [35] Haynes HW, Parcher JF, Helmer NE. Hydrocracking polycyclic hydrocarbons over a dual-functional zeolite (faujasite)-based catalyst. Ind Eng Chem Process Des Dev (Ind Eng Chem Res) 1983;22:401–9. [36] Spath P, Aden A, Eggeman T, Ringer M, Wallace B, Jechura J. Biomass to hydrogen production detailed design and economics utilizing the Battelle Columbus Laboratory indirectly-heated gasifier. NREL/TP-510-37408. National Renewable Energy Laboratory, Golden, Colorado, May 30, 2005. [37] SRI International. PEP yearbook international, 2007. Menlo Park (California): SRI International; 2007. [38] Meyers RA. Handbook of petroleum refining processes. 3rd ed. New York: McGraw-Hill; 2004. [39] Aspen Technology, Inc., Aspen Plus 2006.5 user manual. Cambridge, Massachusetts; 2007. [40] Hydrocarbon Processing. Refining processes 2006 handbook. Houston (Texas): Gulf Publishing Co.; 2006. [41] U.S. EIA. Annual energy outlook 2010 with projections to 2035. DOE/EIA0383(2010), U.S. Energy Information Administration, Washington (DC). http:// www.eia.gov/oiaf/archive/aeo10/pdf/0383%282010%29.pdf [last accessed in August 2011]. [42] Humbird D, Davis R, Tao L, Kinchin C, Hsu D, Aden A, et al. Process design and economics for biochemical conversion of lignocellulosic biomass to ethanol. NREL/TP-5100-47764, National Renewable Energy Laboratory, Golden, Colorado, May 2011. [43] Zhu Y, Jones SB, Biddy MJ, Dagle RA, Palo DR. Single-step syngas-to-distillates (S2D) process based on biomass-derived syngas – a techno-economic analysis. Bioresour Technol 2012;117:341–51. [44] U.S. EIA. U.S. total gasoline wholesale/resale price by refiners. U.S. Energy Information Administration, Washington, DC. [released at September 2013]. [45] Zhu Y, Frey HC. Uncertainty analysis of integrated gasification combined cycle systems based on frame 7H versus 7F gas turbines. J Air Waste Manage Assoc 2006;56:1649–61.