Negative-carbon drop-in transport fuels produced via catalytic hydropyrolysis of woody biomass with CO₂ capture and storage

Abstract

Process designs for prospective first-of-a-kind (FOAK) catalytic hydropyrolysis facilities converting woody biomass residues into “drop-in” transportation fuels were developed, including some designs incorporating CO₂ capture and storage (CCS). The energetic, carbon, and economic performances of these designs were simulated and analyzed. Estimated greenhouse gas emissions for the resulting fuels are far below those of conventional petroleum-derived fuels. For plant designs with CCS, the biofuels are characterized by strongly negative emissions. The additional capital costs and energy penalties for CO₂ capture range from modest to high, depending on the extent of capture employed. The fuel production cost at a commercial-scale FOAK plant without CCS corresponds to a break-even crude oil price of 95 $ per bbl. At a 120 $ per t CO_2,eq. greenhouse gas (GHG) emission price, the plant design that would capture about half of the CO₂ available for capture would have identical production cost as the design without any CO₂ capture; in both cases the break-even oil price would be 28 $ per bbl. A design maximizing CO₂ capture would produce fuels with a break-even oil price of 44 $ per bbl at this GHG emission price. The prospective economics of drop-in fuels from biomass produced via catalytic hydropyrolysis appear quite favorable relative to other biofuel production systems, but can only be confirmed via demonstrations at scale.

---

1. Introduction

Limiting global warming requires significant reductions in greenhouse gas (GHG) emissions.¹ One way to reduce emissions in the transportation sector is by replacing petroleum-based liquid fuels with sustainably produced biofuels. By integrating capture and underground storage of by-product CO₂ into the production of the biofuels, CO₂ can effectively be removed from the atmosphere. This paper presents results of a detailed assessment of the prospects for one technological route to such fuels involving advanced pyrolysis-based processing of woody biomass residues, in some cases with CO₂ capture and storage (CCS) as part of the design.

Pyrolysis is the thermally-driven conversion of biomass into an organic liquid (pyrolysis oil) in the absence of oxygen. Liquid yields can be high compared to biochemical or gasification-based biofuels production, and the simpler processing involved implies lower capital costs per unit of biomass input and prospectively reasonable economics for relatively smaller-scale plants.^2–4 However, crude pyrolysis oils contain a variety of unwanted molecules (e.g., carboxylic acids, alcohols, benzene) and refining is needed before the oil can be used as a vehicle fuel.^3,5,6 Dedicated refining at a pyrolysis facility would be expensive due to the small scale. As a result, pyrolysis is often foreseen as an intermediate processing step: relatively small pyrolysis facilities convert biomass into energy-dense oils that are then transported to existing petroleum refineries for processing into “drop-in” substitutes for petroleum-derived diesel or gasoline. The benefits of this concept are simplified biomass transportation logistics – there is no need to transport large quantities of biomass over long distances to fuel large centralized conversion plants – and reduced investment costs. However, petroleum refiners are often reluctant to co-process pyrolysis oils because of unknown risks to equipment not designed to process such oils.

It is important to note, however, that not all pyrolysis oils are created equal. The products of pyrolysis depend on the operating conditions of the pyrolysis system. With conventional fast pyrolysis, involving heating rates >1000 °C s⁻¹, the resulting oils (fast pyrolysis oil, FPO) are unstable, acidic and hydrophilic, potentially damaging refinery equipment and negatively impacting refinery reliability.^5,7–9 Catalytic pyrolysis produces oils with improved properties compared with those via fast pyrolysis by passing the troublesome vapor-phase molecules through acidic catalysts (e.g., zeolites) that stabilize the vapors and break down carboxylic and hydroxyl groups.^10,11 Depending on the severity of the reaction conditions and the acidity of the catalyst the resulting catalytic pyrolysis oil (CPO) is stable, almost acid-free and hydrophobic. However, the acidic catalysts increase coke formation, which reduces yield.¹² CPO is also rich in aromatics, including benzene, which demands extensive hydrogenation to produce a (drop-in) vehicle fuel. The hydrogenation requires large amounts of hydrogen, increasing refining costs. Also, if the hydrogen is derived from fossil fuels, as is typically the case at existing refineries, the lifecycle GHG emissions associated with the final fuels will be higher than if the hydrogen were from a non-fossil source.

Catalytic pyrolysis in a pressurized hydrogen atmosphere (catalytic hydropyrolysis) is a more recent development that seeks to take advantage of the benefits of catalytic pyrolysis while reducing its disadvantages.^7,13,14 With catalytic hydropyrolysis the immediate exposure of initial pyrolysis vapors to H₂ minimizes formation of aromatics and oxygenated species without increasing coke formation.^7,13–15 Additionally, since the hydrogenation is integral to the process (rather than being done at a separate refinery), there is the potential to generate H₂ from the non-condensable by-products of the pyrolysis process and thereby to maintain low lifecycle GHG emissions of the fuel products. Considering the added complexities and costs of operating reactors in pressurized hydrogen atmospheres, larger scales would likely be favored for catalytic hydropyrolysis facilities than for conventional fast or catalytic pyrolysis facilities. Catalytic hydropyrolysis oil (CHPO) properties would not be unfamiliar to petroleum refiners, lowering the risks that any needed final upgrading would entail for a conventional refinery. Key properties of different pyrolysis oils and of crude oil are compared in Table 1.

Table 1 Properties of non-upgraded oils from different pyrolysis processes.^a The last column displays properties of a generic crude oil

		FPO^b^7–9	Mild CPO^c^16–18	Severe CPO^c^18	CHPO^7,15,19	Crude oil^12
a The abbreviations stand for fast pyrolysis oil (FPO), catalytic pyrolysis oil (CPO) and catalytic hydropyrolysis oil (CHPO). Note that chemical properties and composition for each of these will vary somewhat depending on feedstock and pyrolysis conditions used in its production. b FPO characteristics listed here are for non-upgraded FPO. When FPO is upgraded, its properties and composition will move towards those of CPO. If the upgrading is severe enough, an oil quality similar to severe CPO might be reached. c The difference between mild and severe CPO is the severity of the reaction conditions in the catalytic pyrolysis reactor (e.g., higher temperature, longer residence time, stronger catalyst). d In the petro-chemical industry acidity is often expressed as Total Acid Number (TAN). This is the total amount of KOH needed to completely neutralize the oil (mg KOH per g oil). e Viscosity measured at 40 °C. f Viscosity measured at 50 °C.
Miscibility in crude oil		Poor	Good	Excellent	Excellent	Excellent
Thermal stability		Poor	Medium	High	High	High
Moisture content (wt%ar)		15–30	3–11	Similar to crude oil	Similar to crude oil	0.1
Acidity^d		pH = 2.0–3.7 (∼100 TAN)	pH = 3.7–5.0 (∼25 TAN)	<1 TAN	<1 TAN	<1 TAN
Viscosity	Kinematic	<40 000 cP^e	15–90 cSt^e	15–90 cSt^e	Less viscous than crude oil	210 cSt^f
	Dynamic		16–104 cP^e	16–104 cP^e		180 cP^f
Heating value (MJHHV kgwet^−1)		15–18			42	43
Density (t m^−3)		1.2	1.1	1.1 > x > 0.85	0.85	0.86–0.94
N-Content (wt%dry)		0.1–2	0.1–2	<1	<1	<1
O-Content (wt%dry)		40	12–27	<1	<1	<1
Phenolic content (wt%dry)		25	<25	Low (∼0)	Low (∼0)	Low
BTX content (wt%dry)		Low		High

---

Motivated by the prospectively attractive properties of CHPO, this paper presents detailed technical, economic and GHG emissions assessments of catalytic hydropyrolysis with integrated CO₂ capture and storage to produce vehicle fuels with negative lifecycle GHG emissions. Work on catalytic hydropyrolysis has been reported by several researchers but publications associated with work at the Gas Technology Institute (GTI) in Chicago, USA provide the most detailed understanding currently available in the public domain.^{7,13,15,20–23} GTI has conducted lab-scale testing of a catalytic hydropyrolysis reactor since 2011 and a demonstration project (5 t per day biomass input) is under development in India.²⁴ The work of Tan^25,26 and others,^7,13 supplemented by discussions with experts, provided key inputs for our theoretical design of a first-of-a-kind commercial-scale catalytic hydropyrolysis facility. To our knowledge, our work is the first published performance and cost analysis of catalytic hydropyrolysis integrated with CO₂ capture and storage. We first provide an overview of process designs investigated, followed by discussion of methodologies and assumptions used for (1) process simulations, (2) lifecycle greenhouse gas emissions estimations, and (3) economic analysis. This is followed by a presentation and discussion of results.

2. Overview of process designs

Our catalytic hydropyrolysis process designs include a pyrolysis section, a H₂ generation section and a utilities section (see Fig. 1). Opportunities for CO₂ capture within the H₂ generation section or at the point of final exhaust of combustion products to the atmosphere are indicated in Fig. 1 and described further below.

Fig. 1 Schematic overview of an integrated catalytic hydropyrolysis facility. The green box indicates the pyrolysis section, the red box the H₂ generation section and the yellow box the utilities section. Potential locations for CO₂ capture are depicted by the purple boxes. Compressors and heat exchangers that are part of each plant design are not shown here. The catalytic hydropyrolysis reactor is abbreviated to HPyr.

The pyrolysis section converts the input biomass into gasoline and diesel blendstocks. The biomass is first dried and then fed into the fluidized-bed catalytic hydropyrolysis reactor where it decomposes into char, water, condensable organic compounds and non-condensable gases (NCG). The organic compounds, after condensing, constitute the crude pyrolysis oil. The exact role H₂ plays is not entirely understood, but the H₂ environment in the reactor stabilizes the organic compounds and partly saturates them, inhibiting unwanted side-reactions that occur during conventional fast pyrolysis. The most H₂-consuming reaction is the hydrodeoxygenation of the biomass.²² H₂ also induces chain terminating reactions, thereby limiting char and coke formation. The reducing atmosphere may also contribute to keeping the catalyst active. The pyrolysis reactor can be operated at a wide temperature range (336–650 °C), but most of the experiments reported in the literature use 375–450 °C.^13,21,22 A disengagement zone at the top of the reactor ensures that the heavier catalyst particles remain in the reactor. The vapors and entrained ash and char particles exit to a cyclone that separates solids from the vapors. The char is fed to a combustor (located in the utilities section). The vapors are sent to a fixed-bed hydrodeoxygenation (HDO) reactor where the H₂, in combination with a different catalyst, reduces the heteroatom content (e.g., nitrogen, oxygen, sulfur) of the organic compounds further and also increases saturation of the hydrocarbons. The vapors are subsequently cooled to separate aqueous, organic liquid, and NCG phases. The aqueous phase is sent for waste water treatment. The organic phase is separated into a heavy and a light fraction that are transported to a refinery for further processing or blending into diesel and gasoline fuels. The NCG phase, consisting of unreacted H₂ and gases generated in the catalytic hydropyrolysis reactor, is sent to the H₂ generation section where H₂ required by the catalytic hydropyrolysis and HDO reactors is produced.

In the H₂ generation section, the NCG is first processed through a pressure swing adsorber (PSA-1) to separate most of the unreacted H₂ from the remaining gases. The PSA-1 raffinate is sent to a steam reformer (SR), where the organic molecules are converted into CO and H₂. The product from the SR is sent to a water–gas shift (WGS) reactor where the CO reacts with H₂O to form more H₂. In some of the designs with CO₂ capture investigated here, CO₂ is removed by physical absorption between the WGS and the second PSA (PSA-2). H₂ separated by PSA-2 is combined with the H₂ from PSA-1, recompressed and recycled back to the pyrolysis section. The PSA-2 raffinate is sent to the combustor (located in the utilities section).

The utilities section provides all of the required process heat, steam and electricity as well as some additional electricity for export. The PSA-2 raffinate and the char are burned in a combustor that provides heat to the SR. For reasons discussed below, some of our process designs consider the use of some natural gas in the combustor. After heating the SR, the still-hot combustion gases are used to raise steam for Rankine cycle electricity production. The gases leave the steam generator with sufficient heat to be used for biomass drying prior to exhausting to the atmosphere. In some of the designs with CO₂ capture, CO₂ is removed from the final plant exhaust after the biomass dryer using a chemical absorption system. All the CO₂ produced at the facility is found in this exhaust stream. Thus, the designs that capture CO₂ at this location do not also include the physical capture unit in the H₂ generation section.

3. Methodologies

3.1. Process simulations

Steady-state mass and energy balance simulations were developed using Aspen Plus software for each process design. The technical model for the pyrolysis section is based on lab-scale experimental data. It is assumed that a commercial-scale facility would perform similarly. Key features of the simulations are discussed here, with additional details provided in Table 2.

Table 2 Technical assumptions for process simulations using Aspen Plus software

Process	Assumptions
a The electricity consumption of the biomass handling and feeding is based on Liu et al.^39 Heat required to dry the biomass is assumed to be the heat required to heat the biomass from 15 °C to 104 °C plus the heat required to evaporate the water removed when drying the biomass from a moisture content of 30 wt% to 10 wt%. b The performance of the pyrolysis reactor is based on experimental data of run 9/3 given by Marker et al.^15 A more detailed explanation of how output yields and compositions were determined is given in Appendix C of the ESI. c Based on Meerman et al.^40 d The performance of the HDO was estimated by comparing results of experiment run 9/3 (consisting of only the pyrolysis reactor) and experiment run 8/23 (consisting of a pyrolysis and HDO reactor).^13,15 The HDO simulation is run such that the O-atom content of the condensable organic phase leaving the HDO block is reduced to 1 wt%.^15 A more detailed explanation of how yields and compositions of the various fractions were determined is given in Appendix C of the ESI. e Yields are based on Marker et al.,^13 with modifications as listed in Appendix C of the ESI. f Based on the baseline SMR with CO2 capture case given by Rath et al.^41 g The H2O : carbon ratio is minimized to achieve maximum overall efficiency, while still generating enough H2 for the pyrolysis reactor. The resulting ratio of 1.7 is on the low side for SMR facilities. Engineers from Haldor Topsøe pointed out that the combination of a 500 °C pre-reformer, developments in SR and WGS catalysts, and the low concentration of higher hydrocarbons in the feed should keep coke formation within acceptable levels while still using the low H2O : carbon ratio.^33 h To avoid metallurgical limits, the combustion gas temperature was kept below 1400 °C by controlling the combustion air flow.^37 i Except for the last stage turbine, none of the steam turbine stages are allowed to have condensation in the exit steam. The maximum condensation concentration for the last stage is taken from Kreutz et al.^42 j The performance of the physical CO2 capture system and CO2 compression to 150 bar are based on the 2-stage Selexol-based CO2 capture system design described by Black et al.^43 This design captures CO2 from pressurized syngas. k The performance of the chemical CO2 capture system and CO2 compression to 150 bar are based on the Econamine FG Plus-based CO2 capture system described by Black et al.^43 This design captures CO2 from the flue gas of a pulverized coal power plant. l Based on Meerman et al.^40
Thermodynamic properties	Peng–Robinson model
Biomass preparation and handling^a	Biomass handling electricity = 5.6 kW h per tar. Lock-hopper electricity = 2.0 kW h per tdry biomass. Heat requirement dryer = 322 MJ per tar biomass
Pyrolysis reactor^b	RYield block. T = 389 °C. Assumed Δp = 0.5 bar. Output based on experimental results
Cyclone^c	Sep block. Adiabatic. Δp = 0.01 bar. Assumed char removal rate = 100%
Hydrodeoxygenation reactor (HDO)^d	RYield block. Adiabatic reactor. Assumed Δp = 0.5 bar. Exit oxygen atom fraction in condensable organic phase is set to 1 wt%
Fractional condenser^e	Sep block. T = 35 °C. Separates the pyrolysis vapors into four components: an aqueous phase (L), a NCG (G), a light hydrocarbon (L) and a heavy hydrocarbon (L). Yields are empirically based
H2-Pressure swing absorber (H2-PSA)^f	Sep block. T = 35 °C. H2 stream Δp = 0.7 bar and raffinate stream p = 1.4 bar. Captures 85% of the H2. Assumed H2 stream purity is 100%
Steam reformer with heat supplied from combustion product gases^g	The steam reformer consists of a pre-reformer and a main reformer. Both reactors are RGibbs reactors and assumed to operate at chemical equilibrium. T = 500 °C for the pre-reformer and 950 °C for the main reformer. Δp = 3.5 bar across the combined reactors. H2O : carbon input ratio = 1.7
Water–gas shift	RGibbs reactor. T = 300 °C. Δp = 2 bar. Chemical equilibrium assumed
Combustor^h	RGibbs reactor. Complete combustion assumed. T combustion gases = 1400 °C
Heat exchangers	Heat loss = 2% of heat transferred, assumed TPinch = 8 °C
Steam turbines^i	IP at 60 bar and 500 °C; LP at 1.5 bar; condensate at 0.11 bar. Minimum 90% vapor fraction at LP exit. For both turbines: isentropic η = 0.83; mechanical η = 0.98
Physical CO2 capture^j	Sep block. CO2 capture fraction = 90%, electricity demand 42.1 kW h per t CO2 captured (excluding CO2 compression), syngas Δp = 0.5 bar
Chemical CO2 capture^k	Sep block. CO2 capture fraction = 90%, heat duty (steam required) = 3556 kJ per kg CO2 captured, electricity duty = 37.6 kW h per t CO2 captured (excluding CO2 compression), flue gas p = 1.0 bar
Pumps^l	Pump η = 0.925, driver η = 0.97
Compressors (multi-stage)^l	Used for NCG and H2 streams. Intercooling to 35 °C, ηpolytropic = 0.83, ηmechanical = 1.
CO2 compressor^j^,^k	Electricity consumption: 68 kW h per t CO2 for physical capture and 82 kW h per t CO2 for chemical capture

---

For all plant designs, we assume an input capacity for woody biomass residues of 3425 metric tonnes per day (dry basis), corresponding to an annual input of about 1 million tonnes, which we consider to be a practical limit for truck delivered biomass.²⁷ The assumed compositions of biomass and natural gas in our simulations are given in Appendix A of the ESI.†

3.1.1. Pyrolysis section. Woody biomass residues are assumed to arrive at the facility with a moisture content of 30 wt%. The biomass is sized to 10 mm chips‡ and dried on a belt dryer to 10 wt%. Heat for drying is provided by combustion flue gases entering the dryer from the utilities section at around 175 °C. A rotary drum dryer may be more efficient than a belt dryer and have lower capital and operating costs, but it would require higher temperature combustion gases (commonly above 230 °C) and as a result would likely entail additional pollution controls (and costs) to limit volatile organic compound and particulate matter emissions.^29,30

The dried chips are fed via a lock hopper system into the catalytic hydropyrolysis reactor operating at 22 bar. Feeding biomass against pressure presents potential technical problems like bridging, clogging, difficulty of pressure-sealing between reactor and lock hopper, spontaneous ignition, and excessive wear and tear of the pressurization equipment.³¹ However, lock hopper systems have proven operating experience and generally acceptable capital and operating costs, despite their need for compressed gas to operate.³² As the pyrolysis process uses H₂ in the pyrolysis reactor, it is assumed that the first steps of the lock hopper (purging of air and initial pressurization) use N₂, while the later steps (final pressurization) use H₂. This reduces the consumption of high pressure N₂ and lowers the amount of N₂ entering the pyrolysis reactor. Alternatively, if the hydropyrolysis facility is equipped with CO₂ capture and storage (CCS) the purging and/or pressurization of the biomass could also be done using captured CO₂.

The pyrolysis reactor is kept at 389 °C. The heat is provided by the exothermic reaction of the biomass with the H₂. Catalyst particles serve as bed material and H₂ acts as fluidizing agent. As noted earlier, conventional pyrolysis results in the production of significant amounts of unwanted molecules. The presence of H₂ suppresses the formation of unwanted molecules, and it is believed that the suppression effectiveness increases with H₂ partial pressure. This dictates operating the pyrolysis reactor at as high a pressure as possible. Also, a high H₂ flow is desired because as some of the H₂ reacts away the bulk H₂ partial pressure drops. However, a high H₂ input rate results in larger equipment and lower efficiency since more H₂ would be recycled. To estimate the appropriate H₂ input rate, we observed that H₂ is a reactant in the pyrolysis reactor, but it is also a coolant and fluidizing agent. The H₂ requirements for reaction, for cooling, and for fluidizing were estimated (see Appendix B of the ESI†) and resulted in an estimated input requirement of 200 kg H₂ per t_dry biomass for our simulation. This is the lowest flow rate that enables all three functions to be met. At this input rate 24% of the H₂ reacts, leading to a drop in H₂ partial pressure from 22 bar at input to 15 bar at the pyrolysis reactor exit.

The cyclone at the catalytic hydropyrolysis reactor exit removes solids to a sufficient level for the pyrolysis vapors to be fed to the catalytic fixed-bed HDO reactor. The solids are collected as char. Performing the HDO step directly after pyrolysis benefits from the high H₂ partial pressure and elevated temperature of the stream. After the HDO reactor, the condensable hydrocarbons (hydrocarbons containing 4 carbon atoms or more) are removed from the gases and separated into lighter and heavier fractions using fractional condensation. The design of the separation step is important as the final refining – onsite or offsite – depends on whether gasoline, kerosene or diesel is desired. In this study it is assumed that the heavy fraction is refined into a diesel substitute and the light fraction into a gasoline substitute. We assume the needed refining can take place at an existing petroleum refinery. This refining was not simulated, but instead was represented as part of the production cost model (see Section 3.4).

3.1.2. H₂-Section. The NCG stream leaving the fractional condensation step is mainly H₂ (90%_mol), with the rest being C₁–C₃ hydrocarbons, CO and CO₂. To improve the economics of the facility, the H₂ needs to be recycled back to the catalytic hydropyrolysis reactor. Also, additional H₂ is required to make up for the H₂ reacted in the catalytic hydropyrolysis and HDO reactors and any H₂ lost during the H₂ recovery process.

The NCG stream is fed into a H₂-pressure swing absorber (PSA-1) where 85% of the H₂ is separated. The H₂ is combined with the H₂ from PSA-2 (discussed later), recompressed, and fed into the catalytic hydropyrolysis reactor. The raffinate of PSA-1 is mainly CO and hydrocarbons and provides the feed gas for producing additional H₂ by steam reforming.

The SR consists of a pre-reformer followed by a main reformer, each simulated as an isothermal reactor. The pre-reformer converts C₂₊ compounds into CO, and subsequently into CH₄ (see eqn (1) and (2)). This reduces the risk of coke formation in the main reformer. The pre-reformer also protects the main reformer's (more expensive) catalyst from sulfur poisoning by sacrificing some pre-reformer Ni-based catalyst if necessary. The pre-reformer is simulated to operate at 500 °C. To increase overall efficiency the NCG is pre-heated to 500 °C by heat transfer from the reformate exiting the main reformer. Heating a hydrocarbon stream to 500 °C can result in coke formation if the fraction of higher hydrocarbons (C₂₊) is high.³³ In the modelled system the NCG stream entering the pre-reformer contains around 15%_mol C₂₊; at this concentration no coke formation is predicted in the process simulations. The required heat to keep the pre-reformer and the main reformer (discussed next) at the desired temperature is supplied by using the heat of the combustor exhaust gas (see Fig. 2). This is an uncommon design in the chemical process industry for steam reforming, but several vendors, including Air Products and Haldor Topsøe offer designs for commercial applications.³⁴

CO + 3H₂ ⇌ CH₄ + H₂O (2)

Fig. 2 Overview of modelled gas heated steam reformer system. The solid black line indicates a non-reformed stream, the short dotted line indicates a partially reformed stream and the large dotted line indicates a fully reformed stream.

After the pre-reformer the NCG is heated to 950 °C using the combustor exhaust gas and fed into the main reformer. Here most of the methane is converted into CO and H₂. The highly endothermic reaction (eqn (3)) dictates high temperatures in order to achieve high CO yields. The reformer operates at 950 °C and is kept at this temperature using the heat of the combustor gas.^35,36 At this temperature part of the CO is shifted to form H₂. However, the exothermic nature of the water–gas shift (WGS) reaction (eqn (4)) means that at this high temperature H₂ yield is low. Therefore, the reformate is fed into a low temperature WGS reactor. Although low temperatures increase H₂ yield, they also reduce reaction rates resulting in a larger WGS reactor. Due to this trade-off a temperature of 300 °C was selected for the WGS reactor. The H₂ is separated from the NCG gas at PSA-2. The PSA-2 raffinate is sent to the utilities section.

CH₄ + H₂O ⇌ CO + 3H₂, ΔH_r = 206 kJ mol⁻¹ (3)

CO + H₂O ⇌ CO₂ + H₂, ΔH_r = −41 kJ mol⁻¹ (4)

The H₂O added to the SR serves two main functions. First, it moves the equilibrium of eqn (3) to the right, increasing H₂ yield. It also limits coke formation in the SR and WGS reactors. Adding H₂O also lowers overall efficiency, mainly due to the increased amount of gases that must be heated, so H₂O addition should be kept to a minimum. The simulation model indicates that a minimum H₂O : carbon ratio§ of around 1.7 in the stream entering the SR reactor is needed to generate sufficient H₂. This ratio is high enough to avoid coke formation in the SR and WGS reactors (see note g in Table 2).³³ As this ratio minimizes efficiency losses and generates sufficient H₂, this ratio is used in the simulations.

In principle, the H₂ generation section could be redesigned with only a single PSA, which would reduce the number of process steps. Such a simplification can in some cases reduce costs. However, this option was rejected here because of the high H₂ content of the NCG stream coming into the H₂ generation section. If this H₂ is not removed prior to the steam reformer the amount of additional H₂ produced in the steam reformer and WGS reactor would be reduced as the equilibrium of both eqn (3) and (4) shifts to the left. Generating enough H₂ in that situation would require a larger H₂O addition to the NCG stream prior to the WGS reactor and/or a lower operating temperature of the WGS reactor. Both options result in larger equipment, lower efficiency and higher cost. The more economic option is to remove the H₂ prior to the steam reformer using a separate PSA, as in Fig. 1.

3.1.3. Utilities section. In the utilities section, the combustion product gas is used directly in the reformer. To avoid metallurgical limits, the combustion gas temperature is kept below 1400 °C by controlling the level of combustion air.³⁷ To limit the reformer size, and thereby cost, the combustion gas temperature inside the reformer was kept above 1000 °C. After the reformer the combustion gas is fed into a two pressure level heat recovery steam Rankine cycle (HRSC) for electricity production. After the HRSC the remaining heat in the combustion gases is used to dry the feedstock biomass, as described earlier. The HRSC system is designed to ensure that the temperature of the combustion gases leaving the drier is above 90 °C to avoid condensation in the stack.

3.1.4. CO₂ capture section. CO₂ is captured for underground geological storage in some of the plant designs. The only CO₂ emission point of the facility is the flue gas leaving the biomass dryer. At this point the gas stream is around atmospheric pressure and has a CO₂ concentration of around 10 mol%. This makes a CO₂ capture system based on chemical absorption the most suitable capture option. The performance of the chemical CO₂ capture system was modeled after commercial MEA-based technology.

A drawback of a chemical CO₂ capture system is the relatively high parasitic energy demand for solvent regeneration. A much lower energy penalty is incurred if physical solvents are used. However, CO₂ capture by physical absorption only works efficiently at high CO₂ partial pressures.³⁸ For that reason plant designs with physical CO₂ capture place the capture system between the WGS and PSA-2 blocks. The performance of the physical CO₂ capture system was modeled after commercial Selexol technology.

The required steam for regeneration of the CO₂ capture solvents in either case is extracted from the steam turbines. The captured CO₂ is compressed to 150 bar. Downstream transport, injection and storage of captured CO₂ were not physically modelled, but are represented in the economic model as a function of the CO₂ capture rate. It is assumed that the CO₂ is transported 100 km without need for boost compression and injected into an appropriate saline aquifer for permanent storage.

3.2. Process configurations

Six process designs were developed and analyzed (see Table 3). The base case design is given by the layout in Fig. 1 when neither CO₂ capture block is present. In this design all the char produced in the pyrolysis reactor is used as fuel in the combustor, along with the PSA-2 raffinate gas. This design is similar to that of Tan.^25,26 The other five cases all utilize some degree of carbon capture and storage.

Table 3 The investigated pyrolysis configurations

	Char is combusted	Char is sequestered
No CO2 capture		NG-Base
Physical CO2 capture
Chemical CO2 capture		NG-Chemical

---

The CCS cases include two variants of the base design. One of these is the physical case, which includes the physical CO₂ capture block between the WGS and PSA-2 blocks. This results in capture of only a portion of the facility-wide CO₂ emissions. The second variant is the chemical case incorporating chemical CO₂ capture after the biomass dryer block. This results in capture of the maximum amount of facility-wide CO₂ emissions.

The three remaining CCS cases are variants of the base, physical, and chemical cases and are identified as NG-base, NG-physical, and NG-chemical, respectively. In these latter 3 designs, the char produced by the pyrolysis reactor, which has a carbon intensity of 27 kg C per GJ_LHV char, is sequestered by landfilling it or by using it as a soil amendment, and less carbon-intensive natural gas (15 kg C per GJ_LHV NG) is used as fuel for the combustor. This results in lower GHG emissions for the NG-cases relative to their counterpart non-NG cases.

Because the NG-base and NG-chemical cases were found not to provide any new insights, only the red highlighted cases in Table 3 are reported in this paper. Results for the two cases not reported here can be found in Appendix E of the ESI.†

3.3. Lifecycle greenhouse gas emissions model

To determine the GHG performance of the biofuels the lifecycle (LCA) GHG emissions of the biofuels (expressed in kg CO_2,eq. per GJ_{biofuel LHV}) and the GHG emissions index (GHGI), a dimensionless value first introduced by Liu, et al.,³⁹ of the biofuel system are calculated. The LCA metric is used to compare biofuel emissions with those of equivalent petroleum-derived fuels. The GHGI metric is used to compare emissions of the biofuel system with those of a reference system producing an equivalent amount of vehicle fuel and electricity using conventional means (see eqn (5)). A reference year is required to estimate the GHG intensity of the displaced vehicle fuel and electricity. The year 2005 was selected as a benchmark because U.S. national carbon-mitigation goals (28% reduction by 2025 and more than an 80% reduction by 2050) are expressed relative to 2005 emissions.⁴⁴

The numerator includes all GHG emissions from the biofuel system. The denominator consists of the estimated lifecycle GHG emissions involved in producing and using gasoline, diesel and electricity in the US in 2005. A GHGI₂₀₀₅ of 1 means that the biofuel process has GHG emissions equal to those of the reference system. A GHGI₂₀₀₅ of 0 means that the biofuel system has no GHG emissions, i.e. any positive GHG emissions are compensated by removal of atmospheric CO₂ by photosynthesis.

Both metrics include emissions associated with the production and delivery of biomass and natural gas feedstocks, the emissions at the facility itself, and the emissions associated with the distribution and combustion of the biofuels. Emissions associated with transportation of the char to a landfill (in the NG-cases) are neglected. Since it is expected that only relatively modest offsite refining will be needed to produce vehicle-worthy fuels, emissions associated with offsite upgrading of the light and heavy fractions are also neglected. The uptake of atmospheric CO₂ for photosynthesis is considered a negative emission.

A difference between the LCA and GHGI metrics is how emissions are allocated to the by-product electricity. The LCA metric assumes that the by-product electricity displaces grid electricity, and the grid emissions thereby avoided are subtracted from the emissions of the biofuels. The average 2005 grid emissions are assumed for this by-product credit. No allocation is required in estimating the numerator of the GHGI. Instead the reference system emissions in the denominator includes the emissions associated with both the production of liquid fuels and electricity. The assumptions used for the lifecycle assessment are given in Table 4.

Table 4 GHG emissions assumptions

	Unit	Value
a The biomass has a carbon content of 0.497 kg C per kgdry (or 1.82 kg CO2,eq. per kgdry) and a heating value of 19.64 MJHHV per kgdry. b The emissions include those associated with processing of Pennsylvania forest residue chips and transportation for 150 km round-trip by truck.^45 c Upstream emissions for natural gas are 2.48 kg Ceq. per GJLHV,^46 and the LHV : HHV ratio of the used natural gas is 1.11. d The emissions to transport the gasoline (0.91 kg CO2,eq. per GJLHV) and diesel (0.80 kg CO2,eq. per GJLHV) from the conversion facility to fueling stations are from Gerdes et al.^47 The gasoline : diesel ratio is set equal to the ratio produced by the biofuel plants (2 : 1 ratioLHV). e Emissions due to losses during transport and injection of the pressurized CO2. No leakage during storage is assumed. Transport losses are set at 0.0081 kg CO2 lost per (t CO2 captured × km transported), assuming a transport distance of 100 km. Injection losses are assumed to be 5 kg CO2 per t CO2 delivered.^48 f This is the estimated lifecycle grid-average emissions in the US in 2005. Direct power plant emissions and the composition of feedstocks in the electricity mix are taken from the EPA.^49 Upstream emissions are taken from Gerdes et al.^47 for oil and from the Greet model^46 for NG and coal. g The well-to-wheels emissions of gasoline (91.4 kg CO2,eq. per GJLHV) and diesel (90.1 kg CO2,eq. per GJLHV) are taken from Gerdes et al.^47 The gasoline : diesel ratio is set equal to the ratio produced in the biofuel plants (2 : 1 ratioLHV).
Photosynthesis^a	kg CO2,eq. per GJHHV biomass	−92.7
Upstream biomass cultivation, collection and transport^b	kg CO2,eq. per GJHHV biomass	3.58
Upstream NG production, processing, and transport^c	kg CO2,eq. per GJHHV NG	8.2
Downstream fuel delivery to user^d	kg CO2,eq. per GJLHV fuel	0.87
CO2 transport and injection^e	kg CO2,eq. per t CO2	5.81
2005 reference system grid-electricity emissions^f	kg CO2,eq. per MW h	661
2005 reference system transport fuel emissions^g	kg CO2,eq. per GJLHV fuels	91.0

---

3.4. Economic model

To evaluate the economic performance of the different cases the levelized production costs of the biofuels and break-even crude oil prices (BEOP) were calculated using the methodologies described in detail in the online ESI material accompanying Hailey et al.⁵⁰ The BEOP is the crude oil price at which the production cost for the equivalent petroleum-derived fuels equals the production cost of the biomass-derived fuels. Assumed values for financial parameters are given in Table 5. All costs and prices are in constant 2014 U.S. dollars.

Table 5 Assumed values of financial parameters

Parameter		Abbrev.	Unit	Value
a The capital cost factor is a multiplier applied to the scoping-study estimate to estimate the cost of first-of-a-kind facilities. Its rationale is explained in the Capital costs section. b This study focusses on producing fuels with low or even negative GHG emissions. Effective carbon mitigation incentives are required to produce such fuels. It is assumed that these policies effectively value GHG emissions at 100 $ per t CO2,eq.. c This is the average U.S. natural gas price for industrial users in the Reference case of the Annual Energy Outlook 2016 for the period 2020–2040.^51 d This is the average U.S. wholesale electricity generation price in the Reference case of the Annual Energy Outlook 2016 for the period 2020–2040.^51 e It is assumed that the produced char has no economic value and needs to be disposed. For comparison, the average landfill tipping fee in the U.S. in 2013 is estimated at 55 $ per t.^52 f The refining cost is based on refining margins of 0.67 $ per gal gasoline and 0.32 $ per gal diesel. These are the levelized (using 7% discount rate) differences between the U.S. annual average refiner's acquisition cost of imported crude oil and the U.S. annual average wholesale price of gasoline projected for 2021–2040 in the Annual Energy Outlook 2016 Reference Scenario.^51 Using conversion factors of 747 kg m^−3 gasoline, 42.4 MJLHV per kg gasoline, 847 kg m^−3 diesel and 42.6 MJLHV per kg diesel results in refining costs of 5.59 $ per GJLHV gasoline and 2.34 $ per GJLHV diesel. Using the gasoline : diesel ratio as produced in the pyrolysis facilities modelled in this study results in a weighted-average refining cost of 4.51 $ per GJLHV fuel. g The annual O&M cost – excluding feedstock cost – is assumed to consist of 4% of TCI plus the cost for make-up catalyst. The latter, according to Tan,^25 is 52.89 lb h^−1 for a 2000 metric tdry per d biomass input design (equivalent to 0.288 kg catalyst per tdry biomass). Tan's catalyst price is 14 000 $2007 per short ton (equivalent to 16.9 $2014 per kg catalyst) this translates to a catalyst make-up cost of 4.9 $2014 per tdry biomass.
Capital cost factor^a		CCF	—	2.0
Construction time		CT	Year	3.0
Equity fraction		EF	%	45
Debt fraction		DF	%	55
Real cost of equity		RCE	—	0.102
Real cost of debt		RCD	—	0.044
State tax rate		Taxstate	%	6.5
Federal tax rate		Taxfederal	%	35
Property taxes & insurance		PTIrate	%	2.0
Economic lifetime of plant		N	Year	20
GHG valuation^b			$ per t CO2,eq.	100
Biomass price			$ per GJHHV	5.00
Natural gas price^c			$ per GJHHV	5.46
Electricity price^d			$ per MW h	68
Char disposal cost^e			$ per t	55
Refining cost^f			$ per GJLHV	4.51
O&M cost^g	Pyrolysis catalyst		$ per tdry biomass	4.87
	Other O&M cost		$ per year	4% of TCI
Capacity factor			%	80

---

3.4.1. Capital costs. The total capital investment (TCI) for each plant design was developed using factored estimation,^39,53 whereby the cost of each major component is estimated using reference costs for similar components from published sources and expert interviews. The reference costs correspond to specific reference scales and are adjusted to the required scale using eqn (6).where C is the cost of the component, S is the capacity (or size) of the component and f is a component-specific scaling exponent. The subscripts Model and Ref refer to the modelled component and the reference component, respectively. The required size of the components is provided by the Aspen Plus process simulation results.

Indirect costs (IC) and balance of plant costs (BOP), representing e.g., infrastructure, overhead and engineering, are added to the scaled component capital cost. The TCI is the summation of the individual scaled component costs, including BOP and IC (see eqn (7)). The reference scale, scaling exponent, capital cost and BOP and IC values for each component are given in Table 6.

TCI = ∑C_Model × (1 + IC_Model) × (1 + BOP_Model) (7)

Table 6 Reference parameter values for component capital cost estimation

Component	Scaling parameter	S Ref	f	C Ref (M$2014)^a	BOP^b	IC
a If the cost estimate from the literature is in a different year than 2014, the cost estimate is indexed to 2014 using the Chemical Engineering Plant Cost Index. b A variable BOP indicates that the BOP is assumed to be dependent on the scale of the facility, based on the following equation: BOP (%) = 88.67 × (plant scaleFeedstock MWHHV)^0.2096.^39 At the scale used in this study the BOP is 22.0% for the cases that combust char and 21.4% for the cases that sequester the char. c From the supplemental material document published online by Liu et al.^39 According to Liu et al.,^39 the maximum size for a preparation and feeding train is 110 t per h biomass (15 wt% moisture), or 2200 tdry per d biomass. Our design has a capacity of 3425 tdry per d biomass. To keep within the Liu et al.'s maximum train size, we use two identical pre-treatment trains, each processing half the total feedstock. d Based on Tan.^25,26 Consists of the catalytic hydropyrolysis reactor and the hydrodeoxygenation reactor. Tan uses a single train with a capacity of 2000 tdry per d biomass.^25,26 Our design has a capacity of 3425 tdry per d biomass. To keep within Tan's single-train capacity, our design is equipped with two identical pyrolysis and HDO-trains, each train with an operating capacity of 1713 tdry per d biomass. e Based on Tan.^25,26 Consists of an absorption tower, distillation tower, sour water stripper, amine stripper and ammonium sulfate oxidizer. f Based on the baseline SMR with CO2 capture case given by Rath et al.^41 g Based on Floudas et al.^54 The cost estimate by Floudas et al., is valid for a standard reformer where the heat for the endothermic reaction is provided by combustion of a gaseous fuel inside the reformer. In this study a modified steam reformer is used where the combustion of the fuel takes place outside the reformer and the heat is supplied by the hot combustion gas. It is assumed that the modified reformer is more expensive than a conventional reformer due to the lower temperature of the combustion gas. This requires more heat exchange surface area to supply the necessary heat. To compensate for this increase in surface area – and therefore cost – the heat duty of the reformer is added to the scale of the heat exchangers component. h From the supplemental material document published online by Liu et al.^39 i Based on the Shell gasifier with CO2 capture case (case 6) given by Black et al.^43 The referenced case uses a double stage Selexol system to separate the CO2 from the H2S. For the modelled systems a single stage Selexol system would be sufficient due to the low sulfur content of the biomass. This would reduce capital cost, but to what extent is unknown. Therefore, this discount has not been applied here. j Based on the subcritical PC with CO2 capture case (case 10) given by Black et al.^43
Biomass prep. and handling^c	Biomass feed, twet h^−1	64.6	0.77	13.44	Variable	32%
Pyrolysis + HDO reactor^d	Biomass feed, tdry d^−1	2000	0.67	19.40	Variable	35%
Product upgrading^e	Biomass feed, tdry d^−1	2000	0.67	10.49	Variable	35%
H2-PSA^f	Gas flow, 10^6 ft^3 h^−1	0.033	0.70	0.91	Variable	32%
Steam reformer^g	Reformate, kmol h^−1	31 733	0.60	62.45	15.5%	32%
Water–gas shift^h	Input gas, MWLHV	1188	0.67	19.39	15.5%	Incl.
Heat exchangers^h	HRSC + HX duty, MWth	355	0.70	56.14	Variable	27%
Steam turbines^h	ST gross output, MWe	136	0.67	57.45	15.5%	Incl.
Selexol system^i	CO2 captured, t h^−1	442	0.63	129.7	Variable	32%
Econamine plus system^j	CO2 captured, t h^−1	596	0.67	284.8	Variable	32%
Compressors^h	Compr. duty, MWe	10	0.67	6.91	Incl.	32%
CO2 compressor^h	Compr. duty, MWe	13	0.67	10.42	Incl.	32%

---

A TCI derived by factored estimation is often referred to as a scoping-study estimate. The difference between scoping-study estimates and the actual capital costs for construction a first-of-kind (FOAK) facility has been investigated by Merrow et al.,⁵⁵ and Greig et al.,⁵⁶ who found for the cases they investigated that the as-built costs were on average 1.5 and 2.1 times the scoping-study estimates, respectively. Since the pyrolysis systems described in this paper have not yet been built commercially, we multiply the scoping-study TCI by a capital cost factor of 2 to estimate the capital requirement for a FOAK plant. Repeated commercial deployments of the technology can be expected to lower the capital cost factor, though how far it might fall is unknown. The impact of different values of the capital cost factor on fuel production costs is analyzed later in this paper.

To estimate biofuel production costs, the TCI is converted into an annual capital charge using the methodology described in detail by Hailey et al.⁵⁰ The resulting annual capital charge rate is 16% per year of the TCI.

3.4.2. Operating and maintenance costs. Other components of the production cost are the O&M cost and, in the presence of a GHG valuation, the cost of emitting GHG. The O&M cost include salaries, maintenance and consumables. The yearly non-feedstock O&M cost is assumed to consist of 4% of TCI plus the cost of catalyst make-up for the hydropyrolysis reactor. For financial analyses that include a non-zero GHG emissions value, the LCA GHG emissions of the biofuels are used to calculate the cost of the GHG emissions. When the biofuel LCA GHG emissions are negative, they provide an added source of revenue for the facility. In calculating break-even crude oil prices when the GHG valuation is non-zero, the cost of petroleum-derived fuels includes GHG emissions charges based on the 2005 reference system transport fuel emissions given in Table 4.

4. Results

4.1. Energy efficiencies

Results of mass and energy balance simulations are given in Table 7 and carbon balances in Table 8. With a biomass feedstock input of 3425 t_dry per day, the liquid fuel production capacity of the modelled facilities is 7500 bbl per d. In all cases 65% of the biomass energy input (LHV) is converted to biofuels in a gasoline-to-diesel ratio of 2 : 1. This corresponds to a yield of 144 gal ethanol-equivalent per short ton_dry biomass.¶ Detailed process stream data for the physical case are given in Appendix D of the ESI.†

Table 7 Mass and energy balance selected cases

	Base	Physical	NG-Physical	Chemical
a The LHV is based on a moisture content of 30 wt%.
Input (MW)
Biomass, LHV^a (HHV)	687 (778)	687 (778)	687 (778)	687 (778)
Natural gas, LHV (HHV)			99 (110)
Output (MW LHV )
Light hydrocarbon liquids	297	297	297	297
Heavy hydrocarbon liquids	149	149	149	149
Total hydrocarbon liquids	446	446	446	446
Electricity	55	48	39	13
Char			131
Efficiency (LHV basis)
Liquid eff.	65%	65%	57%	65%
Liquid + electricity eff.	73%	72%	62%	67%
Biomass intensity (GJ biomass per GJ liquid)	1.5	1.5	1.5	1.5
Electricity production (MW)
Gross production	66	66	57	39
Parasitic load	11	18	18	26
Net export	55	48	39	13

---

Table 8 Carbon balance selected cases (t C per h)

	Base	Physical	NG-Physical	Chemical
Input
Biomass	71	71	71	71
Natural gas			5
Output
Liquid fuels	34	34	34	34
Char product			13
CO2 captured		17	17	34
Emitted	37	20	13	4

---

The facilities also produce excess electricity. The highest net electricity production is obtained in the base case. The parasitic load of the CO₂ capture and compression equipment reduces electricity exports from 55 MW_e in the base case to 48 and 13 MW_e in the physical and chemical cases, respectively.

The reduction in electricity export is larger for the chemical case than for the physical case because the chemical case captures almost twice as much CO₂ (63 vs. 123 t CO₂ per h), has a slightly higher specific parasitic electricity requirement per t CO₂ captured,|| and a much higher parasitic steam requirement that reduces the gross steam turbine power output.|| The difference in electricity export between the physical and the NG-physical case is due to the differences in the heat duty of the combustor. The hydropyrolysis process produces 12 t per h char. In the physical case all the char is combusted. However, combusting only 80% of the char would be sufficient to meet the heat demand of the facility. The additional 20% goes to increase steam production for electricity generation. In the NG-physical case just enough natural gas is added to the combustor to supply the required process heat, so there is less heat available for the steam cycle.

4.2. Lifecycle GHG emissions

Lifecycle GHG emissions per unit of liquid fuel are shown in Fig. 3, along with GHGI₂₀₀₅ values. The base case has a GHGI₂₀₀₅ of 0.06, i.e., GHG emissions for the system are only 6% of those for the 2005 reference system. Although the GHGI calculation assumes emissions originating from carbon in biomass are exactly offset by photosynthesis occurring during regrowth of an equivalent amount of biomass, the GHGI is not zero due to up and downstream emissions, e.g., during cultivation and transport of the biomass and distribution of the produced transportation fuels.

Fig. 3 Lifecycle (LCA) GHG emissions (left axis) and GHGI₂₀₀₅ values (above each bar) for the systems investigated.

All cases that include CO₂ capture have net negative GHG emissions, with GHGI₂₀₀₅ values ranging from −0.29 to −0.71. These cases effectively remove CO₂ from the atmosphere. Comparing GHGI values for the physical and NG-physical cases shows that replacing the char with NG as combustor fuel reduces net GHG emissions. This is a result of the lower carbon content per unit of energy of NG vs. char, combined with the sequestering of the carbon in the char away from the atmosphere.

The LCA emissions in Fig. 3 include an emissions credit for grid electricity displaced by the electricity exported from the plant. As noted earlier (Table 4), the assumed credit is the U.S. grid-average emissions in 2005 (661 kg CO_2,eq. per MW h). Even if zero credit were assumed for displaced electricity, however, the LCA GHG emissions for the biofuels would still be far below those for equivalent petroleum-derived gasoline and diesel (91 kg CO_2,eq. per GJ_LHV), and for the three cases using CCS the LCA emissions would remain negative.

4.3. Economics

The scoping-study level of estimated capital costs are given in Fig. 4. The capital costs estimated for FOAK plants and used to estimate fuel production costs are the values in Fig. 4 multiplied by the capital cost factor of 2. For the base case the pyrolysis section (pre-treatment, pyrolysis + HDO and product upgrading) accounts for 47% of the scoping-study estimate, the H₂ generation section for 33%, and the utility section for 20%. Adding CO₂ capture increases the capital cost estimate by 24% in the physical case and by 58% in the chemical case. The larger increase in the chemical case is due to the more expensive capture system as well as the higher quantity of CO₂ captured. Replacing char with natural gas (NG-physical case) slightly reduces capital costs as the heat input to the utility section is decreased, resulting in a slightly smaller combustor and steam turbines.

Fig. 4 Breakdown of the scoping-study estimate for the capital cost (excludes capital cost factor). HRSG + ST = heat recovery steam generators + steam turbines. Product upgrading includes distillation columns, scrubbers and other bio-oil purification equipment.

The levelized total production cost of the finished fuels from FOAK plants are given in Fig. 5, both in terms of $ per GJ_LHV and an equivalent $ per bbl break-even crude oil price. Because CO₂ capture would not be considered absent an incentive, this figure assumes a GHG valuation of 100 $ per t CO_2,eq.. The levelized cost in the base case and the physical cases are within 4 $ per bbl. The higher capital cost and lower electricity revenues in the physical case are roughly offset by the difference in GHG emissions credits. Replacing the char with natural gas increases the production cost by about 6 $ per bbl (physical vs. NG-physical cases), as the increase in GHG credits are insufficient to compensate for the added char disposal cost, NG purchase costs and reduced electricity credits. The levelized cost in the chemical case is still higher: although revenues resulting from negative GHG emissions are the highest of any of the cases, they are insufficient to compensate for the much higher capital cost and for the reduced electricity revenues that arise from the larger energy penalty for CO₂ capture.

Fig. 5 Levelized liquid fuel production costs with input assumptions as in Table 5. The corresponding break-even crude oil price is shown at the top of each bar.

4.3.1. Literature comparison. A catalytic hydropyrolysis facility design similar to the base case design has been investigated by Tan,^25,26 and a comparison of these two systems is informative. Tan's process configuration differs from the base case most significantly in that it only produces enough electricity to satisfy internal demand. Instead Tan exports steam. Tan's economics assume the same level of on-site refining as in the base case, but no offsite refining. Tan concluded that wood would be converted into gasoline plus diesel at a rate of 87 gal fuel per t_dry wood. In our base case the conversion efficiency is similar at 90 gal per t_dry wood.

The TCI in Tan's study is 255 M$₂₀₁₄ for an “N^th plant”, compared to our estimate of 612 M$ for a FOAK base case plant. At first glance, these cost estimates seem at odds with each other. However, if the capital cost factor of 2 embedded in our base case estimate is removed, which may or may not lead to a good N^th plant cost estimate, and we further revise our estimate to be for a plant with a capacity equal to that in Tan's study, the resulting capital cost estimate for our design using our cost estimating methodology is 199 M$. The base case contains steam turbines which are absent in Tan's estimate. Removing the cost of the steam turbines further reduces our capital cost estimate to 177 M$ (excluding any contingencies). In Tan's estimate, 37% of his TCI is a contingency (either process or project). Removing these contingencies reduces Tan's capital cost estimate to 160 M$. Thus Tan's methodology and ours give capital cost estimates (excluding contingencies and capital cost factors) within 10% of each other for plants with similar design.

4.3.2. Sensitivity analysis.
Impact of GHG valuation. The plant designs involving CCS would only be considered in lieu of ones without CCS if the valuation of GHG emissions is sufficiently high to offset the increased capital cost and reduced electricity revenues when CCS is included. Fig. 6 shows break-even crude oil prices as a function of GHG valuation. With zero GHG valuation, the base case would have a BEOP of 95 $ per bbl, while BEOP for the physical and NG-physical cases would be 117 and 129 $ per bbl. The chemical case has a BEOP approaching 150 $ per bbl.

Fig. 6 Effect of GHG valuation on break-even oil price of selected cases.

Because all cases have a carbon footprint which is lower than that of petroleum-derived fuels, the BEOP drops with increasing GHG valuation in each case. The lower the emissions for a particular case, the sharper the decline in BEOP. The base case is the most economically attractive option up until a GHG valuation of 120 $ per t CO_2,eq., when it is essentially at parity with both the physical and NG-physical cases. The BEOP for these cases at this point is 28 $ per bbl. The chemical case becomes the most attractive case only when the GHG valuation exceeds 300 $ per t CO_2,eq..

Valuation of char. For the NG-cases, a baseline assumption was that the char would be landfilled at a cost of 55 $ per t (Table 5). However, if the char were shown to be an effective soil amendment, i.e., “biochar” (see Appendix F of the ESI†), its market price might be as high as 200 $ per t (ref. 57–59) while it still also acts to sequester carbon from the atmosphere. It is unknown if the char would have “biochar” properties, but if it could be sold for 200 $ per t, the BEOP for the NG-physical case would drop from 48 $ per bbl (for the reference assumptions in Table 5) to 30 $ per bbl, making it more cost competitive than the physical case, where the char is used for fuel rather than for added revenue. Fig. 7 shows how the competitiveness of the NG-physical case would change relative to that of the physical case for different char prices, NG prices and GHG valuations. The higher the GHG emissions valuation, the lower the char price required for the NG-physical case to produce transport fuels at lower cost than the physical case. This is because either a higher char selling price or a higher GHG valuation improves the economics of the NG-physical case relative to the physical case.

Fig. 7 Effect of char, NG and GHG valuations on the relative competitiveness of the physical and NG-physical cases. The red dot indicates the reference conditions (−55 $ per t char and 100 $ per t CO_2,eq.). For combinations of char price and GHG valuations that fall above a given line the levelized fuel production cost for the NG-physical case is lower than for the physical case.

Impact of bio-oil yield. The base case has a biofuel yield of 0.27 kg biofuel per kg_dry biomass. This yield is based on experiments using mixed wood as feedstock, as described in Appendix C of the ESI.† Other experiments have shown that different feedstocks give different biofuel yields.^13,15 To assess what impact yield differences might have on overall economics, two additional cases – reduced yield, producing 0.20 kg biofuel per kg_dry biomass, and increased yield, producing 0.30 kg biofuel per kg_dry biomass – were developed. See Appendix G in the ESI† on why this range was selected and how the simulation model was adjusted.

A lower biofuel yield results in increased production of NCG in the pyrolysis reactor. This leads to larger H₂ generation and utilities sections and a smaller upgrading area, with a resulting net increase in TCI for the reduced yield case of 7% compared to the base case. However, revenues for the additional electricity produced significantly reduce the negative effects of lower biofuel production, making the net production cost only slightly higher than for the base case. There is similarly only a small difference in net production cost between the base and increased yield cases (see Fig. 8).

Fig. 8 Effect different transport fuel yield on levelized production cost.

Impact of main financial parameters. Additional sensitivity analyses were performed to investigate the effect of different financial assumptions on the levelized production cost (Fig. 9). For illustration, the physical case was selected.

Fig. 9 Impact of several key financial parameter assumptions on the production cost for the physical case. 100% on the x-axis corresponds to the baseline financial parameter valuations in Table 5. Break-even oil prices are displayed by the dotted lines.

The two most influential parameters are the capacity factor and the capital cost factor. Knowing these values a priori is difficult, especially for a FOAK facility. However, even in what might be considered a pessimistic scenario (capacity factor of 70% and capital cost factor of 3, corresponding to 87.5% and 150%, respectively, of the reference assumption in Fig. 9), the BEOP would still be under 100 $ per bbl with other assumptions as in Table 5. Note that this combined effect is not shown in Fig. 9.

5. Discussion and conclusions

In this study the performance and economics of six prospective catalytic hydropyrolysis plant configurations for producing biomass-based transportation fuels have been analyzed. Three of these designs have integrated CO₂ capture and storage (CCS) into the system design. Several important conclusions can be drawn from the analysis.

First, based on what might be considered a conservative capital cost estimate, a first-of-a-kind (FOAK) facility processing 3425 t_dry per d woody biomass without CCS (base case) would produce transportation fuels with significantly reduced lifecycle GHG emissions compared to conventional transportation fuels. This facility would be cost competitive under the assumed baseline assumptions (including a GHG valuation of 100 $ per t CO_2,eq.) with petroleum-derived fuels when the crude oil price is less than 40 $ per bbl. Without a GHG valuation the break-even crude oil price for this design is less than 100 $ per bbl.

Second, adding CO₂ capture enables the production of liquid fuels with strongly negative lifecycle GHG emissions. Capturing CO₂ from a gas stream containing a high partial CO₂ pressure enables capture by physical absorption (physical case), which results in only a modest increase in total production cost. At a GHG valuation of 120 $ per t CO_2,eq. the physical case and the base case exhibit comparable economics. However, the addition of physical CO₂ capture increases process complexity and could potentially reduce plant availability, particularly for a FOAK plant. To justify the risk of adding physical CO₂ capture may require a more significant reduction in net production cost relative to a design without CCS, e.g., when the GHG valuation is much higher than 120 $ per t CO_2,eq..

Third, the design that maximizes CO₂ capture (chemical case) involves considerably more capital cost and energy penalty associated with CO₂ capture. This leads to liquid fuels with strongly negative emissions, but less favorable economics than any of the other cases unless the valuation of GHG exceeds 300 $ per t CO_2,eq..

Fourth, if the by-product char proves able to be used as “biochar” (soil enhancer) both the economics and the GHG mitigation potential of these biofuel systems would improve, but it is currently unknown whether the char can work as a “biochar”.

Results of sensitivity analyses suggest that the above conclusions are robust. However, it is important to note some key uncertainties that cannot be resolved without further research, development, and demonstration. Most importantly, catalytic hydropyrolysis is pre-commercial technology and the assumed technology integrations in our process simulations are untested at any scale, and so some assumptions were necessary. For example, we assumed that a moisture content of 10 wt% would not change the bio-oil yield from those reported for tests with moisture contents up to 5.6 wt%, on which our model is based. We also assumed that a biomass particle size of 10 mm allows optimal performance in the pyrolysis reactors. To date, successful lab-scale tests have been completed with particle sizes up to 5 mm. This smaller size could be used in commercial plants, but biomass processing complexity and parasitic electricity consumption would be considerably increased. Another important technology uncertainty is the effectiveness of a char-fired external combustor to provide heat for steam reforming. Steam reforming with external combustors are offered by some commercial vendors, but operating experience with these steam reformers appears to be very limited. A final, important, uncertainty is the assumed quality of the liquid product. It remains to be demonstrated whether refining to finished fuels will be viable at existing petroleum refineries, as has been assumed in this study (see Appendix H of the ESI† for comparisons of the properties of catalytic hydropyrolysis-derived and conventional fuels). The relatively attractive technical, environmental, and economic performance results found here for low-GHG liquid fuels production via catalytic hydropyrolysis suggests that further RD&D to help resolve these and other uncertainties is warranted.

Acknowledgements

The authors thank the Global Climate and Energy Project at Stanford University and the Carbon Mitigation Initiative at Princeton University for financial support and Michael Desmond and Martin Høj for helpful discussions. The authors also thank two anonymous reviewers for their, which improved the quality of the originally-submitted manuscript.

References

1. IPCC, Climate Change 2014, Synthesis Report. Contribution of Working Groups I, II and III to the Fifth Assessment Report of the Intergovernmental Panel on Climate Change, ed. Core Writing Team, R. K. Pachauri and L. A. Meyer, IPCC, Geneva, Switzerland, 2014, p. 151.
2. R. Brown, Thermochemical processing of biomass, Conversion into Fuels, Chemicals and Power, John Wiley & Sons, West Sussex, 2011.
3. A. Bridgwater, Review of fast pyrolysis of biomass and product upgrading, Biomass Bioenergy, 2012, 38, 68–94.
4. D. Radlein and A. Quignard, A Short Historical Review of Fast Pyrolysis of Biomass, Oil Gas Sci. Technol., 2013, 68, 765–778.
5. Personal communication with Mike Desmond. Senior scientist BP, contacted on 2014-06-10.
6. M. Bertero, G. de la Puente and U. Sedran, Fuels from bio-oils: Bio-oil production from different residual sources, characterization and thermal conditioning, Fuel, 2012, 95, 263–271.
7. T. Marker, M. Linck and L. Felix, Integrated Hydropyrolysis and Hydroconversion (IH²) Process for Direct Production of Gasoline and Diesel Fuel from Biomass, Biomass 2010 Conference, Gas Technology Institute, March, 2010.
8. M. Talmadge, R. Baldwin, M. Biddy, R. McCormick, G. Beckham, G. Ferguson, S. Czernik, K. Magrini-Bair, T. Foust, P. Metelski, C. Hetrick and M. Nimlos, A perspective on oxygenated species in the refinery integration of pyrolysis oil, Green Chem., 2014, 16, 407–453.
9. L. Rosendahl, Biomass Combustion Science, Technology and Engineering, Woodhead Publishing, 2013, ch. 7, p. 320, ISBN: 978-0-85709-131-4.
10. C. Liu, H. Wang, A. Karim, J. Sun and Y. Wang, Catalytic fast pyrolysis of lignocellulosic biomass, Chem. Soc. Rev., 2014, 43, 7594–7623.
11. N. Thegarid, G. Fogassy, Y. Schuurman, C. Mirodatos, S. Stefanidis, E. Iliopoulou, K. Kalogiannis and A. Lappas, Second-generation biofuels by co-processing catalytic pyrolysis oil in FCC units, Appl. Catal., B, 2014, 145, 161–166.
12. T. Dickerson and J. Soria, Catalytic Fast Pyrolysis: A Review, Energies, 2013, 6, 514–538.
13. T. Marker, L. Felix, M. Linck and M. Roberts, Integrated Hydropyrolysis and Hydroconversion (IH²) Process for Direct Production of Gasoline and Diesel Fuel from Biomass, Part 1: Proof of Principle testing, Environ. Prog. Sustainable Energy, 2012, 31, 191–199.
14. F. Melligan, M. Hayes, W. Kwapinski and J. Leahy, Hydro-Pyrolysis of Biomass and Online Catalytic Vapor Upgrading with Ni-ZSM-5 and Ni-MCM-41, Energy Fuels, 2012, 26, 6080–6090.
15. T. Marker, M. Roberts, M. Linck, L. Felix, P. Ortiz-Toral, J. Wangerow, E. Tan, J. Gephart and D. Shonnard, Biomass to Gasoline and Diesel Using Integrated Hydropyrolysis and Hydroconversion, Gas Technology Institute, 2013,  DOI:10.2172/1059031.
16. O. Mante and F. Agblevor, Catalytic pyrolysis for the production of refinery-ready biocrude oils from six different biomass sources, Green Chem., 2014, 16, 3364–3377.
17. D. Dayton, Catalytic Biomass Pyrolysis Technology Development for Advanced Biofuels, Research Triangle Institute, TCBiomass 2013 conference, Chicago, Il, 3–6 Sep, 2013.
18. Personal communication with Robert Brown. Director Center for Sustainable Environmental Technologies at IOWA State University, contacted on 2014-06-06.
19. Personal communication with David Dayton. Director Biofuels at RTI, contacted on 2014-05-30.
20. S. Thangalazhy-Gopakumar, S. Adhikari and R. Gupta, Catalytic Pyrolysis of Biomass over H+ZSM 5 under Hydrogen Pressure, Energy Fuels, 2013, 26, 5300–5306.
21. D. Dayton, J. Carpenter, J. Farmer, B. Turk and R. Gupta, Biomass Hydropyrolysis in a Pressurized Fluidized Bed Reactor, Energy Fuels, 2013, 27, 3778–3785.
22. D. Dayton, J. Hlebak, J. Carpenter, K. Wang, O. Mante and J. Peters, Biomass Hydropyrolysis in a Fluidized Bed Reactor, Energy Fuels, 2016, 30, 4879–4889.
23. R. Agrawal and N. Singh, Synergistic Routes to Liquid Fuel for a Petroleum-Deprived Future, AIChE J., 2009, 55, 1898–1905.
24. J. Lane, IH2 Deep-Dive: Breakthrough Biofuel Technology has Commercial-Scale in Sight, Renewable Energy World, www.renewableenergyworld.com/articles/2016/01/ih2-deep-dive-breakthrough-biofuel-technology-has-commercial-scale-in-sight.html, last accessed: 03-1-2017.
25. E. Tan, Techno-Economic Analysis of the Integrated Hydropyrolysis and Hydroconversion Process for the Production of Gasoline and Diesel Fuels from Biomass, National Renewable Energy Laboratory, 2011.
26. E. Tan, T. Marker and M. Roberts, Direct Production of Gasoline and Diesel Fuels form Biomass via Integrated Hydropyrolysis and Hydroconversion Process – A Techno-economic Analysis, Environ. Prog. Sustainable Energy, 2014, 33, 609–617.
27. E. Larson, G. Fiorese, G. Liu, R. Williams, T. Kreutz and S. Consonni, Co-production of decarbonized synfuels and electricity from coal + biomass with CO₂ capture and storage: an Illinois case study, Energy Environ. Sci., 2010, 3, 28–42.
28. T. Marker, M. Roberts, J. Wangerow, P. Ortiz-Toral, M. Linck and D. Swanson, Biomass to Gasoline and Diesel using IH²® Key Pilot Plant Tests on the Road to Commercialization, Biomass 2010 Conference, Gas Technology Institute, Arlington, VA, 30–31 March, 2010.
29. M. Worley, Biomass Drying Technology Update, BioPro Expo & Marketplace, Atlanta, GA, 14–16 March, 2011.
30. W. Amos, Report on Biomass Drying Technology, National Renewable Energy Laboratory, NREL/TP-570-25885, 1998.
31. A. Rautalin and C. Wilén, Feeding biomass into pressure and related safety engineering, Technical Research Centre of Finland, 1992.
32. J. Dai, H. Cui and J. Grace, Biomass feeding for thermochemical reactors, Prog. Energy Combust. Sci., 2012, 38, 716–736.
33. Personal communication Martin Skov Skjøth-Rasmussen. Vice-president R&D Haldor Topsøe on steam reforming on 2015-08-25.
34. M. Wesenberg, Gas Heated Steam Reformer Modelling, PhD thesis, Norwegian University of Science and Technology, 2006.
35. J. Rostrup-Nielsen and T. Rostrup-Nielsen, Large-scale hydrogen production, CATTECH, 2002, 6, 150–159.
36. G. Hawkings, Steam Reforming – Practical Operation, GBH Enterprises, LTD, www.slideshare.net/GerardBHawkins/steam-reforming-practical-operations, last accessed: 22-9-2016.
37. G. Hawkings, Steam Reforming – Types of Reformer Design, GBH Enterprises, LTD, www.slideshare.net/GerardBHawkins/steam-reforming-types-of-reformer-design, last accessed: 22-9-2016.
38. B. Metz, O. Davidson, H. de Coninck, M. Loos and L. Meyer, Carbon Dioxide Capture and Storage, IPCC, Cambridge University Press, 2005.
39. G. Liu, E. Larson, R. Williams, T. Kreutz and X. Guo, Making Fischer-Tropsch Fuels and Electricity from Coal and Biomass: Performance and Cost Analysis, Energy Fuels, 2011, 25, 415–437.
40. J. Meerman, A. Ramírez, W. Turkenburg and A. Faaij, Performance of simulated flexible integrated gasification polygeneration facilities, Part A: A technical-energetic assessment, Renewable Sustainable Energy Rev., 2012, 15, 2563–2587.
41. L. Rath, V. Chou and N. Kuehn, Assessment of Hydrogen Production with CO₂ Capture – Volume 1: Baseline State-of-the-Art Plants, National Energy Technology Laboratory, DOE/NETL-2010/1434, August 2010.
42. R. Kreutz, E. Larson, G. Liu and R. Williams, Fischer-Tropsch Fuels from Coal and Biomass, 25th Annual International Pittsburgh Coal Conference, Pittsburgh, PA, 29 Sep–2 Oct, 2008.
43. J. Black, J. Haslbeck, N. Kuehn, E. Lewis, L. Pinkerton, J. Simpson, W. Turner, E. Varghese and M. Woods, Cost and Performance Baseline for Fossil Energy Plants – Volume 1: Bituminous coal and natural gas to electricity, Revision 2a, National Energy Technology Laboratory, DOE/NETL-2010/1397, September 2013.
44. Office of the Press Secretary, Fact Sheet: U.S. Reports its 2025 Emissions Target to the UNFCCC, The White House, Washington, DC, March 2015.
45. D. Ortiz, A. Curtright, C. Samaras, A. Litovitz and N. Burger, Near-Term Opportunities for Integrating Biomass into the U.S. Electricity Supply: Technical Considerations, RAND Corporation, Santa Monica, CA, 2011.
46. GREET Model V1.8, Argonne National Laboratory.
47. K. Gerdes and T. Skone, Petroleum-Based Fuels Life Cycle Greenhouse Gas Analysis 2005 Baseline Model, National Energy Technology Laboratory, 2009.
48. Life Cycle Greenhouse Gas Analysis of Advanced Jet Propulsion Fuels, Fischer-Tropsch Based SPK-1 Case Study, Air Force Research Laboratory, 2011.
49. eGRID2007 Version 1.1 Year 2005 Summary Tables, U.S. Environmental Protection Agency, 2008.
50. A. Hailey, J. Meerman, E. Larson and Y. Loo, Low-carbon “drop-in replacement” transportation fuels from non-food biomass and natural gas, Appl. Energy, 2016, 183, 1722–1730.
51. J. Conti, P. Holtberg, J. Diefenderfer, A. Larose, J. Turnure and L. Westfall, Annual Energy Outlook 2016-with projections to 2040, U.S. Energy Information Administration, DOE/EIA-0383(2016), August 2016.
52. Clean Energy Projects, http://www.cleanenergyprojects.com/Landfill-Tipping-Fees-in-USA-2013.html, last accessed: 21-9-2016.
53. J. Meerman, A. Ramírez, W. Turkenburg and A. Faaij, Performance of simulated flexible integrated gasification polygeneration facilities, Part B: Economic evaluation, Renewable Sustainable Energy Rev., 2012, 16, 6083–6102.
54. C. Floudas, J. Elia and R. Baliban, Hybrid and single feedstock energy processes for liquid transportation fuels: A critical review, Comput. Chem. Eng., 2012, 41, 24–51.
55. E. Merrow, K. Phillips and C. Myers, Understanding cost growth and performance shortfalls in pioneer process plants, Rand Corporation, Santa Monica, CA, 1981.
56. C. Greig, A. Garnett, J. Oesch and S. Smart, Guidelines for Scoping and Estimating Early Mover CCS Projects, Milestone 5, Final Report, University of Queensland, Brisbane, Australia, 2014.
57. TR Miles Technical Consultants, The Economics of Biochar Production, PNW Biochar Group Meeting, Richland, WA, USA, May 2009.
58. J. Levine, C. Steinder, H. McLaughlin, A. Harley, G. Flora, R. Larson and A. Reed, Assessment of Biochar's Benefits for the United States of America, 2010.
59. M. Ramage, G. Tilman, D. Gray, E. Hiller, W. Ho, D. Karlen, J. Katzer, M. Ladisch, J. Miranowski, M. Oppenheimer, R. Probstein, H. Schobert, C. Sommerville, G. Stephanopoulos and J. Sweeney, Liquid Transportation Fuels from Coal and Biomass: Technological Status, Costs, and Environmental Impacts, National Academy of Sciences, 2009.

---

Footnotes

† Electronic supplementary information (ESI) available. See DOI: 10.1039/c7se00013h

‡ Smaller particle size increases liquid yields and decreases reactor size but the complexity and energy consumption of the sizing equipment also increase. Particle sizes of up to 5 mm have successfully been tested.²⁸

§ This is the molar ratio between H₂O and carbon, regardless of the parent molecule (CO, CO₂, C_xH_y).

¶ For comparison, the yields for cellulosic ethanol and gasification–Fischer–Tropsch are around 80 and 90 gal ethanol-equivalent per short ton_dry biomass, respectively.⁵⁹

|| The required electricity for the CO₂ capture unit is 42 and 38 kW h per t CO₂ in the physical and chemical cases, respectively, and CO₂ compression requires 68 and 82 kW h per t CO₂. Additionally, steam extraction lowers the output of the steam turbine generator by 3 and 218 kW h per t CO₂, respectively.

---