Fuel 196 (2017) 298–305

Contents lists available at ScienceDirect

Fuel
journal homepage: www.elsevier.com/locate/fuel

Full Length Article

Process modeling of hydrodeoxygenation to produce renewable jet fuel

and other hydrocarbon fuels
Pei Lin Chu a, Caroline Vanderghem a, Heather L. MacLean a,b,c, Bradley A. Saville a,⇑
a
Department of Chemical Engineering and Applied Chemistry, University of Toronto, Toronto, Ontario M5S 3E5, Canada
b
Department of Civil Engineering, University of Toronto, Toronto, Ontario M5S 1A4, Canada
c
School of Public Policy and Governance, University of Toronto, Toronto, Ontario M5S 3K9, Canada

h i g h l i g h t s

Composition of feedstock oils affects product distribution and hydrogen demand.

Feedstock affects thermal and electrical energy requirements.

Oil content of the oilseed is a key parameter.

Strategic heat integration reduces thermal energy demand.

a r t i c l e i n f o a b s t r a c t

Article history: The focus of this work is to model the hydrodeoxygenation process to produce renewable jet fuel and co-
Received 7 August 2016 products from low-input oilseeds, specifically camelina, carinata (non-edible oil) and used cooking oil
Received in revised form 27 January 2017 (UCO), taking into account the fatty acid compositions by incorporating the stoichiometric hydrodeoxy-
Accepted 28 January 2017
genation reactions. This methodology provides insight into the effect of feedstock composition and
Available online 14 February 2017
hydrodeoxygenation reactions upon product yields, product distribution, hydrogen consumption and
process utilities. The resulting product slates, hydrogen gas and utility demands are specific to each of
Keywords:
the camelina, carinata and UCO feedstocks.
Synthetic paraffinic kerosene (SPK)
Hydroprocessed renewable jet (HRJ)
The yield of kerosene-range alkanes ranged from 53 to 54% of the incoming oil, with 13–14% diesel
Biojet fuel range alkanes, 13–15% naphtha, and 7–9% liquefied petroleum gas, depending upon the fatty acid com-
Camelina position. The hydrogen consumption rate ranged from 26 to 30 kg per tonne of incoming oil, depending
Carinata upon the degree of bond unsaturation. Thermal energy demand is 2.8 GJ/tonne oil when processing used
Used cooking oil cooking oil, versus 5.2 and 5.7 GJ/tonne of oil for carinata and camelina, respectively, owing to the addi-
tional energy required for oil extraction. Electricity demand was 73 kWh/tonne oil for UCO, versus 170
and 227 kWh/tonne oil for carinata and camelina.
Ó 2017 Elsevier Ltd. All rights reserved.

1. Introduction to lessen the environmental impact of air travel by improving fuel

efficiency by 1.5% annually until 2020; achieve carbon neutral
The aviation industry relies on petroleum-based fuels, but growth beyond 2020; and halve overall aviation emissions from
potential adverse environmental impacts from these fuels have 2005 levels by 2050 [1–3]. The EU Emission Trading
been identified. The combustion of fossil fuels is of particular Scheme (ETS) based on a ‘‘cap and trade” system also aims to
concern due to the impact of greenhouse gas (GHG) emissions on reduce emissions [4]. As of 2012, airlines operating or providing
climate change. Development of low GHG intensity fuels is an services within the European Economic Area (EEA) are subject to
emerging frontier to address growing climate change concerns the ETS, although the aviation industry was temporarily exempted
in the aviation industry. from the ETS while the EU negotiates an agreement with the Inter-
The International Air Transport Association (IATA) has national Civil Aviation Organization (ICAO). The ETS is currently
responded by creating an emission reduction roadmap, which aims only applicable to flights within the EEA; a global market-based
mechanism is being developed by ICAO to cover international
aviation emissions, to be implemented by 2020 [5].
⇑ Corresponding author.
E-mail address: [email protected] (B.A. Saville).

http://dx.doi.org/10.1016/j.fuel.2017.01.097
0016-2361/Ó 2017 Elsevier Ltd. All rights reserved.
 P.L. Chu et al. / Fuel 196 (2017) 298–305 299

Biojet produced from the hydrodeoxygenation (HDO) process is the conversion of fatty acids produces only alkanes, carbon dioxide
currently one of the biofuels used in demonstration and commer- and water [9–13]. These assumptions do not account for the differ-
cial flights. It is also a technology that has been commercialized to ent degree of saturation of fatty acids, which would affect the H2
produce renewable diesel. Due to its technological maturity, it is needed for the hydrogenation reaction that converts TAG into fatty
expected that this pathway will play a major role in producing acids and propane. Stratton’s LCA work to screen biojet production
low carbon intensity jet fuels. The HDO process to produce pathways examined the cracking reaction based solely on n-
hydroprocessed renewable jet (HRJ) from triacylglycerides (TAG) octadecane, a representative C18 molecule, which may not accu-
includes hydrogenation, deoxygenation, isomerization and hydroc- rately represent the co-production of naphtha or other shorter
racking stages. Hydrogenation is the first reaction, saturating all alkanes when a diverse alkane pool is present [9]. Subsequent pub-
TAG bonds. Deoxygenation follows hydrogenation, and consists lications using data from Stratton [9] have incorporated this sim-
of three parallel reactions, including decarboxylation, decarbonyla- plified view of the conversion process. While helpful for an initial
tion and hydrodeoxygenation, as illustrated in Fig. 1 (adapted from evaluation, these simplifying assumptions do not reflect actual
Veriansyah et al. [6]). The alkanes produced are referred to as syn- conversion reactions. This was later addressed by Pearlson
thetic paraffinic kerosene (SPK) or HRJ, and are commonly known [14,15]; using product yields from the literature for diesel products
as hydroprocessed esters and fatty acids (HEFA) SPK. HDO fuel from soybean oil and jet fuel yields from jatropha oil, he calculated
products are molecularly similar to their petroleum counterparts, product yields including co-products such as propane, LPG and
with the notable difference being a lack of aromatic content. The naphtha. However, Pearlson [14] has generalized the yield and
TAG can be from diverse virgin and used oil feedstocks. Based on conversion based upon soybean oil. Han et al. [13] pointed out
the fatty acid composition of the feedstock, the product yields, pro- the need to tailor the hydrogen (H2) requirements based on the
duct distribution, hydrogen consumption and process utilities of oil feedstock composition, but assumed the complete conversion
the HDO process might vary, perhaps considerably. This issue, of diesel into biojet [13], which is unlikely based upon typical
however, has not previously been investigated, and a key objective cracking reactions and process technology.
of this work is to rigorously evaluate the impact of fatty acid com- The aim of this work is to model the HDO process from Camelina
position upon these key operating parameters. sativa (camelina), Brassica carinata (carinata), and used cooking oil
UOP is a leading technology provider for the HDO process and (UCO), quantifying the H2 demand, energy, electricity require-
has partnered with several biofuel producers, such as Altair and ments and product slate from each set of triglycerides. Due to
Ensyn, to convert vegetable oils and pyrolysis oil into liquid fuels, the scarcity of technology specific data such as for the UOP process,
respectively. UOP has also developed their Green Jet FuelTM Process the modeling work is based on published studies describing the
(Fig. 2) using proprietary multifunctional catalysts [7] to target the HDO conversion of fatty acids into paraffins. The methodology
production of HRJ. The catalysts selectively hydrocrack the deoxy- developed in this study will provide insight into the impact of feed-
genated fatty acids, to maximize production of jet fuel range alka- stock composition upon product yield and hydrogen gas consump-
nes while reducing the co-production of lower value lighter tion. These parameters are crucial when modeling the HDO
hydrocarbons. Since the process is patented, the specific reaction process, to quantify the total thermal energy and electricity
conditions, including its product slate, are not publicly available. demands and corresponding GHG emissions for the jet fuel pro-
The UOP process has produced fuels for most of the biojet test duct, when coupled with information on the thermal energy source
flights using vegetable oils. and electricity grid. Results of the study also serve as a benchmark
Most studies assume that the production of biojet from lipids is for comparison against literature data that used simpler modeling
independent of the feedstock oil and fatty acid profiles, and that approaches, with soy oil and jatropha oil as feedstocks.

Fig. 1. The chemistry of hydrocracking of a TAG molecule that produces carboxylic acids and propane. The carboxylic acids subsequently undergo oxygen removal reactions.
Adapted from Veriansyah et al. [6].
300 P.L. Chu et al. / Fuel 196 (2017) 298–305

Fig. 2. Process schematic of the UOP Green Jet FuelTM process to produce HRD from natural oils, fats and grease with the selective hydrogenation process options to maximize
HRJ production. Adapted from UOP [8]. Note: HRJ = hydroprocessed renewable jet, HRD = hydroprocessed renewable diesel.

2. Process simulation scope and methodology Table 1

Oilseed characteristics used in estimating the solvent extraction mass balance, Rispoli
[22].
Biojet production using the HDO process is comprised of four
different processes, including feedstock production, oil extraction, Camelina, wt.% Carinata, wt.%
fatty acid conversion to renewable alkanes and product recovery, Moisture 9.0 9.0
as described below. Oil extraction is omitted from the model when Fibre 13.5 19.0
Protein 42.5 28.0
processing UCO. The processes to produce and recover renewable
Oil 35.0 44.0
alkanes were modeled using the Aspen Plus simulation software
[16]. The reactions and conversions for each fatty acid were mod-
eled in Microsoft Excel, and the aggregate conversion, product dis-
are heated using injected live steam before crushing in roller mills.
tribution and hydrogen demand were calculated based upon the
The cooked seeds are then extruded mechanically at high temper-
fatty acid composition of the oil from each type of feedstock.
ature and pressure to remove most of the oil. The subsequent
chemical process saturates the seeds with a solvent, usually hex-
2.1. Feedstock production
ane, creating a mixture that is recovered from the seeds by solvent
stripping to separate the oil product and recover the solvent for
Camelina and carinata are seasonal crops with growing seasons
reuse.
of 80–120 days. These crops produce oilseeds with high oil content,
The oil extraction process is the most energy intensive step dur-
and can grow on semi-arid land that is normally unsuitable for
ing the oil extraction stage. Although mechanical extraction pro-
conventional crop production [17]. Spring and winter varieties
cesses require less energy, the mechanical + chemical method
are currently under development for use on fallow lands, with
was selected for its higher oil recovery, which is crucial when the
the potential for double cropping [18]. These crops also require less
cost of feedstock dominates the process economics.
agricultural inputs than conventional crops, which lower produc-
tion costs and GHG emissions, while still achieving high productiv-
ity [19]. 2.3. Oil conversion
Used Cooking Oil (UCO) was also selected as it is a near-term
feedstock for biojet production. UCO is produced in the food indus- The conversion of oil into fuels occurs in a pressurized
try, especially in major urban centers, is a traded commodity, and hydrotreating reactor at high temperatures. The extracted oil is
can be used as a feedstock for the HDO process. fed with compressed hydrogen into the reactor, which contains a
nickel–molybdenum (NiMo) or similar catalyst. The selectivity of
2.2. Solvent extraction cracking is difficult to control, influenced by the catalysts used
and the reaction conditions [23,24]. A natural gas boiler provides
Process details and specifications for oil extraction have been the heat needed to maintain the reaction temperature. Light
described in the literature, including the industrial solvent extrac- hydrocarbons generated as co-products of the process may also
tion process for edible oils processing [20] and oil extraction in the be used for this purpose, or fed to a reforming unit to produce
canola industry [21]. The processing and handling of camelina and the hydrogen necessary for the process. The operating parameters
carinata are assumed to be the same as for canola, due to the sim- used during the oil conversion process modeled in Aspen Plus,
ilarity in seed dimensions and oil content. Oilseed characteristics based upon a NiMo catalyst, are outlined in Table 2.
for camelina and carinata are outlined in Table 1. Oil extraction The high pressure and temperature vapor exiting the reactor is
operations include seed cleaning, crushing, cooking with direct fed to a turbine to generate electricity that provides power to the
steam injection, mechanical extrusion, flaking, countercurrent sol- process. The conversion process, including electricity generation,
vent extraction, meal toasting and drying, and solvent stripping is illustrated via the block flow diagram in Fig. 4.
(Fig. 3). UCO does not require an extraction step. Oil conversion in the HDO process revolves around the decar-
When the oilseeds are received, they are initially screened to boxylation, decarbonylation and hydrodeoxygenation reactions
remove field dirt and metals. To prevent seed shatter, the seeds outlined in Table 3. The decarboxylation reaction removes oxygen
 P.L. Chu et al. / Fuel 196 (2017) 298–305 301

Fig. 3. Block flow diagram of conventional solvent extraction process.

Table 2
Input data and assumptions for the HDO process model [6,25]. the selectivity between the oxygen removal reactions. The reaction
conditions and selected catalyst can impact the extent and selec-
Process parameter Value
tivity of the three competing reactions, which can be optimized
Reaction temperature 400 °C for biojet production. The oxygen removal reactions during the
Reaction pressure 92 bar
HDO process occur in the liquid phase to produce gaseous reaction
Residence time 2h
Catalyst type NiMo products. These gases are suspected to react with H2 gas, causing
Conversion, X (batch operation) 0.9994 methanation (Eq. (4)) and water-gas-shift (Eq. (5)) reactions to
Extent of decarboxylation, e1 0.68 produce CO2, methane and water. This leads to the production of
Extent of decarbonylation, e2 0.03
lighter products and increased the demand for H2 gas.
Extent of hydrodeoxygenation, e3 0.29
Aspen reactor model RStoic
The reactions and conversions were modeled in Microsoft Excel
Property method NRTL & Peng-Robinson based upon each fatty acid in the oilseed (Table 4). This approach
overcomes the limited number of fatty acid components in the
Aspen Plus databank. A mass balance for the oil conversion process
through CAC bond scission to form CO2 in the absence of H2 gas was generated using the stoichiometric reactions that would occur
(Eq. (1)), while the decarbonylation reaction uses H2 gas through during the process. These reactions and the mass balance were
CAO bond scission to remove oxygen, generating CO and water used to calculate the amount of hydrogen gas required and the
(Eq. (2)). Both reactions produce alkanes with one less carbon than amount of CO2, CO and water produced. As shown in Table 4, there
the original fatty acid. On the other hand, the hydrodeoxygenation can be substantial differences in the content of saturated, monoun-
reaction uses H2 to remove oxygen and produce water, thereby saturated and polyunsaturated fatty acids, which should affect
preserving the carbon content of the original fatty acid (Eq. (3)). hydrogen demand. Camelina, carinata and UCO oils have high
The hydrogen requirement for the HDO process is dependent on degrees of monounsaturated and polyunsaturated fatty acids.

Fig. 4. Block flow diagram of the oil conversion process.

302 P.L. Chu et al. / Fuel 196 (2017) 298–305

Table 3 The first separation stage is a ‘‘gas-liquid separation” that

Stoichiometric reactions involved in the HDO process [26]. removes non-condensable gases, such as carbon dioxide (CO2), car-
Stoichiometric reaction DH573 bon monoxide (CO), and excess hydrogen from the reactor. These
(kJ/mol) gases leave the top of the separation column and are combusted
Eq. (1) Decarboxylation C n H2n O2 ! C n1 H2n þ CO2 9.2 by a flare to ensure no combustible gases are emitted to the atmo-
Eq. (2) Decarbonylation C n H2n O2 þ H2 ! C n1 H2n þ CO þ H2 O 179.1 sphere. The liquid products are first processed in a prefractionator
Eq. (3) Hydrodeoxygenation C n H2n O2 þ 3H2 ! C n H2nþ2 þ 2H2 O 115.0 column that separates the lighter (C1-C6) and heavier (>C7) prod-
Eq. (4) Methanation CO2 þ 4H2 $ CH4 þ 2H2 O 177.2
CO þ 3H2 $ CH4 þ H2 O 216.4
ucts. Light products are sent to the stabilizer column to isolate LPG
Eq. (5) Water gas shift CO þ 4H2 $ H2 þ CO2 39.2 (C1-C4) products from light naphtha (>C5). The heavy products are
processed in a crude distillation unit to separate the various cuts
into heavy naphtha (C6-C9), kerosene (C10-C15) and diesel
Camelina oil is mainly composed of polyunsaturated fatty acid (>C15). The fuel products are stored onsite.
(54.3 wt.%), in which the main fatty acid is linolenic acid (C18:3; Process heat for the distillation columns is supplied by steam
32.6 wt.%). Carinata oil is mainly oleic acid (C18:1; 43.2 wt.%) fol- produced in a natural gas boiler, while electricity is provided from
lowed by linoleic acid (C18:2; 36.0 wt.%). The majority of fatty the on-site turbine and from the local power grid. Cooling is sup-
acids in UCO are linoleic acid (C18:2; 43.0 wt.%) followed by oleic plied using a standard water cooling tower system. The energy
acid (C18:1; 36.0 wt.%). Fatty acid compositions of soybean, palm, required during product recovery (due to the number of distillation
rapeseed and jatropha oils are included in the table for comparison. columns for separation) is a major contributor to overall process
Soybean oil has a high degree of polyunsaturated fatty acids energy demand. The specifications for the separation columns
(C18:2; 53.0 wt.% and monounsaturated fatty acids (C18:1; 22.0 modeled are summarized in Table 5, and illustrated as a block flow
wt.%). Rapeseed oil is comprised of 70 wt.% monounsaturated fatty diagram in Fig. 5.
acids (C18:1; C20:1 and C22:1). Jatropha oil has a higholeic acid
(C18:1; 45.0 wt.%) and linoleic acid (C18:2; 34.0 wt.%) content. In 3. Results and discussion
contrast, palm oil is highly saturated (C16:0; 40.8 wt.%).
The product distribution was determined based upon the per- 3.1. Product slates
centage of oil processed through each of the decarboxylation,
decarbonylation and hydrodeoxygenation reactions (Table 3), Modeling of the conversion reactions in Microsoft Excel pro-
while accounting for likely/possible cracking at the double bond duced a mass balance as well as the expected product slate for each
sites in each fatty acid (which leads to different products dictated feedstock based upon its fatty acid profile. Mass balance results
by the fatty acid composition). A theoretical minimum amount of from this study for the three feedstocks are presented in Table 6.
H2 gas required for the HRJ facility was calculated, which formed Results from the literature (Pearlson [14] and Han et al. [13]) are
the basis for the subsequent modeling in Aspen Plus [16]. This included in the table for comparison. The HDO process is governed
was replicated for all fatty acids to obtain the overall mass balance, by the reactions outlined in Table 3 and the production of light
H2 gas requirement and the product slate that are specific for each products is inevitable. Based on the molecular composition for
oil fatty acid profile. Using the inputs and yields from the Excel each of the feedstocks examined, there is a minimum amount of
model for camelina, carinata and UCO, the HDO process was simu- hydrogen gas that is required to cleave the glycerol backbone,
lated for each feedstock in Aspen Plus to obtain overall mass and which is subsequently converted into propane, while producing
energy balances. free fatty acids. The free fatty acids can then undergo decarboxyla-
tion, decarbonylation or hydrodeoxygenation reactions. The selec-
2.4. Product recovery tivity of the fatty acids to undergo either of these deoxygenation
reactions is dictated by the catalysts used and process conditions.
The product stream leaving the turbine is a mixture of fuel Considering the chain length of the fatty acids, for instance
products that requires separation, much like a conventional oil C18:0, cracking of the fatty acid will generate a light fraction
refinery. The products are separated using multiple distillation col- (e.g., C1-C6) and a heavy fraction (e.g., C12-C17). Even though
umns to produce fuels that meet the specifications for their the lighter fractions are not desired, in order to generate valuable
intended use. heavier fractions such as kerosene and diesel, it is necessary to

Table 4
Fatty acid profiles of camelina, carinata, and used cooking oil (UCO), with comparison to other oils.

Fatty acid (wt.%) Camelinaa Carinatab UCOc Soybeand Palmd Rapeseedd Jatrophad
<C14 0 0 0 0 5.2 0 0
C14:0 0.1 0 0 0 1.3 0 0
C16:0 6.8 5.4 13.0 11.0 40.8 3.0 13.0
C18:0 2.7 0.2 3.8 4.0 3.7 1.0 8.0
C20:0 1.5 0.0 0 0 0 0 0
C22:0 0.2 0.0 0 0 0 0 0
C18:1 18.6 43.2 36.0 22.0 37.2 17.0 45.0
C20:1 12.4 0 0 0 0 11.0 0
C22:1 2.3 0 0 0 0 45.0 0
C18:2 19.6 36.0 43.0 53.0 10.1 14.0 34.0
C18:3 32.6 15.2 3.6 8.0 0 9.0 0
C20:2 1.3 0 0 0 0 0 0
C20:3 0.8 0 0 0 0 0 0
a
From [27].
b
From [28].
c
From [29].
d
From [13].
 P.L. Chu et al. / Fuel 196 (2017) 298–305 303

Table 5 integration using pinch analysis was performed in order to model

Input data for modeling of hydrocarbon product separation. a realistic production facility. The highest energy recovery was
Unit Aspen Number of Column obtained when the product recovery and oil conversion processes
block stages pressure are integrated. This significantly reduces the overall energy input,
Prefractionator PetroFrac 15 2.5 bar especially thermal energy, compared to previous predictions in
Stabilizer RadFrac 20 15.5 bar the literature (compare, e.g., 5723 MJ/tonne of camelina oil for
Crude Distillation Unit (CDU) PetroFrac 21 1 bar our optimized process with heat integration, to 17,628 MJ/tonne
of camelina oil predicted by Han et al. [13]). The use of oilseed
feedstocks result in higher electricity demand compared to UCO.
The electricity demand for oilseed feedstocks in Pearlson [14]
crack the longer chains. Some would argue that cracking is not nec- and Han et al. [13] is also noticeably lower, perhaps due to differ-
essary for C18:0, but in the HDO reactor, cracking can occur in the ent assumptions regarding (or exclusion of) the oil extraction pro-
midst of decarboxylation, decarbonylation and hydrodeoxygena- cess, which is energy intensive.
tion reactions, and it is difficult to control the degree of cracking. The HDO reactions are highly complex and difficult to model,
Nonetheless, cracking can be minimized by lowering the reaction leading to uncertainty in predictions of process yields. Previous
temperature, altering the reaction pressure and the use of selective studies have constructed simplified HDO process models that
catalysts. Theoretically, it is not possible to avoid producing lighter relied on UOP’s proprietary process data, sufficient for the purpose
fractions or processing the heavier fractions. If the heavy fractions of estimating process utilities, and relied on generic literature data
are further processed, more light hydrocarbons would be observed for H2 requirements and the product slate. The analysis in this
in product distribution. In a process that targets the production of study was based on the selectivity of the deoxygenation reactions
HRJ, where the diesel fraction is further cracked into kerosene- and the location of the double bond within each fatty acid chain.
range hydrocarbons, additional light hydrocarbons are inevitably This allows prediction of product slates, H2 demands and utility
produced. Historically, diesel has a slighter higher market value demands that are specific to each of the camelina, carinata and
than kerosene, and therefore, producers have little to no motiva- UCO feedstocks.
tion to invest more capital and convert a higher value product, There are limited literature data available on process utilities
i.e., diesel, into a lower value product. Therefore, it is not realistic and product yields for the HDO conversion process. Currently,
for a refinery to continuously crack diesel into lighter fractions to the majority of the publications on HDO conversion are based on
maximize kerosene output. a single model developed by Pearlson [14], where process utility
Although the differences in fatty acid composition have yielded estimates have been based on similar unit operations in conven-
different mass balance results, these differences for an individual tional petroleum processing. While Pearlson [14] has provided a
oilseed/oil are small relative to the compositional differences benchmark, the estimated process energy does not include oilseed
between different oilseeds, and thus, the observed trends illustrat- feedstock handling and processing, which has been found to con-
ing differences in the product slate and hydrogen requirements tribute significantly to the overall process energy demand. More-
between feedstocks should remain consistent. By comparison, the over, the reported product yields have been determined from
product slate and mass balances in Pearlson’s work [14] are inde- diesel production from soybean oil, and jet fuel production has
pendent of feedstock composition. It should be noted that kerosene been inferred from data using jatropha oil. These values have since
yields in Han et al. [13] are higher compared to our work and Pearl- been cited and widely used to estimate biojet production from var-
son’s work, due to Han’s assumption of complete conversion of die- ious feedstocks, without including the effects of reaction condi-
sel into kerosene. CO2, CO and water values were not reported by tions, feedstock compositional differences such as chain lengths,
Han et al. [13]. and degree of bond unsaturation, which collectively impact pro-
cess hydrogen requirements and product yields. When these val-
3.2. Process energy demand ues are used generically without considering differences due to
crop and process yields, process conditions and hydrogen require-
The heating and cooling requirements for the HDO process were ments, the evaluation of the associated GHG emissions could be
calculated using Aspen Plus, and are presented in Table 7. Heat affected, especially for land use change (LUC), which is strongly

Fig. 5. Block flow diagram of the product purification process.

304 P.L. Chu et al. / Fuel 196 (2017) 298–305

Table 6
Mass balances for HDO conversion of camelina, carinata and UCO to produce HRJ. Results from Pearlson [14] and Han et al. [13] are presented for comparison.

Camelina Carinata UCO Pearlson Han et al. Han et al. Han et al. Han et al. Han et al.
Soyoil Soybean Palm Rapeseed Jatropha Camelina
Input
Oil 1000 1000 1000 1000 1000 1000 1000 1000 1000
H2 gas 30 26 26 40 27 22 27 25 30
Output
CO2 101 95 104 54 N/R N/R N/R N/R N/R
CO 2.7 2.5 2.7 N/R N/R N/R N/R N/R N/R
Water 36 34 37 87 N/R N/R N/R N/R N/R
LPG or propane mix 88 79 69 102 78 70 62 78 77
Naphtha 127 145 147 70 57 63 72 57 57
Kerosene 535 537 529 494 740 740 760 740 750
Diesel 140 132 138 233 N/R N/R N/R N/R N/R

Values expressed per tonne of oil, and mass reported in kg.

N/R – not reported.

Table 7 Thermal energy demand is 2.8 GJ/tonne oil when processing

Thermal energy and electricity requirements for kerosene production via the HDO used cooking oil, versus 5.2 and 5.7 GJ/tonne of oil for carinata
process using camelina, carinata and UCO. Results from Pearlson [14] and Han et al.
[13] are presented for comparison.
and camelina, respectively, owing to the additional energy
required for oil extraction. Electricity demand was 73 kWh/tonne
Total process energy Thermal energy, Electricity, oil for UCO, versus 180 and 227 kWh/tonne oil for carinata and
MJ/tonne oil kWh/tonne oil
camelina. The lower oil content of camelina compared to carinata
Camelina 5723 227 contributes to the higher thermal and electrical energy demand
Carinata 5185 180
UCO 2820 73
for the oil extraction process, when expressed per tonne of oil
Pearlson, Soyoil 12,473 66 product.
Han et al., Soybean 17,628 45
Han et al., Palm 7013 45
Han et al., Rapeseed 11,036 44 Declaration of interest
Han et al., Jatropha 11,084 44
Han et al., Camelina 10,809 46 The authors have no financial interest in the outcome of this
research.

dependent upon crop yields. Han et al. [13] acknowledged the lim- Acknowledgements
itations in the HDO process when variations in oil composition are
overlooked, and examined the effects of composition on hydrogen This research was financially supported by the Centre for
requirements for soybean, palm, rapeseed, jatropha and camelina. Research in Sustainable Aviation, the Natural Sciences and Engi-
In Han’s study, an increase in hydrogen consumption was observed neering Research Council, and the BioFuelNet Network Centre of
with an increasing quantity of unsaturated bonds, as expected. Excellence.
Collectively, our predictions of thermal energy demand are sub-
stantially lower than prior literature estimates based upon soy oil, References
palm oil, jatropha and camelina, illustrating the benefits of com-
prehensive heat integration among the oil extraction, oil conver- [1] IATA - Halving emissions by 2050 - aviation brings its targets to Copenhagen;
2009. <http://www.iata.org/pressroom/pr/Pages/2009-12-08-01.aspx> [accessed
sion and distillation operations. Compared to the literature, we July 20, 2016].
predict higher electricity demand for processing camelina and car- [2] van Renssen S. Climate battle for the skies. Nat Clim Chang 2012;2:308–9.
inata, due to the additional energy required for oil extraction, http://dx.doi.org/10.1038/nclimate1493.
[3] Resolution on the implementation of the aviation ‘‘CNG2020” strategy 2010.
whereas our estimate of electricity demand for UCO is in line with
<http://www.iata.org/pressroom/pr/Documents/agm69-resolution-cng2020.
that of the literature. The different thermal and electrical energy pdf> [accessed February 17, 2016].
requirements predicted by our study would affect the GHG inten- [4] The EU emissions trading system (EU ETS) 2013. <http://ec.europa.eu/clima/
sity of the HDO process to produce renewable jet fuel. policies/ets/index_en.htm> [accessed July 20, 2016].
[5] EU action-reducing emissions from aviation n.d. <http://ec.europa.eu/clima/
policies/transport/aviation/index_en.htm> [accessed July 20, 2016].
[6] Veriansyah B, Han JY, Kim SK, Hong SA, Kim YJ, Lim JS, et al. Production of
4. Conclusions renewable diesel by hydroprocessing of soybean oil: effect of catalysts. Fuel
2012;94:578–85. http://dx.doi.org/10.1016/j.fuel.2011.10.057.
[7] McCall MJ, Kocal JA, Bhattacharyya A, Kalnes TN, Brandvold TA. Production of
In this study, the effects of feedstock compositional variations aviation fuel from renewable feedstocks. PAT – US 8039682 B2; 2011.
on product yields and process utilities were examined, using a [8] UOP Green Jet FuelTM Process n.d. <http://www.uop.com/green-jet-fuel/>
model to calculate cracking reactions and the alkane product dis- [accessed July 20, 2016].
[9] Stratton SW. Life cycle assessment of greenhouse gas emissions and non-CO2
tribution based upon the fatty acid profile of each feedstock. The combustion effects from alternative jet fuels. Massachusetts Institute of
yield of kerosene-range alkanes ranged from 53 to 54 wt% of the Technology; 2010.
incoming oil, with 13–14% diesel range alkanes, 13–15% naphtha, [10] Shonnard DR, Williams L, Kalnes TN. Camelina-derived jet fuel and diesel:
sustainable advanced biofuels. Environ Prog Sustain Energy 2010;29:382–92.
and 7–9% LPG, with the balance comprised of water, CO2, and CO.
http://dx.doi.org/10.1002/ep.10461.
Feedstocks with a higher degree of bond unsaturation generally [11] Agusdinata DB, Zhao F, Ileleji K, Delaurentis D. Life cycle assessment of
consume more hydrogen compared to feedstocks with less unsat- potential biojet fuel production in the United States. Environ Sci Technol
urated bonds, although the degree of cracking is also a factor; 2011;45:9133–43. http://dx.doi.org/10.1021/es202148g.
[12] Fan J, Shonnard DR, Kalnes TN, Johnsen PB, Rao S. A life cycle assessment of
the consumption rate ranged from 26 to 30 kg per tonne of incom- pennycress (Thlaspi arvense L.)-derived jet fuel and diesel. Biomass Bioenergy
ing oil. 2013;55:87–100. http://dx.doi.org/10.1016/j.biombioe.2012.12.040.
 P.L. Chu et al. / Fuel 196 (2017) 298–305 305

[13] Han J, Elgowainy A, Cai H, Wang MQ. Life-cycle analysis of bio-based aviation [21] (S&T)2 Consultants INC. Lifecycle analysis canola biodiesel; 2010.
fuels. Bioresour Technol 2013;150:447–56. http://dx.doi.org/10.1016/j. [22] Rispoli K. Life cycle and supply assessment of aviation biofuels from oil
biortech.2013.07.153. feedstock, in the Canadian context. University of Toronto; 2014.
[14] Pearlson MN. A techno-economic and environmental assessment of [23] Cheng J, Li T, Huang R, Zhou J, Cen K. Optimizing catalysis conditions to
hydroprocessed renewable distillate fuels. Massachusetts Institute of decrease aromatic hydrocarbons and increase alkanes for improving jet biofuel
Technology; 2011. quality. Bioresour Technol 2014;158:378–82. http://dx.doi.org/10.1016/j.
[15] Pearlson M, Wollersheim C, Hileman J. A techno-economic review of biortech.2014.02.112.
hydroprocessed renewable esters and fatty acids for jet fuel production. [24] Bezergianni S, Voutetakis S, Kalogianni A. Catalytic hydrocracking of fresh and
Biofuels, Bioprod Biorefining 2013;7:89–96. http://dx.doi.org/10.1002/ used cooking oil. Ind Eng Chem Res 2009;48:8402–6. http://dx.doi.org/
bbb.1378. 10.1021/ie900445m.
[16] Aspen Technology Inc., Aspen Plus simulation software n.d. [25] Kim SK, Brand S, Lee HS, Kim Y, Kim J. Production of renewable diesel by
[17] Marillia EF, Francis T, Falk KC, Smith M, Taylor DC. Palliser’s promise: Brassica hydrotreatment of soybean oil: effect of reaction parameters. Chem Eng J
carinata, An emerging western Canadian crop for delivery of new bio- 2013;228:114–23. http://dx.doi.org/10.1016/j.cej.2013.04.095.
industrial oil feedstocks. Biocatal Agric Biotechnol 2014;3:65–74. http://dx. [26] Snåre M, Kubičková I, Mäki-Arvela P, Eränen K, Murzin DY. Heterogeneous
doi.org/10.1016/j.bcab.2013.09.012. catalytic deoxygenation of stearic acid for production of biodiesel. Ind Eng
[18] Gesch RW, Archer DW. Double-cropping with winter camelina in the northern Chem Res 2006;45:5708–15. http://dx.doi.org/10.1021/ie060334i.
Corn Belt to produce fuel and food. Ind Crops Prod 2013;44:718–25. http://dx. [27] Moser BR, Vaughn SF. Evaluation of alkyl esters from Camelina sativa oil as
doi.org/10.1016/j.indcrop.2012.05.023. biodiesel and as blend components in ultra low-sulfur diesel fuel. Bioresour
[19] Cardone M, Mazzoncini M, Menini S, Rocco V, Senatore A, Seggiani M, et al. Technol 2010;101:646–53. http://dx.doi.org/10.1016/j.biortech.2009.08.054.
Brassica carinata as an alternative oil crop for the production of biodiesel in [28] Dorado MP, Ballesteros E, López FJ, Mittelbach M. Optimization of alkali-
Italy: agronomic evaluation, fuel production by transesterification and catalyzed transesterification of brassica carinate oil for biodiesel production.
characterization. Biomass Bioenergy 2003;25:623–36. http://dx.doi.org/ Energy Fuels 2004;18:77–83.
10.1016/S0961-9534(03)00058-8. [29] Haigh KF, Vladisavljević GT, Reynolds JC, Nagy Z, Saha B. Kinetics of the pre-
[20] Anderson GE. Solvent extraction. Am Oil Chem 2011. <http://lipidlibrary.aocs. treatment of used cooking oil using Novozyme 435 for biodiesel production.
org/OilsFats/content.cfm?ItemNumber=40337> [accessed July 20, 2016]. Chem Eng Res Des 2014;92:713–9.