Renewable Energy 219 (2023) 119591

Contents lists available at ScienceDirect

Renewable Energy
journal homepage: www.elsevier.com/locate/renene

Industrial production of green diesel in Brazil: Process simulation and

economic perspectives
Ian V. Barbosa a, Letícia A. Scapim a, Raquel M. Cavalcante b, André F. Young a, *
a
Departamento de Engenharia Química e de Petróleo, Universidade Federal Fluminense, Rua Passo da Pátria, 156, Niterói, Brazil
b
Escola de Química, Universidade Federal do Rio de Janeiro, Av. Athos da Silveira Ramos, 149, Rio de Janeiro, Brazil

A R T I C L E I N F O A B S T R A C T

Keywords: The purpose of this work is to apply process simulation to infer the technical and economic bottlenecks of the
Green diesel industrial production of green diesel, taking as a context the Brazilian market at the end of 2021/beginning of
Hydrotreated vegetable oil 2022. In the considered economic scenario, a stand-alone green diesel production process does not seem
Hydrotreatment
economically viable in Brazil, as the fuel would need to cost 30 % more than the European market price and
Hydrogenation
Process simulation
twice the conventional diesel price (US$ 2.08/L) to compensate for the investment made. However, significant
Economic analysis price reductions - 10 %–15 % in the considered scenarios (to US$ 1.76–1.87/L) - can be achieved with capital and
energy savings due to the availability of a conventional refinery hydrotreatment unit, and eventual parity with
market prices can be achieved with soybean oil price reductions (of 22–40 %, completely possible in the Brazilian
market). This analysis shows that green diesel may be absolutely possible depending on the economic context
and that there is room for its gradual incorporation in petroleum diesel. This is the first research article to
demonstrate it though process simulation and to economically compare the investment on a stand-alone process
with a refinery retrofit.

1. Introduction Euro 6-type engines, within the scope of the Brazilian Proconve 8
(Vehicle Emission Control Program, in Portuguese), starting from 2023
The traditional renewable substitute for petroleum diesel is bio­ [4]. It tends to make more difficult the continuous increase of the bio­
diesel, whose production involves converting vegetable oils and animal diesel mixture in conventional diesel in the country.
fats through the transesterification of triglycerides with a short-chain An alternative is the adoption of “drop-in” fuels, which can be used in
alcohol, usually methanol. However, as it has a different structural conventional automobiles in any proportion, without the need of mod­
formula from petroleum diesel, biodiesel has some limitations, such as ifications in the engine and other mechanical components, in addition to
higher viscosity, lower oxidation stability and higher solvency. It has not requiring changes in the existing infrastructure for their storage and
implications for the whole current transport infrastructure, which in­ transport [5,6]. This is the case of green diesel or HVO (hydrotreated
cludes not only its direct application in the vehicle/machinery engines, vegetable oil), which appears as a more promising alternative to pe­
but also the fueling stations and distribution networks, fully developed troleum diesel because it is chemically the same thing - the only dif­
for plain and stable liquid hydrocarbons derived from petroleum [1]. For ference is its renewable source.
this reason, the addition of biodiesel to conventional diesel must be Therefore, the objective of this work is to analyze the technical and
limited, as the complete switch from the fossil fuel to the renewable one economic feasibility of an HVO production plant in Brazil and validate
cannot be achieved without significant mechanical changes in conven­ the potential of its production from soybean oil, which is the main
tional engines [2]. biodiesel production feedstock in the country, as it is one of its largest
The key factor that contributes to the lower stability of biodiesel is producers in the world [7]. To this end, the process was simulated in the
the presence of oxygen and unsaturation, since oxygenated molecules Aspen HYSYS® v12.0 software for rigorous solution of the mass and
are more reactive to heat and unsaturated molecules are more likely to energy balances, which allowed equipment sizing, and profitability and
oxidation [3]. In addition, biodiesel combustion has higher NOx gener­ cash flow analyses. Then, the investment on a stand-alone process plant
ation, violating the new maximum limits for pollutants to be adopted in was compared with the use of the hydrotreatment structure of a

* Corresponding author.
E-mail address: [email protected] (A.F. Young).

https://doi.org/10.1016/j.renene.2023.119591
Received 23 June 2023; Received in revised form 27 October 2023; Accepted 4 November 2023
Available online 6 November 2023
0960-1481/© 2023 Elsevier Ltd. All rights reserved.
I.V. Barbosa et al. Renewable Energy 219 (2023) 119591

pre-existing petroleum refinery, and the results were discussed in terms it contains water, free fatty acids and other impurities that impact the
of green diesel competitiveness with conventional diesel and biodiesel, production costs. The energy costs were estimated at 1–7% of the total
besides its gradual incorporation in conventional diesel. This is the first production cost, and raw material costs were 23–33 %. According to the
research article to present such analysis, which constitutes a novelty. authors, if the waste cooking oil hydrogenation unit was incorporated
into an existing petroleum refinery, not only the investment cost would
2. Literature review reduce, but also the raw material costs: it would drop to 41 %, since the
hydrogen and PSA units (adsorption process to remove CO2) would
2.1. Catalysts already be installed. The production costs were estimated in the range of
US$ 0.63/L–0.68/L, showing, according to the authors, that the process
A wide variety of catalysts for this process can be found in the is not economically viable. A possible solution would be favoring the
literature. It is interesting to mention that, in addition to the sulfur occurrence of decarboxylation, which happens at low pressures and
catalysts already used in petroleum refineries and for co-processing - results in less water.
which have the advantage of not requiring many modifications -, spe­ Another alternative for cost reduction would be to reduce the
cific catalysts for green diesel are being developed. Based on literature, it amount of H2, which is an expensive gas and comes from fossil sources.
is possible to conclude that the most conventional support is Al2O3, Between 7 and 16 mol of hydrogen per mole of triglycerides are
having moderate acidity and high stability, although the deactivation of required. The solution could be reusing by-products/waste streams from
the catalyst by the formation of coke is a problem that motivates the the hydrotreatment process as a source of this gas. For example, up to 75
study of other supports. Furthermore, as active material, metals such as % of the H2 needs could be met by steam reforming the propane
Ni are preferable, as they present better activity, bearing in mind that generated during the process, followed by a water-gas shift reaction [1].
the low activity may be linked to the low acidity of the support [8]. Choe et al. [21] carried out a techno-economic and environmental
NiMO/Al2O3 was reported to provide high conversions [9,10], therefore assessment of green diesel production through two different pathways:
it was chosen as the reference catalyst for this study. thermochemical process with CO production from reverse water gas
shift; and electrochemical process with CO production from CO2 elec­
2.2. Economic evaluations of green diesel production trolysis. The authors evaluated hydropower, solar photovoltaic and
onshore wind as energy sources. For a production capacity of 100
Technical-economic analyzes are of crucial importance to ensure tons/day, even considering an uncertainty of 50 %, the authors
that the new technology is competitive with the existing ones. According demonstrated that the green diesel production costs for any of the routes
to Kann et al. [11], cited by No et al. [12] the cost of hydroprocessing is and energy sources were economically unfeasible compared with the
higher than that of the transesterification process, although there are conventional diesel market price. However, a future projection showed
studies that diverge [13]. Anyway, according to the US Energy Infor­ that green diesel production through the electrochemical pathway with
mation Administration (EIA), green diesel supply is expected to surpass onshore wind can become competitive in the next decade, besides being
the biodiesel supply in the United States in the next years, due to gov­ more eco-friendly. Of course, it depends on renewable electricity cost
ernment targets and incentives [14]. reductions. Hydrogen production via water electrolysis was the biggest
Melero et al. [15], studying the NexBTL™ and Ecofining™ processes, fraction of the production costs in every route, therefore, meanwhile,
concluded that these technologies are competitive in relation to the conventional hydrogen sources may be necessary.
biodiesel production process when carried out in medium-sized opera­ These analyses were the references for the economic evaluation
tional units. However, profitability will depend on the market values of performed in this work, and it is interesting to notice how the conclu­
the vegetable oil and petroleum. In addition, Salvador et al. [16] indi­ sions formulated in these works are in line - considering the different
cated the dependence on coproducts, public policies and decentralized times and places - with the discussions presented in the next sections.
hydrogen production for a biofuel-driven biorefinery to be feasible, The contributions of this article to the scientific literature on green
while studying three scales of a biorefinery for green diesel production diesel are: to provide a process simulation that can be used as a basis for
from macauba oil. the rigorous solution of the mass and energy balances to be applied in an
Faraci et al. [17] compared green diesel production from palm and economic evaluation, with sufficient information for its complete
soybean oil. The capital and production costs were the same for pro­ reproduction/adaption; to present the advantages of the use of a pre-
cessing both oils, but differed in the contribution of different factors. existing refinery infrastructure over the construction of a stand-alone
Among these, it is possible to highlight the costs with hydrogen, cata­ facility, even if co-processing is not considered; and to evaluate sce­
lysts and expenses with other chemical products, being 29 % higher for narios of energy savings and gradual incorporation of green diesel in
soybean oil. In this case, palm oil green diesel would be feasible if the petroleum diesel.
petroleum price was at least US$ 52/bbl, and soybean oil green diesel
would be feasible if it was US$ 67/bbl, plus government subsidy. 2.3. Co-processing
Hilbers et al. [18], assuming an internal rate of return of 15 % and
depreciation in ten years, calculated a negative cash flow - i.e. the Based on the compiled economic analyzes, it is noticed that the most
production was not feasible in that economic context. The authors efficient solution to reduce HVO capital and production costs is to make
estimated a production cost of US$ 0.70–0.79/kg of green diesel. use of existing units in petroleum refineries, through the adoption of
Diniz et al. [19] compared the green diesel production price from mixtures of renewable compounds with conventional diesel [22].
carinata oil, camelina oil and jatropha oil. In this study, jatropha oil had Hydrotreatment units in refineries were built to reduce the sulfur con­
the lowest production cost as it was the raw material with the lowest tent of the petroleum fractions, however it is also suitable for the pro­
cost, but, once more, the revenue was lower than the production cost, duction of biofuels [23].
resulting in a negative cash flow. For the camelina oil hydroprocessing It is believed that renewable fuels have a greater chance on the
plant, the equilibrium prices for LPG, naphtha, green jet fuel and green market if they are blended or produced via co-processing with petro­
diesel were US$ 0.50/L, US$ 0.80/L, US$ 0.32/L and US$ 0.79/L; for leum, but there are some considerations to be made, such as the higher
carinata oil, US$ 0.56/L, US$ 0.89/L, US$ 0.36/L and US$ 0.88/L; for content of contaminating metals (Ca, P, K, Na) in vegetable oils. There is
jatropha oil, US$ 0.74/L, US$ 1.18/L, US$ 0.48/L and US$ 1.16/L. a need to implement a pre-treatment reactor for their removal.
Therefore, the carinata oil presented greater advantages. Furthermore, the cold flow properties of the produced diesel should
To reduce the raw material costs, Glisic et al. [20] proposed to study limit the amount of vegetable oil that can be processed as n-paraffin
the use of waste cooking oil for the production of green diesel. However, (first product of hydrotreatment), which will impact the cloud point.

2
I.V. Barbosa et al. Renewable Energy 219 (2023) 119591

Besides, deoxygenation tends to compete with desulfization, which is a eicosanoic (0.43 %), gondoic (0.22 %), palmitoleic (0.11 %) and myr­
risk, given the requirement for low sulfur diesel [17]. istic esters (0.08 %). For its representation, however, the mixture was
There are co-processing studies that make use of fossil raw materials simplified to trilinolein (53.27 %), triolein (24.30 %), tripalmitin (11.13
such as SRGO (straight run gasoil), LAGO (light atmospheric gasoil), %), trilinonelin (6.85 %) and tristearin (4.45 %), which make up 98.75
HAGO (heavy atmospheric gasoil) and HVGO (heavy vacuum gasoil) % of the oil’s fatty acid profile.
[24]. Normally, the catalysts used for HDO will be the ones used for
hydrodesulfurization, since the reactions will coexist [25]. Therefore, 3.1.2. Catalyst definition
there is much room for future studies involving the technical-economic The catalyst chosen for the hydrotreatment of soybean was NiMo/
viability of co-processing gasoil and vegetable oil. Al2O3. This choice was made since alumina is a widely used support and,
But even if vegetable oils are processed alone in a pre-existing re­ according to Veriansyah et al. [9], this catalyst guaranteed better results
finery infrastructure, considerable gains can be achieved in terms of in terms of economics and conversion, compared to other catalysts over
capital and production costs. That is what we want to show in this work. the same support, such as Pd, Pt or CoMo.
Once the catalyst was chosen, it was possible to select the articles
3. Process simulation that would serve as the basis for this work, which contained the neces­
sary data for the project. They were the works of Veriansyah et al. [9],
Prior to the simulation, it is necessary to define the involved com­ Plazas-González et al. [29] and Hancsók et al. [30]. But the catalyst do
ponents - with some considerations - and the thermodynamic package to not participate in the simulations, it is only implicitly considered in the
be used, in order to represent the green diesel production plant in the reaction models and conditions.
most reliable way. In this work, the simulations were performed in the
Aspen HYSYS® v12.0 process simulator. The simulation foresees the 3.2. Thermodynamic package
side production of biokerosene, which was made possible by means of
one additional distillation column in the product separation stage. Due to the sole presence of non-polar components - such as hydro­
carbons - and light gases - such as CO2 and H2 - after reaction, the most
3.1. Chemical components definition appropriate thermodynamic model for this section is a cubic equation of
state, such as Peng-Robinson [31]. Besides that, aiming at the continuity
Most of the components needed for the simulation were already of this work or its mere reproducibility by an interested reader, we find
available in the simulator database. However, among the triglycerides, pertinent to point out that, although suitable for the equipment units
only triolein was present. So, the other constituints of the vegetable oil comprising all the way from the inlet streams to the reactors - due to the
(tripalmitin, tristearin, trilinolenin, trilinolein) and the isomers present highly non-ideal nature of the thermodynamic systems -, activity coef­
in the product (i-pentadecane, i-hexadecane, i-heptadecane and i-octa­ ficient models like UNIQUAC or NRTL, did not generate coherent re­
decane) had to be created as hypothetical components, whose thermo­ sults, even with internal parameter estimation via group-contribution.
dynamic properties values are presented in Table 1. Because of that, the Peng-Robinson equation of state was the only
The density (Ideal Liq Density), boiling point (Normal Boiling Pt) and thermodynamic package used for the entire simulation. However, we do
molecular weight (Molecular Weight) values were obtained from the not consider it a significant deficiency, because, as pointed out by Pla­
CHEMSRC tool [26]. Meanwhile, critical temperatures were extracted zas-González et al. [29], this model satisfactorily predicts the
from the literature [27] for the triglycerides - calculated with the Con­ vapor-liquid behavior of fatty acid compounds at high pressures.
stantinou and Gani method -, and obtained from the CHEMÉO database
[28] for the isomers. Then, the remaining properties were estimated for 3.3. Additional considerations
each component from their UNIFAC structures: acentric factor, critical
pressure and critical volume. 3.3.1. Plant capacity definition
Since a chemical component can present numerous isomers and due As it is a biofuel not yet commercially produced in Brazil, an analysis
to the lack of specificity in the literature regarding the structural formula was made based on the production capacity of a green diesel production
of the components generated in the isomerization reaction, a similar plant to be implemented in the country: a partnership between Vibra
structure was considered for all of them: straight chain alkanes with a and Brasil BioFuels (BBF). The project consists of the construction of a
methyl branch on the second carbon. biorefinery in the Amazon region with a production capacity of 500
million liters per year [32]. Considering an average density of 772.5
3.1.1. Vegetable oil representation kg/m3 [33], the production capacity of the plant would be of about 1.17
Hydrotreatment reactions involve the participation of two compo­ × 106 kg/day (or 48,769 kg/h) of green diesel.
nents: hydrogen gas and the triglycerides that make up the vegetable oil,
in this case soybean oil, selected for its strong potential in the Brazilian 3.3.2. Inlet streams definition
market. Since its composition may vary depending on the region or Both soybean oil and H2 feed streams enter the process at ambient
season, the following representative fatty acid composition was used temperature and pressure, as they are streams which origin is the
[9]: linoleic (52.56 %), oleic (23.98 %), palmitic (10.98 %), external environment, defined as 30 ◦ C and 1 atm (about 101.3 kPa).
alpha-linolenic (6.76 %), stearic (4.39 %), behenic (0.49 %) %), The function of the hydrogen stream is to supply the H2/oil ratio at the

Table 1
Thermodynamic properties of the hypothetical components.
Component Boiling Point (◦ C) Density (kg/m3) Critical Temperature (◦ C) Critical Pressure (kPa) Acentric Factor Molecular Weight

Tripalmitin 759.8 900.0 685.6 243.7 3.4390 807.3

Tristearin 813.0 900.0 703.8 176.9 4.0710 891.5
Trilinolein 816.5 925.0 705.6 212.1 3.7530 879.4
Trilinolenin 814.3 900.0 706.6 175.1 3.7510 873.3
i-Pentadecane 216.6 768.0 430.8 1898.0 0.2466 212.4
i-Hexadecane 282.0 772.0 440.4 1433.0 0.9384 226.4
i-Heptadecane 291.4 776.0 445.9 1386.0 0.8213 240.5
i-Octadecane 304.9 800.0 457.2 1386.0 0.8981 254.2

3
I.V. Barbosa et al. Renewable Energy 219 (2023) 119591

entrance of the reactor, so its value is defined based on its consumption Table 2
in the hydrogenation reactions. Meanwhile, the oil flow rate is calcu­ Hydrodeoxygenation reactions.
lated to satisfy the specified green diesel production defined in the last Triglyceride Reaction
section. The ratio of hydrogen and soybean oil used was 450:1, based on
Tripalmitin C51H98O6 + 12 H2 → C3H8 + 3 n-C16H34(aq) + 6 H2O(l)
the work of Hancsók et al. [30], which describes the simulation of a Tristearin C57H110O6 + 12 H2 → C3H8 + 3 n-C18H38(aq) + 6 H2O(l)
green diesel production plant from different vegetable oils and catalysts, Triolein C57H104O6 + 15 H2 → C3H8 + 3 n-C18H38(aq) + 6 H2O(l)
including sunflower oil, the most similar to soybean oil, and the CoMo Trilinolein C57H98O6 + 18 H2 → C3H8 + 3 n-C18H38(aq) + 6 H2O(l)
catalyst; this information was absent in the articles referring to soybean Trilinolenin C57H92O6 + 21 H2 → C3H8 + 3 n-C18H38(aq) + 6 H2O(l)

oil.
Reaction modeling was done in order to find values similar to the
ones indicated in the used references and in order to obtain a product Table 3
stream that is also consistent with the literature: in addition to C15–C18 Decarboxylation and decarbonylation reactions.
chains corresponding to green diesel, there is the presence of products Triglyceride Reaction
from the cracking reaction, as kerosene - fraction of the products with a Tripalmitin 9 3 3 3
boiling point between 170 and 270 ◦ C - and naphtha - between 40 and C51H98O6 + H2 → C3H8 + 3 n-C15H32(aq) + H2O(l) + CO2(g) +
2 2 2 2
200 ◦ C [9]. Taking into account the complexity of such reactions, a CO(g)
Tristearin 9 3 3 3
stream composed of light hydrocarbons was inserted into the process at C57H110O6 + H2 → C3H8 + 3 n-C17H36(aq) + H2O(l) + CO2(g) +
2 2 2 2
the outlet of the hydrogenation reactor. It was taken as a basis that, CO(g)
according to Krár et al. [34], the composition of C5–C10 hydrocarbons is, Triolein
C57H104O6 +
15 3 3
H2 → C3H8 + 3 n-C17H36(aq) + H2O(l) + CO2(g) +
3
on average, 0.9 %. An equal-mass composition stream of n-pentane - 2 2 2 2
CO(g)
representing the naphtha fraction - and n-decane - representing the Trilinolein 21 3 3 3
C57H98O6 + H2 → C3H8 + 3 n-C17H36(aq) + H2O(l) + CO2(g) +
kerosene fraction, was considered: its flow rate was defined for such 2 2 2 2
CO(g)
composition, but based on a stream absent of hydrogen, since this value
Trilinolenin 27 3 3 3
refers to the stream of products obtained in the hydrotreatment step. C57H92O6 +
2
H2 → C3H8 + 3 n-C17H36(aq) + H2O(l) + CO2(g) +
2 2 2
CO(g)
3.4. Reaction modeling
C17 paraffins.
Given the absence of kinetic data in the references adopted as a basis
Table 4 describes the data used to obtain the reaction yields [34]. It
for this project, two possibilities were considered: using the product
can be concluded that tripalmitin was 97.21 % converted, while the
stream composition data in order to make a yield reactor feasible or
other triglycerides as a whole were 72.15 % converted.
calculating the individual conversions of each reaction in the Aspen
To calculate the yield of each of the reactions, the following values
HYSYS® v12.0 simulator until the composition indicated for the product
were considered: percentage of the triglycerides with the same amount
stream in the literature was reached. In this scenario, due to the lack of
of carbons at the inlet; percentage of the formed hydrocarbon at the
articles on the production of green diesel from soybean oil using NiMo/
outlet; average percentage of the fraction at the outlet that corresponds
Al2O3 as a catalyst, both alternatives proved to be unfeasible, since the
to the products - 72.9 % - and the stoichiometry of the modelled re­
articles did not present data on the isomers and the liquid and vapor
actions (3:1). The calculation synthesized in Equation (1), where HC
product ratio in the hydrotreatment reaction step.
stands for “hydrocarbon”:All the mentioned reactions were added to a
The solution found was the comparison between the types of oil
first conversion reactor with a rank value (0 or 1): as both routes use the
commonly used as raw material, to select the one that had the closest
same reagent, the reactions occur in sequence, according to their rank
possible fatty acid profile to soybean oil, so that it was possible to search
for the necessary data in other articles. Based on Douvartzides et al.
[33], it is concluded that sunflower oil has the greatest similarity to Table 4
soybean oil, since both have trilinolein as their main constituents, fol­ Hydrogenation reaction yields [34].
lowed by triolein, representing more than 70 % of the total mass (
0.729
)
composition. Based on this, the work of Krár et al. [34] - on sunflower oil 3 ∗ HC percentage at outlet(%) ∗
HC molar mass
Yield(%) = (1)
- was used as a basis for data selection to calculate the conversions of ∑ Percentage of triglyceride in the inlet
each reaction, while the work of Veriansyah et al. [9] - on soybean oil - Molar mass of triglyceride

was used as a basis for the reaction conditions. The similarity between
the reaction conditions between these works allowed us to validate the
Triglyceride: Hydrocarbon: Reaction Calculated
adopted strategy. Inlet (%) Outlet (%) Yield (%)
In addition to hydrocarbons (C11–C19) formed by the hydrogenation
Tripalmitin: Pentadecane: Decarboxylation and 85.54
reaction - which represent between 72.7 and 73.1 % of the product -,
6.33 0.80 Decarbonylation
there is about 10.2 % of gases (including CO2, CO, NH3, H2S and hy­ Hexadecane: Hydrodeoxygenation 11.67
drocarbons such as C1–C4), 9.0 % of water and - without specifying the 6.25
amount of each one -, percentages of paraffins of 19 carbons or more, Tristearin: Heptadecane: Decarboxylation and 10.06
carboxylic acids and alkyl esters; this lack of specificity made the use of a 3.45 10.50 Decarbonylation
Triolein: 21.64
yield reactor unfeasible, as the compositions of all products were not Trilinolein:
known. Information on the concentration of light hydrocarbons (C5–C10) 67.78
allowed simulating the formation of products by hydrocracking, while Trilinolenin:
the concentrations of C16 and C18 paraffins served as a basis for calcu­ 0.09
Tristearin: Octadecane: Hydrodeoxygenation 62.09
lating the conversions of the hydrogenation reactions. The hydro­
3.45 68.60
deoxygenation reactions are described in Table 2 for each triglyceride. Triolein: 21.64
The decarboxylation and decarbonylation reactions are described in Trilinolein:
Table 3. As the hydrocarbon obtained in both reactions is the same for a 67.78
given triglyceride, they were combined into a single reaction and their Trilinolenin:
0.09
conversions were calculated based on the concentration of the C15 and

4
I.V. Barbosa et al. Renewable Energy 219 (2023) 119591

value; a value of 0 was adopted for those of hydrodeoxygenation. For the Adjust block.
decarbonylation/decarboxylation reactions, it was necessary to divide The hydrogenation reactor (R-100) has the function of converting
the yield by the percentage of unreacted triglycerides. Thus, for the Feed_Cooled stream into the Product stream, which is mixed with the
example, the input conversion values of the reactions involving light hydrocarbon stream (HC_light) in MIX-102 before entering the
tristearin were equal to 62.09 % and 26.53 %, respectively – it will result isomerization reactor, as described in section 3.3.2. The first reactor
in the 10.06 value given in Table 4. bottom outlet stream (Liquid) returns to the equipment through MIX-
Table 5 describes the isomerization reactions and the value of the 101: composed of the reactor condensate, the Liquid_back stream re­
isomer/paraffin ratio obtained at the end of the reaction. The isomeri­ duces the temperature of the feed stream. The conversions of each re­
zation reactions were added in a second conversion reactor, completing action in the isomerization reactor (R-101) are assured using Adjust
the hydrotreatment module, in order to facilitate the modeling of all the blocks, based on the isomer/paraffin ratio observed for each hydrocar­
hydrotreatment reactions. In this case, each conversion was calculated bon in the selected reference, according to Table 5. The isomerization
using an Adjust block: the target variable was the ratio of isomers in the reactor top outlet stream (Iso-product) is composed of the generated
product stream and the adjusted variable was the conversion of each hydrocarbons and other products: propane, water, carbon monoxide and
hydrocarbon. carbon dioxide, in addition to excess hydrogen and unreacted
triglycerides.
The separation stage is divided into three steps: the three-phase
3.5. Process description
separators, the PSA (pressure swing adsorption) unit and the distilla­
tion columns. The Iso-product stream goes to the last heat exchanger
This section presents the description of the process flowsheet,
(E− 102), where it is cooled to a temperature of 35 ◦ C, assuming the use
regarding its equipment units and stream’s identification. Additional
of cooling water. At this temperature, most of the water is removed from
information can be found in Tables S1 to S3 of the Supplementary Ma­
the gases before going to the first separator and most of the hydrocar­
terial, where the reader may access the stream properties and equipment
bons remain in the Aqueous phase stream, which favors product gener­
details.
ation in the plant. In the V-100 separator, the water is segregated into
Fig. 1 presents the flowsheet generated in Aspen HYSYS® v12.0 for
the Water stream, while the gaseous stream (Vapor) goes to the PSA unit
the green diesel production process. The hydrotreatment module is
and the other liquid stream goes to the second separator. While the first
indicated by a green square. For simulation purposes, the hydrotreat­
separator removes the water present in the product, in order to ensure
ment module is composed of a hydrogenation reactor and an isomeri­
that the water content in the green diesel is within the established
zation reactor, as explained in the previous section, but it constitutes a
standards, the second equipment removes the triglycerides not
single piece of equipment (although the physical plant will require more
consumed in the reactions.
than one vessel, as explained in the next section).
The V-101 separator is responsible for removing the denser fraction
The process starts with two inlet streams: the first corresponds to
of the product stream, after its decompression in the VLV-101 valve, into
soybean oil (Oil_Feed) and the second to the hydrogen make-up stream
the Residue stream. In addition, part of the gases present in the liquid
(H2_Feed). Both streams are at ambient conditions of pressure and tem­
stream due to the high pressure of the product, such as hydrogen and
perature, 30 ◦ C and 1 atm. Since the hydrogen/oil ratio at the reactor
propane, are removed in the Gas stream. Thus, the second separator
inlet must be kept equal to 450, there is a hydrogen recycle stream
operates at atmospheric pressure, while the first at 92 bar, in order to
(H2_Recycle) that mixes with the H2_Feed, which supplies the plant’s H2
reduce the loss of product in the vapor phase.
consumption. The ratio is calculated through a Spreadsheet block, hid­
Before the Vapor stream enters the PSA unit, it is depressurized, since
den in the flowsheet, based on the Feed stream, and is used in the
the unit operates at ambient conditions: 35 ◦ C and 1 atm. This unit (X-
calculation of the hydrogen make-up flow rate, through an Adjust block
100) contains four columns, loaded with selective adsorption materials
(Flow H2).
(zeolite 13X and activated carbon) and is responsible for separating the
The H2 stream leaving MIX-100 is compressed in the first hydrogen
hydrogen gas, which is recovered back into the process, free of other
compressor (C-101) up to a pressure of 4.56 bar and in the second
gases. The impurities removed in the PSA unit are: CO, CO2, propane,
compressor (C-102) up to 20.52 bar. Then it is cooled in the E− 100 heat
water vapor and part of the gaseous products, such as n-pentane.
exchanger in order to reduce the energy consumption required in the
Coming from the second separator, the Liquid product stream flows
third compression stage. Thus, the H2_out stream is compressed in the
towards the first distillation column (T-100), which operates at atmo­
third compressor (C-103), reaching the final pressure of 92 bar, neces­
spheric pressure. The separation process in this equipment generates
sary for the reactions.
two streams: Vapor.out, composed of light hydrocarbons (for example, n-
The Oil_Feed stream enters the process towards pump P-100, having
pentane and propane), and Liquid.out, composed of C15 to C18 hydro­
its pressure raised to 92 bar, the operating pressure of the hydrotreat­
carbons and their isomers, and n-decane. The latter goes to a second
ment module. The compressed stream Oil_P1 goes to MIX-101, where it
distillation column (T-101), leading to the production of two main
is mixed with the stream of hydrogen (H2) and condensate from R-100,
product streams: Green Diesel and Light product (which represents
forming the Feed stream, which needs to be cooled from 487 ◦ C to 400 ◦ C
biokerosene).
in the second heat exchanger (E− 101) before entering the reactor. Thus,
the Feed_Cooled stream is at 400 ◦ C and 92 bar.
4. Equipment sizing
Both reactor blocks operate under the same conditions - 92 bar and
400 ◦ C - as they constitute a single, isobaric and isothermal piece of
In this section, the methodology used for equipment sizing is
equipment. The reactions do not change the pressure, only the tem­
described with the respective references used to support it. All the
perature of the system: to keep it constant, there is a heat stream and an
remaining information was obtained from the simulations results -
process stream conditions and specifications. The equipment sizes and
Table 5 other details can be consulted in Table S3 of the Supplementary
Isomerization reactions [34].
Material.
Reagent → Product Structure Isomer/Paraffin Ratio The reactors, represented in the simulation environment by the two
Pentadecane → 2-methyl-tetradecane C15H32 0.45 conversion reactor blocks, in an industrial plant are equivalent to a set of
Hexadecane → 2-methyl-pentadecane C16H34 0.48 continuous stirred tanks. For their design, a space time (τ) of 2 h was
Heptadecane → 2-methyl-hexadecane C17H36 0.51 assumed for the temperature and pressure conditions of 400 ◦ C and 9.2
Octadecane → 2-methyl-heptadecane C18H38 0.54
MPa [9]. The reactor volume is determined form the process flow rate.

5
I.V. Barbosa et al. Renewable Energy 219 (2023) 119591

Fig. 1. Green diesel production plant flowsheet in Aspen HYSYS® v12.0.

The flow rate considered in the calculation was the liquid phase of the bar, the ratio is 3:1. These are the ratios used for the first and second
feed stream plus a percentage of 20 %, in order to represent the space separators, respectively. Furthermore, the liquid level will depend on
occupied by the vapor phase. It was considered a cylindrical shape and a the holdup - minimum time necessary for the equipment to reach the
height/diameter ratio of 3:1. The number of reactors was chosen to limit normal operating level -, suggested by the authors as 10 min.
the height to approximately 10 m. It resulted in three parallel reactors
with 3.24 m of diameter and 10.27 m of height. 5. Economic analysis
For the heat exchangers design, it was used the methodology and
heuristic rules proposed by Seider et al. [35]. The global heat transfer This section aims to describe the methodology followed to estimate
coefficient was extracted from the same book and depend on the species the investment necessary for the construction of a green diesel produc­
involved in the heat exchange. The heat transfer rate was obtained tion plant, as well as to evaluate its production costs, revenues and
directly from the Aspen HYSYS® v12.0 simulator. Exchangers with a profitability. A cash flow analysis was performed to evaluate the eco­
heat exchange area of less than 200 ft2 were designed as double-tube, nomic feasibility of the process proposed in this work. The economic
while exchangers with an area of more than 200 ft2 were designed as analysis was situated in the Brazilian economic scenario at the end of
a shell and tube with a floating head. Condensers and reboilers of the 2021/beginning of 2022, so that it was not influenced by the geopolit­
distillation columns were considered common heat exchangers. ical context that resulted in large variations in the grain and fuel markets
Stainless-steel serpentines were assumed for the reactors. from the third month of 2022 onwards.
Distillation columns were designed as packed columns. Thier di­
ameters were calculated with the Leva Method, as described by Seider
et al. [35], and its height was calculated through the HETP 5.1. Capital cost
methodology.
In order to obtain hydrogen with maximum purity from the reaction The methodology used to estimate the capital cost was the Lang
product stream, it is necessary to use a PSA unit. Considering the pres­ Method, detailed by Seider et al. [35]. First, the FOB (free on board)
ence of H2, CO2, CO, H2O and light hydrocarbons in the stream, two prices of the equipment must be obtained. They can be calculated by
layers of adsorbents were considered. The first is responsible for empirical formulas present in the literature. As these correlations have
removing heavy contaminants, including water, light hydrocarbons and already been determined for a considerable time ago, it is necessary to
CO2, and it is composed by activated carbon. Above it, a second layer, update the results. For this, the obtained value is multiplied by the Plant
composed of zeolite 13X, removes the other contaminants [36]. In this Cost Index, a factor periodically published by the Chemical Engineering
way, the obtained hydrogen stream is in a high degree of purity to be Magazine. The sum of the updated prices is performed and multiplied by
recycled back to the beginning of the process. a factor defined as 1.05, which considers the transport of the equipment
The PSA unit was sized with four vessels, operating in an interleaved to the plant site. Finally, the obtained value is multiplied by the Lang
manner. Its design was carried out based on the desired CO2 and CO factor (fL), resulting in the fixed investment. This methodology is sum­
removals (in terms of volume flows), with CO being removed by zeolite marized in Equation (2):
13X and CO2 by activated carbon. In this sense, it was possible to ∑ PCI
calculate the volume of adsorbents needed for the individual removals IF = 1.05xfL x Ceq,i (2)
PCIb
and, from it, the volume of each column, as well as its height and i

diameter. For the zeolite 13X adsorbent, it was considered a pore size of
where IF is the fixed investment, fL is the Lang factor, PCI is the Plant
1 nm, a surface area of 0.6 km2/kg, a density of 0.58 kg/L and an
Cost Index for the project year; PCIb is the Plant Cost Index for the base
adsorption capacity of 0.25 kg per kg of adsorbate [37]. While for
year; and Ceq,i is the FOB cost of equipment i. All the formulas used to
activated carbon, it was considered a pore size of 2 nm, a surface area of
obtain the equipment prices in this work were taken from Seider et al.
0.8 km2/kg, a density of 0.45 kg/L and an adsorption capacity of 0.25
[35] and refer to 2013 data, so PCIb is 567. The PCI value used to update
[38].
these prices was 754.7, relative to September 2021.
The process requires two three-phase separation vessels. They were
The Lang factor takes into account installation, control and instru­
sized according to the methodology proposed by Towler and Sinnott
mentation costs, construction and engineering costs, plumbing, elec­
[39]. As they are horizontal separators, the diameter is dependent on the
trical network, auxiliary constructions and involved fees. For a plant
vessel’s length. Therefore, for an operating pressure greater than 35 bar,
that only processes fluids, the value of fL is 5.04. To get the total capital
the length/diameter ratio is 5:1, while for pressures between 0 and 20
cost, fixed investment is summed up with the working capital, estimated

6
I.V. Barbosa et al. Renewable Energy 219 (2023) 119591

at 15 % of the total capital cost. disposal was determined according to Turton et al. [42].
A few observations are important. First of all, two sequences of The operating labor cost was calculated by multiplying the average
centrifugal compressors operating in parallel were considered, each of salary of the operators by the number of workers required per shift. This
them with three units in series, meeting the maximum compression ratio amount can be calculated by Equation (3) [42]:
of six, but all of them with an operational head higher than the limit for ( )0,5
which the cost equation is valid. In the absence of equations covering Nwk = 6.29 + 31.7Ns 2 +0.23Nns (3)
higher heads, it was used, but it is a known source of uncertainty, since
the compressors correspond to the most significant part of the equip­ where Nwk is the number of workers per shift, Ns is the number of
ment costs. equipment that operate with solids and Nns is the number of equipment
Second, the considered PSA adsorbents have the following costs and that do not involve particulate solids handling, which includes reactors,
lifespans: the indicated exchange time for zeolite 13X is six months and columns and heat exchangers. The remaining production costs were
its price is US$ 61.80/kg [37]; the indicated exchange time for activated estimated with the methodology proposed by Turton et al. [42], pre­
carbon is 12 months and it costs US$ 9.26/kg [38]. sented in Table 7.
Third, as for the separation of the catalyst used in the reactor, the
cost of a battery of cyclones was calculated. It is not present in the 5.3. Cash flow and economic evaluation criteria
simulation, since the catalyst was only considered as a cost in the eco­
nomic analysis. And there was no need for sizing it, since its cost can be Based on the production costs, it is possible to carry out a profit­
estimated based on the inlet flow rate. ability analysis of the process, but this is not enough to assess whether
the process is in fact economically viable, as there is no information on
the compensation of the investment made. For this, a cash flow analysis
5.2. Production cost must be carried out. The cash flow is an evaluation of all the company’s
cash inflows and outflows, from the initial investment to the last year of
The total production cost can be divided into direct costs, indirect plant operation, which was considered 20 years in this work.
costs and general expenses. The first one involves operating expenses The fixed investment cost is expended from the first year until the
directly linked to the production level of the plant, such as the cost of end of the project’s construction phase, which, depending on the di­
raw materials, catalysts, solvents, utilities, effluent disposal, mainte­ mensions of the plant, can vary from six months to three years [42]. In
nance and repairs, operating labor, laboratory charges, operating sup­ this work, a period of three years of construction was assumed, being
plies, patents and royalties. Indirect costs are production costs not applied the division of 30 %, 40 % and 30 % of the fixed investment
directly associated to the production level of the plant. These are
packaging and storage costs, taxes and insurance. General expenses, on
Table 7
the other hand, involve administrative costs, products distribution and
Economic results.
sale, and research and development costs. The production cost calcula­
Capital Costs (US$)
tion was performed considering a production factor of 7920 h/year,
which was obtained taking into account that the plant operates 330 days Total Equipment Cost 81,747,170.88
a year, 24 h a day. Fixed Investment (IF) 432,606,028.28
Working Capital (CWK = IT) 76,342,240.28
The main component prices are shown in Table 6. In the base sce­
Total Investment (IT = IF + CWK) 508,948,268.56
nario, it was considered the raw material prices found in the Brazilian Direct costs (US$/year)
market at the end of 2021/beginning of 2022, and, for the products, it Raw materials
was considered the prices of conventional diesel and kerosene, in order Soybean oil 697,426,329.60
Hydrogen 22,005,640.80
to infer the competitiveness of the biofuels in this market. The price of
Catalyst 30,443,212.80
the NiMo/Al2O3 catalyst was found on the market for laboratory scales Utilities
only. However, NiMo and CoMo catalysts, using alumina as a support, Cooling Water 62,054.43
have similar costs on this scale. Therefore, the industrial-scale CoMo/ Fired Heat 658.93
Al2O3 value provided by Zibo Yinghe Chemical Co., Ltd. was considered. Refrigerant 1 0.08
Adsorbents 3,072.12
Catalyst activation costs were not considered, although some kind of
Electricity 251,088,867.23
sulfide activation may be necessary [10,22,40]. The prices of the utilities Effluent Disposal 1,531,864.22
were obtained from the Aspen HYSYS® v12.0 simulator. The cost of Operating Labor (COL) 49,239.41
electricity was determined according to ENEL [41]. The cost of effluent Supervisory and Clerical Labor (CSC = 0.18 COL) 8,863.09
Maintenance and Repairs (CMR = 0.06 IF) 25,956,361.70
Operating Supplies (COS = 0.15 CMR) 3,893,454.25
Table 6 Laboratory Charges (CLC = 0.15 COL) 7,385.91
Component prices. Patents and Royalties (CPR = 0.03 CTP) 39,072,961.31
Subtotal 1,071,549,965.88
Raw Materials, Products, Unit Price Reference
Indirect costs (US$/year)
Catalyst and Utilities
Overhead, Packaging and Storage (COPS = 0.60 (COL + CSC + 15,608,678.52
Soybean Oil US$/kg 1.5120 [43] CMR))
Hydrogen US$/kg 1.5950 [44] Local Taxes (LT = 0.032 IF) 13,843,392.90
Diesel US$/L 0.9767 [45] Insurance (IN = 0.005 IF) 2,163,030.14
Kerosene US$/L 0.7398 [45] Subtotal 31,615,101.56
NiMo/Al2O3 US$/kg 1.5000 Zibo Yinghe Chemical General Expenses (US$/year)
Co., Ltd Administrative Costs (CAD = 0.25 COPS) 3,902,169.63
Cooling Water US$/kJ 6.51 × Aspen HYSYS® v12.0 Distribution and Sellings Costs (CDS = 0.10 CTP) 130,243,204.36
10− 5 Research and Development (CRD = 0.05 CTP) 65,121,602.18
Fired Heat US$/kJ 1.78 × Aspen HYSYS® v12.0 Subtotal 199,266,976.17
10− 5 Total Production Cost (CTP) (US$/year) 1,302,432,043.61
Refrigerant 1 US$/kJ 2.74 × Aspen HYSYS® v12.0 Revenues (US$/year)
10− 6 Green Diesel 679,181,023.96
Electricity US 0.1208 [41] Biokerosene 2,398,519.45
$/kWh Subtotal 681,579,543.41
Effluent Disposal US$/ton 0.0360 [42] Annual Net Profit (US$/year) − 620,852,500.20

7
I.V. Barbosa et al. Renewable Energy 219 (2023) 119591

between these years. Working capital is expended in the fourth year, and CO2 gases, in addition to H2O, propane and pentane in the vapor
which would be the first year of operation. phases. The Gas stream still has a small amount of H2 that was not
The cash flow is the difference between the annual production costs completely separated by the first three-phase separator.
and the revenue obtained with all the products. A product revenue
calculation is the product between its sale price and the amount pro­
duced in a year, which is obtained from the process simulation. The 6.2. Economic results for a new green diesel production plant
implicit hypothesis in this calculation is that it is assumed that 100 % of
what is produced of a given product will be sold. Over the gross profit, it The economic results obtained for the green diesel production pro­
was also considered some Brazilian taxes, such as the Social Contribu­ cess are summarized in Table 7. The first thing that can be concluded is
tion on Net Income and the Income Tax. The result is the net annual that the total investment cost is compatible with those reported in the
profit. literature. For example, Glisic et al. [20] obtained a total equipment cost
With the cash flows obtained for the 3 years of construction and 20 of US$ 34,496,000.00 when simulating a plant with a production ca­
years of operation, it is possible to calculate de net present value (NPV), pacity of 100,000 tons/year. It is about four times smaller than the one
according to Equation (4): simulated in this work. Applying the “six tenths” rule of economy of
scale, it would result in approximately US$ 80 million for a plant of
∑
23
(Cash Flow) similar size, without considering price update, which is very close to the
NPV = t
(4)
t=1 (1 + i)t value obtained in this work.
An analysis of the equipment costs shows that the most expensive
where i is the interest rate and t is the period of time. part of the plant are hydrogen compressors, which account for 92 % of
NPV is always a comparison with another investment, with a known the total equipment cost. This was already indicated in the literature
and constant interest rate: the so-called minimum attractive rate of re­ [20]. The reactors appear in second place, accounting for around 7 %.
turn (MARR). In this work, a MARR of 10 % was considered in NPV The other equipment units contribute with negligible values.
calculations. If NPV is positive, investing in the process is more advan­ Fig. 2 shows the production cost distribution. It can be seen that raw
tageous than this other possible investment. materials correspond to 55 % of the production costs. Of these, 97 %
The payback period is an indicator used to assess the viability of an come from soybean oil acquisition. It is due to the expressive mass flow
investment. It corresponds to the time it takes to obtain a positive NPV. rate of soybean oil in the process compared to the flow rate of the H2
Payback is important because, even if the NPV is positive for a certain make-up stream, since most of the hydrogen is being recycled. It is also
venture, economic viability is not sustained with very long payback important to highlight that soybean oil has presented historically high
periods, which do not attract investors. For this work, a payback period prices in Brazil in the last years [47]. In the second place comes elec­
of 5 years of operation was initially stipulated as the limit for economic tricity, with 19 % of the production costs. It is justified by hydrogen
feasibility. compressing and is much more than what was estimated by Glisic et al.
[20]. Although catalyst reuse was not considered, its contribution to the
6. Results and discussion total production cost is not significant, therefore there is much room for
catalyst development.
6.1. Simulation results
6.2.1. Equilibrium prices for a five-year payback
The produced green diesel presents, in terms of mass fractions, 73 % As can be seen, in this base scenario, the process is not profitable.
of C18, 11 % of C17, 9 % of C16 and 1 % C15. In the case of isomers, it is Therefore, it is important to determine the equilibrium prices for green
possible to notice their low expressiveness in the product composition, diesel and its main raw material, soybean oil. In this work, except in the
accounting for the remaining 5 %. It was expected, since the soybean oil
reaction yields were approximated by sunflower oil reaction yields, as
described in section 3.4. According to Krár et al. [34], the product is
obtained with fewer isomers in its composition than would be obtained
with soybean oil. This factor directly impacts the cold flow properties of
the biofuel, but does not impact the economic analysis.
Another important point is the absence of water in the final product,
due to the efficient separation of water in the first three-phase separator,
being below the limit of 200 mg/kg, as specified in ANP Resolution No.
842/2021 [46] (ANP is the Brazilian National Agency of Petroleum,
Natural Gas and Biofuels). The characteristics of the obtained product
are consistent with the product specifications - such as a specific mass of
782.2 kg/m3 (at 20 ◦ C) and kinematic viscosity of 3.654 cSt (at 40 ◦ C).
In addition, the second distillation column made it possible to obtain
kerosene as a by-product with adequate purity, which was also consid­
ered among the plant’s revenues. However, it is clear that its volume
flow is much lower than that of green diesel, with little impact on the
total revenue.
It is worth mentioning that naphtha could be obtained by adding a
third distillation column in the process, which would separate it from
the other components of the Vapor_out stream. However, as naphtha is
lighter than green diesel and kerosene, it was lost in gaseous form
through the Purge and Gas outlet streams. In addition, it was decided not
to do so, given the low mass flow rate of the Vapor_out stream - of only
100.5 kg/h -, which would have a small impact on the total revenue.
In the case of process effluents, the Residue stream basically consists
of unreacted triglycerides, while the Purge and Gas streams contain CO Fig. 2. Production cost distribution.

8
I.V. Barbosa et al. Renewable Energy 219 (2023) 119591

cases analyzed in the next section, the considered equilibrium prices that the capital cost can be considered equal to zero (but keeping CMR,
were those which would make the payback time to be 5 years of oper­ COS, LT and IN). According to the 2021 Statistical Yearbook by ANP
ation (8th year of the cash flow), keeping all the other costs constant. In [50], the production of conventional diesel in nine Brazilian refineries
this case, the equilibrium price of green diesel should be US$ 2.08/L to averaged 4776.31 m3/day between 2011 and 2020, so the green diesel
compensate for the investment made (Scenario A in Fig. 3). This value is production capacity analyzed in this work fits into a scenario of an
2.13 times the cost of S-10 diesel in Brazil in the considered economic existing refinery. Therefore, in this case, the equilibrium price of green
scenario [45]. Still, when comparing with biodiesel, whose value in the diesel must be calculated so that the process is simply profitable. The
Brazilian market was US$ 1.27/L in the considered scenario [48], it is possible cost reductions with the adoption of a pre-existing refinery
1.64 times more expensive. This shows how uncompetitive green diesel infrastructure is shown in Fig. 3.
is against conventional diesel and biodiesel at this moment. The sale price of green diesel that meets this requirement is US$
As green diesel does not yet have a considerable demand in Brazil, 1.87/L (Scenario B in Fig. 3). It is still 1.91 times the cost of S-10 diesel in
the average price obtained in the European market can be used as a basis Brazil, which means that green diesel would still not be competitive with
for comparison. It is US$ 1.60/L [49]. It was then analyzed how much conventional diesel, but it represents a 10 % reduction from the original
cheaper the value of soybean oil should be so that the sale price of green value. When comparing with biodiesel, it is also not competitive, as it
diesel was competitive with the European market and resulted payback represents 1.47 times its market value. However, to compete with green
in 5 years. It was concluded that the cost of soybean oil should be 40 % diesel in the European market, the value of soybean oil should be 78 % of
lower than that found in the Brazilian market at this moment, which is its considered market price, which is compatible with the national price
not too much if we consider recent price increases: the value of soybean history oscillation, as can be seen in Fig. 4. Alternatively, soybean oil
oil in 2020 was US$ 0.72/kg in Brazil and escalated quickly to histori­ could be substituted by a non-edible oil, or some other cheaper raw
cally high levels due to many factors [47], as can be seen in Fig. 4. Green material. If it could be acquired for about US$ 1.15/kg, this would make
diesel would be absolutely possible in this context. However, in the green diesel competitive in this scenario.
considered scenario, it would require government subsidy. Fig. 4 shows historical prices for soybean oil in Brazil. The data is
from Comex Stat [51], the official database for Brazilian foreign trade
6.2.2. Equilibrium prices for longer payback periods statistics. The data was generated for the period from 2010 to 2023
Although a five-year period is commonly accepted as a reasonable (until September) and corresponds to the import and export averages for
term for the return of the investment, it is convenient to analyze how the entire year. The soybean oil equilibrium prices discussed in the
much lower could be the equilibrium price of green diesel, or how much previous paragraphs for payback in 5 years with a stand-alone plant and
higher could be the price of soybean oil if longer payback times were with refinery infrastructure are indicated in the figure, demonstrating
considered, because any new stand-alone green diesel production plant the viability of both alternatives depending on the economic scenario.
is expected to last much longer than 5 years. Therefore, the equilibrium Due to the already known impact of the electricity cost on the total
prices for these components were evaluated for payback periods of 10, production cost of the plant, it could be considered an additional sce­
15 and 20 years, and the results are presented in Table 8. nario with the reduction of this value arising from energy co-generation
From this analysis it is possible to roughly conclude that long-term from the Vapor_out stream. It contains mostly propane and pentane, and
investments in stand-alone green diesel production plants would be a LHV (lower heating value) of 4.57 × 104 kJ/kg. It resulted in an energy
justified for biofuel prices above US$ 2.00/L and raw material prices cost reduction of 0.5 %, not affecting the equilibrium price of the
below US$ 1.00/kg. This is a convenient reference for public policies to product. However, assuming that other energy sources may be available
be stablished in this sense. If these values are not competitive with the in the pre-existing refinery infrastructure, other scenarios could be
considered market values, an alternative is to reduce the capital costs analyzed. For example, considering 25 % energy savings (Scenario C in
associated with the construction of a new facility by making use of pre- Fig. 3), the equilibrium price of green diesel drops to US$ 1.76/L, and,
existing refinery infrastructure. based on the value of diesel in the European market, soybean oil could
be acquired at 87 % of its base cost. But it would require 61 % energy
savings for the cost of green diesel to drop to US$ 1.60/L, which would
6.3. Green diesel production in a pre-existing refinery be finally competitive in the European market. It shows the importance
to search for alternative electric energy sources or to develop more
For this analysis, it is assumed that the existence of a previous efficient hydrogen compression systems for the green diesel production
infrastructure dispenses any investment - most optimistic analysis -, so

Fig. 3. Green diesel equilibrium prices in different scenarios (payback in 5 years).

9
I.V. Barbosa et al. Renewable Energy 219 (2023) 119591

Fig. 4. Price history of soybean oil in Brazil [51] and required prices for payback in 5 years.

green diesel production process is not economically viable. The

Table 8
hydrogen compression system, and the high cost of vegetable oil and
Equilibrium prices of green diesel and soybean oil for different payback periods
electricity can be pointed out as the main factors for this result. Process
of a stand-alone production plant.
feasibility would therefore be conditional on government support for
Payback Period (years) 5 10 15 20
changing the current energy scenario, and on the market’s interest in a
Green Diesel Price (US$/L) 2.08 2.00 1.98 1.96 greener, although more expensive, product. However, considerable
Soybean Oil Price (US$/kg) 0.91 1.02 1.05 1.06 price reductions can be achieved through the use of the existing petro­
leum refining infrastructure - of up to 10 % without considering eventual
plant. energy savings -, which appears as an interesting transition route.
With this, the potential economic benefit of taking advantage of the This article presents the necessary data for its completely repro­
infrastructure of a pre-existing petroleum refinery for vegetable oil duction, and the economic analysis can be easily updated and adapted to
hydroprocessing was demonstrated. That is why, as discussed in section other countries, by associating the corresponding market values.
2, many industrial initiatives have been following this strategy. Although Brazil has recognized expertise in biofuels and is developing
Evidently, the assumption of “zero investment” is an ideal case. For this projects of this type in order to remain at the forefront of the sector,
reason, it is recommended to carry out additional studies in order to green diesel is expected to soon become a worldwide reality, and co-
assess the actual investment needed in this type of application, i.e. how processing seems to be an interesting starting option, especially for pe­
much a conventional hydroprocessing unit of a petroleum refinery must troleum producing countries.
be adapted to receive vegetable oil. This work faced some limitations regarding the absence of experi­
In addition, it is interesting to analyze how much green diesel could mental and process simulation data of soybean oil hydrogenation over
be added to conventional diesel in a petroleum refinery in order to keep NiMo/Al2O3 catalyst in the literature, as well as the absence of data from
it competitive in the Brazilian market. According to the literature, it is commercial plants for results comparison. Besides further investment on
common to find percentages of green diesel in the blend with petroleum these topics, as suggestions for future works, it is proposed to look for
diesel of up to 30 % [52], and the optimal proportion would be of less alternative sources of electricity supply and to optimize the number of
than 50 % [53]. If we take into account the price of conventional diesel hydrogen compression stages and its conditions. It would also be
in Brazil in 2019 - US$ 0.84/L, adjusted for inflation [54] - and consider interesting to perform simulations of soybean oil co-processing scenarios
an eventual return to this level in the next few months, it is possible to with gasoil, and analyze its economic viability. The possible separation
estimate how much green diesel could be added to conventional diesel in of propane from the light hydrocarbon stream, in order to generate
order to keep its market value at US$ 0.98/L, which is the price of diesel naphtha as an additional by-product, should be evaluated as well.
in the considered Brazilian market [45] (both retail values). It would be It is also important to mention that the current Brazilian National
13 % (Scenario D in Fig. 3) without energy savings and 15 % (Scenario E Biofuels Policy (RenovaBio) foresees carbon crediting as a means of
in Fig. 3) with 25 % energy savings (assuming linear contributions). adding value to the renewable fuels’ production chains, but it does not
Therefore, there is an opportunity for gradual green diesel insertion in yet consider green diesel. Besides that, the formal implementation of a
the Brazilian fuels supply chain, which will of course depend on the carbon market in the country is being discussed, but there are still no
international markets of oil and grains. defined metrics. Therefore, we suggest that the effects of possible carbon
credits or taxes on the green diesel equilibrium price be evaluated in
7. Conclusions and suggestions for future works future works, as a way of ensuring its competitiveness in an ever
evolving market.
In this work, the process of producing green diesel (HVO) from
soybean oil hydrogenation was simulated in Aspen HYSYS® v12.0. With CRediT authorship contribution statement
the rigorous mass and energy balance results, the equipment was sized,
the economic feasibility of a new plant was analyzed, and the potential Ian V. Barbosa: Methodology, Investigation, Writing – original
gain with the use of a pre-existing infrastructure of a petroleum refinery draft, Writing – review & editing. Letícia A. Scapim: Methodology,
was assessed. The capacity of the industrial plant was defined based on a Investigation, Writing – original draft, Writing – review & editing.
real project of a biofuel plant in Brazil, and the economic scenario at the Raquel M. Cavalcante: Resources, Supervision. André F. Young:
end of 2021/beginning of 2022 was taken as case study. Conceptualization, Methodology, Writing – original draft, Writing –
It is concluded that, under this economic scenario, the stand-alone review & editing, Visualization, Supervision.

10
I.V. Barbosa et al. Renewable Energy 219 (2023) 119591

Declaration of competing interest [20] S.B. Glisic, J.M. Pajnik, A.M. Orlović, Process and techno-economic analysis of
green diesel production from waste vegetable oil and the comparison with ester
type biodiesel production, Appl. Energy 170 (2016) 176–185, https://doi.org/
The authors declare that they have no known competing financial 10.1016/j.apenergy.2016.02.102.
interests or personal relationships that could have appeared to influence [21] C. Choe, B. Lee, H. Lim, Sustainable and carbon-neutral green diesel synthesis with
the work reported in this paper. thermochemical and electrochemical approach: techno-economic and
environmental assessments, Energy Convers. Manag. 254 (2022), 115242, https://
doi.org/10.1016/j.enconman.2022.115242.
Acknowledgement [22] A.V. Lavrenov, E.N. Bogdanets, Y.A. Chumachenko, V.A. Likholobov, Catalytic
processes for the production of hydrocarbon biofuels from oil and fatty raw
materials: contemporary approaches, Catalogue Index 3 (2011) 250–259, https://
The authors would like to thank the Carlos Chagas Filho Foundation doi.org/10.1134/S2070050411030044.
for Research Support of the State of Rio de Janeiro (FAPERJ) for [23] J. Calero, D. Luna, E.D. Sancho, C. Luna, F.M. Bautista, A.A. Romero, et al., An
financial support (Grant N◦ E-26/211.234/2019). overview on glycerol-free processes for the production of renewable liquid
biofuels, applicable in diesel engines, Renew. Sustain. Energy Rev. 42 (2015)
1437–1452, https://doi.org/10.1016/j.rser.2014.11.007.
Appendix A. Supplementary data [24] S. Bezergianni, A. Dimitriadis, Comparison between different types of renewable
diesel, Renew. Sustain. Energy Rev. 21 (2013) 110–116, https://doi.org/10.1016/
j.rser.2012.12.042.
Supplementary data to this article can be found online at https://doi. [25] A.A. Porsin, E.N. Vlasova, G.A. Bukhtiyarova, A.L. Nuzhdin, V.I. Bukhtiyarov,
org/10.1016/j.renene.2023.119591. Sulfide catalysts for production of motor fuels from fatty acid triglycerides, Russ. J.
Appl. Chem. 91 (2018) 1905–1911, https://doi.org/10.1134/
S1070427218120017.
References
[26] C.A.S. ChemSrc, Number search. https://www.chemsrc.com/en/. (Accessed 9
January 2022).
[1] J.C. Serrano-Ruiz, E.V. Ramos-Fernández, A. Sepúlveda-Escribano, From biodiesel [27] A.C. Petcu, V. Plesu, C. Berbente, Estimation methods for thermophysical
and bioethanol to liquid hydrocarbon fuels: new hydrotreating and advanced properties of Camelina sativa crude oil, UPB Sci. Bull. Ser. B Chem. Mater. Sci. 78
microbial technologies, Energy Environ. Sci. 5 (2012) 5638–5652, https://doi.org/ (2016) 59–70.
10.1039/c1ee02418c. [28] Cheméo. Chemical & Physical Properties. https://www.chemeo.com/(accessed
[2] A.F. Young, F.L.P. Pessoa, E.M. Queiroz, Biodiesel Production Technologies: January 9, 2022)..
Supercritical and Enzymatic Production, Novas Edições Acadêmicas, Saarbrücken, [29] M. Plazas-González, C.A. Guerrero-Fajardo, J.R. Sodré, Modelling and simulation
Germany, 2015. of hydrotreating of palm oil components to obtain green diesel, J. Clean. Prod. 184
[3] S. Boonyasuwat, J. Tscheikuna, Co-processing of palm fatty acid distillate and light (2018) 301–308, https://doi.org/10.1016/j.jclepro.2018.02.275.
gas oil in pilot-scale hydrodesulfurization unit over commercial CoMo/Al2O3, Fuel [30] J. Hancsók, T. Kasza, S. Kovács, P. Solymosi, A. Holló, Production of bioparaffins
199 (2017) 115–124, https://doi.org/10.1016/j.fuel.2017.02.069. by the catalytic hydrogenation of natural triglycerides, J. Clean. Prod. 34 (2012)
[4] Automotive Business. Euro 6/P8 Is Defined by Conama for 2022/23, 2018. http 76–81, https://doi.org/10.1016/j.jclepro.2012.01.036.
s://www.automotivebusiness.com.br/pt/posts/noticias/euro-6-p8-e-definido-pe [31] M.K.A. Hamid, HYSYS: an Introduction to Chemical Engineering Simulation: for
loconama-para-2022-23/. (Accessed 12 December 2020) (in Portuguese). UTM Degree++ Program, University of Technology Malaysia, Skudai, Malaysia,
[5] J. Guo, G. Xu, F. Shen, Y. Fu, Y. Zhang, Q. Guo, Catalytic conversion of Jatropha oil 2013.
to alkanes under mild conditions with a Ru/La(OH)3 catalyst, Green Chem. 17 [32] M. Nogueira, BBF and Vibra Announce Brazil’s 1st Green Diesel Biorefinery for
(2015) 2888–2895, https://doi.org/10.1039/C4GC02406K. 2025, ISTOÉ, 2021 (in Portuguese).
[6] F.P. Sousa, L.N. Silva, D.B. de Rezende, L.C.A. de Oliveira, V.M.D. Pasa, [33] S.L. Douvartzides, N.D. Charisiou, K.N. Papageridis, M.A. Goula, Green diesel:
Simultaneous deoxygenation, cracking and isomerization of palm kernel oil and biomass feedstocks, production technologies, catalytic research, fuel properties and
palm olein over beta zeolite to produce biogasoline, green diesel and biojet-fuel, performance in compression ignition internal combustion engines, Energies 12
Fuel 223 (2018) 149–156, https://doi.org/10.1016/j.fuel.2018.03.020. (2019), https://doi.org/10.3390/en12050809.
[7] T. Barreiros, A. Young, R. Cavalcante, E. Queiroz, Impact of biodiesel production [34] M. Krár, S. Kovács, D. Kalló, J. Hancsók, Fuel purpose hydrotreating of sunflower
on a soybean biorefinery, Renew. Energy 159 (2020) 1066–1083, https://doi.org/ oil on CoMo/Al2O3 catalyst, Bioresour. Technol. 101 (2010) 9287–9293, https://
10.1016/j.renene.2020.06.064. doi.org/10.1016/j.biortech.2010.06.107.
[8] N. Arun, R.V. Sharma, A.K. Dalai, Green diesel synthesis by hydrodeoxygenation of [35] W.D. Seider, D.R. Lewin, J.D. Seader, S. Widagdo, R. Gani, K.M. Ng, Product and
bio-based feedstocks: strategies for catalyst design and development, Renew. Process Design Principles: Synthesis, Analysis and Evaluation, fourth ed., John
Sustain. Energy Rev. 48 (2015) 240–255, https://doi.org/10.1016/j. Wiley & Sons, Inc., New York, United States, 2017.
rser.2015.03.074. [36] F.V.S. Lopes, C.A. Grande, A.M. Ribeiro, V.J.P. Vilar, J.M. Loureiro, A.E. Rodrigues,
[9] B. Veriansyah, J.Y. Han, S.K. Kim, S.-A. Hong, Y.J. Kim, J.S. Lim, et al., Production Effect of ion exchange on the adsorption of steam methane reforming off-gases on
of renewable diesel by hydroprocessing of soybean oil: effect of catalysts, Fuel 94 zeolite 13X, J. Chem. Eng. Data 55 (2010) 184–195, https://doi.org/10.1021/
(2012) 578–585, https://doi.org/10.1016/j.fuel.2011.10.057. je900303y.
[10] L.M. Orozco, D.A. Echeverri, L. Sánchez, L.A. Rios, Second-generation green diesel [37] A.F. Young, H.G.D. Villardi, L.S. Araujo, L.S.C. Raptopoulos, M.S. Dutra, Detailed
from castor oil: development of a new and efficient continuous-production process, design and economic evaluation of a cryogenic air separation unit with recent
Chem. Eng. J. 322 (2017) 149–156, https://doi.org/10.1016/j.cej.2017.04.027. literature solutions, Ind. Eng. Chem. Res. 60 (2021) 14830–14844, https://doi.org/
[11] J. Kann, J. Kriis, H. Rang, Advances in biodiesel fuel research, Proc. Est. Acad. Sci. 10.1021/acs.iecr.1c02818.
Chem. 51 (2002) 75, https://doi.org/10.3176/chem.2002.2.01. [38] J.L. Sorrels, A. Baynham, D.D. Randall, K.S. Schaffner, Carbon Adsorbers, 2018. htt
[12] S.-Y. No, Application of hydrotreated vegetable oil from triglyceride based biomass ps://www.epa.gov/sites/default/files/2018-10/documents/final_carbonad
to CI engines – a review, Fuel 115 (2014) 88–96, https://doi.org/10.1016/j. sorberschapter_7thedition.pdf. (Accessed 9 January 2022).
fuel.2013.07.001. [39] G.P. Towler, R.K. Sinnott, Chemical Engineering Design: Principles, Practice, and
[13] T. Bohl, A. Smallbone, G. Tian, A.P. Roskilly, Particulate number and NOx trade-off Economics of Plant and Process Design, second ed., Butterworth-Heinemann,
comparisons between HVO and mineral diesel in HD applications, Fuel 215 (2018) Boston, United States, 2013.
90–101, https://doi.org/10.1016/j.fuel.2017.11.023. [40] A. Srifa, K. Faungnawakij, V. Itthibenchapong, N. Viriya-empikul,
[14] E. Shi, W. Sommer, A. Smiddy, EIA Projects U.S. Renewable Diesel Supply to T. Charinpanitkul, S. Assabumrungrat, Production of bio-hydrogenated diesel by
Surpass Biodiesel in AEO2022, EIA - US Energy Inf Adm, 2022. catalytic hydrotreating of palm oil over NiMoS2/γ-Al2O3 catalyst, Bioresour.
[15] J.A. Melero, J. Iglesias, A. Garcia, Biomass as renewable feedstock in standard Technol. 158 (2014) 81–90, https://doi.org/10.1016/j.biortech.2014.01.100.
refinery units. Feasibility, opportunities and challenges, Energy Environ. Sci. 5 [41] Enel. Electricity Tariff, 2021. https://www.eneldistribuicaosp.com.br/corporati
(2012) 7393, https://doi.org/10.1039/c2ee21231e. vo-poder-publico/tarifa-de-energia-eletrica. (Accessed 28 December 2021).
[16] G.V. Salvador, V.M.M. Salim, F.S. Toniolo, Sustainability assessment of a [42] R. Turton, R.C. Bailie, W.B. Whiting, J.A. Shaeiwitz, Analysis, Synthesis, and
decentralized green diesel production in small-scale biorefineries, Biofuels, Design of Chemical Processes, third ed., Prentice Hall, Boston, United States, 2009.
Bioprod. Biorefining 16 (2022) 1527–1550, https://doi.org/10.1002/bbb.2419. [43] Canal Rural. Soybean Oil Price. https://www.canalrural.com.br/(accessed January
[17] G. Faraci, C. Perego, C. Gosling, J. Holmgren, T. Marker, R. Marinangeli, New 10, 2022)..
developments in renewable fuels offer more choices: vegetable oil-based diesel can [44] da Silva GN. Green hydrogen in petroleum refining: a case study for the reduction
offer better integration within crude-oil refineries for fuels blending, Hydrocarb. of emissions and efficient use of energy resources. Ens Energético. https://ensaioe
Process. September (2007) 67–72. nergetico.com.br/o-hidrogenio-verde-no-refino-dopetroleo-um-estudo-de-caso-
[18] T.J. Hilbers, L.M.J. Sprakel, L.B.J. van den Enk, B. Zaalberg, H. van den Berg, L.G. para-a-reducao-de-emissoes-e-uso-eficiente-de-recursosenergeticos/(accessed
J. van der Ham, Green diesel from hydrotreated vegetable oil process design study, January 10, 2022) (in Portuguese)..
Chem. Eng. Technol. 38 (2015) 651–657, https://doi.org/10.1002/ [45] ANP - Brazilian National Agency of Petroleum Natural Gas and Biofuels. Price
ceat.201400648. Survey System. https://preco.anp.gov.br (accessed January 9, 2022) (in
[19] A.P.M.M. Diniz, R. Sargeant, G.J. Millar, Stochastic techno-economic analysis of Portuguese).
the production of aviation biofuel from oilseeds, Biotechnol. Biofuels 11 (2018) [46] ANP - Brazilian National Agency of Petroleum Natural Gas and Biofuels. ANP
161, https://doi.org/10.1186/s13068-018-1158-0. Resolution No 842. Brazil, 2021 (in Portuguese).

11
I.V. Barbosa et al. Renewable Energy 219 (2023) 119591

[47] Cepea. SOJA/RETRO, Lower Supply and Firm Demand Support Prices in 2019 [52] A. Prokopowicz, M. Zaciera, A. Sobczak, P. Bielaczyc, J. Woodburn, The effects of
2020, 2019 (in Portuguese). neat biodiesel and biodiesel and HVO blends in diesel fuel on exhaust emissions
[48] Forbes. Government Reaffirms Decision to Maintain Biodiesel Content in Diesel at from a light duty vehicle with a diesel engine, Environ. Sci. Technol. 49 (2015)
10% for 2022, 2022. https://forbes.com.br/forbesagro/2022/01/governo-reafir 7473–7482, https://doi.org/10.1021/acs.est.5b00648.
ma-decisao-de-manter-teor-de-biodiesel-no-diesel-em-10-para-2022/. (Accessed 7 [53] A. Rimkus, J. Žaglinskis, S. Stravinskas, P. Rapalis, J. Matijošius, Bereczky Á,
February 2022) (in Portuguese). Research on the combustion, energy and emission parameters of various
[49] G. Squadrin, NWE HVO, SAF at Records on Feedstocks Prices, Demand, Argus, concentration blends of hydrotreated vegetable oil biofuel and diesel fuel in a
2021. compression-ignition engine, Energies 12 (2019) 2978, https://doi.org/10.3390/
[50] ANP - Brazilian National Agency of Petroleum Natural Gas and Biofuels. Oil, en12152978.
Natural Gas and Biofuels Statistical Yearbook 2021, 2021. https://www.gov.br/an [54] H. Ferrari, M. Barbosa, D. Rodrigues, Gasoline and diesel reach highest real value
p/pt-br/centrais-de-conteudo/publicacoes/anuario-estatistico/oil-natural-gas-an in 20 years, Poder360 (2021). https://www.poder360.com.br/economia/gasoli
d-biofuels-statistical-yearbook-2021. (Accessed 5 April 2022). na-e-diesel-alcancam-maior-valor-real-em-20-anos/. (Accessed 26 January 2023)
[51] MDIC - Brazilian Ministry of Development, Industry, Commerce and Services. (in Portuguese).
Comex Stat. http://comexstat.mdic.gov.br/en/geral (accessed October 18, 2023)..

12