Chemical Engineering Journal 451 (2023) 138709

Contents lists available at ScienceDirect

Chemical Engineering Journal

journal homepage: www.elsevier.com/locate/cej

Kinetic modeling of the hydrocracking of polystyrene blended with

vacuum gasoil
David Trueba a, Roberto Palos a, b, Javier Bilbao a, José M. Arandes a, Alazne Gutiérrez a, *
a
Department of Chemical Engineering, University of the Basque Country UPV/EHU, PO Box 644, 48080 Bilbao, Spain
b
Department of Chemical and Environmental Engineering, University of the Basque Country UPV/EHU, Plaza Ingeniero Torres Quevedo 1, 48013 Bilbao, Spain

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

Keywords: The kinetic modeling of the hydrocracking of a mixture of polystyrene (PS) and vacuum gasoil (VGO) over a
Hydrocracking PtPd/HY catalyst has been carried out. The reactions have been performed in a batch reactor under the following
Kinetic modeling conditions: 380–420 ◦ C; 80 bar; content of PS in the feed, 10 wt%; catalyst/feed ratio, 0.1 in mass; and time,
Waste plastics
30–300 min. Different reaction networks and kinetic models have been studied, in which the evolution of
Polystyrene
Vacuum gasoil
product distribution (unconverted PS, dry gas, liquefied petroleum gases, naphtha, light cycle oil, heavy cycle oil
Catalyst deactivation and coke) with the extent of time has been quantified by considering three different simultaneous deactivation
mechanisms (plastic fouling, coke deposition and metal poisoning). The kinetic model selected (based on a 7-
lump reaction network) has been used for performing a parametric study, determining that 400 ◦ C and 180
min are the optimal conditions for maximizing the yield of naphtha (35 wt%) at the same time that PS is totally
converted. This original kinetic model may act as a basis for scaling-up studies focused on the large-scale
valorization of waste plastics by co-feeding them into a hydrocracking unit of a Waste-Refinery.

[7]. For example, waste PET pyrolysis [8] appears to be successful for
the production of gases (CO, CO2, ethylene), along with commercial
1. Introduction interest solids (benzoic and benzoyl formic acid) and oil. Apart from
that, the technological development of gasification is also remarkable
The growth of plastic demand in the packaging, automotive, con­ [9].
struction and electrical and electronics industries does not come with an The possibility of co-feeding waste plastics or their derivatives ob­
efficient waste management [1]. Consequently, this waste is mainly tained in a fast pyrolysis stage with benchmark streams to conventional
landfilled and incinerated provoking different phenomena that interfere refinery units (Waste-Refinery) has emerged as a promising strategy that
with human health [2,3], such as the presence of microplastics in can be the solution for the major waste plastics mismanagement (spe­
aquifers and oceans and the emission of toxins in uncontrolled inciner­ cifically polyolefins) [6] through their conversion into fuel and chemi­
ation. In this context, the thermochemical technologies suitable for cal. In this way, the valorization of waste plastics within the refineries
treating waste plastics (pyrolysis, gasification, cracking and hydro­ would ease the integration of the oil industry in the circular economy
cracking) [4,5] are attracting attention in recent times since they offer action plan, since the valorization of goods produced with petroleum
the possibility of recovering monomers and producing fuels or derivatives would be performed resulting in an oil consumption reduc­
hydrogen. One of the main difficulties when establishing the best tion. The production will be optimized according to the capacity of the
valorization route is the heterogeneous character of the waste plastics, refinery units in order to adapt the composition of the products to legal
whose average distribution in the European Union urban solid wastes is requirements and, subsequently, promote their commercialization and
the following [6]: polypropylene (PP), 26 wt%; high density poly­ distribution by the conventional distribution channels. The refinery
ethylene (HDPE), 15 wt%; low density polyethylene (LDPE), 13 wt%; units with a potentially key role within the Waste-Refinery are those of
linear low density polyethylene (LLDPE), polyvinyl chloride (PVC), 20 catalytic processing. Thus, the potential of fluid catalytic cracking (FCC)
wt%; polystyrene (PS), 7 wt%; polyethylene terephthalate (PET), 6 wt%, has been explored by feeding HDPE dissolved in vacuum gas oil (VGO)
and; others, 5 wt%. Pyrolysis has revealed itself as a straightforward [10] and mixtures of HDPE pyrolysis waxes and VGO [11]. Palos et al.
technology of adequate application for the valorization of waste plastics

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

https://doi.org/10.1016/j.cej.2022.138709
Received 20 June 2022; Received in revised form 26 July 2022; Accepted 17 August 2022
Available online 22 August 2022
1385-8947/© 2022 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-
nc-nd/4.0/).
D. Trueba et al. Chemical Engineering Journal 451 (2023) 138709

Nomenclature np number of experimental data

R universal gas constant, kJ/mol
Abbreviations SSE sum of square errors
DG dry gas t reaction time, min
HCO heavy cycle oil T temperature of reaction, K
LCO light cycle oil tb catalyst pores blockage time, min
LPG liquefied petroleum gases Tref reference temperature, K
PS polystyrene W catalyst mass, g
Yi yield of i lump, wt%
Symbols
Cc carbon content deposited on the catalyst, gcoke g-1
catalyst Superscripts
Ej activation energy of the reaction j, kJ/mol n number of deactivation types
F1-α Fisher’s distribution critical value αj order of reaction j
FA-B Fisher’s test value
kd,c catalyst deactivation rate constant related to coke Subscripts
deposition A, B compared models
kd,m catalyst deactivation rate constant related to metals c coke deactivation
poisoning i certain lump
kj reaction rate constant of reaction j j certain reaction
kj,Tref reaction rate constant of reaction j at temperature of k certain type of deactivation
reference m deactivation by metals poisoning
M metal loading of the catalyst, kg of metal (kg of catalyst)-1 p deactivation by PS deposition
M0 maximum metal loading of the catalyst, kg of metal (kg of s steady state
catalyst)-1 Greek symbols
mc coke deactivation rate order α deactivation constant of coke deposition
mm metals poisoning rate order φ activity function
nd number of experimental data ψ global activity
ne number of experiments υ degrees of freedom
nl number of lumps

[12] have developed a kinetic model for HDPE waxes cracking. a calculation methodology far more complex than the lump-based
Regarding waste plastics hydroprocessing, it has been studied prin­ models [22]. These models composed of discrete lumps are simpler
cipally with virgin plastics [13], aiming to maximize transport fuel and allow for more easily quantifying phenomena that, given their
production [14]. Vance et al. [15] have studied polyethylene hydro­ complexity, are normally empirically quantified, such as the diffusional
cracking with a catalyst and under specific operating conditions that limitations of the components in the reaction media and the deactivation
promote the isomerization obtaining isoparaffinic gasoline as the main of the catalysts [23]. The lumps are groups of compounds with similar
product, while Choi et al. [16] have studied the hydrocracking of py­ kinetic behavior that are treated as pseudo-components in the modeling.
rolysis waxes. The attractiveness of the hydroprocessing refinery units to In the hydroprocessing of heavy petroleum fractions, the lumps are
add value to the waste plastics or their pyrolysis waxes lies in their established according to: (i) boiling point criteria basis into dry gas
versatility for treating heavy and aromatic streams, obtaining products (DG), liquefied petroleum gases (LPG), naphtha, light cycle oil (LCO),
with a composition similar to that of commercial fuels by using the heavy cycle oil and coke (carbon deposited on the catalyst), using
appropriate operating conditions and catalysts [17]. Vela et al. [18,19] different strategies for establishing the appropriate reaction network
have checked that the co-feeding of HDPE with VGO has the interest of [24], performing sensibility analyses to simplify the models [25], and
increasing the conversion of the VGO and reducing the concentration of relating kinetic parameters values with those in the literature [26]; or
aromatics in the naphtha fraction produced. The same authors [19] have (ii) SARA (saturates, aromatics, resins and asphaltenes) composition
compared the results of the different strategies of co-feeding HDPE or [27,28], using separate models to predict liquid, gas and coke yields
plastic pyrolysis oil (PPO) with VGO. However, making the upgrading of [29].
waste plastics in refinery hydroprocessing units come true it is required The modeling of the hydrocracking of VGO has been carried out
the knowledge of kinetic models that would allow for (i) predicting the using kinetic models with between 4 and 9 component lumps [24,30]. In
effects of the operating conditions in the product distribution, (ii) contrast, the precedent works in the literature about the kinetic
optimizing the operating conditions and (iii) defining the possible modeling of the hydrocracking of plastics are very scarce. Bin Jumah
revamping requirements of the units. et al. [28] proposed for the hydrocracking of light-density polyethylene
The kinetic modeling of catalytic processes such as cracking and (LDPE) over a Pt/Hbeta catalyst a 4-lump (LDPE, heavy liquids, naphtha
hydrocracking of hydrocarbons (which present a complex reaction and gases) scheme of reactions. They obtained that the limiting step was
network) shows particular difficulties predicting the evolution over time the conversion of the LDPE into heavy liquid, given the diffusional re­
of the product distribution due to deactivation. Consequently, well- strictions that the macromolecular chains of LDPE encountered within
established software such as Advances Kinetics and Technology Solu­ the structure of the zeolite.
tions (AKTS) cannot be used as it is in the cases of thermokinetic results In a previous work [31], we investigated the co-feeding of poly­
of waste plastics pyrolysis [20]. The fundaments for the kinetic styrene (PS) and VGO under different operating conditions (temperature
modeling of the hydroprocessing of heavy refinery streams have been and reaction time) in a semi-batch reactor using a PtPd/HY catalyst.
developed by means of studies carried out in batch reactors with slurry- Indeed, an exhaustive characterization of the deactivated catalysts was
phase catalysts [21]. The kinetic models that consider the reactions of also carried out to deeply understand the deactivation phenomena
the individual components [17] require a bigger experimental base and involved in the hydrocracking of PS/VGO blend and its consideration for

2
D. Trueba et al. Chemical Engineering Journal 451 (2023) 138709

kinetic modeling. The aromatic nature of PS requires (just like VGO) volatile species. Furthermore, prior to the reaction, the catalyst has been
hydrocracking in order to maximize naphtha production with the reduced to achieve its active form. The reduction has been carried out
adequate composition for its commercialization as gasoline. In this ex-situ in a fixed bed reactor at 400 ◦ C for 4 h under a mixture of gases
work, the data obtained in those experiments have been used to estab­ consisting of a 30 mL min− 1 flow of H2 and a 50 mL min− 1 flow of N2.
lish an innovative lump-based kinetic model. In addition, each deacti­
vation mechanism has been modeled using each own equation, in order 2.4. Analysis of products
to predict the evolution of the product distribution. Thus, four different
models have been tested and the election of the optimal one has been The products have been submitted to different analyses in order to
made by statistical methods. Finally, the best kinetic model has been properly characterize them. But, first of all, it should be detailed which
used to find the optimal operating conditions to maximize the conver­ has been the procedure followed for differentiating them, i.e. for prop­
sion of PS and the production of naphtha, which can be assimilated by erly closing the mass balance. The amount of gas has been obtained as
commercial gasoline. The interest of the kinetic model lies on its the difference between the weight of the reactor before and after the
application to the hydrocracking unit and the proposed methodology reaction. For determining the amount of non-converted plastic a pro­
will be used as basis for the study of the hydrocracking with different cedure composed of two consecutive extractions has been required
plastics and catalysts. (Fig. S1). The amount of coke deposited on the spent catalysts has been
quantified by means of temperature-programmed oxidation (TPO).
2. Materials and methods Finally, the amount of liquid products has been obtained as the differ­
ence between the mass of the feedstock loaded into the reactor and the
2.1. Charges mass of the other products obtained according to the aforementioned
calculations.
The vacuum gasoil (VGO) has been provided by Petronor refinery The gas products have been analyzed by gas chromatography in an
(Muskiz, Spain). The techniques used for its characterization have been Agilent Technologies 6890 GC equipped with an HP-PONA capillary
already detailed in a previous work [31]. The main physicochemical column (50 m × 0.2 mm) and an FID detector. For properly separating
properties of the VGO can be found in Table S1 in the Supplementary the lightest hydrocarbons, the analysis has been carried out at cryogenic
Material. In short, it consists of a mix of hydrocarbons, mainly paraffins temperatures (-30 ◦ C) using CO2 for reaching that temperature. The gas
(49 wt%) and with a remarkable content of 3+-ring aromatics (24.5 wt fraction has been separated into dry gas (DG, C1-C2) and liquefied pe­
%), with a boiling range (408–503 ◦ C) that corresponds to a light VGO. troleum gases (LPG, C3-C4).
The polystyrene (PS) has been purchased from Dow Chemical in The liquid products have been divided into three different lumps
Tarragona (Spain). It has been received in 4 mm pellet form and prior to according to the following boiling ranges (TB): naphtha (TB less than
being blended with the VGO it has been milled to dust by cryogenic 216 ◦ C), light cycle oil (LCO, 216 < TB < 343 ◦ C) and heavy cycle oil
methods. The main properties of the PS (molecular weight, 311.6 kg (HCO, TB greater than 343 ◦ C). For performing this differentiation, the
mol− 1; density, 1.030 g mL− 1; and dispersity, 2.39) have been provided liquid products have been submitted to a simulated distillation analysis
by the supplier. according to the procedure specified in the ASTM D2887 Standard.
Likewise, they have been analyzed in an Agilent Technologies 6890 gas
2.2. Catalyst chromatograph equipped with a DB-2887 semi-capillary column (10 m
× 0.53 mm) and an FID detector.
A PtPd catalyst supported on a Y zeolite has been used for the hy­ The methodology used has allowed to quantify the total amount of
drocracking runs. The nominal content for each metal is 0.5 wt%. The coke as well as to distinguish between the coke located on the external
synthesis procedure, as well as the characterization techniques used can surface of the catalyst and that deposited within the pore system of the
be found in a previous work [31]. Shortly, the textural properties have zeolite [33]. Apart from that, scanning electron microscopy (SEM) im­
been obtained through N2 adsorption–desorption isotherms; the acidic ages of the spent catalysts have been recorded to ratify which is the
properties have been measured by means of ammonia temperature- predominant deactivation phenomenon at each temperature. The
programmed desorption (TPD) and pyridine Fourier-transform apparatus used has been a Schottky-type field emission scanning elec­
infrared spectroscopy (FTIR) spectra; the actual metal content has tron microscope (JEOL JSM-7000F) equipped with a secondary electron
been quantified by inductively coupled plasma-optical emission spec­ detector working at 10 kV. Finally, the content of metals on the spent
trometry (ICP-OES); and metal dispersion has been calculated by CO catalysts has been also measured by ICP-OES analysis in an Agilent
pulse chemisorption. From the properties (Table S2), it should be Technologies 7200-ES equipment, after submitting the samples to
highlighted the relatively high specific surface area (548 m2 g− 1), which digestion in an HNO3, HCl and HF mixture.
is a consequence of the contribution of the micropores surface (368 m2
g− 1). Moreover, it has a total acidity of 0.465 mmolNH3 g-1and a 3. Kinetic modeling
Brønsted/Lewis (B/L) ratio of 0.98. Finally, it should be remarked the
dispersion of the metals on the support (47.8 %). To the best of our knowledge, there are no other kinetic studies about
the hydrocracking of PS/VGO blends. Therefore, the different models
2.3. Reaction equipment and procedure proposed for the hydrocracking of heavy oil [34] have been taken as the
starting point.
The hydrocracking runs have been carried out in an already
described experimental setup [32], which consists of a 50 mL batch 3.1. Reaction network
reactor that operates in semi-continuous regime and under the following
operating conditions: 380–420 ◦ C, 80 bar, 30–300 min and with The discrimination of the different kinetic models has been focused
catalyst-to-feed and plastic-to-feed ratios of 0.1 in mass basis in both on the two reaction networks collected in Fig. 1. Model A (Fig. 1a)
cases. The stirring speed of the reactor has been established at 1300 rpm consists of a 6-lump reaction network, quite similar to that proposed by
in order to minimize the external diffusion limitations. Note that the Martínez and Ancheyta [34] for the hydrocracking of heavy oil, to which
stirring has been switched on once the established temperature and the PS has been added. As it can be seen, the network considers a
pressure conditions have been reached, considering that very moment parallel-series system of reactions in which the hydrocracking of the
the zero time of reaction. The experimental setup is also equipped with a heavy species into lighter ones occurs sequentially. In addition, it has
condensation system that ensures the proper separation of the light been considered that just the compounds within the HCO lump could

3
D. Trueba et al. Chemical Engineering Journal 451 (2023) 138709

3.2. Mass conservation equations

The methodology used for the kinetic data analysis is in concordance

with the stages and the recommendations established in the literature
for catalytic processes with complex reaction networks that consider the
catalyst deactivation [35]. Furthermore, to establish the mass conser­
vation equations of the different lumps, the following assumptions have
been made: (i) external diffusivity is negligible due to the stirring speed
used in the experiments [27]; (ii) gas and liquid phase concentrations
are uniform in the reactor; (iii) a high heat transmission is achieved
avoiding the formation of temperature gradients in the reactor [36],
considering the reactor as isothermal; (iv) the continuous feeding of H2
creates an excess of this reactant that allows considering that pressure is
constant within the reaction medium over time; and (v) the catalyst
deactivation is due to three simultaneous causes, namely plastic fouling,
coke deposition and metal poisoning (as explained in Section 3.3).
This way, the evolution over time of the yield of each lump i within
the reaction network (Yi) is represented by the following conservation
equation:

∑ j
dYi
= ψ kj (Yi )nj (1)
dt 1

where kj is the kinetic parameter, nj is the order of reaction j and ψ is

the global deactivation function, considering that the effect of deacti­
vation in the individual reaction kinetics is non-selective. In concor­
dance with the common treatment in the literature [37], the individual
reactions have been considered of first order (nj = 1) except for those in
which the HCO lump (heavy fraction of the VGO) is involved, which are
of second order (nj = 2).
In order to express the temperature dependence of the kinetic pa­
rameters, they have been defined according to the reparameterized
Arrhenius equation [38].
[ ( )]
Ej 1 1
kj = kj,Tref exp − − (2)
R T Tref

being kj,Tref the kinetic parameter of reaction j at the reference

temperature, Ej the activation energy of reaction j, R the universal gas
constant, T the temperature and Tref the reference temperature (673 K).

3.3. Deactivation causes and kinetics

The rigorous quantification of the deactivation is one of the main

challenges of the kinetic modeling of the hydrocracking of heavy oil
fractions. However, this goal faces the difficulty of establishing the ki­
netic models of various simultaneous causes. In hydrocracking re­
actions, the principal deactivation causes are the formation of coke
(carbonaceous material that covers sequentially the active sites and
micropores of the acid support) and the poisoning of the metallic sites
caused by the presence of metals in the feed [23]. As an illustrative
Fig. 1. Proposed 6-lump (model A) (a) and 7-lump (model B) (b) reac­ example of pore blockage, nitrogen adsorption–desorption isotherms of
tion networks. spent catalysts at the intermediate temperature have been collected in
Fig. S2, observing both a general decrease in the surface properties of the
condensate forming coke. However, some coke precursors can be also fresh catalyst that evidences the blockage on meso- and micro-pores
hydrogenated making this reaction a two-way pathway. Hence, the re­ [39]. In a previous work [31] about the hydrocracking of the PS/VGO
actions in model A account for a total amount of 11 kinetic constants. blend, another deactivation cause at low reaction temperature was
The second proposal (model B in Fig. 1b), more complex, splits the gas identified. In these conditions, the unconverted PS also contributed to
lump into DG and LPG lumps, becoming a 7-lump reaction network. catalyst deactivation, since the partially degraded PS molecules depos­
Additionally, it has been also considered the formation of coke from the ited on the particles of the catalyst totally blocked its porous structure.
PS, without taking into account the inverse route. The enlargement of Equally, it was verified that an increase in the temperature reduces the
the reactions system adds 6 new kinetic constants (accounting for a total significance of this deactivation cause, but increases the importance of
amount of 17 in model B), 5 corresponding to the formation of dry gas the presence of carbonaceous material (internal coke) in the channels of
from the rest of the lumps and the remaining one corresponding to the the HY zeolite. The formation of the internal coke lies in the conden­
polystyrene-to-coke pathway. sation reactions of the aromatics in the reaction medium, which take
place in the acidic sites of the zeolite and are boosted at high tempera­
tures [40].

4
D. Trueba et al. Chemical Engineering Journal 451 (2023) 138709

The SEM images of the spent catalysts obtained at the different re­ their respective expressions have been combined for determining the
action temperatures (Fig. 2) allow for assessing the incidence of this global deactivation function (ψ in Eq. (1)).
variable in the relative significance of the different deactivation causes. For the deactivation function corresponding to fouling caused by the
At 380 ◦ C, the surface of the catalyst presents a smooth envelope that sedimentation of the partially degraded PS chains (φp) the kinetic
covers almost the totality of the catalyst particles (Fig. 2a). By means of deactivation model proposed by Elizalde and Ancheyta [45] has been
EDX analysis, it has been confirmed that the covering has a carbona­ applied. It represented sequentially the coverage of the sites and pores
ceous nature but is different from the common composition of the cat­ blockage by heavy refractory compounds in the feed by means of two
alytic coke. Thus, it can be concluded that this material corresponds to expressions that must be applied before (Eq. (3)) and after (Eq. (4)) the
partially degraded PS molecules that have been deposited on the catalyst blockage time (tb):
particles. Furthermore, this problem has not been observed at higher For 0 < t ≤ tb.
temperatures, exposing that a temperature of 380 ◦ C is not sufficient to
φp = φs + (1 − φs ) e - kd,p t (3)
propagate the hydrocracking of the PS. Consequently, this particular
phenomenon of the hydrocracking of PS must be implemented in the For t > tb.
deactivation kinetic, distinguishing it from the deactivation by catalytic
coke. φp = φs + (1 − φs ) e - kd,p (t - tb ) (4)
At 400 and 420 ◦ C, the plastic deposition phenomenon is no longer
where kd,p is the deactivation kinetic constant for plastic deposition
observed since clearer images of the catalyst particles have been ob­
and φs the steady-state catalyst activity function, in concordance with
tained (Fig. 2b and c, respectively), and also the drastic catalytic activity
the consideration made by Monzón et al. [46]. These authors established
decay did not occur [31]. Last, the image displayed in Fig. 2c exposes
different deactivation kinetic equations considering that deactivation is
that heavy metals poisoning (also found at 400 ◦ C) is relevant at 420 ◦ C.
usually not complete and that the catalyst achieves a pseudo-stable state
This phenomenon has been previously observed in the hydroprocessing
with a constant remaining activity.
of vacuum residue [41]. Indeed, the images show that the highest atomic
The deactivation related to the formation of coke has been quantified
weight metals do stand out over the low molecular weight species. The
through a hyperbolic function (Eq. (5)) derived from the Langmuir-
brilliant metal particles observed in the spent catalyst (not detected in
Hinshelwood concepts, introducing a constant α, which represents a
the fresh catalyst) have been assigned by EDX to Fe and Ni particles,
proportional ratio between the deposited coke on the catalyst (Cc) and
which are found in VGOs [42]. The content of these metals in the
the poisoned active sites [23]. The deposited coke on the catalyst was
deactivated catalysts has been measured by ICP-OES, finding that their
measured in the previous work [31] by means of TG-TPO, whose results
content is noticeable and that it follows an increasing trend both with
are exemplified in Fig. S3 and Table S3, which show TPO profiles and
temperature and with time. The highly brilliant particles in Fig. 2d
coke contents, respectively. TPO results demonstrate the special in­
constitute a sort of metal alloy whose formation is possible at 420 ◦ C
crease in the coke amount deposited on the catalysts when coke depo­
[43,44]. The formation of these alloys might have a strong influence in
sition is the main deactivation route (400 and 420 ◦ C), as well as the
catalyst deactivation due to the pores blocked by these deposited metals.
displacement on the combustion temperature peak that is related to a
This way, it has been verified that three deactivation causes can be
higher development of coke structures when temperature is increased.
distinguished: (i) fouling caused by the external deposition of PS chains
(at low temperature); (ii) internal deposition of catalytic coke; and, (iii) 1
φc = (5)
metals poisoning. Hence, three kinetic deactivation equations have been (1 + αCc )mc
adapted to quantify the deactivation related to each of the causes and

Fig. 2. SEM images of the spent catalysts obtained at 380 ◦ C (a), 400 ◦ C (b) and 420 ◦ C (c and d).

5
D. Trueba et al. Chemical Engineering Journal 451 (2023) 138709

being α the deactivation constant for coke deposition, Cc the carbon SSEA − SSEB
fractional content of the catalyst and mc a fitting parameter (deactiva­ FA− = SSEB
> F1− α (υA − υB , υB ) (9)
B
tion order by coke).
υA − υB
υB
The equation used for the activity decay related to the metals
where υ are the degrees of freedom of each kinetic model, α is
poisoning is based on an empirical proposal [34] when analyzing the
considered as 0.05 (that corresponds to a level of significance of 95 %),
hydrocracking in a slurry reactor. The corresponding deactivation
and F1-α is the critical value for Fisher’s distribution and it is calculated
function is dependent on the concentration of the metals in the catalyst
by using the function finv available in MATLAB software. The basis of
(M) and the total metal loading that the catalyst can adsorb (M0). In this
this method is that if the inequality of Eq. (14) is fulfilled, then the
case, the metal loading has been measured from the spent catalysts of
improvement provided by a kinetic model B with respect to a kinetic
the previous work by ICP-AES, as collected on Table S4, which exposes
model A is significant. This methodology can be applied when two
both the trends with time and temperature.
models have different degrees of freedom, which are calculated as
( ) mm
M follows:
ϕm = 1 − kd,m (6)
M0
υ = nl ne − np (10)
where kd,m is the deactivation kinetic constant for metals poisoning,
being nl the number of lumps, ne the number of experiments and np
M is the metal loading for a t time, M0 is the maximum metal loading
the number of parameters.
that the catalyst can take and mm is the corresponding deactivation
order.
4. Results and discussion
Thus, a global deactivation equation (as defined in Eq. (1)) can be
established considering the simultaneous effect of the different causes in
4.1. Model discrimination
the deactivation function [47]. Gayubo et al. [48] consider the synergy
between the reversible deactivation (due to coke deposition) and the
In order to establish the most appropriate kinetic model, different
irreversible one (dealumination) of a HZSM-5 catalyst in methanol
stages have been followed, comparing the fitting to the experimental
conversion to hydrocarbons with the following expression:
data of each of the models described above in Table 1. Firstly, the kinetic
∏
n models corresponding to the models A and B (explained in Section 3.1
ψ = φk (7) and described in Fig. 1) have been compared. To simplify, it has been
1
considered that the catalyst deactivation is only by coke deposition
being n the number of deactivation causes and φk the mathematical (main cause of catalyst deactivation at 400 and 420 ◦ C, temperatures of
expression that describes each deactivation cause. greater interest due to the high conversion of PS achieved). Secondly,
In order to apply Eq. (7) after having analyzed which deactivation after having selected the best reaction network, it has been assessed the
causes are involved at each temperature, the catalyst deactivation at interest of considering the three deactivation causes (model B-2). In
380 ◦ C must be modeled considering the three deactivation causes. addition, the possibility of considering the effect of non-catalytic stages
Thus, Eqs. (3–5) will be used to quantify the φp, φc and φm deactivation occurring in parallel with the catalytic mechanism has been evaluated
functions, respectively. Equally, following the observations about the (model B-3). The statistical parameters calculated in the model
deactivation causes involved at 400 and 420 ◦ C, the deactivation func­ discrimination have been collected in Table 1.
tion for these temperatures will consider the deposition of coke (φc) and In the first stage, which is the comparison of the reaction networks
the deposition of metallic compounds (φm), which are defined by Eqs. displayed in Fig. 1, model B has been taken as the benchmark because its
(5) and (6), respectively. corresponding kinetic model presents the highest degrees of freedom.
Thus, it has been checked if the reduction of the degrees of freedom by
3.4. Calculation methodology applying model A is statistically significant. Attending to the results
(Table 1), it can be seen that the SSE obtained with model A has been
The optimization of the apparent values of kinetic constants of the lower than that of model B (19.35 vs 21.22, respectively), but the cri­
individual reactions, activation energies and deactivation constants terion detailed in Eq. (9) has not been fulfilled. Therefore, the reaction
established in the kinetic model has been accomplished by adjusting the network proposed in model B remains a better option.
experimental data of the evolution with the extent of time of the yield of To improve the kinetic model, it has been established the kinetic
each lump (determined in a previous work [31]) and the predicted model B-2 that uses the same reaction network as model B but it con­
values. For that purpose, it has been used an in-house developed MAT­ siders the three different deactivation causes in the computing of the
LAB code that searches for the minimization of the sum of squares of the deactivation function. The degrees of freedom of kinetic model B-2 have
errors between experimental and predicted values: been reduced to 62, so kinetic model B has been taken once again as
nl ∑
∑ nd
( )2 reference for applying the Fisher’s test (Eq. (9)). As it can be seen in
SSE = Yiexp − Yicalc (8) Table 1, the SSE obtained with kinetic model B-2 is notoriously inferior
1 1 to that obtained with model B (8.12 vs 21.22, respectively). In addition,
where nl is the number of lumps in the reaction network, nd is the the value of FB-A is higher than that of F1-α (16.67 vs 2.25, respectively).
number of experimental data, Yexp
i is the experimentally obtained yield Hence, the inclusion of the different deactivation routes, i.e. kinetic
and Ycalc
i is the corresponding computed yield. model B-2, statistically improves the description made by kinetic model
Additionally, the code also calculates the residual values for each B, with the same reaction network (model B in Fig. 1).
parameter, the Jacobian matrix and the correlation coefficient matrix. The last proposal (kinetic model B-3) has consisted on including the
The mathematical procedure followed for their determination is
explained by Amin et al. [49]. Table 1
The discrimination between the different reaction networks in Fig. 1 Statistical comparison between the proposed kinetic models.
has been realized by comparing not only the SSE of the corresponding Model Lumps Deactivation functions v FA-B F1-α SSE
kinetic models but also performing a significance test based on Fisher’s
A 6 1 63 1.22 2.36 19.35
method. For this purpose, the following criterion must be fulfilled [35]: B 7 1 68 – – 21.22
B-2 7 3 62 16.67 2.25 8.12
B-3 7 3 44 – 1.84 14.76

6
D. Trueba et al. Chemical Engineering Journal 451 (2023) 138709

thermal reaction pathways to Model B-2, by the addition of supple­

mentary first-order reactions for all the individual reactions in the
network reaction of model B in Fig. 1, except for the reactions in which
the HCO lump is involved, as they have been modeled with second-order
reactions. Once again, the results of the fitting have been summarized in
Table 1. This time, the SSE obtained with kinetic model B-3 has not
improved the results previously obtained with model B-2 in spite of its
higher number of kinetic parameters. Therefore, the inclusion of the
thermal routes has not improved the fitting.
To sum up, the best results have been obtained with kinetic model B-
2, which considers the reaction network of 7-lumps (model B in Fig. 1)
and the three different deactivation causes described in Section 3.3. The
apparent kinetic constants of this model have been related in Table 2.
The subsequent sections will explore the fitting of this model, the pre­
dicted values with respect to those experimentally obtained, the evo­
lution of the different activity functions with time, and it will be used to
find the optimal operating conditions.

4.2. Fitting and model validation Fig. 3. Parity plot of experimental vs predicted yields with kinetic model B-2
for the different lumps.
The optimized parameters corresponding to kinetic model B-2
(Table 2) have been used for calculating the yields of each lump at the the model [50]. However, as aforementioned (Section 4.1), in this work
three studied temperatures and different reaction times. The comparison the inclusion of the thermal routes did not improve the overall fitting of
between experimental data and the predicted response has been the model.
collected in a parity plot displayed in Fig. 3. Overall, a good accuracy of
the model has been achieved, even considering the fitting difficulty at
4.3. Analysis of kinetic parameters
380 ◦ C, when a rapid deactivation occurs because of the blocking of the
active sites and porous structure by the deposition of the partially con­
By comparing the values obtained for the apparent kinetic parame­
verted PS chains.
ters (Table 2), it can be extracted the importance of the steps in the
This fitting can be observed in more detailing Fig. 4, where the
reaction network (model B in Fig. 1). First, the kinetic parameters of
experimental data (symbols) are shown over the predicted data (lines) in
steps #7 and #8 expose the contribution of the HCO lump to the desired
different graphs that collect the evolution of the yield of each lump with
products, i.e. LCO and naphtha. The higher value of apparent kinetic
the time. Apart from the good fit observed on the graphs, two main
constant of step #7 is in agreement with the sequential mechanism
considerations can be extracted from these graphs. The deviations are
established in the literature for the hydrocracking of heavy oils [51].
mainly attributed to experimental errors in the measure of the yield of
This mechanism explains that middle distillates are preferentially
PS (unconverted fraction) at 380 ◦ C, as it is a viscous mixture whose
formed from heavy lumps (HCO), which in turn lead to the formation of
evolution with time is difficult to quantify. Thus, the experimental error
lighter species and gases. In the same way, the apparent kinetic pa­
at 380 ◦ C is noteworthy if the focus is moved to HCO and naphtha yields.
rameters in the steps that LCO acts as a reactant (steps #11, #12 and
On the other side, the fitting of the HCO and DG lumps at 420 ◦ C from
#13) also evidence the primary formation of the naphtha lump.
120 min onwards could have been improved by including thermal
Concerning naphtha formation, the apparent kinetic constant of
routes. Indeed, the production of gases through thermal routes in the
steps #2 and #3 allow concluding that the preferred destination of the
hydrocracking of hydrocarbon streams has been already observed [18]
molecules obtained in the hydrocracking of the PS is the naphtha lump.
and even the consideration of these routes has improved the quality of
This fact is consistent with a previous work [13], in which compounds
with a boiling range within that of the naphtha fraction were obtained in
Table 2
the hydrocracking of neat PS. Indeed, the formation of not only 1-ring
Apparent kinetic parameters at the reference temperature and corresponding
aromatics but also of saturated cyclic compounds was observed.
apparent activation energies calculated with kinetic model B-2.
The reaction network of kinetic model B-2 (Fig. 1b) contemplated the
Reaction Parameter kj,Tref Units Ej (kJ/mol) formation of coke from HCO and PS lumps (steps #1 and #6, respec­
PS → Coke k1 (2.47 ± 0.2) 10-2 g-1
cath
− 1
122 tively) because the formation of coke from plastics was previously re­
PS → LCO k2 (3.19 ± 0.3) 10-1 g-1
cath
− 1
49 ported in the literature [52]. Attending to the apparent kinetic constants
PS → Naphtha k3 (1.01 ± 0.1) 100 g-1 h − 1
224
cat
of these steps it can be observed that this consideration has been
PS → LPG k4 (4.27 ± 0.1) 10-2 g-1
cath
− 1
355
PS → DG k5 (5.17 ± 0.4) 10-3 g-1
cath
− 1
413 appropriate since the kinetic parameters of both steps #1 and #6 are of
HCO → Coke k6 (4.74 ± 1.0) 10-5 g-1 -1
cat gHCOh
− 1
1 the same magnitude that other routes of those lumps. However, coke is
HCO → LCO k7 (1.82 ± 0.2) 10-2 g-1 -1
cat gHCOh
− 1
281 preferentially formed from HCO lump given the content of 3+-ring ar­
HCO → Naphtha k8 (1.82 ± 0.1) 10-4 g-1 -1
cat gHCOh
− 1
298 omatics on it (Table S1), although the kinetic parameters cannot be
HCO → LPG k9 (5.19 ± 0.8) 10-5 g-1 -1
cat gHCOh
− 1
337
HCO → DG k10 (2.72 ± 0.1) 10-5 g-1 g -1
h − 1
480
directly compared because of the different order of the reactions. In
cat HCO
LCO → Naphtha k11 (6.92 ± 0.2) 10-1 g-1
cath
− 1
114 addition, kinetic model B-2 also contemplates the possibility of coke
LCO → LPG k12 (1.45 ± 0.3) 10-1 g-1
cath
− 1
121 molecules being converted into HCO molecules (step #17). Even though
LCO → DG k13 (2.19 ± 0.1) 10-2 g-1
cath
− 1
207 this step is residual, it has a certain incidence at 420 ◦ C, as the yield of
Naphtha → LPG k14 (9.09 ± 0.5) 10-2 g-1
cath
− 1
43
coke goes through a maximum at 120 min (Fig. 4c).
Naphtha → DG k15 (8.03 ± 0.3) 10-4 g-1
cath
− 1
65
LPG → DG k16 (2.34 ± 0.3) 10-2 g-1 − 1 Attending to the values of the apparent activation energy of the
cath 12
Coke → HCO k17 (6.73 ± 0.5) 10-3 g-1
cath − 1
1 formation steps toward the undesirable lump dry gas, the order of these
Plastic fouling kd,p (4.29 ± 0.3) 10-2 h− 1 – parameters is remarkable: E10 > E5 > E13 > E15 > E16. This highlights the
Coke deposition αd,c (4.69 ± 0.5) 101 – – slight effect of temperature on DG formation from the reactants (HCO
Metals poisoning kd,m (3.66 ± 0.1) 10-1 188
and PS), as it is the product of the overcracking of primary (LCO,
–

7
D. Trueba et al. Chemical Engineering Journal 451 (2023) 138709

Fig. 5. Coefficient correlation matrix of the kinetic parameters of model B-2.

notice that the correlation matrix is perfectly symmetrical, which means

that half of the correlation coefficients shown in the matrix are redun­
dant and unnecessary. Thus the principal diagonal of the matrix is equal
to 1.0 because exposes the correlation of each parameter with itself.
This way, strong interdependence between parameters can reveal
avoidable steps and therefore modeling parameters, which will help to
optimize the number of freedom degrees involved in the model. The
interdependence between parameters identify possible over­
parametrized routes, as there is a certain degree of uncertainty in the
values obtained for those parameters when this correlation is above the
critical one [54]. The most noteworthy coefficients are those relating k4-
k12, k9-k14 and k12-k14, all related to the formation of LPG lump. The
main reason for this high interrelation can be found in the formation of
lumps in a cascade way, as it can be seen on the reaction network in
Fig. 1b, the two last parameters connect the formation of LPG from LCO
and naphtha, respectively. The formation of naphtha from LCO is pre­
cisely one of the most rapid and clear pathways, so the influence of this
reaction on LPG formation from both LCO and naphtha is evident.
Something similar can be attributed to the interdependence of k4-k12
coefficients, as the PS is mainly converted into LCO and naphtha so the
subsequent production of LPG is directly influenced by reaction #4.
However, the suppression of any of these paths seems unreasonable as
the direct formation of LPG compounds from PS, LCO and naphtha has
been proven in literature [55].

4.4. Activity evolution and importance of deactivation causes

The adequate consideration of catalyst deactivation (Section 3.3) has

a great impact on kinetic model fitting. Fig. 6 shows the evolution over
time of the catalyst activity for the different temperatures of reaction.
Fig. 4. Fitting of the predicted values (lines) with respect to the experimental
Continuous lines represent global activity (ψ) and dashed lines describe
yields (scatter). the deactivation functions related to coke (φc) and metal (φm) deposition
defined in Eqs. (5) and (6), respectively. Deactivation due to PS fouling
(φp) has not been displayed on Fig. 6 because of being the main deac­
naphtha) and secondary (LPG) formed lumps.
tivation only at 380 ◦ C and at this temperature the global deactivation is
In order to study the influence and codependence between the ki­
almost equal to PS fouling deactivation. At 400 and 420 ◦ C this function
netic parameters, the correlation matrix of the 17 parameters considered
value is very close to the unity and therefore it is not shown on in Fig. 6.
in model B-2 has been obtained (Fig. 5). The matrix depicts the linear
Two significant conclusions can be extracted from the activity pro­
association between all the possible pairs of kinetic parameters, allow­
files. First, the activity decay related to plastic deposition at 380 ◦ C is
ing discerning between the intrinsic relationships of the proposed
very quick and it is halved in less than 30 min. Moreover, the remaining
mechanism or a strong interrelation due to a bad design of experiments
activity from 120 min and onwards is marginal. In fact, the pore
[53]. The values collected in the matrix are within the range –1.0 to 1.0
blockage time, defined as one of the optimizing parameters (Eq. (9)),
and the further away the correlation coefficient is from 0, the stronger
agrees with the experimental observation, resulting in tb = 28 min. Even
the linear association between the two kinetic parameters. One should
though coke and metals poisoning have been also considered in the

8
D. Trueba et al. Chemical Engineering Journal 451 (2023) 138709

deposition, its activity decay doubles its magnitude when increasing

temperature from 400 to 420 ◦ C, confirming its strong temperature
dependence.
The pseudo-stable remaining activity observed for coke deactivation
function at 400 and 420 ◦ C is characteristic of this type of deactivation in
hydroprocessing, as the rate of condensation of the coke precursors is
compensated by their hydrocracking rate, favored by high hydrogen
pressures and the catalyst activity. This situation has been observed in
the hydrodeoxygenation of bio-oil [56] and has been considered for the
kinetic model of this reaction over an acid activated carbon based
bifunctional catalyst [57].

4.5. Operation maps: Optimization

The kinetic model allows performing reactor simulations to find the

optimal operating conditions according to the established goals. The
simulation has been carried out above 400 ◦ C to avoid the rapid deac­
tivation of the catalyst caused by the unreacted PS chains. In this way,
the temperature range used has been 400–440 ◦ C and reaction time has
been extended up to 480 min, to obtain a full picture of the framework of
the operation. Fig. 7 shows the predicted values over temperature and
through reaction time of the yields of PS, naphtha, LPG and DG lumps, as
Fig. 6. Evolution of the catalyst activity functions with the reaction time with they are the ones with the highest interest to be optimized. The opti­
the separate contribution of deactivation by coke deposition and metals mization strategy has been established according to different issues that
poisoning. ψ* (380 ◦ C) = φp. will be relevant when facing the scaling-up of the process (in which a
continuous reactor will be used). Thus, plastic conversion should be
kinetic modeling at 380 ◦ C, the PS fouling eclipses their effect and maximized (by obtaining the minimum yields of PS) in order to priori­
thereby they are not shown in Fig. 6. In contrast, at 400 and 420 ◦ C, tize the elimination of the waste plastic and avoid solid deposits from
when the PS fouling is insignificant because of the evidence discussed in unconverted plastic that would result in operation stops [58]. Moreover,
Section 3.3, different levels of the catalyst activity prevail after 300 min. naphtha is a priority in a hydrocracking unit as the production of high-
Thus, the remaining activity is of 0.50 and 0.28 at 400 and 420 ◦ C, quality gasoline-like streams from alternative feeds is gaining attention
respectively. due to oil depletion [59]. Last, the extent of the formation of the dry gas
On the other hand, the comparison of the deactivation functions by overcracking must be kept under control to maximize selectivity
evolution due to coke and metals poisoning at 400 and 420 ◦ C confirms towards naphtha or LCO, which are the products of commercial interest
the different importance of the two deactivation causes. Although they as fuels [60].
present considerably different remaining activities, coke deposition is Fig. 7a shows that above 400 ◦ C PS yields are less than 1 wt%, which
the main cause of deactivation at both temperatures. As for metal means plastic conversions of 90 %, for times higher than 60 min. The
line corresponding to 0.1 wt% yield represents a conversion of a 99 %,

Fig. 7. Prediction with the kinetic model of temperature and reaction time effect on the yields (wt%) of PS (a), naphtha (b), LPG (c) and DG (d) lumps.

9
D. Trueba et al. Chemical Engineering Journal 451 (2023) 138709

demonstrating that almost full conversion can be achieved at 400 ◦ C Government (grant RTI2018-096981-B-I00), the European Union’s
above 120 min. Besides, at 400 ◦ C and for times higher than 180 min ERDF funds and Horizon 2020 research and innovation program under
naphtha yield (Fig. 7b) can be maximized with a total yield of 34 wt%. the Marie Skłodowska-Curie Actions (grant No 823745) and the Basque
Therefore, the commitment between the yields of naphtha and gases will Government (grant IT1645-22). David Trueba thanks the University of
be critical within the scope of full plastic conversion. Fig. 7c and d evi­ the Basque Country UPV/EHU for his PhD grant (PIF 2018).
dence the well-known effect of temperature on overcracking reactions, The authors thank for technical and human support provided by
promoting the production of gases when the temperature is risen from SGIker of UPV/EHU and European funding (ERDF and ESF). The authors
400 ◦ C to above. Taking into account that the maximum yield of naphtha also acknowledge Petronor refinery for providing the feed used in the
is achieved at 400 ◦ C and for times equal to or greater than 180 min, work.
with full conversion of the plastic and that reaction times and temper­
atures higher produce higher amounts of lower-value gases (increasing Appendix A. Supplementary data
their production in almost a 40 wt% in one hour at 400 ◦ C), optimum
conditions must be established at 400 ◦ C and 180 min. Supplementary data to this article can be found online at https://doi.
org/10.1016/j.cej.2022.138709.
5. Conclusions
References
The kinetic parameters of a 7-lumps model have been estimated for
the joint valorization of PS and VGO in a batch reactor operating at [1] I. Antonopoulos, G. Faraca, D. Tonini, Recycling of post-consumer plastic
380–420 ◦ C and with reactions of 30–300 min. Different deactivation packaging waste in the EU: recovery rates, material flows, and barriers, Waste
Manag 126 (2021) 694–705, https://doi.org/10.1016/j.wasman.2021.04.002.
causes have been contemplated depending on the reaction temperature. [2] Z. Yang, F. Lü, H. Zhang, W. Wang, L. Shao, J. Ye, et al., Is incineration the
This way, catalyst activity loss can be a consequence of plastic deposi­ terminator of plastics and microplastics? J. Hazard. Mater. 401 (2021), 123429
tion, coke formation and metals poisoning. The application of each https://doi.org/10.1016/j.jhazmat.2020.123429.
[3] A. Xu, M. Shi, X. Xing, Y. Su, X. Li, W. Liu, et al., Status and prospects of
deactivation mechanism to the different reaction temperatures has been
atmospheric microplastics: a review of methods, occurrence, composition, source
substantiated by experimental evidence. Moreover, the kinetic model and health risks, Environ. Pollut. 303 (2022), 119173, https://doi.org/10.1016/j.
has been used to predict the optimal operating conditions, finding that envpol.2022.119173.
[4] R.-X. Yang, K. Jan, C. Chen, W.-T. Chen, K.-C.-W. Wu, Thermochemical conversion
reactions carried out at 400 ◦ C and for 180 min provide: (i) full con­
of plastic waste into fuels, chemicals, and value-added materials: a critical review
version of the PS (avoiding at the same time PS fouling as a deactivation and outlooks, ChemSusChem (2022) e202200171, https://doi.org/10.1002/
cause); (ii) a minimal fraction of the products suffering from over­ cssc.202200171.
cracking towards gas products; and (iii) a selectivity of naphtha and [5] F. Zhang, F. Wang, X. Wei, Y. Yang, S. Xu, D. Deng, et al., From trash to treasure:
chemical recycling and upcycling of commodity plastic waste to fuels, high-valued
middle distillates of ca. 50 wt%, with a yield of naphtha of 34 wt%. Coke chemicals and advanced materials, J. Energy Chem. 69 (2022) 369–388, https://
deposition is the main cause of the catalyst deactivation, but at 400 ◦ C doi.org/10.1016/j.jechem.2021.12.052.
the catalyst activity achieves a pseudo-stable value of the 50 % of the [6] R. Palos, A. Gutiérrez, F.J. Vela, M. Olazar, J.M. Arandes, J. Bilbao, Waste Refinery:
the valorization of waste plastics and end-of-life tires in refinery units. A review,
initial one from the 120 min of reaction. Energy and Fuels 35 (2021) 3529–3557, https://doi.org/10.1021/acs.
The kinetic model is attractive for the simulation of continuous re­ energyfuels.0c03918.
actors with the aim of scaling up the process of the hydrocracking of PS [7] G. Lopez, M. Artetxe, M. Amutio, J. Bilbao, M. Olazar, Thermochemical routes for
the valorization of waste polyolefinic plastics to produce fuels and chemicals. A
and VGO, which is a rational solution for the management of waste review, Renew. Sustain Energy Rev. 73 (2017) 346–368, https://doi.org/10.1016/
plastics. The application of this model to different plastics is now to be j.rser.2017.01.142.
applied, with optimistic predictions based on its consideration as an [8] Osman AI, Farrell C, Al-Muhtaseb AH, Al-Fatesh AS, Harrison J, Rooney DW.
Pyrolysis kinetic modelling of abundant plastic waste (PET) and in-situ emission
individual lump, in such a way that the kinetic parameters can be
monitoring n.d. 10.1186/s12302-020-00390-x.
recalculated in accordance with each polymer’s properties and their [9] G. Lopez, M. Artetxe, M. Amutio, J. Alvarez, J. Bilbao, M. Olazar, Recent advances
synergistic effects with the VGO. Even though the obtained kinetic pa­ in the gasification of waste plastics. A critical overview, Renew. Sustain. Energy
Rev. 82 (2018) 576–596, https://doi.org/10.1016/j.rser.2017.09.032.
rameters correspond to the hydrocracking of a PS/VGO (10 wt%)
[10] E. Rodríguez, A. Gutiérrez, R. Palos, F.J. Vela, M.J. Azkoiti, J.M. Arandes, et al.,
mixture, the methodology here employed can be extended to different Co-cracking of high-density polyethylene (HDPE) and vacuum gasoil (VGO) under
compositions and feeds (including different plastics and refinery refinery conditions, Chem. Eng. J. 382 (2020), https://doi.org/10.1016/j.
streams). In addition, the applicability of the model does not only focus cej.2019.122602.
[11] E. Rodríguez, R. Palos, A. Gutiérrez, D. Trueba, J.M. Arandes, J. Bilbao, Towards
on naphtha production but also on targeting different products. Last, the waste refinery: Co-feeding HDPE pyrolysis waxes with VGO into the catalytic
great effect of deactivation in the results encourages deepening in the cracking unit, Energy Convers Manag. 207 (2020), 112554, https://doi.org/
kinetic modeling of deactivation mechanisms based on the experimental 10.1016/j.enconman.2020.112554.
[12] R. Palos, E. Rodríguez, A. Gutiérrez, J. Bilbao, J.M. Arandes, Cracking of plastic
study of the individual incidence of coke and metals deposition on the pyrolysis oil over FCC equilibrium catalysts to produce fuels: kinetic modeling,
catalyst activity decay. Fuel 316 (2022), 123341, https://doi.org/10.1016/j.fuel.2022.123341.
[13] D. Munir, M.F. Irfan, M.R. Usman, Hydrocracking of virgin and waste plastics: A
detailed review, Renew. Sustain. Energy Rev. 90 (2018) 490–515, https://doi.org/
Declaration of Competing Interest 10.1016/j.rser.2018.03.034.
[14] V.L. Mangesh, P. Tamizhdurai, P. Santhana Krishnan, S. Narayanan, S. Umasankar,
The authors declare that they have no known competing financial S. Padmanabhan, et al., Green energy: Hydroprocessing waste polypropylene to
produce transport fuel, J Clean Prod 276 (2020), 124200, https://doi.org/
interests or personal relationships that could have appeared to influence 10.1016/j.jclepro.2020.124200.
the work reported in this paper. [15] B.C. Vance, P.A. Kots, C. Wang, Z.R. Hinton, C.M. Quinn, T.H. Epps, et al., Single
pot catalyst strategy to branched products via adhesive isomerization and
hydrocracking of polyethylene over platinum tungstated zirconia, Appl. Catal. B
Data availability
Environ. 299 (2021), 120483, https://doi.org/10.1016/j.apcatb.2021.120483.
[16] I.-H. Choi, H.-J. Lee, G.-B. Rhim, D.-H. Chun, K.-H. Lee, K.-R. Hwang, Catalytic
The authors are unable or have chosen not to specify which data has hydrocracking of heavy wax from pyrolysis of plastic wastes using Pd/Hβ for
been used. naphtha-ranged hydrocarbon production, J. Anal. Appl. Pyrol. 161 (2022),
105424, https://doi.org/10.1016/j.jaap.2021.105424.
[17] J.W. Thybaut, G.B. Marin, Multiscale aspects in hydrocracking: from reaction
Acknowledgments mechanism over catalysts to kinetics and industrial application, Adv. Catal. 59
(2016) 109–238, https://doi.org/10.1016/bs.acat.2016.10.001.
[18] F.J. Vela, R. Palos, J. Bilbao, J.M. Arandes, A. Gutiérrez, Effect of co-feeding HDPE
This work has been carried out with the financial support of the on the product distribution in the hydrocracking of VGO, Catal. Today 353 (2020)
Ministry of Science, Innovation and Universities (MICIU) of the Spanish 197–203, https://doi.org/10.1016/j.cattod.2019.07.010.

10
D. Trueba et al. Chemical Engineering Journal 451 (2023) 138709

[19] F.J. Vela, R. Palos, D. Trueba, J. Bilbao, J.M. Arandes, A. Gutiérrez, Different supported catalysts, Energy Fuels 25 (2011) 3389–3399, https://doi.org/10.1021/
approaches to convert waste polyolefins into automotive fuels via hydrocracking ef200523g.
with a NiW/HY catalyst, Fuel Process. Technol. 220 (2021), 106891, https://doi. [41] P. Torres-Mancera, J. Ancheyta, J. Martínez, Deactivation of a hydrotreating
org/10.1016/j.fuproc.2021.106891. catalyst in a bench-scale continuous stirred tank reactor at different operating
[20] C.I. Akor, A.I. Osman, C. Farrell, C.S. McCallum, W. John Doran, K. Morgan, et al., conditions, Fuel 234 (2018) 326–334, https://doi.org/10.1016/j.
Thermokinetic study of residual solid digestate from anaerobic digestion, Chem. fuel.2018.06.122.
Eng. J. 406 (2021), 127039, https://doi.org/10.1016/J.CEJ.2020.127039. [42] E. Rodríguez, R. Palos, A. Gutiérrez, J.M. Arandes, J. Bilbao, Scrap tires pyrolysis
[21] A. Quitian, J. Ancheyta, Experimental methods for developing kinetic-models for oil as a co-feeding stream on the catalytic cracking of vacuum gasoil under fluid
hydrocracking reactions with slurry-phase catalyst using batch reactors, Energy catalytic cracking conditions, Waste Manag 105 (2020) 18–26, https://doi.org/
Fuels 30 (2016) 4419–4437, https://doi.org/10.1021/acs.energyfuels.5b01953. 10.1016/j.wasman.2020.01.026.
[22] P.J. Becker, N. Serrand, B. Celse, D. Guillaume, H. Dulot, Comparing hydrocracking [43] L. Li, S. Zuo, P. An, H. Wu, F. Hou, G. Li, et al., Hydrogen production via steam
models: continuous lumping vs. single events, Fuel 165 (2016) 306–315, https:// reforming of n-dodecane over NiPt alloy catalysts, Fuel 262 (2020), 116469,
doi.org/10.1016/j.fuel.2015.09.091. https://doi.org/10.1016/J.FUEL.2019.116469.
[23] E. Rodríguez, G. Félix, J. Ancheyta, F. Trejo, Modeling of hydrotreating catalyst [44] J. Zhang, X. Hu, B. Yang, N. Su, H. Huang, J. Cheng, et al., Novel synthesis of PtPd
deactivation for heavy oil hydrocarbons, Fuel 225 (2018) 118–133, https://doi. nanoparticles with good electrocatalytic activity and durability, J. Alloy. Compd.
org/10.1016/j.fuel.2018.02.085. 709 (2017) 588–595, https://doi.org/10.1016/J.JALLCOM.2017.03.202.
[24] Z. Till, T. Varga, L. Szabó, T. Chován, Identification and observability of lumped [45] I. Elizalde, J. Ancheyta, Modeling catalyst deactivation during hydrocracking of
kinetic models for vacuum gas oil hydrocracking, Energy Fuels 31 (2017) atmospheric residue by using the continuous kinetic lumping model, Fuel Process.
12654–12664, https://doi.org/10.1021/acs.energyfuels.7b02040. Technol. 123 (2014) 114–121, https://doi.org/10.1016/j.fuproc.2014.02.006.
[25] Z. Till, T. Varga, J. Sója, N. Miskolczi, T. Chován, Reduction of lumped reaction [46] A. Monzón, E. Romeo, A. Borgna, Relationship between the kinetic parameters of
networks based on global sensitivity analysis, Chem. Eng. J. 375 (2019), https:// different catalyst deactivation models, Chem. Eng. J. 94 (2003) 19–28, https://doi.
doi.org/10.1016/j.cej.2019.121920. org/10.1016/S1385-8947(03)00002-0.
[26] B. Browning, P. Alvarez, T. Jansen, M. Lacroix, C. Geantet, M. Tayakout-Fayolle, [47] J. Corella, A. Monzón, Modeling of the deactivation kinetics of solid catalysts by
A review of thermal cracking, hydrocracking, and slurry phase hydroconversion two or more simultaneous and different causes, Ind. Eng. Chem. Res. 27 (1988)
kinetic parameters in lumped models for upgrading heavy oils, Energy Fuels 35 369–374, https://doi.org/10.1021/ie00075a001.
(2021) 15360–15380, https://doi.org/10.1021/acs.energyfuels.1c02214. [48] A.G. Gayubo, A.T. Aguayo, M. Olazar, R. Vivanco, J. Bilbao, Kinetics of the
[27] G. Félix, J. Ancheyta, Comparison of hydrocracking kinetic models based on SARA irreversible deactivation of the HZSM-5 catalyst in the MTO process, Chem. Eng.
fractions obtained in slurry-phase reactor, Fuel 241 (2019) 495–505, https://doi. Sci. 58 (2003) 5239–5249, https://doi.org/10.1016/J.CES.2003.08.020.
org/10.1016/j.fuel.2018.11.153. [49] S.S. Amin, H. Abdollahi, A. Naseri, The influence of coupled kinetic-equilibrium in
[28] A. Bin Jumah, M. Malekshahian, A.A. Tedstone, A.A. Garforth, Kinetic modeling of breaking the correlation between kinetic parameters, Chemom. Intell. Lab. Syst.
hydrocracking of low-density polyethylene in a batch reactor, ACS Sustain Chem. 215 (2021), 104349, https://doi.org/10.1016/j.chemolab.2021.104349.
Eng. 9 (2021) 16757–16769, https://doi.org/10.1021/acssuschemeng.1c06231. [50] H.H. Pham, N. Thuy Nguyen, K.S. Go, S. Park, N. Sun Nho, G.T. Kim, et al., Kinetic
[29] G. Félix, J. Ancheyta, Using separate kinetic models to predict liquid, gas, and coke study of thermal and catalytic hydrocracking of asphaltene, Catal. Today 353
yields in heavy oil hydrocracking, Ind. Eng. Chem. Res. 58 (2019) 7973–7979, (2020) 112–118, https://doi.org/10.1016/j.cattod.2019.08.031.
https://doi.org/10.1021/acs.iecr.9b00904. [51] H.H. Pham, K.H. Kim, K.S. Go, N.S. Nho, W. Kim, E.H. Kwon, et al., Hydrocracking
[30] L. Han, X. Fang, C. Peng, T. Zhao, Application of discrete lumped kinetic modeling and hydrotreating reaction kinetics of heavy oil in CSTR using a dispersed catalyst,
on vacuum gas oil hydrocracking, China Pet Process Petrochemical. Technol. 15 J. Pet Sci. Eng. 197 (2021), 107997, https://doi.org/10.1016/j.
(2013) 67–73. petrol.2020.107997.
[31] D. Trueba, R. Palos, J. Bilbao, J.M. Arandes, A. Gutiérrez, Product composition and [52] S. Liu, P.A. Kots, B.C. Vance, A. Danielson, D.G. Vlachos, Plastic waste to fuels by
coke deposition in the hydrocracking of polystyrene blended with vacuum gasoil, hydrocracking at mild conditions, Sci. Adv. 7 (2021) eabf8283, https://doi.org/
Fuel Process. Technol. 224 (2021), 107010, https://doi.org/10.1016/j. 10.1126/sciadv.abf8283.
fuproc.2021.107010. [53] G. Schultz, R. Alexander, F.V. Lima, R.C. Giordano, M.P.A. Ribeiro, Kinetic
[32] R. Palos, A. Gutiérrez, J.M. Arandes, J. Bilbao, Upgrading of high-density modeling of the enzymatic synthesis of galacto-oligosaccharides: Describing
polyethylene and light cycle oil mixtures to fuels via hydroprocessing, Catal. Today galactobiose formation, Food Bioprod. Process. 127 (2021) 1–13, https://doi.org/
305 (2018) 212–219, https://doi.org/10.1016/j.cattod.2017.06.033. 10.1016/j.fbp.2021.02.004.
[33] T. Wang, M. Chen, X. Liu, Z.G. Zhang, Y. Xu, Distinguishing external and internal [54] F. Ketzer, F. de Castilhos, An assessment on kinetic modeling of esterification
coke depositions on micron-sized HZSM-5: Via catalyst-assisted temperature- reaction from oleic acid and methyl acetate over USY zeolite, Microporous
programmed oxidation, New J. Chem. 43 (2019) 13938–13946, https://doi.org/ Mesoporous Mater. 314 (2021), 110890, https://doi.org/10.1016/j.
10.1039/c9nj02899d. micromeso.2021.110890.
[34] J. Martínez, J. Ancheyta, Kinetic model for hydrocracking of heavy oil in a CSTR [55] A. Bin Jumah, A.A. Tedstone, A.A. Garforth, Hydrocracking of virgin and post-
involving short term catalyst deactivation, Fuel 100 (2012) 193–199, https://doi. consumer polymers, Microporous Mesoporous Mater. 315 (2021), 110912, https://
org/10.1016/j.fuel.2012.05.032. doi.org/10.1016/j.micromeso.2021.110912.
[35] T. Cordero-Lanzac, A.T. Aguayo, A.G. Gayubo, P. Castaño, J. Bilbao, Simultaneous [56] T. Cordero-Lanzac, R. Palos, J.M. Arandes, P. Castaño, J. Rodríguez-Mirasol,
modeling of the kinetics for n-pentane cracking and the deactivation of a HZSM-5 T. Cordero, et al., Stability of an acid activated carbon based bifunctional catalyst
based catalyst, Chem. Eng. J. 331 (2018) 818–830, https://doi.org/10.1016/j. for the raw bio-oil hydrodeoxygenation, Appl. Catal. B Environ. 203 (2017)
cej.2017.08.106. 389–399, https://doi.org/10.1016/j.apcatb.2016.10.018.
[36] C.J. Calderón, J. Ancheyta, Modeling of CSTR and SPR small-scale isothermal [57] T. Cordero-Lanzac, I. Hita, F.J. García-Mateos, P. Castaño, J. Rodríguez-Mirasol,
reactors for heavy oil hydrocracking and hydrotreating, Fuel 216 (2018) 852–860, T. Cordero, et al., Adaptable kinetic model for the transient and pseudo-steady
https://doi.org/10.1016/j.fuel.2017.11.089. states in the hydrodeoxygenation of raw bio-oil, Chem. Eng. J. 400 (2020),
[37] S. Sánchez, J. Ancheyta, Effect of pressure on the kinetics of moderate 124679, https://doi.org/10.1016/j.cej.2020.124679.
hydrocracking of Maya crude oil, Energy Fuels 21 (2007) 653–661, https://doi. [58] O. Dogu, M. Pelucchi, R. Van de Vijver, P.H.M. Van Steenberge, D.R. D’hooge,
org/10.1021/ef060525y. A. Cuoci, et al., The chemistry of chemical recycling of solid plastic waste via
[38] M. Schwaab, J.C. Pinto, Optimum reference temperature for reparameterization of pyrolysis and gasification: State-of-the-art, challenges, and future directions, Prog.
the Arrhenius equation. Part 1: problems involving one kinetic constant, Chem. Energy Combust. Sci. 84 (2021), 100901, https://doi.org/10.1016/J.
Eng. Sci. 62 (2007) 2750–2764, https://doi.org/10.1016/j.ces.2007.02.020. PECS.2020.100901.
[39] T. Cordero-Lanzac, I. Hita, A. Veloso, J.M. Arandes, J. Rodríguez-Mirasol, J. Bilbao, [59] M.S. El-Sawy, S.A. Hanafi, F. Ashour, T.M. Aboul-Fotouh, Co-hydroprocessing and
et al., Characterization and controlled combustion of carbonaceous deactivating hydrocracking of alternative feed mixture (vacuum gas oil/waste lubricating oil/
species deposited on an activated carbon-based catalyst, Chem. Eng. J. 327 (2017) waste cooking oil) with the aim of producing high quality fuels, Fuel 269 (2020),
454–464, https://doi.org/10.1016/J.CEJ.2017.06.077. 117437, https://doi.org/10.1016/j.fuel.2020.117437.
[40] A. Gutiérrez, J.M. Arandes, P. Castaño, A.T. Aguayo, J. Bilbao, Role of acidity in [60] Z. Cao, X. Zhang, C. Xu, X. Huang, Z. Wu, C. Peng, et al., Selective hydrocracking of
the deactivation and steady hydroconversion of light cycle oil on noble metal light cycle oil into high-octane gasoline over bi-functional catalysts, J. Energy
Chem. 52 (2021) 41–50, https://doi.org/10.1016/j.jechem.2020.04.055.

11