1 s2.0 S0009250919306876 Main

Chemical Engineering Science 210 (2019) 115195
Contents lists available at ScienceDirect
Chemical Engineering Science

journal homepage: www.elsevier.com/locate/ces
Hydrogenation and hydrodesulfurization in gas phase of light

hydrocarbons from hydrocracking, desulfurization and delayed coking.
I catalyst deactivation
Roberto Galiasso Tailleur
Simon Bolivar University, Dep. Thermodynamics, Sartenejas, Baruta, Miranda 8900, Venezuela
HyPro Consultant 4250 Corrine Dr., Suite 204, Orlando, FL 32814, USA
h i g h l i g h t s g r a p h i c a l a b s t r a c t
Co-process gas phase hydrocracking

products and Delayed Coker products. Cat. Make up
Use a fluidized bed to measure two
catalysts deactivations.
Determine kinetic and deactivation Hydrogenation Recycle
rates of reactions using three active
sites model.
A reactor model with three
compartments, mas transfer,
crossflow and backflux.
DCN
GPR
Data about benzene diffusivity, XPS,
SEM-AES, MAS 13C NMR and CS2
Fractionation
HPHTS
extraction.
HCK
Study the relative stability of feed and
products at different temperatures. Prod.
Discuss catalyst deactivation by
Cat. Regen.
olefins base on solids and liquid TBR.
analyses.
a r t i c l e i n f o a b s t r a c t
Article history: Hydrogenation and hydrodesulfurization reactions of hydrocarbons in gas phase was studied in a
Received 21 December 2018 fluidized-bed reactor using PdNiW/TiO2-Al2O3 and NiMo/TiO2Al2O3 catalysts. Catalyst deactivation was
Received in revised form 15 August 2019 performed at different temperatures and using hydrocracking product with and without delayed coking
Accepted 31 August 2019
naphtha feeds. Delta of species and catalysts activities were measured, and catalyst characterized by
Available online 11 September 2019
physical and chemical properties, metals and carbon dispersion, acid sites content and benzene diffusiv-
ity in hydrogen. Coke was extracted by CS2 and analyzed by MAS 13C NMR. There are different rates of
Keywords:
sites deactivation when processing gas feed with or without olefinic hydrocarbons on these catalysts.
Kinetic and deactivation rates
Hydrogenation
Olefins and diolefins produce carbon deposits fast; carbon distribution across the radius of particle
Coke deposition depend on temperature and amounts of aromatics in the feed. Reaction and deactivation rates constants
Reactor simulation were obtained for each of catalysts. The simulation model used only predicted well products properties
Coke composition without olefins and diolefins. The mechanism of reaction are discusses based on solid characterizations.
Ó 2019 Elsevier Ltd. All rights reserved.
E-mail address: [email protected]
https://doi.org/10.1016/j.ces.2019.115195
0009-2509/Ó 2019 Elsevier Ltd. All rights reserved.
2 R.G. Tailleur / Chemical Engineering Science 210 (2019) 115195
Nomenclature
a deactivation constant [–] Lmf minimum fluidization height [cm]

At cross sectional area of the reactor [cm2] Lr length of reactor [cm3]
Cc coke concentration wt%, or [mmol/cm3] N number of compartments in series [–]
CCH4 concentration of methane as a function of time (tracer) n2 number of compartments in series in the core region [–]
[mmol/cm3] P pressure [MPa]
Ci mean concentration of species mmol/cm3 P objective function
Deff effective hydrogen diffusivity [cm2/s] QFC volumetric rate of crossflow [mol/cm3/s]
db average bubble diameter [cm] R universal gas constant (8.314) [J/(mol K]
Dfb diameter of the freeboard [cm] r radial position [cm]
dp particle average diameter [microns] ri reaction rate of specie i [mmol/(cm3 s)]
Dr reactor diameter [cm] t contacts time [h]
E(t) RTD pulse function [–] T temperature [K]
E0 dense bed surface flux [kg/(m2 s)] u forward solid flowrate [cm3/s]
g gravitational acceleration [cm/s2] Ub bubbling velocity [cm/s]
GPSV gas phase space velocity [h1] Ue emulsion velocity [cm/s]
HL bed height [cm] Umf minimum fluidization velocity [cm/s]
k0i pre-exponential reaction rate constant [mmol/cm3/s/ V, Vb, Ve volume of reactor, bubble, and emulsion [cm3]
Pa0.3] w backward solid flowrate [cm3/s]
KBE Interphase mass exchange coefficient between bubble W masses flowrate [g/s]
and emulsion [s1] Xj delta of species [–]
1. Introduction hydrogenation and hydrodesulfurization of diesel gas oil and vac-

uum residue were studied in pilot plant long term test in 2008
The light products of VGO and residue hydrocracking, of (Galiasso Tailleur, 2008). Data for the dynamic gas liquid equilib-
hydrodesulfurization of kerosene and diesel, of delayed coking rium and the process scheme was developed previously (Galiasso
and pyrolysis of biomass require and additional hydrogenation Tailleur, 2017).
and desulfurization to fulfill the refinery specifications. The con- The HPHTS produce a gaseous hydrocarbon stream that can be
cept of deep hydrogenation and desulfurization using a high- process in fixed bed or moving bed or a fluidized bed reactor to
pressure, high-temperature separator (HTHPS) and a gas and obtain the product specification to the refinery specification
trickle-bed reactors operating in parallel was introduced Galiasso (olefins, benzene, diaromatics sulfur content.) See process scheme
Tailleur et al. (2002), see Fig. 1, and then the concept were in Fig. 1. The author had demonstrated that in gas phase, without
extended by Galiasso Tailleur (2006) to mild-hydrocracking. The liquid phase wetting the solids, the same catalysts used in the trickle
use of the HTHPS and two in-parallel reactors, one for vaporized bed (NiMo/Al2O3 or PdNiW/Al2O3) have higher activity and selectiv-
feed (gas-phase reactor) and another for non-vaporized feed ity in sulfur nitrogen and aromatics conversion, as well as they pre-
(trickle bed reactor), were simulated in 2008 (Galiasso Tailleur sent lower deactivation than in trickle bed (Galiasso Tailleur et al.,
et al., 2008). The long-term deactivation that occurred during the 2008; Galiasso Tailleur, 2014). Thus, the use of one or several
Fresh catalysts
Pyrolysis Naphtha
Coker Naphtha
Light Catalytic Naphtha
GPR
Spent catalysts
Regeneration
Vac H2 to recycle
HPHTS
Gas Oil
Naphtha
HCK
Diesel
H2
TBR
FRACTIONATION
Recycled VGO
Fig. 1. Process scheme to hydrotreat the gas phase of a hydrocracking or Mild-hydrocracking unit together with other external Feeds.
R.G. Tailleur / Chemical Engineering Science 210 (2019) 115195 3
high-pressure high temperature separators (HPHTS) make possible genated to use these hydrocarbons as a fuel. There are few refer-
to treat gas (vapor) phase independently of liquid phase. This tech- ences about the selective hydrogenation of olefins and diolefines
nology can be also used after one or two stages hydrocracking or in commercial application. For example, Bunyard (1999) discuss
hydrodesulfurization. The feed can be combined with others the need to hydrotreat DCN before desulfurization and present
gaseous streams from the refinery and treated in the same reactor very useful correlations to predict the extent of hydrogenation of
(e.g. naphtha from pyrolysis, bio-oils, delayed coking and cracking olefins; diolefins, and aromatics, as well as sulfur and nitrogen
products.) The catalyst can convert different molecules present in removals. Chernysheva et al. (2005) propose the use of naphthas
gas phase that may not access the catalyst in trickle-bed operation. as Components of Hydrotreating Feedstock.
Gas phase reactors are currently in use for different reactions Fundamental catalyst deactivation by coke was analyzed by
(hydrogenation, desulfurization and reforming, among many many authors, some of them mentioned in Argyle and Bartholmew
others) that treat light hydrocarbons (gases, naphtha from biomass review (Argyle and Bartholomew, 2015). Galiasso Tailleur group
pyrolysis, delay coking, and others). Commercial experience in develop new catalyst and study its catalyst deactivation by aro-
hydrogenation of diolefins and olefins, light naphtha from Delayed matics during cracked gas oil upgrading to high quality diesel
Coker (DC), catalytic cracking (FCC), pyrolysis oil from biomass and (Galiasso Tailleur, 2007); they also model the catalyst deactivation
Hydrocracked product from heavy residue streams, show that it is (Galiasso Tailleur, 2005) and measured the effective diffusivity in
a difficult tasks. When the feed contains olefinic-hydrocarbons, spent hydrotreating catalysts (Galiasso Tailleur, 2010). Galiasso
they are fast polymerized and plug pores of catalyst and the inlet et al. (1983) demonstrated how critical is the effects of pores size
of fixed bed reactor. More than that, they cannot be cooled and distribution in long term test in pilot plant. The influence of cata-
re-heated because also plug the heat exchangers. They need to lyst acidity, pore distribution and feedstock quality in coke accu-
be fed directly from the process to the HPHTS where temperature mulation was studied by several authors, for example see the
is adjusted by the other streams (see Fig. 1) Recently, I had demon- results obtained by Marafi and Stanislaus (2001). Wivel et al.
strated in commercial operation that by including more than 5% of (1991) studied the effects of initial coking and deactivation of
pyrolysis oil from biomass or more than 10% of delayed coke naph- hydrotreating catalysts, and Wood and Gladden (2003) discussed
tha in the hydrotreating feed decrease by more than half the life of the effect of coke deposition upon pore structure and self-
hydrodesulfurization catalysts and plug the fixed- bed reducing diffusion in deactivated industrial hydroprocessing catalysts.
the cycle length. Pacheco et al. (2011) propose an accelerated deactivation of
There are few papers dedicated to the hydrogenation of hydrotreating catalysts by coke deposition. In this paper we
Delayed-Coker naphtha (DCN), pure or in blends; among them explore the idea of accelerating catalyst deactivation to understand
are those of the group of Miller et al. (2000) and Diez et al. the impact of hydrotreat a hydrocracking product plus some
(1990). Badawi et al. (2010) reported the kinetic of olefin hydro- delayed coke product.
genation on hydrotreating catalysts. Magyar et al. (2008) studied There are many fundamental studies about the mechanisms
the conversion of olefins and diolefins on commercial NiMo and that form the so called ‘‘coke”. For example, Guisnet and
CoMo catalysts and Toba et al. (2005) described the selective Magnoux (2001), among others, review the mechanism of coke for-
hydrodesulphurization of FCC gasoline over CoMo/Al2O3 sulfide mation from different species of hydrocarbons at different temper-
catalyst. Lengyel et al. (2010) signaled the problem of catalytic atures, and Bartholomew (2001) discuss the different types of coke
coprocessing of Delayed Coker Light Naphtha with Light Straight- formed, but did not correlated the information with their origins.
run Naphtha/FCC Gasoline. They do not disclose about mechanism Zeuthen et al. (1994) studied the formation of coke on hydrotreat-
and rates of active sites during catalyst deactivations. Xin et al. ing catalysts, and Díez et al. (1990) analyzed the deactivation of
(2018) studied the hydrogenation of olefins in Bitumen-Derived Ni–Mo/c-Al2O3 catalyst. Vogelaar et al. (2007) among many others,
Naphtha over a Commercial Hydrotreating Catalyst; they describe report the radial deposition of coke measured by Raman Spec-
in detail the instability associated with the presence of olefins and troscopy. Everybody agrees about the effects of diffusion of species
the selective hydrogenation of olefin-rich fractions and rank the in coke distribution inside the particle. Despite the abundant liter-
reactivities of the different olefin/diolefin compounds. ature, control of coke deposition on acid and metals sites is an open
Catalyst deactivation during hydrotreating of other cracked subject of research. Our group study the effects of processing a gas-
products has been studied by many authors. For example, Diez eous fraction of a hydrocracking products with and without
et al. (1990) study the deactivation of NiMo catalyst by LCCO and delayed coking naphtha to determine the rates of catalyst
observe a progressive deposition of coke by 13C NMR during 20 h deactivation.
of operation. The hydrotreating of these feeds were performed in a fluidized
Hydrogenation in fluidized bed reactor had been mentioned for bed reactor for two catalysts having different active phase but the
different applications. Diao et al. (2005), for example, studied the same support. The kinetic and deactivation rates, as well as the
gaseous catalytic hydrogenation of nitrobenzene to aniline in a fluid dynamic were determined and a new sets of rates constants
two-stage fluidized bed reactor. Another application is for carbon obtained. Different temperatures have been used to accelerate
dioxide hydrogenation (Kim et al., 2001). Nevertheless, there is the catalysts deactivations. Complementary information (CI) sup-
no information published about fluidization in hydrogen at high ply description of the method used to study the reactor and the dif-
pressure and high temperature. fusivity in spent catalyst.
Thermal cracking produces naphtha, kerosene and diesel frac-
tions that contains variable amount of sulfur and nitrogen, and 2. Experimental
proportionally high amounts of olefins and diolefins respect to
similar straight run feeds. Commercial practice treats these prod- 2.1. Catalyst preparation
ucts diluted in straight run naphtha and diesel to avoid severe
deactivation of catalyst and bed plugging. To achieve other product PdNiW/TiO2-Al2O3 and NiMo/TiO2-Al2O3 catalysts were used to
specifications (aromatic, sulfur, and so on) a high pressure hydro- demonstrate the effects of accelerated deactivation in the perfor-
treater is needed. This unit seldom use a ‘‘guard” reactor to pre- mance of fluidized bed reactor pilot plant reactor. The support of
convert diolefins and olefins at low temperature and then remove catalysts was prepared according to a previously developed proce-
sulfur, nitrogen, aromatics in a trickle bed reactor. Delayed coke dure (Galiasso Tailleur and Ravigli, 2005) that coprecipitated an
(DCN) and Pyrolysis gasoline (PG) must be completely hydro- alkoxide of titanium and alumina in the presence of ammonium
hydroxide to produce a sol-gel. This sol gel was then filtered, aged, sample. EDS detect all metals present on surface with an average
dried and agglomerate in a continuous rotating-plate using mother depth of 30 Å. 20 radial positions were explored to determine the
liquid as a binder. The particles of 80–120 mm of diameter obtained average carbon and main metal content and then used six radial
were dried at 390 K in a fluidized bed continuous dryer, calcined in position to determine the profile of C/RMe and C/Cav ratios. These
air at 450 K and treated with steam and ammonia at 573 K for six values were calculated using different beam energies and samples
hours, all in the same equipment to form a structured mesoporous with different deactivation times on stream and temperatures.
system. Final solid was either impregnated by: Effective diffusivity of benzene in hydrogen was determined
using a modified Wicke-Kallemback method (shallow bed, high-
(1) Tungstate salt in water. Then the solid was dried at 320 K pressure micro-reactor) at different level of coke on catalysts; the
and re-impregnated with a palladium and nickel nitrate salt method measured at 603 K the pulse of benzene, the rates of
in water. The catalyst A was calcined at 643 K and then the adsorption desorption (Galiasso et al., 1983) (see CI) using a stream
active phase pre-sulfided at 643 K for 6 h using HCK-G feed of 1% of benzene in hydrogen. The ratios of the tail and main peak
in hydrogen spiked with 1000 ppm of S2C at 6 MPa. areas were measured, and the first, second, and third momentum
(2) Molybdenum salt in water. Then the solid was dried at 320 K of E(t) curve (age function) were calculated from data at different
and then re-impregnated with nickel nitrate salts. The cata- level of catalysts deactivations; the values were used to evaluate
lyst B was calcined in air at 643 K, and then the active phase the effects of coke on average effective molecular diffusivity (Deff)
pre-sulfided at 643 K for 6 h using HCK-G feed in hydrogen and external pore mouths plugging (Galiasso et al., 1983).
spiked with 1000 ppm of S2C at 6 MPa. The families of hydrocarbon distributed by carbon were deter-
Properties of fresh catalysts A and B are depicted in Table 1 and mined by extended PIONA analyses (GC-FID, GC-AED, GC–MS and
on surface dispersion of metals (XPS) for fresh and spent samples elemental analyses.) GC-FID analyses of liquids were carried out
in Table 2. in HP 5970 equip using a Restek 15,118 column of 60 m in length,
0.2 mm in diameter, and 0.5 lm in fill thickness; the carrier was
2.2. Analyses of gases and solids hydrogen. GC–MS analyses of feed and product used a HP 6890,
equiped with an EliteTM column of 30 m 0.25 mm 0.25 lm.
Chemical composition of the catalysts was determined by atomic Mass spectra were obtained in a HP 5921 with gas delay time of
absorption (Table 1). The average values of carbon and sulfur con- 6 min, a scan time of 0.20 sec and a interscan delay time of 0.05
tent on catalysts were measured by micro-combustion analyses, sec. The spectrum of each analyte has a unique ‘‘fingerprint” that
textural properties of solids using ASAP 2400 Micromeritics equip, were analyzed using Ion Fingerprint Detection algorithm (Medford,
acidity by adsorption-desorption of pyridine at two temperatures, MA) based on a NIST Spectra Library. These analyses were cali-
and metal dispersion by X-ray-photoelectron-spectroscopy (Lay- brated with model molecules to verify the MS resolutions. All the
bous Hereus apparatus, Mg cathode, 50 eV of power). Metals disper- information is consolidated base on C, S, N and H mass balances
sions were measured by deconvolution of metals signals using Al as in Tables 3–5.
reference (Vogelaar et al., 2007) (Table 2) and radial distribution of Total conjugated dienes (DOL) content in DCN and hydrotreated
metals and carbon content analyzed using Jeol JSM 5800 scanning products were measured by ASTM D2930, despite of criticism that
microscope and INCA EDS. The apparatus uses an electron beans maleic anhydride (MA) may react with some diaromatics. Olefins
of 500 Å in diameter to generate secondary electron images of the were measured by Bromine number (ASTM D1159-07(2017),
Table 1
Catalyst properties.
Properties type of Cat. A B

Fresh 40 h Deact*. Fresh 40 h Deact*.
PdO wt% 0.1 –
MoO3 wt% – 13.5
NiO wt% 5.4 4.8
WO3 wt% 12.3 –
Coke wt% (613 K) 0 1.8 2.9 0 2.0 3.3
Surf Area m2/g 189 122 96 195 113 90
Pore Vol. cm3/g 0.45 0.41 0.39 0.44 0.4 0.38
Av mesop diam lm 88 89 87 90 91 90
Py lmol/g (340 K) 2.23 1.2 0.45 2.37 1.05 0.34
Deff cm/s (Benz/H2) 3.0 106 – 1.8 106 2.6 106 – 1.2 106
*
HCK feed 603 K, 10 MPa.
Table 2
On surface catalysts properties.
Properties, feeds hours on stream PdWNi (A) HCK (A) 623 K MoNi (B) HCK (B) 623 K
40 292 40 292
IS/RIMe (XPS) % 2.0 2.1 2.3 2.6 2.7 2.8
IW/RIMe (XPS) % 6.7 4.2 3.3 – – –
IMo/RIMe (XPS) % – – – 7.1 5.8 3.4
INi/RIMe (XPS) % 4.4 2.4 1.6 4.2 3.2 2.1
IC/RIMe (XPS) % 0.1 1.3 3.8 0.1 1.5 4.1
mmol Py/g (370 K) 1.3 0.9 0.2 1.6 0.8 0.2
Deactivated (623 K) HCK + DCN (A) HCK + DCN (B)
IC/RIMe (XPS) % 4.2 4.3
Table 3
Inlet Carbon distribution, family for a (90 HCK-G/10% DCN.
Properties wt% HCK-G Py Blend 0.9 + 0.1 Olefins blend Diolefins blend
q (kg/L) 0.753 0.685 0.749 – –
HCS1 C1 0.2 0.5 0.215 – –
C2 0.7 3 0.815 – –
C3 1.18 5 1.371 0.05 –
C4 2.21 7 2.4495 0.07 0.1
PSC2 C5 5.12 12 5.464 0.2 0.3
C6 6.12 21 6.864 0.1 0.4
C7 12.4 20 12.78 0.05 –
C8 16.3 17.2 16.345 0.1 0.1
C9 9.8 9.5 9.785 0.64 0.12
PSC3 C10 6.2 4.3 6.105
C11 3.6 1.2 3.48 Types OL Types DOL
C12 4.5 0.7 4.275 1-Butene C==
4
C13 5.9 0.4 5.605 isobutene C==
5
C14 5.8 0 5.51 2-butene C==
6
PSC4 C15 6.7 0 6.365 1-pentene Cyclo-C==

8
C16 6.8 0 6.46 Cyclo-pent. Cyclo-C==
9
C17 2.6 0 2.47 Cyclo-hex. –
C18 1.8 0 1.71 Heptenes –
C20 0.7 0 0.665 Al-Cyclo-h. –
PSC5 C+21 1.37 0 1.3015
Table 4
Species distribution in the Feed HCK-G and Products of HCK-G.
Feed (HCK1)
PSC’s Par Isop MN DN MAr DiAr Ben Th* BTh* DiTh* Ol* Dol*
PSC1 2.3 1.1 0 0 0 0 0 0 0 0 300 150
PSC2 3.4 4.5 8.5 0.1 3.8 0 3.4 150 230 0 150 80
PSC3 3.8 8.9 4.3 7.4 4.6 2.4 0 50 120 230 0 0
PSC4 5.7 12 4.0 4.5 10.4 4.2 0 140 220 300 0 0
PSC5 0.9 2.1 0.3 3.4 1.8 3.2 0 0 140 2000 0 0
Product (603 K 1.4 MPa, GSPV:1; H2/HC:23)
PSC1 3.2 1.8 0 0 0 0 0 0 0 0 2 0
PSC2 3.2 8.4 4.6 0.2 2.8 0 0.9 1 0 0 1 0
PSC3 4.8 15 3.8 6.4 3.6 1.8 0 0 0 2.3 0 0
PSC4 4.9 15 6 3.9 6.9 2.2 0 0 0 3 0 0
PSC5 0.8 1.9 0.3 3.4 0.8 1.9 0 – – 8 0 0
*
Ppm.
Table 5
Species distribution in Feed and Products of HCK-G + 10% DCN.
PSC’s Par Isop MN DN MAr DiAr Ben Th* BTh* DiTh* Ol* Dol*
PSC1 2.2 1.0 0 0 0 0 0 0 0 0 300 150
PSC2 3.3 4.4 8.4 0.1 3.7 0 3.4 150 230 0 150 80
PSC3 3.7 8.8 4.2 7.3 4.5 2.3 0 45 117 222 0 0
PSC4 5.6 11.9 3.9 4.4 10.3 4.1 0 1330 211 290 0 0
PSC5 0.9 2.1 0.3 3.3 1.7 3.1 0 0 124 1950 0 0
Product (603 K 1.4 MPa, GSPV:1; H2/HC:23)
PSC1 3.2 1.8 0 0 0 0 0 0 0 0 2 0
PSC2 3.2 8.4 4.6 0.2 2.8 0 0.9 1 0 0 1 0
PSC3 4.8 15 3.8 6.4 3.6 1.8 0 0 0 2.3 0 0
PSC4 4.9 15 6 3.9 6.9 2.2 0 0 0 3 0 0
PSC5 0.8 2.2 0.3 2.8 1.7 2.3 0 0 0 32 0 0
*
Ppm.
1
electrometric titration). That determine the average olefins (OL) H and 13C NMR Bruker 100 kHz) were used to determine the
species content. Detailed characterization of olefins and diolefins average degree of paraffins isomerization and Mass 13C NMR to
was performed for feed and some of products using a supercritical analyze carbon compounds present of catalysts.
CO2 fractionation, followed by elution though different type of col- Solid sample used for analyses was statistically pick-up from
umns to determined OL and DOL species by GC–MS and 1H NMR. five samples of 1 g, taken in different positions in the reactor
The products of the gas phase reactor must fulfill the pool spec- because the reactor behaves as several compartment in-series with
ifications (ca: less that 15-ppm sulfur less than 1% of olefins, ben- backflux; the values of five samples measures three times were
zene and polyaromatics, and without diolefins content.) averaged. Feeds were composed by either:
(1) Hydrocracking gas-phase product obtained from the HPHTS Fluidized bed reactor
(Fig. 1, HCK1). HCK1 is a hydrocracking commercial product,
condensed and stored under nitrogen. HCK unit operated to bubbles bubbles
produce low sulfur diesel using vacuum Gas Oil as Feed at q
75% of heavy gas oil conversion. It contains five pseudo- STR f
components (PSC1-5), see Table 3 and read point 2.4. STR STR
(2) 0.9 HCK1 with 5 wt% of C+5 DCN, mentioned as HCK2.
(3) 0.9 HCK1 with 10 wt% of C+5 DCN mentioned as HCK3.
Fig. 3. Reactor model used to simulate the fluid dynamic of the fluidized be reactor.
Hydrocarbons are fed with hydrogen (H2, 98% purity) contain-
ing 1200 ppm H2S. The HCK1 gas phase was overheated and sent
to the fluidized bed reactor in gas phase while DCN are blended (2) Gas phase plug flow reactor; gas leaving the previous
at the inlet of reactor without preheating (Fig. 2); liquid phase from compartment (I), move upward in the reactor in bubbles of
HPHTS was stored for further study in a TBR (not cover in this constant size;
paper, see reference Galiasso Tailleur et al. (2008)). (3) Gas-solid compartments (II and III). Dense phase (gas and
solid), leaving the previous compartment (I), move up into
2.3. Fluid dynamic model of the fluidized bed reactor (SBFBR) these compartments in series (well mixed) with some
degree of backflow (f:vf/vbf);
Fig. 2 shows the pilot plant flows scheme (reactor of 0.08 m (4) Mass transfer between bubbles (2) and dense phase (3).
diameter and 1.2 m high.) The fluidized bed reactor contains These zones exchange gas and solid by crossflow qc
500 g of catalyst in the range of 80–100 mm. It was operated at near (mol/m2 h) and by mass transfer (kb-e CH2 and kb-e CH2);
minimum linear gas velocity (0.1 cm/s), 14 MPa and 573 K. To
understand the movement of the gas phase and the solids, a pulse Reactions take place in compartments (I, II and III); reaction in
of gas methane (0.1 cm3) were added to the hydrogen at the inlet bubbles and freeboard are neglected. The mass transfer of gases
of the fluidized bed reactor by an automatic valve; the pulse was is calculated using Eq. (1) the degree of backflow by Eq. (2), the
measured at the inlet and outlet streams (see Fig. 2 and CI); the average bubble size by Eq. (3) and mass transfer by Eq. (4), all of
method allows to measure the movement of gases trough bubbles them determined in the previous study (Vogelaar et al., 2007):
and dense region. Samples of gas and solid were taken in the four 0:5
emf ub De
regions along the axis of the bed (pitot tube) and at the outlet of K be be
H2 ¼ K HC 7:6 ; ð1Þ
db
reactor. The gas and solid fluid dynamic information, measured
using tracer input-output concentrations, and the methodology 0:2
u uo umf
used to process the data are well described in reference Galiasso f ¼ ¼ 0:2 þ 0:025 ð2Þ
wu db
Tailleur and Farina (2019a) and will not be discussed further here
(see results and discussion in second part of this paper (Galiasso
db ðzÞ ¼ 8ðU o U mf Þ0:33 T 0:2 P0:2 dp lnz0:77 ;
0:02
ð3Þ
Tailleur and Farina, 2019b).) Fluid dynamic model that represents
the behavior of this reactor is described in Fig. 3. The elements 0:5
used in the model for the reactor are: emf ub De
K be be
H2 ¼ K HC 7:6 ð4Þ
db
(1) Gas-solid compartment (I); it is a well-mixed reactor that
The SBFBR reactor model requires information about the
represent the gas distribution zone in boundary conditions);
amounts of gases and solid in the emulsion phase and in the bub-
ble size, as well as gas and solid properties at operating conditions
Solid Flow (bed expansion, porosity at minimum fluidization emf, minimum
Fresh catalyst fluidization velocity umf, and average bubble size db) to calculate
(make up) coke concentration in each of compartments and at the outlet of
SBFBR. Fluid dynamic information is obtained from experimental
Data Logger, flow Filter bag
and pressure data because correlations available in the literature are not appli-
Gas to GC-MS cable for hydrogen-solid fluidization at high pressure (14 MPa)
controller
Tracer
and medium temperature (603–643 K).
detector
Oven, Temp.
control 2.4. Equations for the adiabatic fluidized bed reactor with continuous
flow of gases
Laser bubble Mass balance equations for the above-mentioned model con-
detector
sider a homogeneous distribution of acid sites across the radius
CH4 injection and average effective diffusivity.
For hydrogen and hydrocarbon in the compartment (I) (work as
a boundary condition).
H2 or Bubble
H2+HC dNH2bI X
N2 GCMS ¼ v o C H2o v f C H2I es gH2 ai ri ð5Þ
V I dt
S2C Injection
To vacuum dNHCI
¼ v o C HCo v f C HCI es gHC ai ðtÞr i ð6Þ
Solid To CNMR and C analyses V I dt
Fig. 2. Pilot plant fluidized bed hydrotreater. Sensors used to control the fluid T S;I ¼ T g;I ð7Þ
dynamics are included in the figure (Galiasso Tailleur and Ravigli, 2005).
dC HC;e2
For hydrogen and hydrocarbons in bubbles, exchanging gas and
¼ t1 C HC;1 þ uC HC;3 wC HC;2 þ K be
HC ðC HC;e2 C HC;b2 Þ
solid with dense phase of compartment (II): V s dt
Z
dNH2bII 0:65
dNH2b v CF2; C HC;e2 es gHC E2 ri ð13Þ
¼ ub ðzÞ þ K be
H2 eb ðC H2;bII C H2;eII Þ þ P CF;H2 C H2;eII
V bII dt 0:10 dz
For solid phase components in dense phase (compartments III)
ð8Þ with backflow, and exchanging gas and solid with bubbles by
Z 0:65
crossflow and mass transfer
dNHCbII dNHC b
¼ ub ðzÞ þ K be
HC ðC HC;eII C HC;bII Þ þ P Ceb;II C HCe;II ð9Þ dC H2;eIII
dz ¼ t1 C H2;4 þ uC CH4 wC H2;III þ K be
V bII dt H2 ðC H2;bIII C H2;eIII Þ
0:1
V s dt
For hydrogen and hydrocarbons in bubbles, exchanging gas and v CFIII; C H2;eIII es gH2 ai ðtÞri ð14Þ
solid with dense phase of compartments (III):
Z 1:15 dC HC;eIII
dNH2bIII dNH2b
¼ ub ðzÞ ¼ K be
H2 eb ðC H2;bIII C H2;eIII Þ þ P CebIII C H2e;III
¼ t1 C HC;4 þ uC HC;4 wC HC;4 þ K be
HC ðC HC;eIII C HC;bIII Þ
dz V s dt
V bIII dt 0:65
v CF3; C HC;eIII es gHC aiðtÞri ð15Þ
ð10Þ
Z 1:15 dV bII dV bIII
dNHCbIII dNHC b ¼ ¼0 ð16Þ
¼ ub ðzÞ þ K be
HC ðC HC;eIII C HC;bIII Þ þ P Ceb;III C HCe;III dt dt
V bIII dt 0:65 dz
The solid balance is:
ð11Þ
dW II dW III
For solid phase in dense phase (compartment II) with backflow, ¼ ¼0 ð17Þ
dt dt
and exchanging gas and solid with bubbles by crossflow and mass
transfer: Apparent rates of reactions are defined based on elemental
reactions in Tables 5 and 6. The reactor is pseudo-isotherm and
dC H2;e2
¼ t1 C H2;1 þ uC H2;3 wC H2;2 þ K be the boundary conditions are:
H2 ðC H2;b2 C H2;e2 Þ
V s dt
v CF2; C H2;e2 es gH2 E2 r i ð12Þ t : 0; at inlet : Ci : Cio and Ti ðtÞ : Tio ; ai : 1; P : 14 MPa; F i : F io ; ð18Þ
Table 6
Stoichiometry of proposed expression for kinetic and deactivation rates.
Reactions (gas) Gas phase rate of reactions

HDS Th + H2-H2S + C4 k1 C Th þ k2 C BTh þ k3 C DBT
ðcHCSSH2 Þ ¼ (21)
BTh + H2-H2S + Aar ½1 þ 1:3e1200=RT C RS
DBTh + H2-H2S + DAr
HDNi ArNi + H2-Aar k4 C ArNi C 0:3
H2
ðcArNiArsNi Þ ¼ (22)
½1 þ K Ar C RAr þ K NAr C NAr
HDS (Hid) DBT + H2-CHSB ? DFB fast k4 C DBT C 0:3
H2
ðcDBTArsN Þ ¼ (23)
HDAr ADAr + 2H2 ? ANAr k6 C DAr C 0:3
H2
ðcDARNAr Þ ¼ (24)
HYD ANAr + 3H2 ? ADN k7 C NAr C 0:3
H2
ðcANARADN Þ ¼ (25)
RO ANAr + 1.1 H2 ? AAr k8 C NAr
ðcNArAr Þ ¼ (26)
½1 þ K Ar C RAr
RO ADN + 1.1 H2 ? AN k9 C DN
ðcADNA0 N Þ ¼ (27)
½1 þ K Ar C RAr
HYD AAr + 3H2 ? AN k10 C AAr C 0:3
H2
ðcAArA0 N Þ ¼ (28)
0
CRACK AAr ? Ar + A k11 :C AAr
ðcAArAr Þ ¼ (29)
½1 þ 3:0eð1200=RTÞ C RAr
CRACK iP-A0 + C3 k11 :C AAr
ðcAArAr Þ ¼ (30)
½1 þ 3:0eð1200=RTÞ C RAr
CRACK AN ? N + A0 k12 C AN
ðcANrN Þ ¼ (31)
½1 þ 3:0eð1200=RTÞ C DAr
HYD B + 3H2-CH k13 C BT C 0:3
H2
ðcBCH Þ ¼ (32)
ISOM nP ? iP k14 C nP k15 C iCp
ðcnPiP Þ ¼ (33)
HYD A0 + H2-iP k16 C OL C 0:3
H2
ðcOLAAR Þ ¼ (34)
HYD OL + H2 ? AAr k16 C OL C 0:3
H2
ðcOLAAR Þ ¼ (35)
HYD DOL + H2 ? OL k17 C DOL C 0;3
ðcDOLOL Þ ¼ H2
(36)
POLYM 2DOL ? COKE k18 C DOL
ðcDOLcoke Þ ¼ (37)
½1 þ K Ar C RAr
POLYM 2OL ? Coke k19 C OL
ðcOLCoke Þ ¼ (38)
½1 þ K Ar C RAr
POLYM 3A0 ? Coke k19 C OL
ðcA0 Coke Þ ¼ (39)
½1 þ K Ar C RAr
t : t; at outlet : CiðtÞ : Ci f ðtÞ; Ti ðtÞ : Tio ai : aiðtÞ; P : 14 MPa; ð19Þ These species represent all the families of compounds with dif-
ferent amounts and sizes of alkyl-groups (lumped by families).
Wi ðtÞ : Wio ; ð20Þ Since the hydrogenation in gas phase produce small dealkylation,
the hydrogenated PSSs, present in one of the PSC in the feed,
remains in the same PSC in the product. Table 4 shows as an exam-
2.5. Apparent kinetic and deactivation rates constants
ple the carbon distribution analyses of feeds and blend of them,
and Tables 5 and 6 the amount by weight of PSS’s present in each
The kinetic rates of hydrogenation of HCK1 feed were obtained
of PSCs. The amounts of nitrogen, olefins and diolefins in HCK1 are
previously (Galiasso Tailleur et al., 2008), without olefins in the
very small, but their concentration in DCN are important given its
feed. The same kinetic and deactivation expression is used here
reactivity.
but the constants were calculated taken in account the catalyst
The stoichiometry and the expression of the rates of appear-
deactivation with and without olefins in the feed. Current data
ances or disappearances of species (ci) are depicted in Table 5
was developed in a fluidized bed reactor because the olefins (OL)
(Eqs. (21)–(39).) They are expressed in mol g1 h1 and concentra-
and dio-olefins (DOL) were fast polymerized, reactions that pro-
tions Ci in mol/L for i species in each of PSCs. The mass and energies
duce heterogeneous coke distribution along the fixed bed of pellets
balances (Eqs. (40)–(59)) are depicted in Table 7 for these parallel
and across of catalyst particle radius.
and in-series types of reactions.
The hydrocarbons present in gas phase in the feeds were
There are three different sites on the catalyst that perform the
divided in five pseudo-components (PSC’s) and thirteen pseudo-
reactions:
species (PSS’s):
PSC’s: PSC1, Gas (C1-C4), PSC2: Naphtha (C5-C10); PSC3: Kerosene
(a) Sites 1 produces direct desulfurization of Th, BTh and DBTh
(C11-C14); PSC4: Diesel (C14-21) and PSC5 (C+21).
species into hydrogen sulfide and hydrocarbons (Eq. (21)).
PSS’s: Par: paraffins, IP (iso-paraffins), AMN (alkyl-mono-
These reactions seem to occur on the border of metallic clus-
cyclo-paraffins), ADN (alkyl-di-cyclo-paraffins), AMAr (alkyl-
ters (PdWNi or MoNi); a1 is its activity (),
mono-aromatics, other than benzene), ADAr (alkyl di-aromatics),
(b) Sites 2 hydrogenates olefins, diolefins, diaromatics and
B (benzene); Th (thiophene), BTh (benzo-thiophene), DiTh
monoaromatics and benzene; they are on metallic cluster
(dibenzo-thiophene), Ol (olefins), DOl (diolefins), Ni (nitrogen)
containing S; in addition they hydrogenated molecules
and A0 (olefins formed by cracking and dealkylation). In addition,
containing S and N hetero-atoms; a2 (–) is its activity,
there is H2, H2S, H3N gaseous components.
Table 7
Parallel in series sets of rates reactions used in the mass balances.

r Bth ¼ cBthðMArÞ a1 f 1 g1 (40) r Th ¼ cthðC 4 Þ a1 f 1 g1 f 1 (41)

rDBT ¼ f cDBthðArArÞ a1 þ cDBThArSNÞ :a2 g1 f 1 (42) r DAr ¼ ðcDArNAr Þ a2 g2 f 1 (43)
rNAr ¼ ððcDArNAr Þ:a2 ðcNArDN Þ:a2 Þg2 cNARMAr Þa3 g3 Þf 1 (44) r DN ¼ ððcNArDN Þ:a2 g2 ðcDNMN Þ:a3 g3 Þf 1 (45)
rMN ¼ ðcDNMN Þ a3 g3 f 1 (46) r B ¼ ððcMArB Þ ðcBCH ÞÞa3 f 1 g3 (47)

rIPq ¼ ½ cnPq iPq :a2 g2 ðciPqiPp Þ:a3 g3 f 1 (48) r Pq ¼ cnPq iPq a3 f 1 g3 (49)
rRHDS ¼ r H2 S (50) r Coke ¼ ðcDOlCoke Þ þ ðcOlCoke Þ ðcA0 Coke Þ (51)

rPp ¼ ½ cPq Pp þ cMNPp a3 f 1 (52) r OL ¼ ½ðcDOlOl Þ cOlPp g2 ðcOlCoke Þg3 f 1 (53)
rDOl ¼ ½ðcDOlCoke Þg3 þ ðcDOlOl Þg2 f 1 (54) r H2 ¼ Rr i (55)
rd1 ¼ kd1 aðb þ cC 1coke Þ (56) kd2 a
r d2 ¼ ð1þcC H2 Þ
(57)
qffiffiffiffiffiffiffiffiffiffiffiffi
rd3 ¼ kd3 aðb þ cC 1:5
Coke Þ; (58) gi ¼ 1 th/ with /i ¼ 1r ci
(59)
/ Deff C H2
Fig. 4. Flow scheme of the program used to obtain the kinetic and deactivation rates constants.
(c) Site 3; produces cracking, ring opening and dealkylation on 633 K). The catalysts were deactivated during four successive peri-
Ti-Al Bronsted and Lewis acid sites; a3 (–) is its activity. ods of 48 h (Cond 2: 750 GPSV, 613 K, 0.9 MPa and 500 NL/L of H2/
HC) to accelerate the catalyst deactivation for another 192 h.
kodj and Ed,j are respectively the pre-exponential factor and the Activity tests. Deltas of PSS’s were determined at 603, 613 and
energy of catalyst deactivation, Cc and CH2 (mol/L), are the coke 623 K operating at Cond. 1 for 12 h, after each of above-mentioned
and hydrogen concentration in the reactor, t (days) the time in 48 h periods of deactivation and gases were analyzed. Deltas of
operation, T (K) the temperature. The apparent kinetic and deacti- PSS’s were compered to their initial values (values before deactiva-
vation rates constants were adjusted by the simulation program tion.) Spent catalysts samples were obtained at the end of each
(Fig. 4, see details below.) deactivation period at different temperatures; they were analyzed
(elemental analyses, metals dispersions (XPS), carbon distribution
2.6. Catalytic test that measure the rates of catalysts deactivation in a (SEM-EDS and MAS 13C NMR), CS2 extraction and pyridine
SBFBR adsorption).
Similar tests were performed for catalyst A and B using HCK2
750 g of particles of either A or B catalysts (80–100 mm) were and HCK3 as feed.
fluidized in hydrogen, using a three holes gas-distributor at Carbon and sulfur were determined in samples of catalysts and
593 K, 14 MPa, 500 GPSV. Then, hydrogen was replaced by the the results compared with those of previous periods. These proce-
hydrocracking gaseous stream produced in the HPHTS (HCK1 dures generated information about catalyst deactivation in more
+ H2) or by the blend (either HCK2 or HCK3 and H2) at 623 K and severe conditions than normal operation.
14 MPa. DCN was added in pilot plant at 283 K, 14 MPa to HCK1 feed
The catalysts were pre-conditioned by processing HCK1 feed at preheated at higher temperature than required, for example to
603 K, 500 GPSV, 2000 NL/L (Cond. 1) of H2/HC ratio for 40 h. Com- 628 K, to obtain a blend of feeds at reaction inlet temperature; that
mercial experience demonstrated the importance of this period avoid the polymerization of HCN during heating.
that produces an initial coke deposition in controlled conditions. Olefins and diolefins conversion in HCK1, HCK2 and HCK3
Gas phase (H2, H2S, H3N and hydrocarbons) moves up in bub- feeds. Olefins and diolefins present in the blends are very fast con-
bles from the bottom of the fluidized bed reactor to the gas-solid verted at 623 K and 14 MPa. To be able to measure their rates of
interphase; then, it goes through a cyclone located in the freeboard reactions, the spent catalyst was tested with HCK1, HCK2 and
(20 cm above the interphase) and leaves the unit (Fig. 2). The wall HCK3 at lower temperature (either 543, 563 or 583 K, but the same
of the reactor is maintained at the reaction temperature using five other Cond. 1). Thus, the activity of the sample deactivated at 613,
electric heaters. The fluid dynamic and the amounts of coke on cat- 623 and 633 K were tested at lower temperature.
alyst attained a plateau after a period of 40 h (catalyst conditioning
period.) Deltas of species were measured after this period at three 2.7. Verification of the kinetic and deactivation rates parameters
temperatures (603 and 613 and 623 K) for 12 h on-stream. These
values are used as reference for catalyst deactivation period. Mass balance Eqs. (5)–(20) were solved using a numerical
Coke, sulfur, carbon and metal dispersion, acid sites and average method (a modified predictor-corrector Runge-Kutta) and the sim-
benzene effective diffusivity were measured in samples of catalyst ulation program, written in VB6, calculated the new sets of kinetics
after 40 h on stream. and deactivation rates parameters (Fig. 4) This program used
Gases were analyzed on-line by GC–MS, and activities calcu- seeded values of kinetic and deactivations rates parameters,
lated by the program based on delta of PSSs. Hydrocarbon were deactivation equations, calculated fluid dynamic parameters
condensed, and PSCs obtained by micro-distillation; density, vis- (Eqs. (1)–(4)), other properties (reported in Table 8), the operating
cosity and surface tension were measured in each of cuts by ASTM conditions and the reactor model (Fig. 3) to predict deltas
methods. ((inlet-outlet)/inlet) of properties and catalysts activities.
Deactivation. Catalysts deactivation were performed in four The simulation program adjusts the kinetic and deactivation
tests, after this conditioning stage, at either 603, 613, 623 or rates constants (seeded) using a Genetic Algorithm (GA) until Eq.
(59) pass by a minimum after less than N iterations. That obtain
the best deltas for the 289 experimental data points. Genetic algo-
Table 8 rithm mutates the seeded parameter until the minimum specified
Properties of the fluidized bed used for the simulation program.
is obtained, or the maximum number of mutations is reached.
Properties PtSOxTiZr/SiO2 Activity 573 K, Deactivation 603 K, When the program does not achieve the minimum, the genetic
14 MPa GHSV 540 9 MPa GHSV:750 algorithm adjusts the guessed values of the two kinetic rates
H2 + HC Hydrogen Hydrogen constants that present the largest deviations (P and Q objective
qs g cm3 1.34 1.34 function (Eqs. (61) and (62)). The process continues until the
dp microns 80–120 80–120
minimum value of Q and P are obtained.
db (m) (Galiasso Tailleur 0.055 0.065
X X .X
et al., 2008)
d (–) bed fraction 0.19 0.19
DeltaPred;Nþ1
f DeltaxfPred;N DeltaPred;N
f < 1% ð59Þ
ee (–) emulsion voidage exp 0.34 0.38
Kbe or Kbc/Kbe/Kce (s1) (3) 0.3 0.3 !
X
i¼3
aSim acalc;Exp
umf cm s1 (4) 0.08 0.072 Objective func: P: min; i i
100 < 1:5% ð60Þ
ubr cm s1 0.3 0.5 acalc;Exp
i¼1 i
ub cm s1 (2) 0.5 0.55
ue cm s1 0.1 0.11 !
L/Lmif (–) exp. 1.023 1.033 X
i¼2
DeltaSim DeltaExp
nor freq. of bubbles at the 14.1 19.3 Objective func: Q : min; i i
100 < 1:5%
gas-sparget #/sec i¼1 DeltaExp
i
w5 gh1 solid after cyclone 0. 12 0.22 ð61Þ
f backflux (–) Eq. (2) 0.25 0.33
VI cm3 boundary reactor 200 230 The GA is a stochastic optimization method for solving non-
VII cm3 810 790 differentiable highly nonlinear set of differential equations. The
VI I I cm3 810 790
GA starts the integration of mass balances using the seeded values
of rate parameters with an initial random population which The simulation program verifies the kinetics parameters at
presents the first generation. Each population consists of some different level of deactivation, because when the minimum of P
individuals and individual is made of number of genes. These genes function is not achieved the program adjust the kinetic rate con-
are decision or control variables (temperature, concentrations, resi- stants. The user of the program analyzes the rate of convergence
dence time, time on stream). The objective function (P, Q) are calcu- and, if needed, run the program with another criteria. The program
lated for each of ‘‘individual” based on its genes (operating was run using different guessed values to check the stability of
variables). Populations will be repeatedly modified to reach the solutions. Deltas of species are reported in Fig. 5, activities in
solution which would be the best collective values of catalyst Fig. 6 and the values of the kinetic rates constants found reported
activities. in Table 9. The heterogenous distribution of coke is shown in Fig. 7,
Fig. 5. Example of accelerate deactivation (613 K, 9 MPa, 750 GPSV, 15 H2/HC ratio) as a function of time (four period) Delta of properties are measured at 603 K, 14 MPa, 500
GPSV and 23 NL/L H2/HC ratio); 5a for A and 5b for B catalysts (deactivated using HCK1 feed), and Fig. 6(c and d) for A and B catalyst using HCK2 as feed.
CAT A, HCK1 CAT B, HCK2

1.0 1.0
a2
Coke wt. %
4.0 4.0
Activity ai
a1 a2
Coke wt %
Activity ai
0.99 3.0 0.99 a1 3.0

a3
Cond 2
2.0 Cond 2 a3 2.0
0.98 1.0 0.98 1.0

Cond 1 deviations Cond 1 deviations
0 48 96 144 192 0 48 96 144 192

a) Time on accelerate deactivation (h) b) Time in accelerate deactivation (h)
CAT A, HCK2 CAT B, HCK2

1.0 1.0
Coke wt.%
Activity ai
4.0
Coke wt. %
4.0
Activity ai
deviations
0.99 a2 3.0
0.99
3.0
a1
Cond 2
2.0 Cond 2 a2 2.0
a3 a1
0.98 1.0 0.98 1.0
Cond 1
Deviations
Cond 1
a3
0 48 96 144 192 0 48 96 144 192

c) Time on accelerate deactivation (h) d) Time on accelerate deactivation (h)
Fig. 6. A and B Catalyst activity as a function of deactivation time operating at 613 K (Cond 2). Activities are determined at (623, 14 MPa, 500 GPSV and 23 NL/L H2/HC ratio).
(a and b) (left Y axis) shows the activities for A and B catalyst deactivated by HCK1 feed; (a and b), right Y axis shows coke content in A and B catalyst after treating these two
feed stocks as a function of deactivation time.
Table 9
Kinetic and deactivation rates constants for A and B catalyst (HCK-G). Kinetic rates constants for olefins and diolefins conversion determined at 543 and 563 K 10 MPa.
Reaction constants Units of ki, kio kdjo and kdj koi (A) c.u Ei (A) kcal/mol koi (B) c.u. Ei (B) kcal/mol
ðkThC4 Þ L h/lmol 8.5E+09 18,100 9.5E+09 18,300

ðkBThAB Þ L h/lmol 9.0E+09 19,000 7.2E+09 18,400
ðkDBTDAr Þ L h/lmol 1.5E+09 17,500 2.5E+09 17,600
kDBTArSN Þ L h/lmol MPa0.3 5.2E+10 21,000 1.2E+10 20,700
ðkBNNiAAr Þ L h/lmol MPa0.3 5.0E+08 18,300 1.0E+08 17,900
ðkDArNAr Þ L h/mol MPa0.3 7.1E+09 20,300 4.6E+09 20,000
ðkNArDN Þ L h/mol MPa0.3 2.8E+09 19,000 7.9E+08 18,500
ðkNArAr Þ L h/mol 1.1E+07 14,000 2.3E+07 14,400
ðkDNN Þ L h/mol 1.2E+07 15,300 3.1E+07 15,000
ðkArN Þ L h/mol MPa0.3 2.2E+07 16,000 7.3E+06 15,500
ðkArB Þ L h/mol 1.7E+05 12,000 1.3E+05 12,200
ðkBCH Þ L h/mol MPa0.3 8.2E+07 16,300 4.2E+07 16,000
ðknPiP Þ L h mol 8.5E+08 18,300 4.5E+06 17,000
ðkiPnP Þ L h/mol 2.2E+05 12,000 1.2E+06 14,000
ðkiPiP0 Þ L h/mol 8.5E+04 12,000 8.5E+04 12,000
ðkDOLOL Þ L h/mol MPa0.3 6.7E+07 15,000 1.3E+08 16,000
ðkOLAr Þ L h/mol MPa0.4 1.9E+06 18,000 3.7E+05 17,300
ðkOLCoke Þ L h/mol 3.4E+08 28,000 1.5E+07 27,100
ðkDOLCoke Þ L h/mol 3.4E+08 32,000 2.9E+06 30,000
KMAr l/mol 2.0E+01 1350 1.7E+01 1200
KDAr l/mol 1.8E+01 1150 2.1E+01 1300
kd1 1/day 1.90E+00 8049 4.50E+00 8292
kd2 1/day 2.00E01 4878 3.00E+00 7365
kd3 1/day 6.60E+00 8536 1.20E+01 9024
Deact. 1 (–) n1: 1 c1: 0.2 n1: 1.2 c1: 0.27
Deact. 2 (–) n2: 0.9 c2: 0.29 n2: 1.1 c1: 0.35
Deact. 3 (–) n3: 1.2 c3: 0.38 n3: 1.4 c1: 0.47
C External layer of
Spent A deactivated
By HCK2 at 633 K
Intensity (a.u.)
O
Al
Ni
W
S
a) 0 E (eV) 1 b)
Fig. 7. (a) EDS and (b) SEM (50,000) at the edge of particle (catalyst A after being deactivated at 623, 9 MPa, 750 GPSV and 15 NL/L of H2/HC ratio).
predicted deltas of species and values of activities have been com- 3. Results
pared them with measured ones in Fig. 8 and the study of the
effects of temperature on deactivation constants shown in Fig. 9. 3.1. Catalyst composition
Table 10 describes the products quality and compares them with
their specification. The values of C/Cav and C/RMe ratios (Ka inten- Table 1 shows that catalyst A contains 0.1 wt% of Pd, well dis-
sities of Me: W(A), Mo (B)) are shown in Tables 11 and 12 as well as persed on TiO2-Al2O3 surface; XPS analyses of metal dispersion
the amounts of hydrocarbons soluble in CS2 and benzene. (Table 2) on catalyst A there is 39% higher dispersion of W than
Ni on surface and S/(W + Ni) ratio is of 0.35 (mol) in the capping
2.8. Thermal stability of the feed layer. The support contains 15% of TiO2 on Al2O3 and present 1.3
times higher Ti dispersion than the sum of other metals on surface
The feeds and product of hydrogenation were placed in an auto- (XPS).
clave, pressurize in hydrogen at 14 MPa and heated to different The NiMo/TiO2-Al2O3 (B catalyst) presents a metals composition
temperatures and the amount of gum and solid measured follow- likes the typical commercial HDS catalyst (Table 1), but current
ing the protocol of ASTM D4625. The results of the relative product active phase was prepared by successive impregnation of Mo and
stability can be seen in Fig. 11. Ni. This procedure generated 40% higher Mo dispersion than Ni
Deactivated with HCK1 (623 K) Deactivated with HCK1 (623 K)

1.0 0.4
A B A B
Predicted HDDAr conversion

Predicted HDS conversion
603 K 603 K 613 K 613 K
623 K 623 K 623 K 623 K
0.95 0.35
0.9 0.3
0.85 0.25
0.85 0.9 0.95 1.0 0.25 0.3 0.35 0.4
a) measured HDS conversion b) Measured HDDAr conversion
A B
603 K 603 K
5
Predicted Coke wt. %
623 K 623 K
HCK1
4
2
2 3 4 5
c) measured Coke wt. %
A B A B
603 K 603 K 603 K 603 K
Predicted a1 (-)
Predicted a3 (-)
0.99 623 K 0.99 623 K

623 K 623 K
0.98 0.98
0.97 0.97
HCK1 Feed HCK1 Feed

0.96 0.96
0.96 0.97 0.98 0.99 0.96 0.97 0.98 0.99
d) Measureda1(-) e) Measured a3 (-)
Fig. 8. Prediction of delta sulfur (a), aromatics (b) in the products and (c) coke on catalyst A and B deactivated and tested by HCK1 feed. Dashed lines are used to shows the
trend. (d and f) present predicted activity of direct desulfurization (a1) and cracking (a3) versus measured ones.
on surface (Table 2). Again, Ti dispersion on surface (XPS) is 1.4 Catalyst A showed in previous studies (Galiasso Tailleur, 2007)
times higher than other metal species, and Sulfur/(Mo + Ni) ratio higher (1.2) hydrogenation, (1.7) isomerization and (1.38 times)
is 0.29 (mol) on surface. ring opening activity, but present lower (0.78 times) direct desul-
Both catalysts have similar mesoporous support and were furization than B catalyst.
equally treated by steam ammonia to increase the interconnection The measure of benzene pulse signal dispersions (inlet-outlet
of pores and to build Ti-Al acid centers on surface. This procedure (E(t)) at 363 K and 14 MPa provides information about the average
reduces the micropore volume. Thus, both pre-sulfided catalysts effective molecular diffusivity (Deff) in meso and micropores, as
have 0.3 cm3/g of micropores volume, with average diameter of well as describes the effect of coke deposition on the catalysts.
9 Å. This type of support has a moderate surface area, 24% of them Catalyst A adsorb at the equilibrium 1.4 times higher amount of
in mesopores of 60 Å (average pore diameter.) Benzene per g and present 1.1 times higher rate of benzene
A and B catalysts adsorb pyridine on Lewis and Bronsted sites diffusivity than B (Table 1), despite the similar pores size distribu-
(Table 2). The amount desorbed at 430 K and 530 K shows the pres- tion of both catalysts. The relative diffusivity of benzene in hydro-
ence of 70% of medium and 30% strong acid sites; almost 1.2 times gen were calculated using equations described in reference
of strong acid sites than, for example, commercial FCC catalysts. Galiasso et al. (1983).) The tail is assignee to micropores and the
Deactivated HCK2 Deactivated HCK2

1.0 0.4
A(603K ) A (603 K) deact A(603K) A (603 K) deact
Predicted HDDAr conversion

Predicted HDS conversion
603 K 603 K
623 K 623 K
0.9 0.3
0.8 0.2
0.7 0.1
0.7 0.8 0.9 1.0 0.1 0.2 0.3 0.4
a) measured HDS conversion b) Measured HDDAr conversion
A B
603 K 603 K
Predicted Coke wt. %
5 623 K 623 K
HCK1
4
2
2 3 4 5
c) measured Coke wt. %
Fig. 9. Prediction of delta of sulfur (a), aromatics (b) in products and coke on catalyst (8c) on catalyst A and B deactivated by HCK2 and tested by HCK1.
Table 10
Products quality after deactivation at 613 K, 9 MPa, 750 GPSV, 800 NL/L H2/HC.
PSC2-Naphtha
Quality for de pool Sulfur ppm Benz. % OL + DOL % AMAr % Octane
A HCK-G 3 <1 <0.1 10 71
B HCK-G 1 <1 <0.1 12 64
A HCK-DCN 1 <1 <0.1 12 64
B HCK-DCN 1 2.5 <0.1 21 64
PSC3-Kerosene
Quality for the pool Sulfur Ppm Mar % Dar % S P (–) Cetane (–)
A HCK-G 8 9 4.8 18 39
B HCK-G 3 10 8.3 16 38
A HCK-DCN 8 12 6.8 18 39
B HCK-DCN 3 16 11.1 16 38
PSC4 – Diesel
Quality for the pool Sulfur ppm Mar % Dar % Gun Mg/L Cetane (–)
A HCK-G 14 10 8.5 – 40
B HCK-G 7 14 14.5 – 39
A HCK-DCN 14 12 12.5 – 40
B HCK-DCN 7 19 18.5 – 39
PSC5 – Gas oil
Quality for the FCC Sulfur Ppm Mar % Dar % Nit. ppm CCR wt%
A HCK-G 32 18 15.4 3 <1
B HCK-G 22 20 22.3 4
A HCK-DCN 32 22 18.7 6
B HCK-DCN 22 26 25.1 9
Table 11
EDS measure of carbon (C/Cav) and metal C/CMe (Mo or W) ratios distribution in six positions across the radius. Cav is the average value of carbon measured in 20 points across the
diameter (360°).
(C/Cav)/C/RMe, Temp K/Feed Center 0.2 0.4 0.6 0.8 Edge

543/HCK2 1.0/2.4 0.9/2.1 1.0/2.1 1.0/2.0 0.9/2.0 1.0/1.9
613/HCK1 0.9/3.6 0.9/3.7 1.0/3.6 1.0/3.7 1.0/4.0 1.1/4.3
623/HCK1 0.7/3.8 0.8/3.9 0.9/4.1 0.9/4.2 1.0/4.6 1.1/5.2
623/HCK2 0.6/3.1 0.7/3.5 0.9/3.9 1.1/4.8 1.2/5.9 1.5/6.3
633/HCK2 0.5/2.9 0.6/3.1 0.7/3.3 1.3/4.9 1.9/7.1 2.5/8.7
Table 12
Effect of deactivation temperature on the amount of soluble coke, effective molecular diffusivity and ratio tail/main peaks measured by pulse of benzene in hydrogen.
Cat – Feed Temp K A – HCK-G A – HCK-G + DCN

CS2 sol. (%) Deff * 106 cm2/s Tail/main Peak (%) CS2 sol. (–) Deff * 106 (cm2/s) Tail/main %
543 98.5 3.3 2 45.4 2.8 3
613 57.5 1.8 6 32.3 1.2 2.8
623 45.3 1.2 7 18.5 1.0 11
633 24.3 0.8 15 9.1 0.2 29
second of peaks to mesopores (CI). The ratio of area in E(t) curve the fractionation unit (properties in Table 3, third of columns.)
(output-input) is higher in both fresh catalysts than on spent. DCN was kept refrigerated (283 K) and under nitrogen atmo-
sphere; in pilot plant it was fed directly to the inlet of HPHTS
(Fig. 1) because contains important amounts of olefins and diole-
3.2. Feeds
fins, sulfur, nitrogen, benzene and diaromatics. Notice that there
are different types of olefins and diolefins in the small amount of
Three feeds were studied here. HCK1 is the gas phase fraction
PSC1 than in PSC2 (Tables 3 and 4). PSC1 contains butene, pentenes
obtained at 623 K, 14 MPa in the HPHTS using a Gulf coast refinery
and butadiene, while PSC2 has significant amount of styrene, iso-
hydrocracked product (pilot plant, Fig. 1). The HCK commercial
prene, 3-pentadiene, 2,3-pentadiene and others penta- and hexa-
unit operated with 30% of Coker Gas oil and 70% of straight run
dienes, according to GC–MS and liquid 13C NMR and 1H NMR anal-
gas oil at 70% conversion of Gas Oil fraction. Table 3 shows the car-
yses. The butadiene and cyclo-pentenes are the most reactive
bon distribution of the HCK1 determined by GC–MS and 1H and 13C
molecules that form coke. DCN feed also has significant amounts
NMR analyses. The hydrocarbon distribution is composed by 3.4%
of nitrogen (indole, quinoline and carbazole), and sulfur (thiophene
of PSC1 (C1-C4), 23.7% of PSC2 (Naphtha), 31.4% of PSC3 (kerosene),
and benzo-thiophene.)
37.1% PSC4 (Diesel) and 5.1% of PSC5 (Gas Oil). The carbon distribu-
Fluid dynamic of reactor. The fluid dynamic Eqs. (1)–(4),
tion in gas phase stream is affected by the temperature of vaporiza-
developed in previous work (Galiasso Tailleur and Farina, 2019a)
tion in HPHTS (at constant pressure and H2/HC ratio). Vaporization
for this reactor operating at high-pressure-high-temperature in
in the HPHTS decrease with the molecular weight of the HCK prod-
presence of hydrogen, were confirmed here at 603 K and Cond. 1
ucts and increase with temperature (Galiasso Tailleur, 2017).
and Cond. 2. Notice in Table 8 that the change in operating condi-
The ratio H2/HC used in hydrocracking is three times those used
tions from Cond 1 to Cond 2, used to deactivate the solids, pro-
in conventional HDT or MHCK units. This favorable high ratio is
duces an increasing in bubble size, backflow and crossflow due
imposed by the HPHTS temperature of operation and the H2/HC
to higher backmixing at lower total pressure. The results obtained
ratio in the hydrocracking or hydrotreating reactors. Previous stud-
in the fluid dynamic study will be discussed in the second part of
ies demonstrated that hydrogenation rates of reactions and the
this paper (Galiasso Tailleur and Farina, 2019b). The values of the
rate of coke built-up is slightly affected by the level of hydrogen
fluid dynamic constants calculated at 603 K and Cond. 1 or Cond.
partial pressure (order of reaction 0.3, see hydrogenation equations
2, are reported in Table 8. The gas behavior in the fluidized be reac-
in Table 6). The HCK1 feed has negligible amounts of olefins and
tor of the was determined by a pulse of CH4 in hydrogen (CI).
diolefins.
Conditioning of catalyst. (stabilization of the catalysts.) The
conditioning of the catalysts has been performed at the start up
PSC2 contains small amounts of benzene, nitrogen and sulfur
temperature (603 K, Cond. 1)) using HCK1 feed for 40 h; this proce-
compounds (thiophene and benzo-thiophene, see Table 4) and
dure avoid the initial (fast) heterogeneous deposition of coke in the
its quality is out-of-gasoline specifications. Br number is 0.9
catalysts; then, deactivations were measured at four temperatures
and diene value 0.2 (g I2/100 mL).
in accelerated deactivation conditions (lower hydrogen partial
PSC3 (28% by wt.) contains low amounts of alkyl-diaromatics,
pressure and higher throughput) than those used for commercial
alkyl-di-naphthenes, benzo- and dibenzo-thiophenes type of
application. The procedure deposited additional amount of coke
sulfur (Table 4). This kerosene is out-of-jetfuel specification.
by cracking and polymerization of olefins formed in-situ on the
PSC4 (20% by weight) has important amount of alkyl-
strongest Bronsted acid sites and by hydrogenolysis on some
diaromatics and dibenzothiophene (Table 4). This diesel frac-
metallic nano clusters (Galiasso Tailleur and Ravigli, 2005). During
tion quality is out of low-sulfur diesel specifications.
this 40 h of operation, bubble gas linear velocity was 0.5 cm/s (umf:
The rest is composed by small amount of gasoil hydrocarbons
0.08 cm/s) and the minimum fluidization porosity of 0.58 (Table 8).
(PSC5), whose are regularly sent to lube oil production or used
After this period, 1.8% of carbon on A and 2.0% on B catalyst
as FCC feed. It contains high amount of dibenzo-thiophene
were deposited, values that produce small deactivation. Delta %
and alkyl-diaromatics than straight run feeds.
of properties (defined respect to fresh catalyst) for both catalysts
The second of feeds used here is a blend of HCK1 and delayed can be seen in Fig. 5(a and b) (HDS, HDNi, HDAr, in left Y axis,
coking naphtha (DCN). DCN was obtained at the refinery from and HDB and HDDAr in right Y axis) at time of t = 40 h. The results
confirm (for HCK1 feed) that A catalyst produce higher reduction of catalyst conditioning at 613 K. Deltas of ANAr and AND species
nitrogen and aromatics than B catalyst, while the later produces on sites 3 (RO reactions) were 12.5% and 6.5% on A, and 6.5% and
higher desulfurization than the former one. Fig. 6(a and b) shows 3.2% on B catalyst; 3% of iso-paraffins and 2% of cyclo-paraffins
the activity of catalysts after 48 h on stream. were cracked by acid sites of spent A catalysts at 613 K according
Carbon accumulated is 10% lower in catalyst A than in B to C3-C5 generated; ring-opening reactions are important because
(Table 1), probably due to slightly lower acid sites strength and improve the quality of diesel but their rates values only can be esti-
the presence of Pd in the former of catalysts; Pd decreases cracking mated by the model (in series-parallel reactions). The rates of all
and increases hydrogenation capability. Most of coke is deposited reactions, other than olefins and diolefins hydrogenation, are very
in micropores since their volume is reduced in both catalysts to slow at 543 K.
less than 0.1 cm3/g (T-plot analyses.) Coke deposition produces a Kerosene and diesel fractions are both in specification after
reduction of 35% in (A) and 42% in (B) surface areas, and 8.8% in hydrotreating at 603 K (Cond. 1) on spent catalysts; naphtha must
(A) and 9.1% in (B) pore volumes; the pore distribution in both be sent to Reforming due to low octane number and gasoil fraction
spent catalysts are quite similar (Table 1.) Effective molecular dif- can be processed by FCC (Table 10).
fusivity of benzene in hydrogen has been reduced 24 (A) and 28% The simulation program determined that at 603 K average cat-
(B) by carbon deposition; carbon reduces 35% the adsorption of alyst efficiency factors (g, calculated using Eq. (58)) for H2 and HC,
benzene on A and 48% on B-catalyst at 423 K. E(h) curve presents other than those for OL and DOL, are close to 1. The diffusional con-
slightly lower tail/main peaks ratio respect to fresh catalyst value trol of reactions can be neglected in agreement coke profile in the
because the elimination of micropores by carbon deposition. particle (Table 12).
Pyridine adsorption at 430 K shows a reduction of 46% in A and Above description characterize the catalyst at the start of the
55% in B moderate acid sites (Table 1) but most of strong sites in operation. Now, to determine the catalyst deactivation as a func-
both catalysts were covered by deposits. Reduction in cracked tion of time on stream, these catalysts were deactivated at Cond
products species that formed (C3-C5) are not proportional to the 2 (lower pressure-higher throughput) in comparative ways, and
total amount of carbonaceous material deposited on catalyst, as then tested again at Cond 1 to check the catalyst deactivation.
expected based on previous studies (Galiasso Tailleur et al.,
2008; Galiasso Tailleur and Ravigli, 2005). Most of coke deposits
are built-on strong acid sites at the mouth and inside micropores. 3.3. 3.3. Catalyst deactivation
The detailed analyses of the products produced at the end of
stabilization period (obtained at operating at Cond. 1) are reported The activities were measured at three temperatures for 12 h,
in Table 4 for the HCK1 feed. HDS (sulfur removal) rate is 0.95% after the preconditioning, and then deactivation of both catalysts
lower on A-catalyst but 1.11, 1.7 and 1.3 times higher respectively performed during four periods of 48 h for each of feeds. Detailed
for HDAr (hydrogenation of diaromatics) and HDNi (removal of analyses of the hydrogenated products (tested at 603 K, Cond. 1)
nitrogen) than B-catalyst. A-catalyst depicts higher RO rate of reac- indicated that products quality, obtained with catalysts deacti-
tion than B-catalyst because of Pd content (Galiasso Tailleur and vated at 623 K and HCK of reaction after 232 h on stream, are close
Ravigli, 2005) and its higher number of free-of-coke acid sites. to the specifications (Table 10), while those tested with HCK2 and
Coke deposited were extracted by CS2 from both spent catalysts HCK 3 are out of specifications.
(40 h) and analyzed by GC–MS; the CS2 extracted 85% of carbona- Deactivation with HCK1 feed. Catalysts A and B were deacti-
ceous deposits, mostly composed by C12-C20 heavy paraffinic dimers vated at 603, 613, 623 and 633 K (Cond. 2). Catalyst A and B, for
and trimers. They are formed by slow cracking reactions of iso and example those deactivated at 613 K, accumulate moderate
cyclo-paraffins into olefins (A0 ), whose are then polymerized in- amounts of carbon (Fig. 6(a and b)) and sulfur on surface. They
situ on acid sites or hydrogenated on metal sites. The cracking rate end up after 232 h with 3.1% on A and 3.2% by wt. of coke on B;
is evaluated based on C3–C5 hydrocarbons produced by b- and in addition, they content 3.6% (A) and 3.7% (B) by wt. of sulfur.
c-scission. The insoluble in CS2 deposits from A and B catalysts C/RMe ratio (average metal dispersion measured by XPS) is depict
present an incipient aromaticity (MAS 13C NMR analyses, see CI). in Table 2; the ratio increases 37% on A and 40% on B-catalyst.
The SEM analyses of spent A-catalyst (40 h) and EDS (C/Cav Dispersion of W and Ni are reduced 50% and 63% on A, while on
information in 2d of rows in Table 11, see the method used in CI) B, Mo and Ni decreases 52% and 50%. Carbon on surface reduces
confirm that coke formed by HCK1 feed was homogeneously the numbers of acid sites 24% on A and 20% on B (measured by pyr-
deposited across the particle and well dispersed in cluster lower idine adsorption at 430 K.) Notice that the acid numbers measured
than 1 mm, probably at the mouths of micropores (intersection are qualitative on spent catalyst because pyridine slowly desorb
micro-meso pores); there, the strongest acid sites (Ti-Al) are hydrocarbons from the surface as a function of contact time; thus,
formed according to DFT simulation of the surface. C/Cav (EDS) ratio it tend to over-estimate the number of acid sites really available for
is quasi-constant across the radius in both catalysts when using reactions. The ratio of CS2-soluble/insoluble coke is decreased 43%
HCK1 as feed. (A) and 38% (B). All the values reported above are referred to those
Metals dispersions (XPS), defined as IMe/RMe, have been measured on these catalysts at the start of deactivations.
decreased respect to those of fresh catalysts due to coke and sulfur Soluble in CS2 deposits are composed by dimers and trimers of
accumulation on surface (Table 2). W is reduced 27% and Ni 38% on paraffines (1H and 13C NMR analyses) as it is shown in the second
A, and Mo is decreased 18% and Ni 23% on B-catalyst. part of this paper (Galiasso Tailleur and Farina, 2019b). MAS 13C
Despite of carbonaceous deposits, delta of sulfur is higher than NMR analyses of spent catalysts are shown in CI.
98% and delta nitrogen higher than 94% at 603 K (Cond. 1) on both Deff, the effective diffusivity of benzene, is less reduced in
catalysts. Less than 1% of iso-paraffins are cracked into lighter spent A (40%) than in B (52%) catalysts, despite to have similar
paraffins (there is a small shifts in carbon number distribution) meso-pore size distributions and micropore volumes. The ratio
and more than 22% of n-paraffins transformed into iso-paraffins tail/main peak in E(t) signal, assignee to mesopores (see CI for
(compare values in wt% in Tables 4 and 5). The products obtained details), is slightly increased in both catalysts deactivated at
with both catalysts do not contain olefins and diolefins, hence, 623 K; in addition the tail attributed to the micropores adsorption
when formed, they are either polymerized or hydrogenated. desorption is strongly reduced; benzene adsorption at 363 K
Deltas of di- and mono-aromatics species on sites 2 were 70% decreases 7% on A and 12% on B deactivated catalysts during
and 30% on spent A, and 51% and 22% on spent B catalyst after 234 h respect to those of 40 h.
Radial profile of C/Cav ratio is still flat (EDS third of rows of ing 0.1 of DCN to 0.9 of HCK feed (compare deltas of species in
Table 11) because iso-olefins in HCK1 feed formed A0 olefins, whose Fig. 5(a and c) (for A-catalyst) and Fig. 5(b and d) (for B-catalyst)
are polymerized in-situ to form a carbonaceous deposit. Notice as a result of additional coke deposition. A-catalyst contain 4.2
that coke affect more the activities of B than A catalyst, due to and B 4.9 wt% of carbon formed after 192 h of deactivation in pres-
the former lower number and types of hydrogenating sites. Pd is ence HCK2 feed; sulfur content is also higher in both catalysts (3.8
a component of active sites that improve the hydrogenation func- on A and 4.1 wt% on B). Olefins and diolefins had been totally con-
tions, help paraffins isomerization, ring opening and decrease the verted after all period of deactivation at 613 K.
polymerization on the latter catalyst (Galiasso Tailleur and A-catalyst, after been deactivated at 613 K (Cond. 2), were
Ravigli, 2005). tested with HCK2 at 543 and 563 K (Cond. 1) to measure OL and
Deltas of species measured at 623 K (Cond. 1), after each of peri- DOL conversions at low rates; spent catalyst shows reductions of
ods of deactivation at 623 K, are shown in Fig. 5, as an example. The 65% (A) and 79% (B) in deltas BrN and 57 (A) and 65 (B) % in deltas
same graphics were obtained at other deactivation temperatures MA, respect to those measured before the deactivation (40 h). Effi-
(543, 563, 623, and 633 K.) Delta of total sulfur (direct plus indirect ciency factor is 0.98 in A and 0.99 in B for OL, while this factor is
route) is higher on B than on A catalyst products, but the opposite 0.90 in A and 0.84 in B for DOL hydrogenation at 543 K. These reac-
occurred with nitrogen removal (not shown), aromatics and ben- tions are controlled by preferential adsorption of aromatics that
zene hydrogenation; to see these compare deltas of species in reduce the rate of polymerization. Any attempt to process higher
Fig. 5a vs. b. Higher the temperature, lower is the efficiencies mea- than 10% of DCN in the feed end up with external deposition of
sured for both catalysts after deactivation, as it is shown in Fig. 6, coke in the particles.
and higher the amount of coke uploaded. Clearly, at temperature in the range of 603–633 K the OL and
Compare Fig. 6b to a to see that the rate of activity decreases DOL polymerization reactions are so fast that mass transfer and
faster on B than on A for the three sites; that agrees with the trends hydrogenation reaction cannot avoid the heterogenous formation
observed in metal dispersions, acid sites and effective diffusivity, of a exponentially deceasing of carbon deposition across the radius
but their loss of activities are not proportional to reduction of these of the particle (see analyses of apparent rates constants below.)
properties. Hydrogenation and ring opening are in series reaction, Thus, polymerization reaction mostly occurred at the mouth of
therefore the analyses of the values of delta of species are only an micro and mesopores located in the external layer of the particle.
approximation; better understanding is obtained by using kinetic OL and DOL pseudo-species transfer rate in pores are almost 5
rates constants (Table 9, see below). times slower that the rate of polymerization generating a radial
The largest difference in catalyst deactivation between A and B profile of coke, acid site and metal dispersion. The other species
occurred on acid sites, where the reduction of rates of isomeriza- are not affected by the mass transfer rate because 95% of reactions
tion, ring-opening reactions is higher for the latter catalyst; that takes place on mesopores at slow rates, but micropores are soon
affect the quality of products; notice that the presence of uncon- completely plugged by insoluble material; At temperatures of
verted aromatics affects activity and selectivity because they are deactivation higher than 623 K, the polymerization is so intense
preferentially adsorbed on acid sites (Galiasso Tailleur and than even the mesopores become plugged. E(h) (see CI) shows a
Ravigli, 2005). quasi-complete elimination of the benzene tail because the reac-
The products of hydrotreating at 603 K, after being deactivated tion only take place in mesopores with small changes in Deff.
for 192 h at 623 K, do not contain olefins and diolefins. The quality The profile of carbonaceous deposits induces a decreasing pro-
of products decrease respect to those at the start of deactivation; file of number of acid and metals sites content across the particle
there is slightly higher amount of thiophene, benzo-thiophene, radius. For example, polymerization of OL and DOL present in
benzene and nitrogen, and important amounts of dibenzo- HCK2 feed are 322.3 and 381.5 times faster at 623 K than hydro-
thiophene and diaromatics in the products. The higher the temper- genation of OL and DOL, and 87 times higher than polymerization
ature of deactivation, the lower the quality of the hydrogenated of cracked iso-paraffins formed from iso-paraffins. Comparatively
products. Thus, at the end of deactivation naphtha, kerosene and A and B catalyst shows quasi similar profile of coke above 323 K.
diesel fraction are out of specification (Table 10) and the deactiva- SEM-EDS analyses (Fig. 7a) of A sample, deactivated at 633 K for
tion must be compensated by increasing the inlet temperature. The 232 h, indicated the presence of thin coke layer outside of particle.
efficiency factor for hydrogen and hydrocarbon at this level of con- The matrix of A was dissolved in HF and the remaining solid
versions, particle size and deactivation temperature are close to presents at the microscope analyses filaments, formed in microp-
0.98 for all the reactions in the range of 603–623 K and this type ores, and layer of carbon built outside of particles. EDS radial anal-
of feed, in agreement with quasi homogeneous distribution of yses of C/Cav ratio, measured in six point (7l2), confirm the sharp
small amount of coke and the main role of mesopores (see discus- increase near the edge of particles respect to the average of 20
sion below). points determined across the particle (Table 11, last of rows.) Sim-
When temperature of deactivation is higher than 633 K, patches ilar coke distribution was found on B catalyst. Deactivation at
of coke start to be formed on surface. SEM analyses (not shown) of 623 K in presence of HCK2 and HCK3 feed produce deposits with
HF dissolved samples present incipient filaments of coke with low ratio soluble/insoluble in CS2 hydrocarbons due to intense
diameter higher than 1.5 mm; the C/Cav and C/Me(W) ratios (EDS, polymerization. The higher the deactivation temperature, the
fourth of rows in Table 11) largely varied across the particle. The higher the coke content and aromaticity of deposits (see MASS
13
analyses show more than 15% deviation in carbon content respect C NMR analyses in CI); that will be discussed in detail in the
to the average; 78% of this deposit are insoluble in CS2. second part (Galiasso Tailleur and Ravigli, 2005).
HCH2 and HCK3. DCN is added to HCK1 to understand the Molecular effective diffusivity of benzene and adsorption are
effects of olefins in catalysts deactivations at different tempera- reduced progressively when deactivation temperature was
tures. Catalysts were deactivated with HCK2 and HCK3 feeds at increased from 603 to 623 K (Table 12) in both catalysts. Compare
either 603, 613, 623 or 633 K (Cond. 2.). The activity was tested in this table the difference in Deff measured on spent catalyst A
using HCK 1 (Cond 1.) Fig. 5, and Fig. 6 show as an example, the deactivated using HCK1 feed to those carried out in HCK2 feed.
deltas of species, activities for the three sites and coke content The rate of effective diffusivity in mesopores is reduced 25% in A
on catalysts as a function of time in the four period of deactivation; and 23% on B catalyst deactivated at 623 K. The ratio of tail/main
deltas were measured at 603 K (Cond 1) and deactivation per- peak (see CI) during desorption of benzene have also decreased
formed at 613 K (Cond 2.) All of Deltas have been reduced by add- 45% by this increase in deactivation temperature. Benzene adsorp-
tion at equilibrium is reduced 12% (A) and 7.8% (B), but it is not the solution, but the values found are reproducible and the solu-
proportionally affected by coke built-up due to above-mentioned tion stable.
coke profile across the particle radius. The presence of aromatics First column in the table 8 describes the type of reaction, the
in the feed reduce the deactivation due to olefins and diolefins. 5 second ones reports the units for ki, kdi and the third and fourth
or 10% of DCN were added to an hydrotreated HCK1 product and the values of kinetics and deactivation energies for A-catalyst
A-catalyst was deactivated at 623 K. The results show a faster and HCK feed; the fourth and fifth of columns depicts the same
deactivation of catalyst than using HCK2 and HCK3, at the same constants for B-catalyst.
operating conditions, due to lower aromatics content (see CI). The final values of constants reported in Table 9 reproduce del-
Average coke/metal ratio (dispersion measured by XPS) on A tas with average deviations lower than ±0.7% in 95% confidence
catalyst, (deactivated at 623 K) decreases 68% and average disper- range for both catalysts and HCK1 feed. Fig. 8(a and b) represent,
sion of W and Ni increased 50 and 39% respect to the same sample as an example, deltas predicted for sulfur and aromatics vs. mea-
deactivated at 593 K. On B catalyst the same comparison shows sured at 613 K (Cond. 1) after deactivation at 613 or 623 K. The
that dispersion of Mo and Ni increase 59 and 40% respectively. simulation does not include data of deactivation at 633 K because
Average number of acid sites are now 35% and 42% higher on A the deactivation starts to follow other mechanism and coke depo-
and B catalyst respectively. These is due to change in the radial sition is start being heterogeneous. The simulation model predicts
profile of coke. Lower the temperature flatter is the carbon radial well the conversion of coke Fig. 8c and activity for A and B catalyst
distribution. (Fig. 8d–e).
Clearly most of coke is deposited in cluster in the external layers The kinetic rates constants of olefins and diolefins conversion
of the catalysts and there are many internal acid and metal sites for HCK2 and HCK3 (feeds with 5 and 10% of DCN) was measured
free of coke. at lower temperature and they are also reported in Table 9. The
The above information confirmed the inevitability of heteroge- simulation model predicts with deviation lower than ±1.5% the ole-
neous radial deposition of carbon when the catalyst is deactivated fins and diolefins conversion at 543 K for the catalyst deactivated
by HCK2 at temperatures higher than 583 K with olefinic feed- with HCK1 at 623 K.
stock, despite the small particle diameter, presence of high hydro-
gen partial pressure and aromatics in the feed and a well dispersed 3.4.1. Kinetic and deactivation rate constants for HCK2 and HCK3 feeds
Pd on A catalyst. The simulation program was run using different criteria to pre-
Notice that A catalyst treated at 533 K for 240 h with HCK2 and dict deltas of properties and coke built up after the deactivation of
Cond 2 shows a homogeneous profile of coke across the radius. catalyst at 603 and 623 K using HCK2 as feed. The sequential calcu-
lation used decreasing values of P function to help the program to
3.4. Kinetics and deactivation rate constants converge and obtain the new kinetic and deactivation rates con-
stants. After 20 mutations the program stop and report the best
HCK1 feed. Simulation program of Fig. 4 determined the value values obtained.
of kinetics and deactivation rates constants that reproduce deltas
of species and coke built up as function of time at different temper- (1) The kinetics constants of Table 9 were used, and the simula-
atures, time of deactivation and catalysts. The program predicted tion program calculated (GA) a new set of deactivation con-
well the conversion of species present in HCK1; Figs. 8 and 9 com- stants. They were calculated using data obtained at 603 and
pare the values predicted with those measured at 593, 603, 623 623 K (Cond 1) for HCK1 feed using A and B catalyst deacti-
and 633 K (Cond 1) for each of catalysts deactivated at 603 and vated with HCK2 at 603 and 623 K (Cond 2). Thus, the pro-
623 K (Cond 2). The program was seeded with values of properties, gram (GA) only mutated the deactivation constants. The
operating conditions, kinetic and deactivation rates constants, fluid constants were only obtained after 11 mutations using the
dynamic constants, and delta of species measured during the inter- range 7 < P < 5%. Then the program predicts sulfur and aro-
media tests and during deactivation. Thus, 88 data points were matics (28 data points, Fig. 9(a and b)) with average devia-
used by the program for these two feeds and for each of catalysts tion higher than ±5% and ±7% respectively, deviation that
to solve the mass equation balances and adjust the constants. increases with temperature. Fig. 9c shows that the program
The program, for example in the case of catalyst A and HCK1 feed, under predict the amounts of coke formed respect to those
adjusted the initial delta values in each of compartments in five measured. The higher the deactivation temperature, the
iterations. Then value of P function was obtained and used to higher the deviation observed.
adjust deactivation constants by mutating these parameters. GA (2) The same simulation than 1) but the data was obtained test-
converge after 7 and 8 mutations to achieve intermediate values ing the catalyst A and B with HCK2 as feed (tests at 603 and
of P function (3 < P < 1.5) for A and B catalyst and HCK1 feed. Then 623 K, Cond 1) using catalyst deactivated in HCK2 (Cond2.)
the simulation program was run again changing only the two sets The program only converged after 18 mutations using a
of ki and Ei that represent the major deviation (guessed-obtained). broad objective function (8 < P < 6%). The program predicts
This second simulation converged to lower P values (2 < P < 1.5; sulfur and aromatics content in product with deviation of
Q < 1.5%) after 8 additional mutations (4 min in dual processor ±6 and ±8% respectively, without random distribution.
lab-computer.) The results indicated that hydrogenation of di- (3) The same simulation than 1 but the program adjusts both
aromatic and mono-aromatics rates constants produce the large kinetic and deactivation rates constants using data obtained
deviation in the predicted-measured deltas (Q), thus they are the with HCK1 (Cond 1) feed and the catalysts deactivated with
reason why minimum differences specified for P were not achieved HCK2 (Cond 2). The program converged after 13 mutations
in the first run, but they were obtained in the second of runs. This with a broad objective function (6 < P < 4%) and predict sul-
methodology can be applied to the other constants in successive fur and aromatics content in products with deviation ±5 and
runs. When the program did not converge after a given maximum ±7% with skewness of +0.43 and 0.69 respectively. The
number of mutations, it stops and reported the last values of ki, kdj model under predict coke formed.
Ei and Edj, and delta of conversions to take decisions and change the (4) The same simulation than (2) but the GA adjusted both
strategies used for convergence. The constants found are reported kinetics and deactivation rates constants using data
in Table 9 for both catalysts and HCK1 feed. The program obtained obtained with HCK2 as feed (tests at 603 and 623 K, Cond
a faster convergence when the original values assumed are close to 1) and the catalysts deactivated by HCK2 feed (Cond 2).
The program shows improved deviation in the prediction of Eqs. (58) and (60) for both catalysts at 613 and 623 K (HCK1).
sulfur (±3%) but still present large deviation in aromatics Deviation are lower than ±1% in the 95% range of confidence.
(±6%) with skewness of + 0.23 and 0.58 respectively The same results were observed for a2 site and both catalysts. Devi-
(Fig. 9(a and b)). The model under predicts coke formed ation increase by including data obtained at 633 K.
(Fig. 9c). Catalyst deactivation for HCK2 and HCK3. Logarithm of deacti-
vation constants (kdi) for the three sites of catalyst A (obtained by
The higher the olefins content in the feed the larger the devia- the program using case 4 discussed above) were plotted against
tions observed in P values. The program cannot obtain kinetic the inverse of deactivation temperature for catalyst A in Fig. 10.
and deactivation parameters if data obtained at 633 K was The results show a linear correlation of ln kdi, obtained by the sim-
included in the simulation. ulation program, versus the inverse of deactivation temperature for
Above results show the impact of active sites distribution across 593 and 603 K. The four period samples of spent catalysts A were
the particles. Fast coke deposition in the mouth of pores at high tested at 623 K using HCK1; the average values of kd used in
temperature cannot be simulated using the current model with Fig. 10 show important deviations (±4.5%), even a low temperature.
average Deff and homogeneous distribution of active sites. All the values obtained by deactivation in the range 603 to 633 K
Catalyst deactivation for HCK1. The logarithm of deactivation present higher deviation and a non-linear behavior as a function
constants (kdi, values in Table 9), for the three sites of catalyst A of temperature (only the value for 613 K is shown in Fig. 10). The
were plotted against the inverse of deactivation temperature in higher the temperature of deactivation and the amount of DCN in
Fig. 10. The results show for deactivation and testing in HCK1, a the feed, the higher the deactivation of the three sites because of
quasi-similar deactivation energy in the period 40–232 h on carbon deposition mentioned above. The same happened for B
stream in the range of 593–623 K, not including the data obtained catalysts.
at 633 K points. Thus, it seems that there are different values in the The reaction model used here for all the reactions is based on
preexponential factor due to a global reduction in the number of average catalyst efficiency factor (Eq. (61)) and deactivation
active sites but not to changes in their apparent deactivation (Eqs. (58)–(60)); clearly, the model is less and less useful to predict
energies. A typical behavior with homogeneous coke and active product properties and catalyst deactivation when the amounts of
sites distribution (sees discussion below). Fig. 8(d and e) shows olefinic-diolefins in feed at temperatures increases. That happened
that the model predicted well activities (a1 and a3) using despite that the model fits well the experimental results in HCK2
and HCK3 at low temperature (543–563 K), when the rate of car-
bon deposition is moderate and quasi homogenous across the radii
of particles. The model fails progressively to predict conversions of
sulfur and aromatics for deactivation in the range of 593–633 K.
There is a non-lineal change in the number of acid and metal
sites with the amounts of carbon deposited during the deactivation
with HCK2. Butt and Petersen (1988) and Froment and Bishoff
(1961), among many others authors discussed the case of a non-
linear catalyst deactivation across the radius of particle; the cur-
rent case is similar to catalytic cracking in a riser reactor
(Forzatti and Lietti, 1999), where coking occurred in parallel to
the cracking reaction on the fraction of un-poisoned surface. Thus,
gi : f ½cðr; tÞ; aðr; tÞ; Dðr; tÞ is a complex function of apparent reac-
tion rates and catalyst deactivation of particle, and very different
than Eq. (61) used by the current model. Moderate temperature
(593–613 K) produce a fast deposition of coke when using HCK2
that contain aromatics, olefins and diolefins; they form coke in
the external layers of the particle with small effects in mass trans-
Fig. 10. Logarithm of deactivation constants calculated by the model for the three fer in mesopores. The reactants that found the active sites covered
active sites as a function of deactivation temperature. A-catalyst deactivated by by carbon at the external layer proceed into the next internal layer
jHCK1 and by dHCK 2.
of active sites (a moving front of reactions) and react there. Thus,
Feeds Hydrogenated products Cat A

1.4 1.4
1.4 MPa H2
10% DCN
+90% HCK Hydrog. Prod.
three measures
Gum mg/Lt
Gum mg/Lt
1.0 a) 1.0
Deviation in
+ Styrene
Deviation in
t: 30 min three measures 10% DCN
0.4 0.4 without deact.
HCK-G 10% DCN
with deact.
0% DCN
0.0 0.0
273 373 473 573 273 373 473 573
b) Heating temperature K c) Heating temperature K
Fig. 11. (a) Equip used to determine the relative ‘‘storage” stability of feed and product of reaction at 14 MPa of hydrogen pressure (30 min residence time, glass beads). (b)
gun and solid formation in HCK-G liquid products with and without DCN, as a function of temperatures at constant residence time of 30 min. (9c) Gun and solid formation in
hydrogenated products.
olefins, diolefins, sulfur and aromatics diffuse in the pores until 1961; Forzatti and Lietti, 1999; Moulijn et al., 2001), taken in
their found free-of-carbon active sites. In the range 613–633 K account mass transfer, by a sets of apparent kinetic and deactiva-
the olefins and diolefins polymerization is so fast that occurred tion rates functions, but it is important to determine by a dedicated
at the pore edge reducing the mass transfer of other reactants in experimentation the windows of operating variable where the
mesopores. The reason why the mesopores mouth start to be par- model can be applicable and use the simulation to select the best
tially plugged by carbon is the participation, at high temperature, technology to be used.
of some species of aromatics in building carbon deposits together Alzaid et al. (2018) study the hydrogenation of a series of linear
with olefins and diolefins. The amount of carbon is higher than ole- and cyclic diolefins diluted in decalin, at lower temperature than
fins and diolefins in the feed. Additional resistance to the diffusion used in this work (503 K and 3.4 MPa with 20% of olefins and
of the reactants built up and the activity is reduced faster. 5 wt% diolefins) over a NiMoS/Al2O3 catalyst. They found stronger
deactivation than current one; they mention that dimers are
3.4.2. Hydrogenation of olefins and diolefins present in DCN formed by conjugated double bonds and 4-methylstyrene had
Hydrogenation tests were performed at 534 and 536 K with the highest oligomer yield compared to all other diolefins were
HCK2, HCK3 and with a blend of hydrogenated HCK1 plus 10% of examined. Zhou et al. (2007) determine the kinetics of the selective
DCN; the tests have demonstrated that the faster coke-producer hydrogenation of pyrolysis gasoline providing the rates of coke
feed is the later one that contain very low amounts of aromatics built- up at lower temperature than current application and with-
(see CI) becuase they were hydrogenated. Notice that DCN contain out aromatics. Their kinetic rate of hydrogenation is almost 100
in PSC1 small amount of butene and butadiene a very reactive times faster than those measured here for olefins and diolefins
molecules that polymerize fast. Ratio of Hydrogenation to Poly- hydrogenation on both catalysts but in present of aromatics.
merization rates of reactions on A-catalyst with HCK2 feed is 0.3 Toba et al. (2007) studied the reactivity of various olefins contained
at 534 and 0.14 at 536 K (t: 232 h). This ratio is 0. 22 at 534 and in FCC gasoline on CoMoP/Al2O3. They suggest that hydrogenation
0.09 at 536 K (232 h) for HCK3 in presence of quasi similar of olefins depends on Mo and olefin structures. Brussaneze and
amounts of aromatics. These ratios are always reduced by rising Galiasso Tailleur (2001) develop a model to take in account the
the temperature (see respective activation energies in Table 9) effect of temperature and concentration in parallel catalytic deac-
When di-aromatics are reduced to 1.3% in the feed the rate of poly- tivation in fixed bed reactor; a new set of equation will be dis-
merization is doubled (CI). These tests confirm the importance of cussed in the second part of the paper.
aromatics to control the polymerization. The literature does not disclose about the effect of alkyl-
Thermal stability of the feeds. To understand why olefin and aromatics and Pd content on the rate of olefins and diolefins poly-
diolefins produce important deactivation, a thermal stability test merization and the plugging of micropores; they are important
was developed to measure, in relative ways, the amounts of gum because both parameters affect the rates of carbon built up and
and solid formed after half an hour storage at different tempera- its distribution (Galiasso Tailleur and Ravigli Nascar, 2008).
tures; the device shown in Fig. 11a, operating at 14 MPa of hydro- HCK1 produces, via cracking on acid sites, small size adsorbed
gen pressure, were used. The tests were carried out for HCK1, HCK2 intermediaries (A0 ) from large paraffinic molecules; these frag-
and HCK3 in presence of glass beads (80 mm), a solid with equal ments polymerize in situ to form dimers and trimers or they are
external surface than the catalysts but without acid sites and hydrogenated in nearby metals. These reactions happen very
hydrogenation capabilities. Fig. 11b shows that the rate of change slowly at temperatures below 583 K in presence of heavy aromat-
in gum and solid formed increase with the amount of DCN. The ics that control the adsorption. Different situation happens with
higher the storage temperature and olefin content, the higher the DCN in the feed, because it contain more reactive olefins and dio-
amount gum and solids formed, as expected base on commercial lefins, whose need to be transported across the particle radius and
experiences with DCN storage. The filter used retained two kind with different molecular configuration than those generated in situ
of materials: gum, most of them soluble in S2C, and insoluble by catalytic cracking (A0 ).
solids. Their ratio of them decreases with temperature, as well as A-catalyst have one active acid sites formed by Ti-Al (Bronsted)
the H/C ratio in carbonaceous deposits. When the feed contains clusters, and almost two active metals sites. The strongest Bronsted
PSC1, the gum and solid formed are more of 70% soluble in CS2, and some metals nano-dispersed are early coked by the feed
mostly of them formed by polymers of iso-olefins; in absence of (during preconditioning of the catalyst), thus, only the medium
PSC1 the major part of solids is formed by polymers of alkyl ben- strength sites accumulate carbon during catalyst cycle length.
zenes. Notice that the amount of coke is always higher than the There are several metals configuration formed by W, well
olefins and diolefins converted during the storage. The reactions dispersed on surface, and Ni and Pd are mostly located on top and
involved other molecules. in the border of clusters based on the sequential method of
The relative high thermal stability of hydrogenated products preparation. We speculate that some of them are able to hydro-
(C+5), obtained using catalyst A and feeds with and without olefins, genate p-bonded molecules (aromatics); other one interact with
are shown in Fig. 11c. Notice that all the products have similar sta- sigma bonded molecules (probably associate to the border of the
bility during heating; differences are lower than the reproducibil- W-Ni metals-cluster) via sulfur deficient sites; the metal clusters
ity of the measures, indicated by a vertical segment in Fig. 11c. and Bronsted sites participate in ring opening and isomerization.
When the hydrogenated products are blended with PSC1 the stabil- B catalyst has similar acid sites and different metallic cluster base
ity tests show that important amounts of gum and solid are formed on well dispersed Mo, with Ni on top or in the border of clusters.
(Fig. 11c). These tests confirm the well-known fact that gum and Coke is formed by polymerization on sites with low hydrogenation
solid formation are due to unsaturated components present in capabilities, likes in narrow micropores. Most of micropores cannot
PSC1 and the unviability of heating light components of DCN with- be access by nitrogen and benzene adsorption after the precondi-
out polymerization. tioning stage; as a result, only mesopores are active for the
Discussion. Hydrotreating catalyst deactivation by carbona- reactions.
ceous deposits depends on the number and types of acid sites The HCK1 produce small amounts of olefins via cracking, gener-
and metals present on surface, mass transfer, operating conditions ating a well spread carbonaceous deposits. HCK2 contain olefins
(temperature, hydrogen partial pressure) and presence of highly and diolefins that diffuse inside the mesoporous structure produc-
reactive molecules in the feed. It is possible to predict catalyst ing a relative well distribution of carbon via polymerization at
deactivation (Butt and Petersen, 1988; Froment and Bishoff, temperatures below 563 K. Above this temperature the reaction
of olefins and diolefins become so fast (40 times faster than other 192 h on stream using A than B catalysts. When 5 or 10% of
hydrogenating reaction at 593 K) that carbon is deposited with a DCN was added to the HCK feed, both catalysts, are fast deacti-
decreasing profile across the radius of the particle on A and B cat- vated at this temperature.
alyst. Polymerization occurred despite the preferential adsorption Higher the temperature used for catalyst deactivation (HCK1
of mono and di-aromatics on acids and metals sites. Current exper- feed), higher the coke built up and lower the metals and coke
iments confirm the high thermal and catalytic reactivity of DCN dispersion, number of acid sites, benzene effective diffusivity,
components during reaction with hydrogen above 593 K or during benzene adsorption, surface area and micropores volume on
the storage in hydrogen without catalyst at temperature below both catalysts. The three active sites deactivated at slower rate
423 K. The change in selectivity during deactivation in presence in catalyst A than in catalyst B. There are important differences
of olefins and diolefins probe the presence of almost three active in selectivity and final product quality before and after deacti-
sites that are affected differently by deactivation. vation due to the type of metal active phase and type of coke
The unsaturated hydrocarbons present in DCN produce differ- built up, for a similar porous structure.
ent types of gums (soluble in CS2) and solids (insoluble in CS2.) dur- Spent catalyst deactivated by HCK1 at temperature below 633 K
ing the storage tests. The thermal polymers have different degree shows homogeneous distribution of carbon without appreciable
of polymerization and contain more aromatics than those observed change appreciable in apparent molecular diffusivity of benzene
on the catalysts above 543 K. in mesopores. Carbon deposition block most of micropores in
DCN type of olefinic components can be hydrogenated in pres- both catalysts.
ence of large amounts of aromatics with a relatively small deacti- HCK1 feed form carbon by cracking of paraffins with ulterior
vation at temperatures (around 543–563 K) in presence of A polymerization and hydrogenation of olefins in-situ (a series-
catalyst. Two stages of hydrotreating seems the obvious solution, parallel type of mechanism); carbonaceous deposits are com-
already been used in the industry for heavy residue hydrocracking; posed by dimers and trimers of iso- and cycloparaffins, 73%
first reactor is working a very low temperature to remove the ole- (A) and 65% (B) of them are soluble in CS2.
fins and diolefins (a ‘‘guard” reactor), and second one in-series is Reaction rate of olefinic and diolefins reactions present in HCK2
set at higher temperature to desulfurize and hydrogenate the other feed can be measured at 543–563 K, but they react too fast in
components. Nevertheless, this solution introduces the need to use the range of 603–623 K. Carbonaceous deposits are quasi-
lower temperature in the HPHTS (Fig. 1) and then reheat of the homogeneously deposited on A catalyst at temperature below
feed for the second reactor, losing the HPHTS process advantages. 593 K due to preferential adsorption of aromatics and the effect
Catalyst operating in moving or ebullated-bed (one-stage), work- of Pd content. B catalyst deactivated faster at this condition and
ing at constant temperature (543 K), can operate properly with present small decreasing radial profile of carbon content across
limited amount of DCN by adding continuously regenerated cata- the particle at temperature above 593 K. The higher deactiva-
lyst and withdrawing the same amount of spent catalyst. The level tion temperature and contact time, the stepper the profile of
of activity at steady state is depending on temperature and solid carbon distribution across the radius. The moving front of poly-
throughput used. The higher the amount of DCN in the feed and merization reactions produce a progressive decrease of coke,
temperature used, the higher the solid throughput need to main- metal and sites content across the radius of particles (deactiva-
tain the average activity and the cost of the operation. The pres- tion by shells).
ence of Pd on surface is also critical (Galiasso Tailleur and Metals sites deactivated slower than acid sites, as expected, in
Ravigli, 2005), but the problem of fast external deposition of coke both catalysts. Hence, the selectivity of catalyst changes accord-
was not solved by increasing the well-disperse Pd content on ingly with the amounts of carbon deposited. Catalyst based on
WNi/TiO2 catalyst. Hence, only catalysts with high external con- PdWNi shows lower deactivation and higher isomerization
centration of hydrogenating metals (either Pt or Pd) may help to hydrogenation and ring opening than NiMo catalyst due to Pd
control coke built up at the surface of the particles (see coke in content.
the external layer in Fig. 7b). Catalyst with Pd well dispersed in The fluid dynamic tests confirmed that ebullated-bed reactor
the external layer (onion type of metal deposition) (Betti et al., can be simulated using a combined model of two compartment
2016; Wen et al., 2013) seems to achieve in part the fast hydro- in series with backflow and crossflow with the bubble phase
genation that reduce the catalyst deactivation by olefins and diole- operating in piston flow.
fins; other possibility is the use of hydrogenating catalysts The simulation program adjusted previous kinetic and deactiva-
deposited on thin film on structured-packing in a swing guard type tion rates parameters using 88 data point for each of catalyst
reactor, as was described by Galiasso Tailleur et al. (2008). and HCK1 feed at different deactivation temperature. The opti-
mum sets of constants can predict delta of components, other
than coke on catalyst, with deviation lower than ±0.7% for
4. Conclusions HCK-G feed and both catalysts. Coke is predicted with deviation
lower than ±1.5%.
High-Pressure-High-Temperature-Separator produces a The simulation program could not adjust kinetic and deactiva-
gas-phase hydrocarbon blend at high temperature and hydrogen tion rates constants for HCK2 and HCK3 feeds, and fail to predict
partial pressure. These products, plus other gaseous streams from delta of properties and coke content with reasonable accuracy
the refinery, are sent to a gas-phase reactor for additional in the range 593–623 K. The simulation program used a model
hydrogenation. This hydrogenation can be carried out in different that calculated an average efficiency for the particle and deacti-
types of gas-phase reactors and combination thereof depending vation of sites independent of carbon distribution across the
on the amounts of olefins in the feed. Two catalysts were studied particle. Both are affected by the radial profile of coke. Thus,
with different hydrogenation capabilities, but similar porous struc- the simulation needs another model.
ture and acid sites. They were tested with and without olefinic type There are three different types of sites on both catalysts. The
of feeds in accelerate deactivation tests at different operating metals perform the direct desulfurization on site 1 and the
temperatures. The main results are: hydrogenation of aromatics in sites 2. Acid sites carried out
the ring opening, dealkylation and cracking and polymerization.
Deactivating the catalysts at temperature of 623 K (Cond 2), Accelerate deactivation of sites at different temperatures follow
HCK1 feed produces smaller reduction of conversions after the order:
sites 3 > sites 1sites 1. Galiasso Tailleur, R., 2008. Low-emission diesel production by upgrading LCO plus
SR diesel fractions. Catal. Today 130, 492–500.
Weight of coke is always higher than the amounts of olefins dis-
Galiasso Tailleur, R., 2010. Effective diffusivity measurement in coke-deactivated
appeared, proving that other molecules (likes aromatics) are hydrotreating catalyst. Aiche meeting March 21, 2010.
involved in the polymerization. Galiasso Tailleur, R., 2014. Deep hydrotreating diesel fractions for low-sulfur fuel
production using gas-phase and trickle bed reactors systems operating in
parallel. Catal. Todays 220–222, 133–152.
Galiasso Tailleur, R., 2017. Effect of hydrocarbon vaporization and gases solubility
Declaration of Competing Interest on performance and cycle length of gas-phase – trickle bed reactors system
used for deep diesel hydrotreatment. Fuel 199, 298–323.
Galiasso Tailleur, Farina, C., 2019. Regeneration of PtSO4ZrTi/SiO2 by hydrogen
The authors declare that they have no known competing finan- treatment. Part II Kinetic rate of regeneration. Appl. Cat. (in preparation).
cial interests or personal relationships that could have appeared Galiasso Tailleur, R., Palmisano, E., Arreaza, G., Quenza, S., Ramnarine, S., 2002.
to influence the work reported in this paper. Process scheme for sequentially treating Diesel and vacuum gas oil. US patent #
6444116.
Galiasso Tailleur, R., Ravigli Nascar, Cortés, J., Ramírez, J., 2008. Desulfurización e
Acknowledgement hidrogenación profunda de LGO+LCO. Desactivación del catalizador WNiPd/
TiO2- Al2O3; Preprints of XXI SICAT. Málaga-Benalmádena-Costa. España.
Galiasso Tailleur, R., Farina, C., 2019b. Hydrogenation and hydrodesulfurization in
The author wants to acknowledge the experimental works of gas phase of hydrocracking and delayed coking products. II. Coke
Eng. C. Farina, J.A. Ravigli and P. Restiffo, the 13C NMR analyses of characterization and catalyst regeneration. Chem. Eng. Sci. (in preparation)
Dr A van Hoff laboratory, and the valuable discussions with profes- Galiasso Tailleur, R., Ravigli Nascar, J.A., 2008. The effect of aromatics on paraffin
mild hydrocracking reactions (WNiPd/CeY–Al2O3). Fuel Process. Technol. 89 (8),
sors G. F. Froment, C. H. Bartholemewy, and B. Delmony.
808–818.
Galiasso Tailleur, R., Hernandez, J., Rojas, A., 2008. Selective hydrogenation of olefins
Appendix A. Supplementary material with mass transfer control in a structured packed bed reactor. Fuel 87 (17–18),
3694–3705.
Galiasso Tailleur, T., Ravigli, J., 2005. Catalyst for ultra-low sulfur and aromatic
Supplementary data to this article can be found online at diesel. Appl. Catal. A 282, 227–235.
https://doi.org/10.1016/j.ces.2019.115195. Guisnet, M., Magnoux, P., 2001. Organic chemistry of coke formation. Appl. Catal. A
212 (1–2), 83–96.
Kim, J.S., Kim, H.K., Lee, S.B., Jae Choi, M.J., Lee, K.W., Kang, Y., 2001. Characteristics
References of carbon dioxide hydrogenation in a fluidized bed reactor. Korean J. Chem. Eng.
18 (4), 463–467.
Alzaid, A., Adjaye, J.W., Kevin, J., Smith, J., 2018. Impact of molecular structure on Lengyel, A., Magyar, Kalló, D., Hancsók, J., 2010. Catalytic coprocessing of delayed
the hydrogenation and oligomerization of diolefins over a Ni-Mo-S/c-Al2O3 coker light naphtha with light straight-run naphtha/FCC gasoline. Pet. Sci.
catalyst. Fuel 221, 206–215. Technol. 28, 946–954.
Argyle, M.D., Bartholomew, C.H., 2015. Heterogeneous catalyst deactivation and Magyar, Sz, Hancsók, J., Kalló, D., 2008. Reactivity of several olefins in the HDS of full
regeneration: a review. Catalysts 5 (1), 145–269. boiling point range FCC gasoline over PtPd/USY. Fuel Process. Tech. 89 (2008),
Badawi, M., Vivier, L., Duprez, D., 2010. Kinetic study of olefin hydrogenation on 736–739.
hydrotreating catalysts. J. Mol. Catal. A: Chem. 320, 34–39. Marafi, M., Stanislaus, A., 2001. Influence of catalyst acidity and quality of feedstock
Bartholomew, C.H., 2001. Mechanisms of catalyst deactivation. Appl. Catal. A: Gen. in hydrotreating catalyst deactivation by coke. Pet. Sci. Technol. 19 (5–6), 697–
212, 17–60. 710.
Betti, C., Badano, J., Lederhos, C., Maccarrone, M., Carrara, N., Coloma-Pascual, F., Miller, J.T., Reagen, W.J., Kadluk, J.A., Marshall, C.L., Kropf, A., 2000. Selective
Quiroga, M., Vera, C., 2016. Kinetic study of the selective hydrogenation of hydrodesulphurization of FCC naphtha with supported MoS2 catalysts: the role
styrene over a Pd egg-shell composite catalyst. Reac. Kinet. Mec. Cat. 117, 283. of cobalt. J. Catal. 193, 23–131.
Brussaneze, A., Galiasso Tailleur, R., 2001. Effect o temperature and concentration in Moulijn, J.A., van Diepen, A.E., Kapteijn, F., 2001. Catalyst deactivation: is it
catalytic deactivation of parallel reactions. Can. J. Chem. Eng. 79 (4), 524–535. predictable? What to do? Appl. Catal. A 212, 3–16.
Bunyard, E., 1999. Removing diolefins from coker naphtha necessary before Pacheco, M.E., Martins, V.M., Pinto, J.C., 2011. Accelerated deactivation of
hydrotreating. Gas Oil J., 17–19 hydrotreating catalysts by coke deposition. Eng. Chem. Res. 50 (10), 5975–5981.
Butt, J.B., Petersen, E.E., 1988. Activation, Deactivation and Poisoning of Catalysts. Toba, M., Miki, Y., Kanda, Y., Matsui, T., Harada, M., Yoshimura, Y., 2005. Selective
Academic Press, London. hydrodesulphurization of FCC gasoline over CoMo/Al2O3 sulfide catalyst. Catal.
Chernysheva, E.A., Usova, T.V., Izmashkina, I.V., 2005. Secondary naphthas as Today 104, 64–69.
components of hydrotreating feedstock. Chem. Technol. Fuels Oils 41 (2), 146– Toba, M., Miki, Y., Matsui, T., Harada, M., Yoshimura, Y., 2007. Reactivity of olefins in
150. the hydrodesulfurization of FCC gasoline over CoMo sulfide catalyst. Appl. Catal.
Diao, Sh., Qian, W., Luo, G., Wei Fe, Wang, Y., 2005. Gaseous catalytic hydrogenation B 70, 542–547.
of nitrobenzene to aniline in a two-stage fluidized bed reactor. Appl. Catal. A Vogelaar, B.M., Dickvan, A., Eijsbouts, L.S., Moulijn, J.A., 2007. Analysis of coke
286 (1), 30–35. deposition profiles in commercial spent hydroprocessing catalysts using Raman
Diez, F., Gates, B.C., Miller, J.T., Sajkowski, D.J., Kukes, S.G., 1990. Deactivation of a spectroscopy. Fuel 86 (7–8), 1122–1129.
Ni-Mo/Al203 catalyst: influence of coke on the hydroprocessing activity. Ind. Wen, X., Li, R., Yang, Y., Chen, J., Zhang, F., 2013. An egg-shell type Ni/Al2O3 catalyst
Eng. Chem. Res. 29, 1999–2004. derived from layered double hydroxides precursor for selective hydrogenation
Díez, F., Gates, B.C., Miller, J.T., Sajkowski, D.J., Kukes, S.G., 1990. Deactivation of a of pyrolysis gasoline. App. Catal. A: Gen. 468, 204.
Ni–Mo/c-Al2O3 catalyst: influence of coke on the hydroprocessing activity. Ind. Wivel, P., Zeuthen, P., Jacobson, A.C., 1991. Initial coking and deactivation of
Eng. Chem. Res. 29, 1999–2004. hydrotreating catalysts by real feeds. Stud. Surf. Sci. Catal. 68, 257–265.
Forzatti, P., Lietti, L., 1999. Catalyst deactivation. Catal. Today 52, 165–181. Wood, J., Gladden, L.F., 2003. Effect of coke deposition upon pore structure and self-
Froment, G.F., Bishoff, K.B., 1961. Chem. Eng. Sci. 16, 189–196. diffusion in deactivated industrial hydroprocessing catalysts. Appl Catal A: Gen.
Galiasso, R., Blanco, R., Gonzales, C., Quinteros, N., 1983. Deactivation of 249, 241–253.
hydrodemetallization catalyst by pore plugging. Fuel 62 (7), 817–822. Xin, Q., Alvarez-Majmutov, A., Dettman, H.D., Jinwen Chen, J., 2018. Hydrogenation
Galiasso Tailleur, R., 2005. Effect of catalyst deactivation on mild hydrocracking of olefins in bitumen-derived naphtha over a commercial hydrotreating
reactions. Comput. Chem. Eng. 29 (11–12), 2404–2419. catalyst. Energy Fuels 32 (5), 6167–6175.
Galiasso Tailleur, R., 2006. Diesel upgrading into a low emissions fuel. Fuel Process. Zeuthen, P., Bartholdy, J., Wiwel, P., Cooper, B.H., 1994. Formation of coke on
Technol. 87 (9), 759–767. hydrotreating catalysts and its effect on activity. Stud. Surf. Sci. Catal. 88, 199–
Galiasso Tailleur, R., 2007. Catalyst deactivation by aromatics during LCO upgrading 206.
to high quality diesel. Proceedings of European Congress of Chemical Zhou, Z., Cheng, Z., Cao, Y., Zhang, J., Yang, D., Yuan, W., 2007. Kinetics of the
Engineering (ECCE-6) Copenhagen, 16-20 September. selective hydrogenation of pyrolysis gasoline. Chem. Eng. Technol. 30, 105–109.

1 s2.0 S0009250919306876 Main

Uploaded by

Copyright:

Available Formats

1 s2.0 S0009250919306876 Main

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

1 s2.0 S0009250919306876 Main

Uploaded by

Copyright:

Available Formats

Chemical Engineering Science 210 (2019) 115195

Contents lists available at ScienceDirect

Chemical Engineering Science

Hydrogenation and hydrodesulfurization in gas phase of light

Co-process gas phase hydrocracking

E-mail address: [email protected]

a deactivation constant [–] Lmf minimum fluidization height [cm]

1. Introduction hydrogenation and hydrodesulfurization of diesel gas oil and vac-

Properties type of Cat. A B

PSC4 C15 6.7 0 6.365 1-pentene Cyclo-C==

Reactions (gas) Gas phase rate of reactions

CAT A, HCK1 CAT B, HCK2

0.99 3.0 0.99 a1 3.0

0.98 1.0 0.98 1.0

0 48 96 144 192 0 48 96 144 192

CAT A, HCK2 CAT B, HCK2

0 48 96 144 192 0 48 96 144 192

ðkThC4 Þ L h/lmol 8.5E+09 18,100 9.5E+09 18,300

Deactivated with HCK1 (623 K) Deactivated with HCK1 (623 K)

Predicted HDDAr conversion

0.99 623 K 0.99 623 K

HCK1 Feed HCK1 Feed

Deactivated HCK2 Deactivated HCK2

Predicted HDDAr conversion

(C/Cav)/C/RMe, Temp K/Feed Center 0.2 0.4 0.6 0.8 Edge

Cat – Feed Temp K A – HCK-G A – HCK-G + DCN

Feeds Hydrogenated products Cat A

You might also like