Você está na página 1de 9

Chemical Engineering Science 59 (2004) 2821 – 2829

www.elsevier.com/locate/ces

Deep desulfurization of diesel oil: kinetic studies and


process-improvement by the use of a two-phase reactor
with pre-saturator
C. Schmitz, L. Datsevitch, A. Jess∗
Department of Chemical Engineering, University Bayreuth, Universitatstr. 30, Bayreuth 95447, Germany
Received 3 December 2003; received in revised form 17 February 2004; accepted 24 March 2004

Abstract

Desulfurization of liquid fuels is usually obtained by heterogeneous catalyzed hydrodesulfurization (HDS). Improvement of this process
is needed, particularly with respect to future low residual S-contents of about 10 ppm. Up to now HDS of diesel oil is performed in a
trickle bed reactor equipped with an expensive H2 -recycle, although the H2 -supply is far beyond the amount chemically consumed, at
least for deep HDS of an already hydrotreated feed. In addition, scale up of trickle bed reactors is in general problematic. In this work
kinetic studies and above all a new HDS-concept using a two-phase reactor is presented. Thereby the oil is externally pre-saturated with
H2 and only the liquid phase is passed over the catalytic :xed bed. The H2 -recycle is then redundant, the intrinsic reaction rate can be
utilized, and scale-up problems do not occur. In addition, this concept can also be used for HDS of oils with a higher content of sulfur
and/or unsaturated hydrocarbons by installing a liquid recycle.
? 2004 Elsevier Ltd. All rights reserved.

Keywords: Fuel; Kinetics; Multiphase reactors; Reaction engineering; Desulfurization; Pre-saturation

1. Introduction CO and NOx . (3) The situation is similar for diesel oil with
respect to new catalysts and :lters to reduce particle emis-
In recent years deep desulfurization of diesel oil has at- sions. (4) New technologies based on fuel cells both for
tracted much attention due to the gradually reduction of the vehicles and stationary systems need fuels with less than at
statutory sulfur content (Table 1). In 2005, the S-limit will least 10 ppm S.
be reduced e.g. in Europe to 50 ppm compared to today’s As a quintessence one can state, that by 2010 the S-limit
value of typically 350 ppm. (In this paper, the S-content is of transportation fuels will probably be equal or lower than
given in ppm by weight.) Fuels with even less sulfur are or 10 ppm, at least in Europe, Japan and in the USA (Table 1).
will be on the market due to tax bene:ts. Desulfurization is up to now obtained by heterogeneous
The main reasons for these future stringent S-limits both catalyzed hydrodesulfurization (HDS), whereby the organic
for gasoline and diesel oil are: (1) Exhaust gases that con- sulfur species are converted to H2 S and the corresponding
tain SOx cause air pollution and acid rain, although e.g. hydrocarbons. H2 S is easily removed from the desulfurized
in western countries transportation fuels only contribute to oil and subsequently converted to elementary sulfur in the
about 3% to the total SOx -emissions; nevertheless, ultra-low Claus-process. Technically, HDS of diesel oil or of higher
sulfur fuels (¡ 10 ppm S) are still needed for urban areas. boiling crude oil fractions is up to now performed in a
(2) For gasoline, a S-content of preferably less than 10 ppm trickle bed reactor, i.e. in a three-phase system with gaseous
is needed for new lean fuel engines in combination with H2 , liquid oil and a solid catalyst. Typical reaction condi-
S-sensitive DeNOx-storage catalysts for further reduction of tions are 350◦ C and 7 MPa. With respect to the increasing
demand for practically S-free fuels (¡ 10 ppm), S-species
with a very low reactivity like dibenzothiophene deriva-
∗ Corresponding author. Tel.: +49-921-55-74 30; tives (Fig. 1) have to be converted, e.g. in a 2nd stage deep
fax: +49-921-55-74 35. HDS step, which leads to additional high investment and
E-mail address: jess@uni-bayreuth.de (A. Jess). operating costs.

0009-2509/$ - see front matter ? 2004 Elsevier Ltd. All rights reserved.
doi:10.1016/j.ces.2004.03.015
2822 C. Schmitz et al. / Chemical Engineering Science 59 (2004) 2821 – 2829

Table 1
Limits of sulfur-content of diesel oil in Europe, Japan and in the USA in ppm (by weight).

1953 1976 1990 1992 1996 1997 2000 2005 2006 2011a

EU — — — 500 — 350 50 — 10
Japan 12000 5000 — 2000 — 500 — 50 — —
USA — — 500 — — — — — 15 —
a Proposal of EU commission.

tion energies, reaction orders etc.). It was therefore also nec-


essary within this study to determine the exact reaction rate
S S parameters for the commercial CoMo-catalyst used, thereby
excluding all mass transfer resistances (e.g. pore diJusion).
Dibenzothiophene 4-Methyl-dibenzothiophene
(DBT) (4-MDBT)
Based on the respective kinetic studies, the intrinsic (maxi-
mum) rate of the chemical reaction of HDS could be deter-
mined and :nally compared with the performance both of
the new two-phase reactor and of the classical trickle bed
reactor.
S S

4,6-Dimethyl-dibenzothiophene 2,4,6-Trimethyl-dibenzothiophene
(4,6-DMDBT) (a typical C3-DBT)
2. New concept for (deep) desulfurization
Fig. 1. Typical organic sulfur compounds remaining in hydrotreated
(pre-desulfurized) diesel oil (for comparison, see also Fig. 8). The new concept for desulfurization by a two-phase re-
actor with pre-saturator is shown in Fig. 2. At :rst, the oil
is externally saturated with hydrogen in a pre-saturator, and
Trickle bed reactors have the following problems: (1) so only the saturated liquid oil is passed over the :xed bed.
They are in general complicate with respect to the mass The desired H2 -concentration at the reactor inlet and outlet
transfer (gas–liquid and liquid–solid). In addition, proper is adjusted by the pressure and—if needed(which is not the
hydrodynamic conditions (wetting of catalyst, distribution case for deep HDS)—by a partial recirculation of the liquid
of gas and liquid phase) are hard to realize. Scale-up is there- product. This process is more economic than the classical
fore not an easy task and pilot plants are often still needed. trickle bed HDS-technology: (1) Only the oil/hydrogen so-
(2) The amount of H2 fed to the reactor—above all in case of lution is fed to a two-phase plug Low reactor packed with the
a 2nd stage deep desulfurization of an already hydrotreated catalyst, where the (dissolved) H2 and the organic S-species
fuel—is much higher than the amount chemically needed react. Therefore, large three-phase trickle bed reactors can
(see below). Therefore, a costly recycle compressor for the be replaced by smaller tubular reactors. (2) H2 -recirculation
(unconsumed) hydrogen has to be installed. becomes redundant. (3) Hydrodynamics are simpli:ed, e.g.
There are several options to improve (deep) desulfuriza- a proper distribution of gas and liquid as demanded for a
tion of fuels: (1) Better catalysts for deep HDS based on trickle bed is not needed. So in case of the two-phase con-
nobel metals are on the market and still the focus of research cept scale-up problems practically cease to exist.
(e.g. Fujikawa et al., 2001; Navarro et al., 1996; Reinhoudt It should be emphasized, that in case of deep HDS,
et al., 1999; Robinson et al., 1999). (2) Alternative processes the S-content of the already hydrotreated oil is anyway
have been developed like the adsorption process SZorb of so low, that independent of the used reactor—trickle bed
ConocoPhillips (Sughrue, 2003), and (3) extraction of or- or two-phase reactor with pre-saturator—practically no
ganic S-compounds by novel extraction media like ionic liq- H2 -depletion occurs along the :xed bed. In other words, a
uids is now also considered (BIosmann et al., 2001; Esser H2 -supply within the reactor is not needed. This is clearly
et al., 2003). proven by Fig. 3, which shows the inLuence of pressure on
This work presents a new concept for deep HDS based the theoretical limit of the initial S-content, where a com-
on a two-phase :xed bed reactor with pre-saturation of the plete HDS in the two-phase reactor is stoichiometrically
oil (Datsevich and Mukhortov, 1997; Schmitz et al., 2003; still just possible. For the calculation depicted in Fig. 3,
Schmitz, 2003). the following assumptions were made: (1) The solubil-
Although several kinetic studies on HDS with diJerent ity of H2 in diesel oil is according to literature (Ronze
catalysts (mainly CoMo and NiMo) were already published et al., 2002) 0:14 mol H2 per kg of oil and MPa (370◦ C);
(e.g. Houalla et al., 1980; Zhang et al., 1996), they strongly almost the same result is obtained for the solubility in
diJer with respect to the kinetic parameters obtained (activa- n-dodecane (Schmitz, 2003), used in this work as the base
C. Schmitz et al. / Chemical Engineering Science 59 (2004) 2821 – 2829 2823

Oil (e.g. 100 ppm S) 500 ppm S, a complete conversion of all S-compounds to
Pre-saturator H2 S would only lead to a minor depletion (−7%) of the
H2
dissolved H2 down to a value equivalent to a H2 -pressure
of about 2:8 MPa (Fig. 3). So the amount of H2 leaving the
pre-saturator exceeds many times the amount needed for
100% HDS and would even be adequate to desulfurize an
Two phase fixed-
bed reactor Recycle pump oil with a much higher initial S-content, according to Fig. 3
(oil + solid cat.) 4600 ppm S for the assumed pressure of 3 MPa. Only in
case of an oil with a still higher S-content and/or rich in
unsaturates reactive for hydrogenation, an oil recycle would
be needed (see Section 4.4). Nevertheless, even then a liq-
Desulfurized oil uid recycle is more favourable than a gas recycle as needed
for a trickle bed: The power consumption of a recycle pump
Fig. 2. New concept for deep desulfurization by a two-phase reactor with for the liquid is much lower than of a hydrogen recycle
pre-saturator (the liquid recycle is an option not needed in case of deep compressor. In addition, the investment costs are higher in
HDS, see text).
the latter case.
General aspects of the pre-saturation concept were already
presented elsewhere (Datsevich and Mukhortov, 1997), and
5000 discussed in previous publications for the hydrogenation of
2 MPa (1) n-octene (Battsengel, 2002; Jess et al., 2002), (2) adipic
(for 100 % conversion by HDS)
Limit of S-content of diesel oil

acid-hexandiole-oligoester (Battsengel et al., 2003), and (3)


4000
ionone (Peter et al., 2003). In all these cases, the concen-
tration of the liquid reactant is in contrast to HDS high, and
3000
therefore a liquid recycle is absolutely needed to reach 100%
4600 ppm S conversion.
2000 In the present paper, the pre-saturation concept is ex-
panded to and discussed for deep HDS, i.e. to a typical case
1000 of a low concentration of the liquid reactant. The advan-
tage of the liquid-phase-only hydrogenation for low initial
0
concentrations is also con:rmed by Stankiewicz (2001), al-
0.0 0.5 1.0 1.5 2.0 2.5 3.0 though only based on theoretical considerations without dis-
Partial pressure of hydrogen in MPa cussing concrete reactions and experimental data, respec-
tively. The new concept for deep HDS of diesel oil was
Fig. 3. InLuence of H2 -pressure in the pre-saturator on the theoretical limit tested both for model oils and “real” diesel oils (see also
of the initial S-content to obtain complete HDS in the two-phase reactor
(process con:g. as in Fig. 2; assumptions see text, e.g. H2 -consumption
Schmitz et al., 2003; Schmitz, 2003).
by hydrogenation of ole:ns and aromatics is not considered).

3. Experimental and data analysis

component for experiments with model oils. (2) The The experiments were carried out in a continuous :xed
H2 -supply only occurs in the pre-saturator, and equilibrium bed two-phase reactor (for details see Schmitz, 2003).
according to Henry’s law is established. (3) 2 mol H2 are The externally heated reactor had a length of 140 cm
needed to convert 1 mol S (organic S) to H2 S. Thereby must and a diameter of 4 mm, resulting in a volume of the
be emphasized that this relative low consumption factor is :xed bed of 18 cm3 . The (axial) temperature pro:le was
only valied for CoMo-catalysts with a low hydrogenation controlled at four points along the bed, indicating a max-
activity (as used in this work); this speci:c point will be imum T -diJerence of ±2 K. The oil was pumped to the
discussed further in Sections 4.1 and 4.4. (4) Ole:ns and pre-saturator, where hydrogen is dispersed into the oil by
other unsaturated (aromatic) hydrocarbons are not present means of a frit. The :ll level of oil in the pre-saturator was
(or at least “inert” for hydrogenation on CoMo) in the controlled and regulated by an inspection glass in combi-
pre-desulfurized hydrogenated oil. So a H2 -consumption by nation with a level controller. From time to time it was
side reactions can be neglected. (5) With respect to kinetics veri:ed, that the oil was really completely saturated with
and an adequate amount of solved H2 within the whole re- H2 : A sample of the oil (at reaction conditions) was taken
actor, respectively, the minimum H2 -content at the reactor behind the pre-saturator by a syringe and the amount of
exit should be at least equivalent to a H2 -pressure of 1 MPa. degasing hydrogen was measured.
For example, in case of a pressure of 3 MPa in The oil saturated with H2 in the pre-saturator under re-
the pre-saturator and an already hydrotreated oil with action conditions (T; p) was then pumped through the :xed
2824 C. Schmitz et al. / Chemical Engineering Science 59 (2004) 2821 – 2829

bed. The total S-content was determined by elementary anal-


ysis (Antek 9000, pyroLuorescence detector) with an error
of about 1 ppm. The calibration was adjusted for a range of
10–300 ppm S.
The content of individual S-species like benzothiophene
(BT), dibenzothiophene (DBT), 4-methyldibenzothiophene
(4-DMDBT) and 4,6-dimethyldibenzothiophene (4,6-
DMDBT) was determined by a S-sensitive gaschromato-
graph (GC Varian CP-3800 with pulsed Lame photometric
detector (PFPD)).
The experiments were conducted with a commercial
CoMo-catalyst (SIud-Chemie C20-6, BET surface area:
245 m2 g−1 ), at H2 -pressures from 1 to 7 MPa, tempera-
tures of 340–400◦ C, and residence times up to 1 g h cm−3
(ratio of mass of catalyst to feed rate of oil). The original
catalyst has a special form of three interlinked cylinders Fig. 4. HDS of model oils (each) with individual sulfur compounds
with a characteristic size of 1:7 mm (equivalent diameter of (CoMo-catalyst, 360◦ C, 3 MPa, feed: n-dodecane and in each case only
one single S-compound; lines indicate :rst order reaction with respect to
spheres with the same ratio of volume to external surface
sulfur compound, see text).
area). Selected experiments were done with a smaller diam-
eter of 0:2 mm by grinding the original catalyst. (Although
in case of the original catalyst “wall eJects” may contribute
to a decrease of the achieavable conversion, this was not ceeds by the so-called direct route under formation of
detected.) Model oils as well as two “real” pre-desulfurized biphenyl and H2 S without preliminary hydrogenation of the
diesel oils were used as feedstocks. Model oils were ob- aromatic ring(s). This is con:rmed by an additional exper-
tained by mixing n-dodecane with BT, DBT, 4-MDBT, iment with biphenyl as feed (1% in dodecane). At typical
and 4,6-DMDBT. The diesel oils with a S-content of 260 HDS-conditions (30 bar, 360◦ C, 0:31 g h cm−3 ), a measur-
and 290 ppm were provided by MiRO re:nery, Karlsruhe, able conversion of biphenyl—e.g. to cyclohexylbenzene—
Germany. was not detected (GC with Lame ionization detector). This
In addition to the experiments with the pre-saturation re- again proves that the CoMo-catalyst used here is relative
actor, comparative experiments were also conducted with inactive for hydrogenation of aromatics. So the assumption
a trickle bed reactor (45 cm3 of catalyst), thereby taking a of the relative low consumption ratio of 2 mol H2 per mol S,
model diesel oil (300 ppm S as 4,6-DMDBT in n-dodecane) as used to compile Fig. 3, is justi:ed for the CoMo-catalyst
as a feed. used in this work.
All experiments with individual S-compounds show that
the HDS-reaction is :rst order with respect to the investi-
4. Results and discussion gated compounds. The inLuence of the H2 -pressure is neg-
ligible (zero order), at least in the investigated range from
4.1. HDS of individual S-compounds (experiments with 1 to 7 MPa (Schmitz, 2003). It was beyond the scope of
model oils) this study to explain this unexpected reaction order of zero
with respect to H2 . According to literature, this may be ex-
All HDS-experiments were conducted with the two-phase plained by the interplay of H2 -consumption and formation
reactor (except those described in chapter 4.3). Fig. 4 sum- of HDS-inhibiting H2 S (Vogelaar et al., 2003).
marizes typical results with model oils (semi-log plot), So the decrease of the concentration of individual
whereby each experiment was conducted with only one S-species is given by
single S-species in n-dodecane. (The values in brackets in
cS = cS; 0 exp(−k1
): (1)
Fig. 4 indicate slightly insecure results with S-contents less
than 10 ppm, i.e. out of the accurate range of calibration Thereby cS and cS; 0 are the residual and initial contents
of the elementary analyser.) As expected, the S-content de- of the S-species (mgS =kgoil = ppm), k1 is the rate constant
creases with increasing residence time, although with strong (m−3 kg−1 s1 ), and
the residence time (kg s m−3 ). Ac-
diJerences with respect to the reactivity of the investigated cording to Eq. (1), the plot of log cS; tot vs.
should give a
S-species. Above all, 4,6-DMDBT is much less reactive straight line, which is con:rmed by Fig. 4. The inLuence of
than 4-MDBT, DBT, and BT. the temperature is given by k1 = k1; 0 exp(−EA =RT).
The analysis of the HDS product(s) by the S-sensitive The kinetic data (reaction order(s), pre-exponential fac-
gaschromatograph clearly showed that no S-containing tors, activation energies) for HDS of DBT, 4-MDBT and
species other than the model S-compounds of the feed 4,6-DMDBT (Table 2) were deduced from experiments with
appeared. So most likely HDS on CoMo exclusively pro- varied reaction temperature, pressure and initial S-content
C. Schmitz et al. / Chemical Engineering Science 59 (2004) 2821 – 2829 2825

Table 2 1000
Activation energy, pre-exponential factor, and rate constant (360◦ C) for 360°C 380°C
HDS of individual S-compounds (340–400◦ C, 1–7 MPa, CoMo-catalyst

Measured S-content in ppm


(C20-6, SIud Chemie))

S-Compound Experiments (own measurement) 100


+ 20 %
EA in k1; 0 in k1 (360◦ C) in For comparison:
kJ mol−1 cm3 g−1 h−1 cm3 g−1 h−1 EA from
literaturea
10
DBT 87 2:9 × 108 19 92
4-MDBT 117 2:3 × 1010 5 121
4,6-DMDBT 132 2:1 × 1011 2.7 138 - 20 %

a Houalla et al. (1980) 1


1 10 100 1000
Calculated S-content in ppm

Fig. 5. Comparison of measured and calculated residual sulfur contents


(details in Schmitz, 2003). In case of BT, kinetic data were (model oil with 4,6-DMDBT in n-dodecane (300 ppm S); dashed lines
not determined because of its high reactivity, virtually lead- indicate 20% aberration).
ing to full conversion even for low temperatures and/or
residence times (Fig. 4). The values of the activation en-
ergy for the HDS of all three DBT derivatives are in good
agreement with typical data reported in literature (Table 2).
Reruns of experiments with model oil as well as with diesel
oil indicate, that no measurable aging of the catalyst occur
within the investigated timeframe (500 h with diesel oil and
800 h with model oils), and that the reproducibility is very
well.
The inLuence of pore or external diJusion on the ef-
fective rate was excluded by variation of particle size
(model compound DBT; original size of 1.7 and 0:2 mm).
It turned out that for the given conditions, the rate is in-
dependent of the particle size, and so the intrinsic reaction
rate was always measured, at least for the S-compounds
investigated and the original particle size of 1:7 mm. Fur-
ther calculations (Schmitz, 2003) indicate, that e.g. in case
of the relative reactive DBT, pore diJusion would only
have an eJect for a particle size (spheres) of more than Fig. 6. HDS of a model diesel oil consisting of a mixture of 4,6-DMDBT,
about 8 mm. 4-MDBT and DBT in n-dodecane (360◦ C, 3 MPa).
The reliability of the kinetic data is given in Fig. 5 by
the comparison of experimental and predicted values of
the residual S-content based on Eq. (1) and the data given (total) S-contents of less than 300 ppm, competing eJects
in Table 2, taking 4,6-DMDBT as example. The aberra- —e.g. by adsorption phenomena—can be neglected.
tion is about 20%, indicating that the parameters given in Further calculations show that the decrease of the total
Table 2 are satisfyingly describing the conversion of the S-content cS; tot (in ppm) can be well approximated by a rate
model S-compounds within a relative wide range of the second-order with respect to total sulfur, i.e.
S-content from 300 to 10 ppm, i.e. within the adjusted range
of calibration of the elementary analysis (Antek 9000). Sim- cS; tot = cS; tot; 0 =[k2
cS; tot; 0 + 1] (2)
ilar results were obtained for the other S-species (Schmitz,
2003). with the rate constant k2 in m3 kg1 s−1 ppm1 .
The result of HDS of a model oil containing three On :rst hand this is in contrast to the :rst-order reaction
S-species (DBT, 4-MDBT and 4,6-DMDBT (about found for the individual S-species, but this eJect can be
90 ppm S each) with n-dodecane) is shown in Fig. 6. For easily explained by the stepwise HDS of DBT, MDBT and
comparison, calculated results (lines in Fig. 6) based on DMDBT, which strongly diJer in reactivity (see Fig. 4). So
Eq. (1) and the kinetic data (Table 2) are also indicated. It the rate of HDS (with respect to total S) superproportionally
is interesting to note, that the HDS of all three DBT deriva- decreases with increasing degree of desulfurization by the
tives can be well described by the kinetics obtained from increasing portion of the remaining, less reactive compounds
experiments with single S-species. So at least for initial like 4,6-DMDBT.
2826 C. Schmitz et al. / Chemical Engineering Science 59 (2004) 2821 – 2829

60
4-MDBT in diesel oil 4,6-DMDBT in diesel oil

(as 4-MDBT or 4,6-DMDBT)


50

S-content in ppm
40
calculated curves for 1st
30 order reaction and the kinetic
data given in Tab.2

20

10

0
0.0 0.2 0.4 0.6 0.8
Residence time τ in g.h.cm-3

Fig. 7. Deep HDS of diesel oil (feed: two already hydrotreated diesel oils Fig. 9. Rest content of selected S-species at diJerent stages of HDS
with 290 and 260 ppm S, 360◦ C, 3 MPa; straight (full) lines indicate of an already hydrotreated (pre-desulfurized) diesel oil (diesel oil with
second order with respect to total S, see Eq. (2)). 260 ppm S, 390◦ C, other conditions as given in Fig. 7; values calculated
based on S-sensitive (PFPD) chromatograms).

4.2. HDS of pre-desulfurized diesel oil


The comparison of Fig. 7 with Fig. 6 indicates that the re-
Subsequent to the experiments with model S-compounds, action rate of HDS of both “real” diesel oil (with respect to
two “real” pre-desulfurized diesel oils with 260 and total S) is lower than in case of a model oil even with the re-
290 ppm S were used as feedstocks for HDS in the fractory 4,6-DMDBT, above all in case of the diesel oil with
two-phase reactor. Typical experimental results are shown an initial S-content of 290 ppm S (Fig. 7). Nevertheless,
in Fig. 7 with respect to the inLuence of the residence time the conversion rate of individual S-species like 4-MDBT
on the total sulfur content. The two straight lines repre- and 4,6-DMDBT is not or only slightly lower in the diesel
sent the plot of the linearized form of Eq. (2) (1=cS; tot vs. oil matrix, as shown by comparison of measured values in

) to con:rm the second-order reaction of HDS. Fig. 8 the diesel oil (260 ppm S) and the calculation based on the
shows characteristic S-species determined by chromatog- respective kinetic data of HDS in n-dodecane (Fig. 9). (The
raphy (PFPD-detector). As expected, only the refractory same result is obtained for the second diesel oil, see Schmitz,
DBT-derivatives are present in the pre-desulfurized feed, 2003.) So at least in case of the two investigated diesel oils,
and mainly the multi-substituted S-compounds remain in inhibition of HDS by N-compounds (here about 90 ppm N
the product oil. in both oils) and by aromatic hydrocarbons (not determined

Fig. 8. S-sensitive chromatograms at diJerent stages of HDS of an already hydrotreated (pre-desulfurized) diesel oil (GC with PFPD, feed: diesel oil
with 260 ppm S from MiRO re:nery in Karlsruhe, Germany), 390◦ C, other conditions as given in Fig. 7.
C. Schmitz et al. / Chemical Engineering Science 59 (2004) 2821 – 2829 2827

300 Oil (5000 ppm S, 20 wt.-% olefins)


2-phase-reactor 3-phase-reactor
Pre-saturator
250
H2
S-content in ppm
(as 4,6-DMDBT)

200
Trend 3-phase-
reactor Two phase fixed-
150 Recycle pump
bed reactor (recycle rate: 1.68)
(oil + solid cat.)
100

50
Trend 2-phase-
reactor
Desulfurized oil
0
0.0 0.2 0.4 0.6 0.8 1.0 1.2 (10 ppm S, no olefins)
Residence time τ in g.h.cm-3
Fig. 11. HDS of a diesel oil rich in sulfur and ole:ns in a two-phase
Fig. 10. HDS in a conventional trickle bed reactor and in a reactor reactor with H2 -pre-saturation (assumptions see text).
with H2 -pre-saturation of oil (model oil with 4,6-DMDBT in n-dodecane,
360◦ C, 3 MPa, SIud-Chemie CoMo-catalyst).

scope of this study), e.g. by adjusting optimal conditions


in this work)—as reported in literature (e.g. Koltai et al., with respect to the velocity of the liquid and the gas phase.
2002; Ramachandran and Massoth, 1982; Kabe et al., 1997) Nevertheless, a higher reaction rate than the limiting intrin-
—is irrelevant. So the main reason for the relative low reac- sic rate reached with the two-phase reactor is in any case
tion rate of the diesel oils (with respect to total S) is there- not possible.
fore most probably the high content of C3 - and C4 -DBTs
with three or four additional carbon atoms as methyl- or 4.4. Outlook: potential extension of the pre-saturation
ethyl-groups (see Fig. 8), i.e. of S-species, which are obvi- technology
ously even less reactive than 4,6-DMDBT.
As in case of HDS of the investigated model S-compounds, The two-phase reactor with pre-saturator is by far not
the inLuence of the pressure is negligible, at least for the limited to the case of deep HDS, and can easily be ex-
investigated range from 1 to 7 MPa. The decrease of the tended to feedstocks rich in sulfur and/or unsaturated hy-
total sulfur content with increasing residence time can be drocarbons. This is illustrated by Fig. 11, taking HDS of
approximated quite well by assuming a second-order reac- a diesel oil with 5000 ppm S and 20 wt% of unsaturated
tion with respect to total sulfur (lines in Fig. 7), which is hydrocarbons for a simpli:ed, but instructive case study
in good agreement with the experimental :ndings with a to underline this suggestion. For the respective calcula-
mixture of model S-compounds. As already expected ac- tion (Fig. 11), the following assumptions were made: (1)
cording to the experiments with model oils (with equal or The pressure in the pre-saturator is 5 MPa, i.e. 0:7 mol H2
even higher intrinsic HDS reaction rates), pore or external leave the pre-saturator per kg of oil. (2) For simpli:-
diJusion have no inLuence on the eJective rate, which was cation, all unsaturated hydrocarbons were considered as
again proven by experiments with varied particle size (1.7 dodecene. (3) The H2 -consumption is 2 mol H2 per mol S
and 0:2 mm). So again, the intrinsic rates were measured. and 1 mol H2 per mol dodecene, respectively. (4) The res-
idence time is suTcient both for complete conversion of
4.3. Comparison of deep HDS in the two-phase reactor all the ole:ns and to reach a residual S-content of 10 ppm.
(with pre-saturator) and in a trickle bed reactor (5) A H2 -concentration at the reactor outlet equivalent to
a partial pressure of H2 of 1 MPa is still suTcient with
Finally, comparative experiments were also conducted respect to kinetics.
with a lab-scale trickle bed reactor, thereby using a model oil In contrast to a trickle bed, where hydrogen is supplied by
with 300 ppm S (as 4,6-DMDBT) in n-dodecane. Compared gas–liquid mass transfer within the reactor, a noticeable de-
to the respective experiment with the two-phase reactor, the pletion of dissolved hydrogen would occur in this case study
reaction rate is by a factor of about 2.5 lower (Fig. 10). within the two-phase reactor. Therefore, a liquid recycle is
So only the two-phase reactor with pre-saturator utilizes the needed to compensate the lack of hydrogen. Nevertheless,
maximum (intrinsic) rate of the chemical reaction. In case the recycle rate of less than two (1.68) needed to ensure the
of the trickle bed reactor, this limiting maximum rate is not above-mentioned minimum H2 -content at the reactor outlet
reached, probably because of improper gas–liquid distribu- is still relatively small. So the liquid-phase-only hydrotreat-
tion and/or wetting of the catalyst. Thereby must be empha- ing is certainly still more favourable than HDS in a trickle
sized that probably the operation of the lab-scale trickle bed bed, e.g. with respect to the power consumption of a recycle
reactor could possibly be improved (which was beyond the gas compressor compared to an oil pump.
2828 C. Schmitz et al. / Chemical Engineering Science 59 (2004) 2821 – 2829

Two aspects should :nally be mentioned: 


and comments, SUDCHEMIE for supplying the catalyst,
(1) If other catalysts, e.g. Ni- or Pt-Pd-based systems spe- and the MiRO re:nery (Karlsruhe) for supplying the diesel
ci:c for deep HDS, are used in the second step of HDS, their oil samples.
hydrogenation properties allow a rather easy hydrogenation
of benzyl rings, and then much more hydrogen than just 2 References
mol per mol of organic sulfur will be consumed. Neverthe-
less, the basic principle of the pre-saturation concept is still Battsengel, B., 2002. Zur Mikro- und Makrokinetik mehrphasiger
valid, but a higher total pressure and/or higher recycle rate heterogen-katalysierter Reaktionssysteme. Ph.D. Thesis, University of
of oil would be needed as outlined before. Technology, Aachen.
(2) In contrast to HDS of diesel and fuel oil, the Battsengel, B., Datsevich, L., Jess, A., MIunnich, C., Peter, S., Turek,
T., 2003. Einsatz eines Zweiphasenreaktors mit VorsIattiger fIur
pre-saturation system has clearly one drawback in case of Mehrphasenreaktionen. Chemie Ingenieur Technik 75, 553–558.
HDS of gasoline rich in ole:nes and/or aromatics (e.g. from BIosmann, A., Datsevitch, L., Jess, A., Lauter, A., Schmitz, C.,
Luid catalytic cracking). If a liquid recycle is needed for a Wasserscheid, P., 2001. Deep desulfurization of diesel fuel
suTcient H2 -supply, the octane number of the product is by extraction with ionic liquids. Chemical Communications 23,
presumably decreased, which is not favorable. 2494–2495.
Datsevich, L., Mukhortov, D.A., 1997. Process for hydrogenation of
So in future experimental work, the extension of the organic compounds. Patent RU 2083540.
pre-saturation technology to feedstocks rich in sulfur and/or Esser, J., Jess, A., Wasserscheid, P., 2003. Deep desulfurization of fuels
unsaturated hydrocarbons by installing a liquid recycle will by extraction with ionic liquids. 226th American Chemical Society
have to be further investigated by respective experiments in (ACS) National Meeting, New York, 7–11, September 2003.
Fujikawa, T., Idei, K., Ohki, K., Mizuguchi, H., Usui, K., 2001.
order to prove the theoretical suggestions outlined above.
Kinetic behaviour of hydrogenation of aromatics in diesel fuel over
silica-aluminia-supported bimetallic Pt–Pd catalyst. Applied Catalysis
A General 205, 271–277.
5. Conclusions Houalla, M., Broderick, D.H., Sapre, A.V., Nag, N.K., de Beer,
V.H.J., Gates, B.C., Kwart, H., 1980. Hydrodesulfurization of
methyl-substituted dibenzothiophenes catalyzed by sul:ded Co–Mo=
A new concept for hydrodesulfurization (HDS) of diesel –Al2 O3 . Journal of Catalysis 61, 523–527.
oil using a two-phase reactor (pre-saturated oil and solid cat- Jess, A., Datsevich, L., Battsengel, B., 2002. Experimental and theoretical
alyst) was investigated under typical reaction conditions of studies on hydrogenation in multiphase :xed bed reactors. Chemical
HDS (340–400◦ C, 1–7 MPa, CoMo-catalyst). Model oils Engineering & Technology 25, 621–626.
with dibenzothiophene derivatives as well as predesulfur- Kabe, T., Akamatsu, K., Ishihara, A., Otsuki, S., Godo, M., Zhang, Q.,
Qian, W., 1997. Deep hydrodesulfurization of light gas oil. 1. Kinetics
ized and therefore hydrogenated diesel oils were used as and mechanisms of dibenzothiophene hydrodesulfurization. Industrial
feedstocks. On the basis of the experiments and of the out- Engineering Chemistry Research 36, 5146–5152.
lined theoretical considerations, the comparison of HDS in Koltai, T., Macaud, M., Guevara, A., Schulz, E., Lemaire, M., Bacaud,
a classical trickle bed reactor with the new two-phase tech- R., Vrinat, M., 2002. Comparative inhibiting eJect of polycondensed
nology clearly shows the advantages of the latter: (1) The aromatics and nitrogen compounds on the hydrodesulfurization of
alkyldibenzothiophenes. Applied Catalysis A General 231, 253–261.
new technology utilizes the maximum (intrinsic) chemical Navarro, R., Pawelec, B., Fierro, J.L.G., Vasudevan, P.T., Cambra, J.F.,
rate of the HDS reaction, whereas in case of a trickle bed Guemez, M.B., Arias, P.L., 1996. Deep hydrodesulfurization of DBT
this rate is an upper limit, which can not or only hardly be and diesel fuel on supported Pt and Ir catalysts. Applied Catalysis A
reached. (2) The amount of H2 dissolved in the pre-saturator General 137, 269–286.
Peter, S., Datsevich, L., Jess, A., Panten, J., Pickenhagen, W.,
is by far enough for HDS of already hydrotreated oil. So
Schatkowski, D., 2003. Chemical engineering aspects of hydrogenation
there is no need of the presence of gaseous H2 within the of :ne chemicals. Fourth European Congress of Chemical Engineering,
reactor to supply additional H2 into the liquid phase, and a Granada, Spain, 21–25 September 2003.
H2 -recycle is obsolete. (3) The scale-up of the two-phase Ramachandran, R., Massoth, F.E., 1982. The eJects of pyridine
reactor is much easier than of a trickle bed reactor. (4) Fur- and coke poisoning on benzothiophene hydrodesulfurization
over cobalt-molybdenum/alumina catalyst. Chemical Engineering
thermore, the liquid-phase-only HDS is also an interesting
Communication 18, 239–254.
option for HDS of diesel oils (and other gas oils) rich in Reinhoudt, H.R., Troost, R., van Langveld, A.D., Sie, S.T., van
sulfur and/or ole:ns. In this case, a liquid recycle has to be Veen, J.A.R., Moulijn, J.A., 1999. Catalysts for second-stage deep
installed, which is still much less expensive than the com- hydrodesulfurisation of gas oil. Fuel Processing Technology 61,
mon hydrogen recycle needed for a trickle bed reactor. 133–147.
Robinson, W.R.A.M., von Veen, J.A.R., de Beer, V.H.J., Moulijn, J.A.,
1999. Development of deep hydrodesulfurization catalysts II. NiW,
Pt and Pd catalysts tested with (substituted) dibenzothiophene. Fuel
Acknowledgements Processing Technology 61, 103–116.
Ronze, D., Fongarland, P., Pitault, I., Forissier, M., 2002. Hydrogen
Financial support by the Deutsche Forschungsgemein- solubility in straight run gasoil. Chemical Engineering Science 57,
547–553.
schaft (Je 257/2) and BP Oil is gratefully acknowledged. Schmitz, C., 2003. Zur Kinetik und zur verbesserten ReaktionsfIuhrung
We thank Nick J. Gudde (Re8ning Technology, BP Oil, der hydrierenden Tiefentschwefelung von DieselIol. Ph.D. Thesis,
Sunbury-on-Thames, Great Britain) for fruitful discussions University of Bayreuth.
C. Schmitz et al. / Chemical Engineering Science 59 (2004) 2821 – 2829 2829

Schmitz, C., Datsevich, L., Gudde, N.J., Jess, A., 2003. Improvement of Vogelaar, B.M., Kagami, N., von Langeveld, A.D., Eijsbours, S., Moulijn,
the deep desulfurization of fuels by the employment of a pre-saturation J.A., 2003. Active sites and activity: the eJect of H2 and H2 S partial
and a recycle. 4th European Congress of Chemical Engineering, pressure. 226th American Chemical Society (ACS) National Meeting,
Granada, Spain, 21–25 September 2003. New York, 7–11 September 2003.
Stankiewicz, A., 2001. Process intensi:cation in in-line monolithic Zhang, Q., Ishihara, A., Kabe, T., 1996. Deep desulfurization of light oil
reactors. Chemical Engineering Science 56, 359–364. (part 4). Hydrodesulfurization of methyl-substituted dibenzothiophenes
Sughrue, E.L., 2003. Chemistry added design of future clean fuels. 226th catalyzed by various Co–Mo=Al2 O3 and Ni–Mo=Al2 O3 catalysts.
American Chemical Society (ACS) National Meeting, New York, Journal Japanese Petroleum Institute 39, 410–417.
7–11 September 2003.

Você também pode gostar