Catalysis Today 50 (1999) 73±85

Production of reformulated gasoline in the FCC unit.

Effect of feedstock type on gasoline composition
Angelos A. Lappas*, Dimitris K. Iatridis, Iacovos A. Vasalos
Chemical Process Engineering Research Institute (CPERI) and Department of Chemical Engineering, University of Thessaloniki,
PO Box 361, 57001 Thermi, Thessaloniki, Greece

Abstract

FCC gasoline is a major component in the total gasoline pool produced in an integrated re®nery, but it contains many
compounds (ole®ns, sulfur and aromatics) which lead to harmful automobile emissions. The objective of the present study is
to determine the effect of feedstock quality on gasoline composition in a range of operating variables with a constant type of
catalyst. The work was carried out in an FCC pilot plant constructed and operated in CPERI. The FCC gasoline was fully
analyzed in a system of GC/MS. Ten different feedstocks were used in the unit in order to investigate the feedstock physical
properties which affect the gasoline yield and composition, the feed conversion and the coke yield as well. The gasoline
components were measured as total hydrocarbon groups: aromatics, normal and branched ole®ns, normal and isoparaf®ns and
naphthenes but special emphasis was given, in this study, for the aromatic and ole®nic content of gasoline. The main
conclusion of the work is that feed conversion, coke yield and gasoline yield and composition are strongly in¯uenced by the
type of FCC feedstock. It was shown that a paraf®nic and an aromatic FCC feedstock produce, respectively, an ole®nic or an
aromatic gasoline. The hydrotreating process plays also an important role in the gasoline composition. For these feed effects
detailed qualitative and quantitative information is given in the paper. Moreover, short form models were proposed for the
prediction of conversion coke yield and gasoline composition as a function of the main feedstock properties. Analytical forms
of these models are presented for gasoline aromatics and ole®ns and total conversion as well. The predictions of the models
were satisfactory for all hydrocarbon groups. The models were also validated with experiments using two additional
feedstocks in the pilot unit under a wide range of experimental conditions. # 1999 Elsevier Science B.V. All rights reserved.

Keywords: FCC; Reformulated gasoline; Gasoline composition; Modeling; Feedstock properties

1. Introduction The new regulations in USA for car exhausts have as

objective the reduction in the emissions of total vola-
The new regulations in USA necessitated the refor- tile components (VOCs), NOx and toxics. In this way
mulation of gasoline composition. In order to achieve the re®neries must modify the components in gasoline
this reformulation the re®neries have proceeded to which affects these emissions [1]. It is generally very
changes in the various gasoline production processes. dif®cult to investigate which of the gasoline com-
pounds (and in which degree) affects the ®nal exhaust
emissions. However, some research projects were
*Corresponding author. Tel.: +30-31-463280; fax: +30-31- carried out mainly in USA (AQUIP) which tried to
464164; e-mail: angel@aliakmon.cperi.forth.gr investigate these parameters [2]. Recently, a project

0920-5861/99/$ ± see front matter # 1999 Elsevier Science B.V. All rights reserved.
PII: S0920-5861(98)00464-7
74 A.A. Lappas et al. / Catalysis Today 50 (1999) 73±85

was carried out in Europe (EPEFE) [3] and the results 2. Experimental
were that changes in gasoline quality must include
mainly the reduction of gasoline ole®ns and aromatics 2.1. Description of FCC pilot plant
(and especially benzene) and the reduction of the RVP
and sulfur as well. All these compounds were proved Based on the results from a cold model unit an FCC
to have an effect on NOx, VOC emissions and toxics. pilot plant was constructed in CPERI. The main parts
So an environmentally friendly gasoline must have of the FCC pilot plant are given in Fig. 1. The unit
mainly low content of ole®ns, aromatics, sulfur and consists of a vertical reactor (riser) with 7.08 mm i.d.,
RVP [4]. a ¯uid bed regenerator, the stripper and the lift line.
Production of reformulated gasoline affects several The feed is injected in the bottom of the riser and it
processes in a re®nery. However, taking into account comes in contact with the catalyst ¯owing through a
that the ¯uid catalytic cracking unit (FCCU) produces slide valve. In the riser the reactions take place and at
the highest portion of the total gasoline pool (about the riser exit the mixture is entering the stripper vessel
50%), the 90% of the ®nal gasoline ole®ns and a high where the separation (stripping) of gases from the
percentage of aromatics, the focus is now on improv- solid catalyst occurs. The solids ¯ow through a second
ing the FCC gasoline yield and composition [5]. The slide valve and through the spent catalyst lift line
FCC process is a very complicated process with many (Fig. 1) return to the reactor bottom following regen-
variables which affect the product yield and the pro- eration. The reaction products, from the stripper exit,
duct selectivity. The main parameters which affect the ¯ow through a heat exchanger for the condensation of
gasoline yield and composition are: the FCC operating heavier products. Then the mixture is led to a stabilizer
parameters, the type of catalyst and the type of feed- column for better separation of liquid and gaseous
stock [6]. The re®neries have large databases which products. The mixture of gasoline, light cycle oil
try to correlate FCC product yields and properties to a (LCO) and heavy cycle oil (HCO) is obtained through
wide range of feedstocks, catalysts and operating the bottom of stabilizer. The ¯uid bed regenerator
conditions. However, these results are considered reactor is used to burn the carbon that covers the
proprietary and therefore typically remain unpub- catalyst surface as a by-product of the cracking pro-
lished. Only limited results are included in literature cess. The ¯uidization gas (air) is introduced from the
from such studies [6±8]. base of the regenerator and its ¯ow rate is controlled
The objective of the present study is to determine by a mass ¯ow controller. The regenerator exit stream
the effect of FCC feedstock parameters on gasoline passes through cyclones that remove any entrained
composition using a current generation catalyst under solids. The volumetric ¯ow rates and the compositions
a range of operating parameters. Feedstock has an of the ¯ue and cracked gases are determined by two
extremely large effect on FCC gasoline composition wet test meters and two GCs, respectively. An on-line
and octane number [5]. The work was carried out, in oxygen analyzer always monitors the excess of oxy-
co-operation with two Greek re®neries, in an FCC gen to obtain good catalyst regeneration. A more
pilot plant constructed and operated in CPERI. The detailed description of the pilot plant is presented
objective is to provide guidance for the two re®neries elsewhere [9]. The pilot plant is fully automated
to prepare them for the new gasoline regulations. In and the process control system of the unit is based
the study a large set of experiments were performed on a special industrial computer control system. The
under a wide range of operating variables using 10 control system collects the values of the input and
different feedstocks supplied by the two re®neries. drives the output signals [10].
The gasoline produced was fully analyzed in a system
of GC/MS. The work includes the effect of various 2.2. FCC pilot unit operation
feedstocks on the total hydrocarbon groups of gaso-
line: n-ole®ns, iso-ole®ns, total ole®ns and aromatics. For the current experiments 10 different types of
Moreover, an attempt was made to investigate the FCC feedstocks, provided by two Greek re®neries,
speci®c properties of VGO which in¯uence these were used. One of the feedstocks was used in the unit
hydrocarbon groups. as reference (Kuwait vacuum gas oil, code name: 372).
 A.A. Lappas et al. / Catalysis Today 50 (1999) 73±85 75

Fig. 1. Schematic view of CPERI FCC pilot plant unit.

The properties of these feeds, measured by standard feedstock 3162: a highly paraffinic feed from
ASTM procedures, are presented in Table 1. The refinery 2 (2-paraffinic)
description of the feedstocks according to its code feedstock 3172: an aromatic feed from refinery 2
name of Table 1 is: (2-aromatic)
feedstock 352: non-hydrotreated feed from refinery
feedstock 342: is a part of total FCC feedstock 1 (1-non HT)
from refinery 2, it does not contain extracts feedstock 3122: hydrotreated feed from refinery 1
(abbreviation: 2-no extracts) without resid (1-HT no resid)
feedstock 362: is the total FCC feedstock from feedstock 4112: hydrotreated feed from refinery 1
refinery 2 (2-total) with 15% resid (1-HT-resid)

Table 1
FCC feedstock properties

Code 4112 3122 352 372 3132 342 362 3142 3162 3172
Sp. gravity 0.9189 0.9056 0.9195 0.9314 0. 913 0.9026 0.893 0.920 0.868 1.003
Sulfur (wt%) 0.66 0.175 2.127 2.98 2.036 1.82 1.31 2.722 0.803 4.686
Visc., Cst 508C 22.17 8.96a 31.66 63.35 24.24 17.48 21.99 25.47 7.44a 221.0
Carb. residue 1.091 0.2 0.691 0.407 0.37 0.07 0.38 0.56 0.08 3.2
BN2 (wt%) 0.035 0.033 0.05 0.033 0.029 0.024 0.027 0.05 0.003 0.05
RI, 708C ± 1.4811 1.492 1.498 1.489 1.483 1.478 1.4955 1.4601 1.552

ASTM D1160 (8C)

MeABP 437.3 466.4 458.6 470.9 445.1 425.4 451.6 438.1 471.0 465.0
T 20% 353.2 423 400.3 453.6 385.3 370.4 396.6 379.7 435.2 428.3
T 80% 503 517 514.6 492.5 492.1 467.8 496.0 489.2 523.1 515.0
a
At 1008C
76 A.A. Lappas et al. / Catalysis Today 50 (1999) 73±85

feedstock 3132: type of feedstock from refinery 2 tigation of some physical or chemical properties of the
(2-a) feeds on catalytic cracking. The more simple approach
feedstock 3142: type of feedstock from refinery 2 for a prediction of gasoline composition (as total
(2-b) paraf®ns, naphthenes and aromatics) was based on
the lump methodology developed in Mobil [15]. The
The ranges of FCC operating variables were:
problem with the lump models is that the relative
Tˆ970±10408F, catalyst/oil ratio (C/O)ˆ4±16, weight
concentrations of the species making up individual
hourly space velocity (WHSV)ˆ40±160 hÿ1 and par-
kinetic lumps can change as the reaction proceeds and
tial pressure of hydrocarbons (PPHC)ˆ10±12 psia.
thus these models cannot be extrapolated to new
The catalyst was a commercial equilibrium catalyst
conditions or feedstocks and they are speci®c for
(E-cat) used in a Greek re®nery with the following
the feedstock, catalysts and operating conditions used.
properties: MATˆ70, UCSˆ24.26, matrix surface
For this reason, Liguras and Allen [16,17] have pre-
areaˆ44 m2/g, total surface area (TSA)ˆ158 m2/g,
sented a new class of lumped kinetic models providing
micropore volumeˆ0.05 ml/g, average bulk
a mechanism for utilizing pure compound data in an
densityˆ0.84 g/ml, average particle sizeˆ75 m,
advanced structural model which describe the entire
Al2O3ˆ39.1 wt%, RE2O3ˆ0.65 wt%, Niˆ163 ppm,
catalytic cracking reactions network. This model
Vˆ362 ppm.
represents an oil mixture by a distribution of pseudo-
The most important work in this study is the ana-
components. Recently, Joshi et al. [18] presented a
lysis of gasoline obtained from the different FCC pilot
computer assisted modeling for FCC based on large
plant experiments. However, the determination of
mechanistic kinetic models and a stochastic approach
gasoline composition is dif®cult because of the pre-
to obtain a molecular representation of FCC feed-
sence of many compounds in different amounts.
stocks. All these advanced models can indeed predict
CPERI has applied a methodology for this analysis
gasoline composition and FCC product selectivity but
based on a system of GC/GC±MS. With this method
require detailed mass spectral data, extensive com-
the gasoline can be analyzed not only as a function of
pound kinetic data base and computational power
the different components group (n-paraf®ns, isopar-
which sometimes are not available.
af®ns, normal ole®ns, branched ole®ns, diole®ns,
Considering the dif®culties mentioned above about
saturated naphthenes, unsaturated naphthenes,
the effect of each feedstock property on FCC products,
naphtheno-aromatics, aromatics) but also as a function
the changes in gasoline yield and composition are
of carbon atoms in each one of the various groups.
examined, in this work, as a function of total FCC
With the same method the benzene content can be
conversion (wt%). In this way the effect of conversion
calculated since benzene analysis is suffering from the
on the gasoline yield is presented in Figs. 2 and 3 for
presence of interfering compounds like 1-methylcy-
seven feedstocks from the two re®neries. These seven
clopentane [11].
feedstocks are selected both for the simplicity of the
®gures and since they cover some extreme cases in the
3. Results and discussion quality of all the 10 feeds of this study. Fig. 2 shows
that the total hydrotreated FCC feed with 14% resid
3.1. Effect of feedstock type on gasoline yield (feed 4112) gives the lowest gasoline yield. The
addition of 14% atmospheric resid causes a high
The effect of each one of the FCC feedstock proper- decline in gasoline yield in relation to the similar
ties on the gasoline composition is dif®cult to be feedstock but without resid (3122). This result can
studied since there are no feedstocks which vary only be easily explained by the fact that the introduction of
in one physical property and keep all other properties resid into the feed increases mainly the concentration
constant. For this reason many studies are described in of condensed aromatics. It is well known from the
literature where model compounds or arti®cially pre- mechanism of catalytic cracking [15±17,19,20,22]
pared gas oil feedstocks were used for the experiments that these compounds cannot be cracked to produce
[12±15]. Corma and Woijciechowski [13] and Corma compounds in the gasoline boiling range but they are
et al. [14] have used model compounds for the inves- mainly coke precursors. Indeed, although it is not
 A.A. Lappas et al. / Catalysis Today 50 (1999) 73±85 77

Fig. 2. Effect of various FCC feedstock from refinery 1 on gasoline yield.

Fig. 3. Effect of various FCC feedstock from refinery 2 on gasoline yield.

presented in ®gures, this type of feedstock produces The hydrogenation of VGO has also a major effect
the highest coke yield. The above effect of FCC on gasoline yield. The hydrotreated feed (3122) gives
feedstocks with resid in gasoline is in agreement with higher gasoline than the non-hydrogenated (352). The
[19,20]. delta yield in gasoline for the two feedstocks is about
78 A.A. Lappas et al. / Catalysis Today 50 (1999) 73±85

‡6 wt% and this value is in accordance with results High aromatics yields are also obtained with the non-
presented in [22,23]. It is remarkable from Table 1 hydrotreated feed (352). The aromatics molecules
that the hydrotreating process reduces not only the produce higher octane and thus the high aromatic
sulfur and nitrogen content but it mainly affects the content of gasoline from the non-hydrotreated feeds
chemical structure of the feeds as it is re¯ected from can lead to higher RON for this gasoline in relation to
the reduction in the refractive index (RI) and speci®c the hydrotreated gasoline [19,21]. The aromatic feed
gravity (SG) (Table 1). The hydrotreated feed has less (3172) gives more aromatics than the paraf®nic (3162)
aromatic carbons (lower RI) and more paraf®nic but because this feed (3172) produces low gasoline
carbons. For re®nery 2 (Fig. 3) the highly aromatic yield this moderates the production of the aromatics
feed (3172) gives the lowest gasoline yield while the (Fig. 4). The feedstocks with the higher amount of
highly paraf®nic feed (3162) gives the highest yield. aromatics make the higher amount of aromatics in the
The two feeds with and without extracts (362, 342) gasoline product. This happens because the aromatic
give about the same gasoline yield. rings with an attached substituent group undergo
reaction such that the substituent group reacts to give
3.2. Effect of feedstock type on gasoline aromatic ring which belongs to the gasoline fraction
composition due to the resultant lowering of boiling point
[15,19,20]. For all feeds the gasoline aromatics
The effect of various feedstocks on gasoline com- increase with increasing conversion (Fig. 4). This is
position is also depicted using the conversion as a result of the mechanisms of the reactions (primary
independent variable. In the following, the gasoline and secondary) which take place in catalytic cracking
hydrocarbon groups are presented in wt% on FCC feed and lead in the production of increased refractory
basis. The gasoline aromatics yields are presented for aromatics with feed conversion. Increasing the process
the seven feeds in Fig. 4. The highly paraf®nic feed severity (higher conversions) the ole®ns (and
(3162) produces the lowest aromatics yields while the naphthenes) can be converted, via hydrogen transfer
feed with the resid (4112) gives the highest yields. reactions, to aromatics. Thus, for the production of a

Fig. 4. Effect of FCC feedstocks on gasoline aromatics.

 A.A. Lappas et al. / Catalysis Today 50 (1999) 73±85 79

reformulated FCC gasoline with less aromatics the yield, while the highly paraf®nic feedstock the highest
FCC unit must operate at more moderate than for (feed 3162). The same trend is valid for the branched
conventional conversion level. This conversion level ole®ns as well (Fig. 6). This behavior is attributed to
can be found from Figs. 2±4 taking into account the the production mechanism of ole®ns. According to
available aromatics from the other gasoline compo- carbenium ion chemistry the ole®ns are the primary
nents and of course the aromatic content which will be products from the catalytic cracking of paraf®ns.
de®ned by the new regulations for the reformulated Thus, paraf®nic feedstocks produce more primary
®nal gasoline pool. Figs. 2±4 can also give to re®- ole®ns than aromatic feedstocks. Although ole®ns
neries guidelines for an optimum feedstock (or an participate in many secondary reactions (hydrogen
optimum mixing of feedstocks) in order to achieve transfer, isomerization, etc.) since paraf®nic feedstock
future aromatics in gasoline. Generally, for the same is more crackable than aromatic it requires lower
conversion the lower the aromatic content in the feed, severity (lower C/O) to obtain the same conversion
the lower the production of gasoline aromatics. The as the aromatic feedstock. This lower C/O favors also
same behavior, as for total aromatics, is observed (not the lower extent of bimolecular hydrogen transfer
presented in ®gure) for a very important (for environ- reactions. In contrary the high severity required for
mental reasons) aromatic compound, the benzene. the highly aromatic feedstock (to obtain the same
Benzene yield increases also with conversion. The conversion) favors hydrogen transfer reaction and thus
results for total aromatics and for benzene are in the decrease in ole®ns yield.
agreement with the result from [5,6,11]. Resid in feedstocks (4112) decreases the ole®ns
The effect of feedstock quality on the ole®ns pro- yield in relation to feedstock without resid (3122).
duced in the FCC gasoline is presented in Figs. 5 and 6 This happens for both types of ole®nic compounds
for normal and branched ole®ns, respectively (in a (Figs. 5 and 6) and is due to the higher gasoline
wt% on feed basis), as a function of conversion. Fig. 5 selectivity of the feedstock without resid than the
reveals that for the same conversion the highly aro- corresponding with resid. The hydrotreated feed
matic feedstock (feed 3172) gives the lowest n-ole®ns (3122) gives more ole®ns (normal and branched) than

Fig. 5. Effect of FCC feedstocks on gasoline n-olefins.

80 A.A. Lappas et al. / Catalysis Today 50 (1999) 73±85

Fig. 6. Effect of FCC feedstocks on gasoline branched olefins.

non-hydrotreated feed (352) at the same conversion. reactions [19]. Brevoord and Wilcox [19] using an
As it was mentioned above hydrotreated feeds contain Arco pilot unit concluded similar results for the
more paraf®ns and less aromatics than the non-hydro- high diole®ns yield from paraf®nic feedstock and
treated feeds. Important differences in the ole®ns yield give a possible explanation of diole®ns reactions.
do not exist for the two feedstocks of re®nery 2 (with The effect of conversion on the gasoline ole®ns
and without extracts: 362, 342, respectively). The yield is different from that of aromatics. Ole®ns are
above effect of feedstock type (hydrotreated or not) intermediate and not stable products and they decom-
on gasoline ole®ns has a similar explanation as above pose at high conversion levels (high severity).
for the paraf®nic and aromatic feeds. Generally, the Depending on the type of feedstock the ole®ns yield
ole®nicity of the gasoline is higher for more paraf®nic increases up to a certain value of conversion and after
feeds. A very important type of ole®ns is the diole®ns this value (higher conversion levels) the ole®ns
which although are in small quantities they are sig- decrease (Figs. 5±7). Although limited literature
ni®cant for environmental reasons and moreover they results exist for the effect of feedstock on various
contribute to the gasoline color and stability since they type of ole®ns the present results are in agreement
create gums (polyole®ns). The effect of feedstock type with [5,12,15,19].
on diole®ns yield is presented in Fig. 7 for some feeds.
The main conclusion of this ®gure is that there is a 3.3. Modeling of the feedstock effects on gasoline
strong effect of feed quality on diole®ns yield. Par- composition
af®nic feeds (3122, 3162) produce higher yields than
aromatic feeds (3172, 4112). It must be also pointed For the development of short form models for the
out that in contrast to other ole®ns the hydrotreated effect of FCC feedstock properties on gasoline com-
feed 3122 gives the highest diole®ns yield. Diole®ns position the following methodology was applied. For a
are referred in literature to be a product of thermal constant feed and the same catalyst the effect of FCC
cracking and sometimes the diole®ns content of the operating variables was determined. The independent
gasoline is used as monitoring tool of thermal coking variables taken into account were: C/O, WHSV, T.
 A.A. Lappas et al. / Catalysis Today 50 (1999) 73±85 81

Fig. 7. Effect of FCC feedstocks on gasoline diolefins.

Detailed results of the effect of these operating con- functions as below have been proposed:
ditions on gasoline composition will be presented in a
component i …wt%† ˆ fi …feed†  zi …oper†
future work. Following this, for the same catalyst and
for the eight different feedstocks, the effect of FCC The component i concerns the following gasoline
feedstock properties was determined. From an exten- hydrocarbon groups: n-paraf®ns, i-paraf®ns, normal
sive literature review in CPERI [6±8,19±22,24], it was ole®ns, branched ole®ns, total ole®ns, aromatics,
concluded that the independent feedstock properties naphthenes and they are referred as yields on feed-
which affect gasoline composition and which can be stock basis. In addition to gasoline composition, mod-
measured on a daily basis in any re®nery laboratory els for the total conversion and coke yield were also
are: the mean average boiling point (MeABP), the developed based on the work of Wollaston et al. [8].
wt% sulfur, the speci®c gravity (SG), the refractive The functions f and g are functions of feed quality (f)
index (RI) and the basic Nitrogen (BN2). For the and operating conditions (z). The functions of feed
parameter determination of the short form models, quality include the VGO physical properties which, as
regression analysis studies were performed and a mentioned above, are easily determined in a re®nery.
suitable software was developed based on the Leven- It must be pointed out that each one of gasoline
berg±Marquard algorithm. It is clear that this attempt components is not affected by all feedstock physical
depends on the choice of the function which will be properties. In the following the short form models for
used. In this study many different mathematical func- the conversion, aromatics and ole®ns in gasoline are
tions were applied but linear functions were ®nally discussed since they are more interesting for the
selected since they were adequate and more simple production of reformulated gasoline.
than the non-linear models. For the determination of As far as the total conversion is concerned it was
the parameters only the statistically signi®cant vari- concluded in the present study that it is a strong
ables were considered while for the best model selec- function of three feed properties: MeABP, SG and
tion the F-test were applied at 95% con®dence level. In BN2. All other properties do not in¯uence signi®-
this way for the prediction of gasoline components, cantly the total conversion (for the range of feedstock
82 A.A. Lappas et al. / Catalysis Today 50 (1999) 73±85

properties used in the present study). The following In contrast to the effect of feedstock quality on
equation was obtained: gasoline, coke yield is in¯uenced strongly by feed-
stock RI. Increasing the aromatics of feedstock the
conversion …wt%† ˆ ‰C=O0:6  WHSV0:4 coke yield increases. The aromatics are the main coke
 DEXP…ÿ17650=1:987  T†Š precursors and their condensation reaction leads to
 ‰4820:8 ‡ 9:19  MeABP ÿ 11900  SG more coke while dry gases are also produced
ÿ 0:2472  BN2 Š; [15,20,22]. Although the actual effect of FCC feed
sulfur content on conversion was found above negli-
where MeABP is in 8F, BN2 in ppmw and T in R. gible, sulfur content seems to increase the coke pro-
From the above equation it seems that by increasing duction. Probably sulfur has a deactivation mechanism
MeABP the conversion increases but increasing SG that involves blockage of the active site than a che-
the conversion decreases. The MeABP is an indication mical reaction [21]. Generally, sulfur impurities are
of the boiling range and types of molecules in the FCC associated with heavier feedstocks. It is reported in
feed [21]. The effect of increasing MeABP on the [22] that a portion (3±28%) of the sulfur compounds
conversion is related to the increased carbon number initially present in the feedstock is deposited in coke.
of the components present in FCC feedstocks. It has These compounds are the uncracked thiophenic sulfur
been now accepted that increasing the carbon chain compounds with two or three rings. The fraction of
length (carbon atoms in the hydrocarbon) the cracking sulfur ending up in coke is related to both sulfur type
activity increases. This fact has been related with the and conradson carbon in feed as well as coke yield.
statistical number of cracking possibilities in the n- Thus, always the sulfur in feed has a positive effect on
alkane molecules and to the increase in the adsorption the coke yield. The results of the present study for the
when increasing the hydrocarbon chain length effects of the feedstock properties predicted by the
[14,16,17,21]. The negative effect of SG on the total models on each of gasoline hydrocarbons and for
conversion is related to the amount of aromatics in the conversion and the coke yield are in agreement with
FCC feedstock. Aromatics with the same boiling point results from [6±8].
are more dense than paraf®ns and then the measure of The gasoline aromatics are in¯uenced by MeABP,
a higher SG (or API) indicates a more aromatic feed. RI, SG, BN2 and sulfur content as follows:
Although speci®c gravity by itself shows a weak
gasoline aromatics …wt% on feed†
correlation with conversion [7] its combination with
other properties can give satisfactory predictions of ˆ ‰ÿ7:45 ÿ 0:0435  WHSV ‡ 0:0171  TŠ
total conversion. Moreover, SG is the easiest para-  ‰11:585 ÿ 0:000557  MeABP ‡ 7:799
meter to measure on any feed. The same negative  SG ÿ 0:0829  S ÿ 11:551  RI
effect for total conversion is observed when the BN2
‡ 0:000399  BN2 Š;
increases in a feedstock. The participation of nitrogen
compounds in the actual FCC cracking process is where MeABP is in 8F, BN2 in ppmw, S in wt%, T in R
limited but the negative effect of basic nitrogen com- and RI at 708C.
pounds as reversible poisons for the FCC catalysts is From the above equation it is clear that although the
well known [7,22,25±28]. Basic nitrogen compounds MeABP has strong positive in¯uence on the total
are strongly adsorbed on catalyst acid sites and so they conversion (proportionality factor 9.18) this in¯uence
deactivate the catalyst and decrease the conversion of is relatively small for the aromatics and also the
FCC process. The degree of catalyst deactivation increasing of MeABP gives lower aromatic yield. This
depends on the degree of poisoning of the available result concludes that feedstocks with higher MeABP
catalyst acids sites which also depends on the proton are suitable for the production of reformulate gasoline
af®nity of nitrogen compounds [26,27]. The model since they give high conversion without increase in the
predicts a decrease of about 1 wt% conversion for an aromatics content of gasoline. For example feedstock
increase of about 200±300 ppmw in basic nitrogen. 362 (Figs. 3 and 4) can be a suitable FCC feedstock
This result is in agreement with literature results which can control the gasoline selectivity along with
[21,28]. the low aromatics production. Although RI seems to
 A.A. Lappas et al. / Catalysis Today 50 (1999) 73±85 83

increase the aromatics (Fig. 4) it reduces much more where MeABP is in 8F, BN2 in ppmw, S in wt% and T
the gasoline yield and so the overall increase of RI in R, Kw is the Watson factor.
reduces the aromatics yield. In contrast SG (which The total ole®ns yield in gasoline is affected by the
along with RI can determine the aromatic content of MeABP and mainly by the Kw of FCC feedstocks. Kw
FCC feedstock) in¯uences positively the production in an index of paraf®nicity and thus increasing the
of aromatics. The effect of sulfur on the aromatics is paraf®ns in the FCC feedstock more ole®ns are pro-
negative since sulfur is related to the production of duced from the primary cracking. The effect of
coke and the lost of catalyst activity. BN2 causes an increasing the MeABP is similar to that of total
increase in the aromatics yield although the basic conversion according to the explanations given above.
components have a negative effect on total conversion. The increase of SG in FCC feedstock implies an
However, basic components in¯uence the hydrogen increase in aromatic content which cannot lead to
transfer reactions which lead from the ole®ns to more ole®ns. The model predictions of the effect of
aromatics [24,25]. These results show that for less BN2 on gasoline ole®ns is opposite (and more severe)
aromatics in gasoline the BN2 content of FCC feed- with that on gasoline aromatics. The basic nitrogen
stock must be minimum. components affect the gasoline ole®ns which are
The derived equation for the total ole®ns in the intermediates in the entire reaction network and gives
gasoline (wt% on feed basis) is more paraf®ns and aromatics [26,27].
Using the short form equations developed in this
gasoline olefins …wt% on feed† study, satisfactory results were obtained for all the
ˆ ‰ÿ5:352 ÿ 0:1826  C=O ÿ 0:0506 gasoline components. The models were validated with
many experiments in the pilot unit using two feed-
 WHSV ‡ 0:0165  TŠ
stocks (3132, 3142) under a range of operating con-
 ‰0:00161  MeABP ÿ 4:1764  SG ‡ 0:0824 ditions. In Figs. 8 and 9 the predicted vs. experimental
 S ÿ 0:00038  BN2 ‡ 0:2868  Kw Š; results are quoted for two main gasoline hydrocarbon

Fig. 8. Model predictions on gasoline aromatics.

84 A.A. Lappas et al. / Catalysis Today 50 (1999) 73±85

Fig. 9. Model predictions on gasoline total olefins.

groups (aromatics and total ole®ns). The results are line are favored at high conversions. Moreover, the
satisfactory for all gasoline components: aromatics, addition of resid causes an increase in gasoline aro-
normal ole®ns, branched ole®ns, isoparaf®ns, normal matics yield while high aromatics are also produced
paraf®ns, saturated naphthenes and unsaturated with aromatic and non-hydrotreated feedstocks. Nor-
naphthenes. In general the relative standard deviations mal, branched and diole®ns in gasoline were also
of model predictions are below 12%. studied as a function of feedstock quality. It was
concluded that the aromatic feeds give the lowest
ole®ns while paraf®nic feedstock the highest. Hydro-
4. Conclusions genation of VGO increases the production of ole®ns in
the FCC gasoline. Gasoline ole®ns in contrast to
The FCC pilot plant of CPERI was used for the gasoline aromatics start to decompose at high conver-
investigation of the effect of FCC feedstock properties sion levels.
on gasoline composition, on gasoline and coke yield In addition to the qualitative analysis, an attempt
and on the feed conversion. The gasoline composition was also carried out for a quantitative prediction of
is referred to total hydrocarbon groups: normal par- gasoline composition (and coke yield and conversion
af®ns, isoparaf®ns, normal ole®ns, branched ole®ns, as well) as a function of feedstock properties. Taking
diole®ns, aromatics and naphthenes. In this work into account the FCC feedstock properties which are
special emphasis was given for the aromatics and easily measured in a re®nery, short form models were
the ole®ns in gasoline since these compounds are developed for the gasoline composition. It was proved
interesting for environmental reasons. It was con- that the speci®c gravity, the refractive index, the basic
cluded that feedstock quality affects strongly the nitrogen, the mean average boiling point and the
gasoline yield and composition. Thus, resid in Watson factor are the parameters which in¯uence
VGO, aromatic feedstocks and non-hydrotreated the gasoline composition. Each one of these para-
FCC feedstock produce lower gasoline than paraf®nic meters has a different effect on the gasoline com-
or hydrotreated FCC feed. The aromatics in the gaso- pounds. For example feed conversion is affected by
 A.A. Lappas et al. / Catalysis Today 50 (1999) 73±85 85

MeABP, SG and BN2. The functions proposed were [11] C.A. Yatsu, T.A. Reid, D.A. Keyworth, R. Jonker, M.A.
®nally validated with experiments using two addi- Torem, Proceedings of 206 ACS Meeting, Chicago, 22±27
August 1993.
tional feedstocks in the pilot unit under a range of [12] R.H. Harding, X. Zhao, K. Qian, K. Rajagopalan, W. Cheng,
experimental conditions. The predictions of the model Ind. Eng. Chem. Res. 35 (1996) 2561.
were satisfactory for all gasoline hydrocarbon groups. [13] A. Corma, B.W. Woijciechowski, Catal. Rev.-Sci. Eng. 24
(1982) 1.
[14] A. Corma, P.J. Miguel, A. Orchilles, Appl. Catal. A 117
Acknowledgements (1994) 29.
[15] S.M. Jacob, B. Gross, S.E. Voltz, V.W. Weekman, AIChE J.
22 (1976) 701.
The General Secretariat of Research and Technol-
[16] D.K. Liguras, D.T. Allen, Ind. Eng. Chem. Res. 28 (1989)
ogy Hellas is gratefully acknowledged for providing 665.
®nancial support. The Greek re®neries HAR and [17] D.K. Liguras, D.T. Allen, Ind. Eng. Chem. Res. 28 (1989)
MOH are also acknowledged for the supply of feed- 674.
stocks. [18] P.V. Joshi, S.D. Iyer, M.T. Klein, Proceedings of the
Symposium on Catalysis in Fuel Processing and Environ-
mental Protection, 214 ACS Meeting, Las Vegas, NV, 7±11
September 1997.
References [19] E. Brevoord, J.R. Wilcox, Proceedings of ACS Meeting,
1995.
[1] R.F. Sawyer, Environ. Health Perspectives Suppl. 101 (1993) [20] M.A. Torem, M.F. Cunha, V. Aleksitch, Proceedings of 206
5. ACS Meeting, Chicago, IL, 22±27 August 1993.
[2] V.R. Burns, J.D. Benson, A.M. Hochhauser, W.J. Koehk, [21] W.S. Letzsch, A.G. Ashton, in: J.S. Magee, M.M. Mitchell
W.M. Kreucher, R.M. Reuter, SAE Paper 912 320. (Eds.), Studies in Surface Science and Catalysis, FCC Science
[3] AUTO/OIL Programme, European Commission, CONCAWE and Technology, Elsevier, Amsterdam, 1993.
REVIEW, vol. 5, October 1996. [22] P.B. Venuto, E.T. Habib, Fluid Catalytic Cracking with
[4] M. Absi-Halabi, A. Stanislaus, H. Qabazard, Hydrocarbon Zeolite Catalysts, Marcel Dekker, New York, 1979.
Process. 2 (1997) 45. [23] W. Ginzel, Proceedings of the Sixth Katalistiks Annual Fluid
[5] E.T. Habib, Symposium on the Hydrocarbon Chemistry of Cat Cracking Symposium, Munich, 1986.
FCC Naphtha Formation, ACS Division of Petroleum [24] H.E. Reif, R.F. Kress, J.S. Smith, Petroleum Refiner 40
Chemistry, Miami, 1989. (1961) 237.
[6] M. Schlossman, W.A. Parker, L.C. Yen, CHEMTECH 2 [25] S.E. Voltz, D.M. Nace, S.M. Jacob, V.W. Weekman, Ind. Eng.
(1995) 41. Chem. Process Des. Dev. 11 (1972) 261.
[7] K.R. Kreider, D.A. Keyworth, T.A. Reid, Proceedings of [26] J.B. Green, E.J. Zagula, L.L. Young, Proceedings of the
AIChE Meeting, Chicago, IL, 11±16 November 1990. Symposium on Petroleum Chemistry and Processing, 210
[8] E.G. Wollaston, W.J. Haflin, W.D. Ford, Hydrocarbon ACS Meeting, Chicago, IL, 20±25 August 1995.
Process. 54 (1975) 19. [27] J.B. Green, T.B. McWilliams, G.P. Sturm, Proceedings of the
[9] I.A. Vasalos, A.A. Lappas, D.K. Iatridis, S.S. Voutetakis, Symposium on Petroleum Chemistry and Processing, 210
Proceedings of Fluidization V, Beijing, China, May 1996. ACS Meeting, Chicago, IL, 20±25 August 1995.
[10] S.S. Voutetakis, A.A. Lappas, D.K. Iatridis, I.A. Vasalos, [28] J. Scherzer, D.P. McArthur, Ind. Eng. Chem. Res. 27 (1988)
Comp. Chem. Eng. 20 (1996). 1571.