STUDIES IN COKING OF ARABIAN MIX

VACUUM RESIDUE

A. N. Sawarkar, A. B. Pandit and J. B. Joshi

Institute of Chemical Technology, University of Mumbai, Matunga, Mumbai, India.

Abstract: Batch reaction experiments pertaining to the coking of Arabian mix vacuum residue
(AMVR) were carried out in the temperature range of 430–4758C and at a pressure of
0.2 MPa in an autoclave bomb reactor. The reaction time was varied in the range of 5–90 min.
An attempt has been made to understand the inception, growth and saturation of coke formation
within the range of temperature and reaction time covered in this work. The study revealed that
the coke formation starts near the wall and then propagates towards the centre of the reactor and
the temperature difference between the reaction mass at the wall and at the centre of the reactor
was found to be as low as 0– 18C and as high as 8–128C depending upon the severity of the
reaction and the rate of coke formation. The present investigation also revealed that the coke
formation levels off after a certain reaction time and thereafter a slight fall in the coke yield
occurs. The reaction mixture was also quantified in terms of gas, distillates and unconverted
vacuum residue. A four lump, three rate parameter kinetic model comprising vacuum residue
(VR), gas (G), distillates (D) and the coke (C) has been proposed. VR was found to undergo
first order decomposition and the apparent activation energy for the reaction pathways involving
the formation of gas, distillates and coke from VR was found to be 51, 39 and 63 kcal mol21,
respectively.
Keywords: coking; AMVR; autoclave bomb reactor; coke; residue upgradation; kinetics.

INTRODUCTION with delayed coker is said to be ‘zero resid

refinery’ which clearly brings out the import-
Petroleum residue upgradation is at the zenith ance of delayed coking process in the refinery
of value addition for obtaining the distillates. set-up. Also, the sheer ability to process vir-
The high demand for distillates and ever- tually any refinery stream has made delayed
rising heavy nature of the crude oil has coking a prime choice for residue upgrada-
resulted into a renewed interest in the tion. Furthermore, the growing trend of pro-


Correspondence to: bottom-of-the-barrel processing using various cessing heavy crude oils has brought about
Professor J.B. Joshi, Institute conversion processes. Among the various a renewed interest in delayed coking (Elliott,
of Chemical Technology, processes employed, thermal processes 1992; Bansal et al., 1994) and the worldwide
University of Mumbai,
viz., visbreaking and delayed coking share a coking capacity has reached to about 210
Matunga, Mumbai-400 019,
India. major portion (about 63%) of the petroleum MMTPA (Swaty, 2005).
E-mail: [email protected] residue upgradation (Shen et al., 1998). In The upgradation of residual feedstocks
visbreaking process, reactions are arrested obtained by processing heavy crudes leads
DOI: 10.1205/cherd06159 so that asphaltenes flocculation does not to higher coke yield because of the higher het-
0263–8762/07/
take place and subsequently a stable fuel oil eroatom, asphaltenes and Conradson carbon
$30.00 þ 0.00 (visbroken residue) is obtained. This results residue (CCR) content in the parent heavy
in low yield of 5008C-material (about 37%) crudes (Trasobares et al., 1999). Coke is
Chemical Engineering and the remaining material accounts for fuel defined as a carbonaceous material that is
Research and Design
oil (Zuba, 1998). On the other hand, in insoluble in an aromatic solvent such as
Trans IChemE, delayed coking process, reactions are given benzene and toluene. Coke formation during
Part A, April 2007 sufficient time at relatively high temperatures the thermal cracking of residual feedstocks
to proceed to completion resulting in high indeed is a scientifically intriguing phenom-
# 2007 Institution yield of 5008C-material (about 79%) (Zuba, enon. Over the past five decades many
of Chemical Engineers
1998) along with coke (about 21%). Thus, in researchers worldwide have put in their qual-
delayed coking process, the aim is an econ- ity time and efforts to know the intricacies per-
omical conversion of residual feedstocks, taining to coke formation during the cracking
especially vacuum residues (short residues) of residual feedstocks.
to lighter, more value added products while Magaril and Aksenova (1968) have
producing a coke material of desired quality observed that the coke formation begins
(Rahmani et al., 2002). Hence, a refinery only after accumulation of considerable

481 Vol 85 (A4) 481–491

482 SAWARKAR et al.

amount of asphaltenes. It was also observed that the time for (8) The changes in the chemical structure of resins and
the inception of coke formation coincides with the time of asphaltenes occurring before and after thermal conver-
maximum yield of asphaltenes (Magaril et al., 1971). Magaril sion of the Shengli vacuum residue have been investi-
and Aksenova (1968) have reported that the coke formation gated by Wang et al. (1998). It was found that during
brings about the formation of a new solid phase and always thermal conversion of vacuum residue, the resins bear-
takes place near the wall. Savage et al. (1985) have ing shorter alkyl chains and more peri-condensed aro-
observed an induction period for the thermolysis of asphalt- matic units are responsible for asphaltenes formation
enes from an off-shore California crude at 4008C that disap- while asphaltenes bearing shorter alkyl chains and
peared when the thermolysis temperature was raised to more peri-condensed aromatic units get converted into
4508C. Coke formation during heavy oil upgrading has coke.
been elucidated by Wiehe (1994) based on the pendant- (9) The effect of solvent properties (with similar solubility
core model. According to this model, the core, which consists parameter such as maltene, 1-methyl naphthalene and
of large aromatics, precipitates when the solubility limit of aro- tetralin) on solubility limit and coking kinetics has also
matics in the medium is reached. Further reactions of the aro- been explored (Rahmani et al., 2002). It was observed
matic molecules lead to a more organized and stacked that the hydrogen donating ability of the solvent and
arrangement of the aromatic sheets, which eventually form the hydrogen accepting ability of the asphaltenes play
a separate liquid phase known as mesophase. This liquid a major role in determining the ultimate yield of the coke.
phase is eventually converted to solid material termed as (10) The effect of structural properties of asphaltenes on the
coke. Similarly, because of the nature of the chemical coking rates and coke yields have also been explored
structure of asphaltene molecules present in heavy oils and by studying the thermal cracking of asphaltenes
bitumen, and their solubility characteristics, asphaltene obtained from different origins. It was found that the
molecules can form coke rapidly during thermal treatment coking rate depends on aliphatic sulphide content of
(Wiehe, 1993). the asphaltene while its aromaticity decides the yield
A systematic analysis of the previous work in this area of the coke (Wiehe, 1993).
brings out the following points: (11) The above coking kinetic models have been reported to
follow first order kinetics with the range of pseudo
(1) The residue comprises saturates, aromatics, resins and activation energies (22–83 kcal mol21) depending
asphaltenes. The coke formation is the consequence of upon the feed properties and severity range.
a series of complex reactions and proceeds from saturates
to aromatics to resins to asphaltenes and finally to coke. A brief summary comprising the reaction conditions, differ-
(2) During thermal cracking, asphaltenes become more ent feeds studied and the findings of the investigators is given
aromatic and at a particular stage of conversion they in Table 1. From the foregoing discussion, it can be found that
undergo phase separation by breaking of colloidal considerable work has been done on coking kinetics with
equilibrium of the whole residue. special emphasis on the mechanism of coke formation,
(3) In the separated phase, the asphaltenes lack in inter-conversion of the solubility class components during
H-abstraction which otherwise would have undergone conversion, role of these components in coke formation, influ-
radical termination by H-abstraction from maltene ence of structural properties on coking rate and yields and so
components. Thus, they undergo condensation and on. The proposed models are based on the mechanistic
polymerization reactions resulting into the coke formation. pseudo components, phase separation, pendant core, and
(4) The model showing the interconversion along with paral- so on, which explains the coke formation during thermal
lel formation of pseudo boiling cuts has been proposed conversion.
by Takatsuka et al. (1989) and has been experimentally However, it may be pointed out that the information with
validated. regard to the actual coke build up (inception, growth and
(5) With the help of the structural changes at the molecular saturation of coke formation) in the reactor (batch mode or
level and using solvent-resid phase diagram, it was continuous mode) has not been reported in the literature.
found that the shift from one class to another such as Qualitative information can be found (Magaril and Aksenova,
maltenes (heptane soluble) to asphaltenes (heptane 1968) with regard to the coke formation. The authors have
insoluble, toluene soluble) to coke (toluene insoluble) reported that coke formation first takes place at the wall.
occur via continuous change in the molecular weight However, there is no quantification reported as to how
and Conradson carbon residue (CCR) within the same much temperature gradient exists between the wall and the
class (Wiehe, 1992). This has been observed and centre of the reactor at different processing severities.
corroborated by Yasar et al. (2001). Hence, it was thought desirable to undertake a systematic
(6) The concept and existence of certain threshold investigation of the coking behaviour of Arabian mix
concentration of asphaltenes as solubility limit (SL) was vacuum residue (AMVR) in terms of inception, growth and
incorporated in the coke formation model (Wiehe, subsequent saturation of coke formation at different proces-
1993). Later, using optical microscopy the onset of sing severity.
neophase separation during thermal cracking was also With regard to the kinetic modeling, it was found that most
experimentally proved (Li et al., 1998). of the models proposed are based on structural changes
(7) Song et al. (1995) have studied the kinetics of coking within the residue and a very few models have been pro-
of Gudao vacuum residue in the temperature range posed which comprise the lumps of industrial relevance
of 400 –4408C and 460 –5008C. The thermal cracking and involve coke as one of the components. After the close
reactions were found to follow first order kinetics over scrutiny of the available literature, it was found that the
the studied temperature range. model, which comes close on this account, is proposed by

Trans IChemE, Part A, Chemical Engineering Research and Design, 2007, 85(A4): 481– 491
 Table 1. Reported work on the coking kinetics.

Authors Experimental details Proposed kinetic model Conclusions

Magaril (1968) Feed Resins – (1) There exists a certain saturation concentration of asphaltenes beyond which
Reactor Quartz test tube they precipitate out from saturate solution.
Temp. (8C) 400 (2) Coke formation is the consequence of precipitation and accumulation of
Reaction 160 asphaltenes. The coking rate is equal to the rate of asphaltenes precipitated.
time (min)
Magaril et al. Feed Asphaltenes isolated from – (1) The effect of solvent properties on the kinetics of
(1971) cracked residue and coke formation was studied.
deasphalted bitumen (2) Assuming the spherical shapes of asphaltene and solvent molecules and
Solvent Transformer oil, using their molecular wt and density, the asphaltene concentration in the
anthracene oil. solution was proposed at various conversions.
Reactor Autoclave without stirring (3) Coking was found to initiate after certain threshold concentration of
Temp. (8C) 380– 410 asphaltene in the solution and this value was found to be higher for solvent
Reaction 5 –120 with good solvating power.
time (min) (4) Coke formation was found to occur after a certain severity of cracking has
reached.
Takatsuka Feed Residual oil (1) Atmospheric equivalent temperature and hydrocarbon (HC) partial pressure
et al. (1989) Reactor Flow reactor, were used to estimate the Arrhenius rate parameters.
semibatch, (2) Higher HC partial pressure decreases the reaction rate of polycondensation
autoclave reactions while cracking rate increases with increasing the reactor pressure.
Temp. (8C) 400– 450 (3) The activation energy for the cracking and polycon
Reaction time (min) 0 –450 densation reactions were estimated to be 60 kcal mol21 and 40–
Pressure (MPa) 0.013–0.45 50 kcal mol21, respectively.

Wiehe (1992) Feed VR and its SARA fractions (1) The solvent-resid phase diagram (plot of molecular weight– hydrogen
Reactor Batch tubing bomb content) was proposed to distinguish one pseudocomponent from another
Temp. (8C) 400 and to track the chemical changes that result their movement from one
Reaction time (min) 60 solubility class to another.
Pressure (MPa) 7 MPa (N2 atm) (2) Elemental analysis and molecular wt. of thermal conversion products of resid
and SARA fraction was studied.

Trans IChemE, Part A, Chemical Engineering Research and Design, 2007, 85(A4): 481–491
(3) Thermal cracking of the SARA fractions resulted to reduction in their
molecular weight, H-content (slightly in the case of saturate and aromatics)
and coke formation.
STUDIES IN COKING OF ARABIAN MIX VACUUM RESIDUE

Wiehe (1993) Feed Cold Lake VR (3 g) – (1) Inhibition of heptane solubles for coke formation, a maxima for the
Reactor Quartz tube asphaltenes formation which matches with the coke induction period, parallel
reactor decrease in the asphaltenes concentration beyond certain heptane solubles
Temp. (8C) 400 (SL, solubility limit).
Reaction time (min) 0 –180 (2) A kinetic model was proposed that explains the above observations and
Pressure (MPa) Open reactor with estimated the disappearance of asphaltenes by first order with rate constant
continuous of 0.026 min21 while SL, was estimated to be 0.49 wt%.
N2 flow

(continued )
483
 Table 1. Continued 484
Authors Experimental details Proposed kinetic model Conclusions
Del Bianco Feed Belaym VR (1) The coke formation was proposed to be formed via a reaction intermediate (I)
et al. (1993) (CCR-20.8 wt%) (2) E1-49.4 kcal mol21; A1-31.97 min-1; E2-63.9 kcal mol21; A2-40.92 min21.
Reactor Batch reactor (3) Structural study shows that thermal cracking of asphaltenes follow
Temp. (8C) 410–470 dealkylation reactions.
Reaction time (min) 0– 120 (4) Condensation reactions prevail at higher severity levels.
Pressure 1 (N2 atm)
(MPa)

Song Feed Gudao VR (5 g) –

et al. (1995) Reactor Batch tubular reactor (1) Assuming the first order kinetics, the Arrhenius parameters were found to
Temp. (8C) 400–500 vary from E400-440 – 170.7 kJ mol21, A400-440 –7.853 e10 to E460-500 –
Reaction time (min) 5 –80 180 kJ mol21, A460-500 – 9.24 e11.
Wang Feed Shengli VR (500þ) – (1) Structural parameters were compared between SARA fractions obtained from
et al. (1998) Reactor Autoclave with the feed and after its thermal cracking.
magnetic stirrer (2) Peri-condensed aromatic resins get converted into asphaltenes, which further
Temp. (8C) 410 condense to give coke formation.
Reaction time (min) 60
Pressure (MPa) 0.1
Li et al. Feed Shengli, Daqing, – (1) The variation in the onset of neophase formation, necessary for the coke
(1998) Gudao (300 g) formation, was studied using optical microscope for three different VR varying
Reactor FYX-05A autoclave largely in their properties.
Temp. (8C) 370–390 (2) A stability function, based on the SARA composition, was determined to
Reaction time (min) 10–70 represent the thermodynamic stability of the VR.
Pressure (MPa) 1.0
SAWARKAR et al.

kH
Rahmani Feed Athabasca asphaltenes H þ !k bAþ þ (1  b)V (1) The solubility limit kinetic model was proposed to study thermal cracking of
A
et al. (nC7 insol. 3 g) Aþ ! cA
þ (1  c)(H
þ V) asphaltenes in maltene, 1-methyl naphthalene and tetralin.
(2002) Reactor Batch microreactor where, H þ ¼ fraction of reac. nC7 sol. (2) Coke formation was found to be strongly influenced by the chemical
(15 mL) H
¼ fraction of prod. nC7 sol. interactions between the asphaltenes and the solvent medium.
Temp. (8C) 430 A þ ¼ fract. of reac. asphaltene (3) The hydrogen donating ability of the solvent and the hydrogen accepting
Reaction time (min) 60 A
¼ fract. of prod. asphaltene ability of the asphaltenes were found to have a pronounced effect on the coke
Pressure (MPa) 0– 4 MPa V ¼ cracked distillate prod yield.
kA, kH ¼ reac. rate const. for the (4) The model proposed was found to be consistent with the data on coke
thermolysis of reac. asphaltene and formation from asphaltenes in studied solvents with low hydrogen donating
nC7 sol., respectively b, ability.
c ¼ stoichiometric coefficients
Rahmani Feed Asphaltenes from AL, – (1) Phase separation model was proposed to study the coking kinetics of
et al. (2003) AH, Maya, GudaoVR asphaltenes having range of structural properties.
Reactor Batch microreactor (2) Structural dependency on coking kinetics showed that sulphide content of
(15 mL) asphaltenes correlated with the cracking rate while the aromaticity decided
Temp. (8C) 430 the coke yield.
Reaction time (min) 60
Pressure (MPa) 9.8 MPa
Schabron Feed Boscon, Llyodminster, – (1) Coke formation involves a complex set of reactions which fall somewhere
et al. (2003) Redwater BC, between zero-order and first-order kinetic mechanisms.
MaxCL2 (5 g) (2) Kinetics followed by coke formation is most likely dependent on the source of
Reactor Tubular reactor the residue.

Trans IChemE, Part A, Chemical Engineering Research and Design, 2007, 85(A4): 481– 491
Temp. (8C) 400–450 (3) The activation energies of the secondary coke formation reactions of the
Reaction time (min) 5– 150 studied feeds were found to be 2.2 to 2.6 times higher than that for the initial
coke formation.
 STUDIES IN COKING OF ARABIAN MIX VACUUM RESIDUE 485

Del Bianco et al. (1993). The model proposed by the authors Experimental Set-Up and Procedure
comprises vacuum residue, distillate fraction, reaction inter-
Figure 1 shows the schematic representation of the exper-
mediate and the coke. A good account of the variation in
imental set-up employed for investigating the coking beha-
the rate parameters has been given in the proposed kinetic
viour of AMVR and subsequently generating the batch
scheme. However, the proposed kinetic model does not
kinetic data of residue cracking under coking conditions.
include gas fraction separately. Gas (C1 –C4) forms one of
The reactor used was SS-316 autoclave (80 mm i.d. and
the major products of delayed coking process with the yield
94 mm o.d.) having 600 mL capacity with a provision of sim-
up to 9.5 wt% (Zuba, 1998) depending upon the processing
ultaneously measuring the temperature of the reaction mass
severity. Hence, it was thought desirable to include this frac-
near the wall and at the centre of the reactor and the vapour
tion in the proposed kinetic scheme so that useful information
temperature. Molten tin bath was used as the heating
can be obtained with regard to the rate variations for the
medium. A cold trap was used for the separation of the lighter
paths involving gas fraction, along with rate variations for
products from the gaseous components formed during crack-
the paths involving distillates, coke and unconverted
ing. A gas meter (wet type) was used to quantify the gas
vacuum residue (VR). Thus, in the proposed kinetic
formed during the course of the reaction. The molten tin
scheme, all the major products of the delayed coking process
bath was heated electrically and the temperature was
have been incorporated and an attempt has been made to
controlled within +28C using a PID controller.
investigate the variation in the rate parameters for the poss-
About 300 g of the feed was charged to the reactor and it
ible paths with the help of the generated experimental data.
was initially flushed with nitrogen and then pressurized to
In the present investigation, the reactor was provided with
0.2 MPa. The pressure was maintained throughout the reac-
an additional thermocouple at the wall of the reactor along
tion period using needle valve. The molten tin bath was kept
with the conventional thermocouple at the centre of the
approximately 50–608C higher than the desired reaction
reactor. In doing so, the temperature –time profiles at the
temperature so as to maintain the temperature gradient and
wall and at the centre were obtained and hence some
provide the heat required for the various thermal cracking
useful information was obtained with regard to the tempera-
reactions. The reactor was placed just above the molten tin
ture difference at these two locations in the reactor over the
bath for preheating up to 3508C. This preheating reduced
studied temperatures and reaction times. AMVR is a widely
the time period required to attain the experimental tempera-
produced residue in petroleum refineries and a feed of
ture. As soon as the reaction mass attained the preheating
commercial interest (Schucker, 1983). Hence, AMVR was
temperature, the reactor was dipped in the bath with constant
employed in this investigation for the coking studies.
manual rocking of the reactor for temperature uniformity. The
desired temperature was achieved within 1.5 –2 min. Once
the desired reaction temperature was attained, the stop-
EXPERIMENTAL SECTION watch was reset at zero time i.e., t ¼ 0. Thus, the reaction
time reported in this work is the time period after the reaction
Arabian mix vacuum residue was obtained from the oper-
mixture attains the desired reaction temperature. Since the
ational vacuum distillation unit of HPCL refinery located in
coking process starts near the wall, the wall temperature
Mumbai. The feed was characterized as per the standard
has been reported as the temperature of reaction. The temp-
ASTM/IP methods. Table 2 summarizes the important
eratures of the reaction mass and bath were measured at the
physico-chemical properties of the feedstock studied.
interval of every 30 s. The cracked lighters and gas fractions
were allowed to escape the reactor using needle valve

Table 2. Physico-chemical characteristicsa of Arabian mix vacuum

residue (AMVR).

Feedstock
Sr No Property AMVR

1 Density, gm cc21 1.023

2 Kinematic viscosity, cSt
at 1008C 1084.10
at 1358C 168.49
3 S, wt% 4.43
4 CCR, wt% 21.10
5 C/H ratio 6.59
6 Mol. Wt. 800.2
7 nC5 insol. wt% 18.72
8 Pour point, 8C þ42
9 Metal content, ppm
V 89.00
Ni 31.65 Figure 1. Schematic diagram for experimental set-up: (1) SS-316
Fe 32.00 reactor (600 mL capacity); (2) molten tin bath; (3) electric heaters
10 H-C type analysis, wt% for furnace; (4) furnace temperature control; (5) thermocouple for
Saturates 15.00 the tin bath; (6) thermocouple for the reaction mass near the wall;
Aromatics 66.28 (7) thermocouple for the reaction mass at the centre; (8) thermo-
Resins 7.6 couple for the vapours of the reaction mass; (9) pressure indicator;
Asphaltenes 11.12 (10) pressure release needle valve; (11) vertical moveable stand;
(12) cold trap for liquid product; (13) liquid product collector; (14)
a
All the tests were performed as per the ASTM/IP methods. gas meter; (15) gas outlet.

Trans IChemE, Part A, Chemical Engineering Research and Design, 2007, 85(A4): 481–491
486 SAWARKAR et al.

RESULTS AND DISCUSSION

The coking behaviour of Arabian mix vacuum residue
(AMVR) has been investigated over the temperature range
of 430 –4758C. Figure 3 shows the temperature profiles of
the reaction mass near the wall and at the centre of the reac-
tor at different severities which represent no coke formation
[Figure 3(a)], inception of coke formation [Figure (3b)],
growth of coke [Figure (3c)] and the final limiting coke for-
mation [Figure (3d)] conditions. TW represents the tempera-
ture of the hydrocarbon mixture at the wall of the reactor
and TC represents the temperature of the hydrocarbon
mixture at the centre of the reactor.
Figure 3(a) shows the temperature profile during the mild
reaction condition, i.e., 4308C and for 5 min reaction time.
From Figure 3(a), it can be observed that there is no
change in the temperatures of the reaction mass at the wall
(Tw) and at the centre (TC) as the coke formation does not
Figure 2. Schematic representation of experimental procedure and
analysis. take place at that reaction conditions (time and temperature)
which was also corroborated by performing the standard
merit number test (IFP-3024-82, severity test for thermal
reactions by spot method and residue stability determination)
maintaining the desired pressure. The lighter fractions were
wherein the 5008Cþ was found to be stable. Figure 3(b)
separated using cold-trap kept at 0 –38C while the volume
shows the temperature profile at a temperature of 4458C
of the gas fraction formed after passing through the cold-
and for 10 min reaction time. From Figure 3(b), it can be
trap was measured using a gas meter. The gas sample
observed that there is no change in the temperatures of the
was collected for its compositional analysis. The desired
reaction mass at the wall and at the centre for the entire reac-
reaction temperature was maintained throughout the run by
tion time of 10 min except for last couple of minutes or so
adjusting the position (lowering or raising) of the reactor in
wherein a difference of about 2 –38C was found between
the tin bath. After the desired reaction time, the reactor was
the wall and centre temperatures, which indicates the
removed from the tin bath and was immediately quenched
inception of coke. The inception of coke was corroborated
using water bath (258C). The quenching period was
by performing the standard merit number test, wherein the
approximately 1 min. The reactor was then slowly depressur-
5008Cþ material was found to be unstable. The coke content
ized to atmospheric pressure and the entrained vapours and
was determined as toluene insoluble fraction using ASTM
gaseous products were collected and quantified separately.
D-4072 from the remaining 5008Cþ material and from the
The volume of nitrogen was subtracted from the measured
hard material (in the case of high severity runs). Figure 3(c)
cumulative volume of the gaseous products so as to obtain
shows the temperature profile at a temperature of 4608C
the volume of the formed gaseous products during the
and for 15 min reaction time. From Figure 3(c), a tempera-
reaction. The leftover (visbroken tar/hard material depending
ture difference of about 4 –68C could be seen after a reac-
upon the severity of the reaction) in the preweighed
tion time of 12 min. An incremental temperature difference
reactor was collected and weighed in a flask for further
at the wall and at the centre of the reactor at these con-
analysis. The reactor was again weighed so as to account
ditions indicates the rise of coke forming tendencies.
for the uncollected material. The material balance was
Figure 3(d) shows the temperature profile at a temperature
accounted in terms of gaseous products, liquid product
of 4758C and for 20 min reaction time. A further increase
and visbroken tar/hard material and was always within
in the temperature difference (8 –128C) at the wall and at
98–99%.
the centre of the reactor can be attributed to the pronounced
effect of polymerization and polycondensation reactions at
such a high temperature. Table 3 shows the yield of coke
Separation Scheme and Analysis when the feed was thermally cracked at different tempera-
tures and reaction times.
The reaction was based on the conversion of the VR
(5008Cþ) to the components boiling below 5008C and coke
(defined as toluene insoluble material). Figure 2 shows the
Product Distribution
complete reaction scheme and analysis of the reaction
products. The feed was thermally cracked for the desired The product lumping procedure has been adopted by
reaction temperature and residence time. The gas was ana- many researchers to describe the kinetic behaviour of com-
lysed by refinery gas analyser (RGA) for its compositional ponents undergoing various types of reactions that include
analysis. In the case of low severity experiments, the tar/ cracking, polymerization, condensation, H-abstraction, and
pitch material was distilled so as to separate the distillate so on. Though such lumping approach has certain pitfalls
fraction (5008C2). The coke content was determined as (Weekman, 1979), it explains the kinetic behaviour to a satis-
toluene insoluble fraction using ASTM D-4072 from the factory level and can be readily used by the refiners. In the
remaining 5008Cþ material. Similarly, the coke content was present case, a four lump parallel reaction model, consisting
determined from the hard material (obtained in case of high of VR, G, D and C has been proposed to describe the kinetic
severity runs). behavior of the AMVR (Figure 4).

Trans IChemE, Part A, Chemical Engineering Research and Design, 2007, 85(A4): 481– 491
 STUDIES IN COKING OF ARABIAN MIX VACUUM RESIDUE 487

Figure 3. Temperature–time profiles at different severities: (a) 4308C, 5 min; (b) 4458C, 10 min; (c) 4608C, 15 min; (d) 4758C, 20 min (— Tw; — Tc).

Table 3. Yield of coke at different severities.

Temperature (8C)
Product (wt%) Reaction time (s) 430 445 460 475

300 0.00 0.20 0.66 13.89

600 0.00 1.09 12.89 21.16
900 0.85 6.54 16.74 27.80
Coke 1200 1.19 13.65 21.09 30.39
1800 3.40 19.22 28.22 30.83
3600 12.41 27.99 30.42 30.13
5400 24.14 29.83 29.74 29.83

The maltene phase has certain solubility limit beyond

which the asphaltenes precipitate out and exist as a separ-
Figure 4. Proposed four lump kinetic model with three rate par-
ate phase. In this phase, the high rate of condensation of ameters: VR—vacuum residue (5008Cþ); G—gas (C1 –C4); D—dis-
the asphaltene radicals which are thus lean by H-abstrac- tillates (IBP-5008C); C—coke.
tion (from the maltene phase) leads to the toluene insoluble
material (coke). The above theory of coke formation has
been used by different authors (Wiehe, 1993; Rahmani
et al., 2002, 2003) to explain the kinetic behaviour of differ- 4758C, markedly high yield of coke (about 14 wt%) was
ent feeds for the range of coking operation severities. From obtained even at a reaction time of 5 min because of the
Figure 5, it can be observed that the coke formation does high rate of polymerization and polycondensation reactions
not start at 4308C unless a reaction time of 15 min is prevailing at this temperature. In fact, the coke formation
attained. At 4458C, the inception of coke formation was was found to level off as early as at 20 min of reaction
found to start at 10 min. At 4608C, coke formation was time, thus indicating that coke formation rapidly proceeds
found to be high (about 13 wt%) even for the reaction to completion at higher temperature. This observation is in
time of 10 min. It was also observed that after a reaction agreement with the results of Schabron et al. (2003). The
time of 60 min at 4608C, coke formation levels off. At second and important observation was that, once the coke

Trans IChemE, Part A, Chemical Engineering Research and Design, 2007, 85(A4): 481–491
488 SAWARKAR et al.

Figure 5. Comparison between experimental and predicted values of Figure 6. Comparison between experimental and predicted values of
coke at different temperatures: S, 4308C; A, 4458C; 4, 4608C; , gas at different temperatures: S, 4308C; A, 4458C; 4, 4608C; ,
4758C (—predicted curves). 4758C (—predicted curves).

has reached saturation and if the reaction time is still raised, Gas concentration
a slight fall in the coke yield was found to occur at 4608C, The analysis of gas by refinery gas analyser (RGA)
90 min and at 4758C, 60 and 90 min. The marginal showed that methane is the major component (about
reduction in the coke formation, over extended time can 42 wt%) followed by ethane (about 22 wt%) (Table 4). Fur-
be attributed to the enhanced devolatilization of the volatiles thermore, processing severity was found to have negligible
trapped in the coke matrix at higher temperatures. This effect on overall gas composition. The gas formation was
observation is in agreement with Gray et al. (2004). Maxi- found to increase with an increase in the processing tempera-
mum coke yield (30.83 wt%) was found to be at 4758C ture. Maximum gas yield was found to be 15.23 wt% at 4758C
and for reaction time of 30 min. The coke near the wall and for 90 min of reaction time (Figure 6).
was found to be hard and had to be removed with the aid
of a chisel and hammer thus corroborating the notion that
polymerization reaction starts progressing from the wall to Distillate concentration
the centre of the reactor. Distillate yield was found to increase with increasing pro-
Along with coke, the reaction mixture was also quantified in cessing severity. At 460 and 4758C, the distillate yield was
terms of gas (G), distillates (D) and unconverted vacuum resi- found to level off and showed asymptotic behaviour
due (VR) and the variation of each of these fractions with time
within the studied temperature range was also investigated.
The yield of the products was calculated using the following
equation:

Wi
yield (wt%) ¼  100 (1)
WVR

where Wi (i ¼ G, D and C) is the fractional weight of the

product specie.

Table 4. Gas composition at various severities.

Gas composition (wt%)

4308C, 4458C, 4608C, 4758C,

Component 600 s 600 s 900 s 1200 s

C1 45.71 44.08 43.72 44.70

C2 22.47 22.17 21.87 22.91
C20 2.85 2.82 3.21 2.98
C3 þ C30 21.71 21.37 22.55 20.17
nC4 5.52 6.04 6.43 5.60
iC4 0.19 0 0.12 0 Figure 7. Comparison between experimental and predicted values of
C40 1.52 3.49 2.05 3.61 distillate at different temperatures: S, 4308C; A, 4458C; 4, 4608C; ,
4758C (—predicted curves).

Trans IChemE, Part A, Chemical Engineering Research and Design, 2007, 85(A4): 481– 491
 STUDIES IN COKING OF ARABIAN MIX VACUUM RESIDUE 489

Figure 8. Comparison between experimental and predicted values of Figure 9. Arrhenius plot for four lump, three parameter model
VR at different temperatures: S, 4308C; A, 4458C; 4, 4608C; , (S, VR ! G; A, VR ! D; 4, VR ! C).
4758C (— predicted curves).

(Figure 7) at higher reaction times (60 and 90 min). Maximum and conversion of vacuum residue have been modelled
distillate yield (53.59 wt%) was found to be at 4758C and for with first order kinetics.
the reaction time of 60 min. The rate equations for different species considered in the
proposed kinetic model (Figure 4) based on first order
kinetics can be given by the following set of differential
VR concentration equations:
The 5008Cþ was found to decrease with increasing sever-
ity. At higher temperatures (460 and 4758C), the conversion dVR
¼ kVR (2)
of VR was found to be pronounced and the maximum conver- dt
sion of vacuum residue was found to be 99.38% at 4758C dG
¼ k1 VR (3)
and for the reaction time of 90 min (Figure 8). dt
dD
¼ k2 VR (4)
dt
dC
Reaction Kinetics ¼ k3 VR (5)
dt
Schabron et al. (2002) have reported that coke formation
involves a complex set of reactions and may fall somewhere where k ¼ k1 þ k2 þ k3.
between zero order and first order kinetics. For the present The analytical solutions for the above equations (2)–(5)
kinetic scheme (which involves formation of gas and distillate are as follows;
along with formation of coke), attempt was made to investi-
gate the order for three reactions as shown in Figure 4. VR ¼ VRo ekt (6)
The correlation coefficients of the parity plots for the 0th, k1 VR0
G ¼ G0 þ (1  ekt ) (7)
0.5th and 1st order reaction are given in Table 5. It can be k
seen that the best fit for the kinetic scheme proposed is k2 VR0
D ¼ D0 þ (1  ekt ) (8)
obtained with first order kinetics for all the reactions. k
Hence, the different reaction pathways possible from the pro- k3 VR0
C ¼ C0 þ (1  ekt ) (9)
posed kinetic scheme i.e., formation of gas, distillates, coke k

where k ¼ k1 þ k2 þ k3 and VR0, G0, D0 and C0 are zero time

Table 5. Correlation coefficient (C.C.) of the parity plots for different concentrations of VR, G, D and C, respectively.
orders of reaction. Temperature range: 430–4758C. The estimation of the rate constants was achieved by
non-linear regression (Kataria et al., 2004), which involves
Correlation coefficient for
minimization of the sum of square of error (SSE) objective
Vacuum function given as follows:
Order of residue Gas Distillate Coke Xm X n
2
reaction concentration concentration concentration concentration SSE ¼ yexpij  Ypredij
i¼1 j¼1
0th order 0.9326 0.9678 0.7857 0.6659
0.5th order 0.7636 0.9021 0.5239 0.4584
1st order 0.9857 0.975 0.8855 0.921
The rate constants were estimated by Simplex method using
the MS-EXCEL spreadsheet programme. The guess values

Trans IChemE, Part A, Chemical Engineering Research and Design, 2007, 85(A4): 481–491
490 SAWARKAR et al.

Table 6. Variation in rate constants for different pathways at different as low as 0 –18C and as high as 8–128C depending
temperatures.
upon the severity of the reaction.
Rate constants, (3) The coke formation levels off after a certain severity of the
s21 (1  1024) T ¼ 4308C T ¼ 4458C T ¼ 4608C T ¼ 4758C reaction and thereafter there is a slight fall in the coke
yield as the volatiles trapped in the coke matrix are
k1 (VR ! G) 0.35 0.68 1.47 3.09
k2 (VR ! D) 1.27 2.36 3.65 7.13 cracked when the reaction time is further increased.
k3 (VR ! C) 0.65 1.69 3.82 9.96 (4) Maximum coke formation (about 30.83 wt%) for Arabian
mix vacuum residue was found at the temperature of
4758C and at a reaction time of 30 min.
(5) Severity was found to have negligible effect on overall
Table 7. Activation energy and frequency factor for different pathways.
gas composition and methane (about 42 wt%) was
Activation energy, cal mol21 Frequency factor, s21 found to be the major component.
(6) A four lump, three rate parameter model has been pro-
E1 (VR ! G) 50 726 (+7013) 1.997E þ 11 (+130)
E2 (VR ! D) 38 997 (+9864) 1.694E þ 08 (+942) posed to explain the mechanism of cracking and coking
E3 (VR ! C) 62 650 (+6981) 1.953E þ 15 (+127) reactions. A very good agreement between the exper-
imental yield and predicted yield of different pseudocuts
was obtained.
were selected using the methodology given by Mosby et al. (7) The value of k3 (rate constant representing formation of
(1986) and Ayasse et al. (1997). Figures 5, 6, 7 and 8 shows coke from VR) at 4758C was found to be about 2.5
the comparison between the experimental and predicted times higher than its corresponding value at 4608C indi-
values for coke, gas, distillate and VR, respectively in the pro- cating that coking reactions are predominantly favourable
posed kinetic scheme. Figure 9 shows the Arrhenius plot for at 4758C.
the proposed four lump, three parameter model. The estimated (8) Over the temperature range studied, VR was found to
rate constants and the apparent activation energy for various undergo first order decomposition and the apparent
product lumps for different reaction pathways are given in activation energy for the reaction pathway involving the
Table 6 and Table 7, respectively. From Table 6, it can be formation of coke from VR was found to be about
seen that the value of k3 (reaction pathway involving formation 24 kcal mol21 higher than the activation energy of for-
of coke from VR) at 4758C is about 2.5 times higher than corre- mation of distillates from VR.
sponding value at 4608C indicating that coking reactions are (9) Over the temperature range studied, the apparent
predominantly favorable at 4758C. This has been corroborated activation energy pertaining to the reaction involving the
by the experimental data (Table 3) wherein it can be seen that at formation of gas from VR was found to be 51 kcal mol21.
4758C, about 14 wt% of coke yield was obtained for the reac-
tion time of 5 min whereas for the same reaction time at
4308C, 4458C and 4608C, the yield of coke was only about NOMENCLATURE
1 wt%. In the present work, the activation energy of the coke AMVR Arabian mix vacuum residue
formation reaction (63 kcal mol21) was found to be higher C coke (defined as toluene insoluble), wt%
than that of activation energy of the distillate formation reaction C0 concentration of coke at zero reaction time, wt%
CCR Conradson carbon residue, wt%
(39 kcal mol21). With regard to the confidence estimate of the D distillates (IBP-5008C), wt%
kinetic parameters, 95% confidence interval was estimated. D0 concentration of distillates at zero reaction time,
The values in bracket (Table 7) indicate 95% confidence inter- wt%
val. As can be seen from confidence intervals in Table 7, the E1, E2, E3 activation energy, kcal mol21
G gas (C1 –C4), wt%
activation energy of the path pertaining to coke formation is sig- G0 concentration of gas at zero reaction time, wt%
nificantly higher than the activation energy of the path pertain- k1, k2, k3 rate constants, s21
ing to distillate formation. It can also be seen that the activation k k1 þ k2 þ k3, s21
energy of the path pertaining to coke formation is comparable to SARA saturates, aromatics, resins, asphaltenes
the activation energy of the path pertaining to the gas formation. SL solubility limit of solvent for asphaltenes, wt%
SSE sum of square of errors
The activation energy for the reaction pertaining to the for- t time, s
mation of coke is in line with the activation energy TC temperature of the reaction mass at the centre of
(63.9 kcal mol21) reported by Del Bianco et al. (1993). The the reactor, 8C
higher activation energy pertaining to the coke formation vis- TW temperature of the reaction mass near the wall, 8C
VR vacuum residue (5008Cþ), wt%
à-vis distillate formation imply that the reactions responsible VR0 concentration of vacuum residue at zero reaction
for the coke formation become relatively more important with time, wt%
the increase in temperature and coke formation proceed to Wi weight fraction of product specie obtained, g
completion rapidly. WVR weight fraction of VR charged, g
yi variable

CONCLUSIONS
(1) A new experimental set-up presented in this work was
able to give the insight of inception, growth and sub- REFERENCES
sequent saturation of coke formation within the severity
Ayasse, A.R., Nagaishi, H., Chan, E.W. and Gray, M.R., 1997, Lumped
range studied. kinetics of hydrocracking of bitumen, Fuel, 76: 1025–1033.
(2) The temperature difference of the reaction mass at the Bansal, B.B., Gentry, A.R. and Moretta, J.C., 1994, Improve your
wall and at the centre of the reactor was found to be coking process, Hydrocarbon Proc, 73(2): 63– 66.

Trans IChemE, Part A, Chemical Engineering Research and Design, 2007, 85(A4): 481– 491
 STUDIES IN COKING OF ARABIAN MIX VACUUM RESIDUE 491

Del Bianco, A., Panariti, N., Anelli, M., Beltrame, P.L. and Carniti, P., Song, F.M., Yang, G.J., Han, S.K., Huang, Z.Q. and Qian, J.L., 1995,
1993, Thermal cracking of petroleum residues. 1. Kinetic analysis Study on coking kinetics of Chinese gudao vacuum residue, Fuel
of the reaction, Fuel, 72: 75–80. Sci Technol Int, 13: 819– 835.
Elliott, J.D., 1992, Maximize distillate liquid products, Hydrocarbon Swaty, T.E., 2005, Global refining industry trends: The present and
Proc, 71(1): 75– 84. future, Hydrocarbon Proc, 84(9): 35– 46.
Gray, M.R., McCaffrey, W.C., Huq, I. and Le, T., 2004, Kinetics of Takatsuka, T., Kajiyama, R., Hashimoto, H., Matsuo, I. and Miwa,
cracking and devolatilization during coking of Athabasca residues, S.A., 1989, Practical model of thermal cracking of residual oil,
Ind Eng Chem Res, 43: 5438–5445. J Chem Eng Jpn, 22: 304 – 310.
Kataria, K.L., Kulkarni, R.P., Pandit, A.B., Joshi, J.B. and Kumar, Trasobares, S., Callejas, M.A., Benito, A.M., Martinez, M.T., Severin,
M.M., 2004, Kinetic studies of low severity visbreaking, Ind Eng D. and Brouwer, L., 1999, Upgrading of a petroleum residue.
Chem Res, 43: 1373– 1387. Kinetics of Conradson carbon residue conversion, Ind Eng Chem
Li, S., Liu, Z., Liu, C., Que, G., Liang, W. and Zhu, Y., 1998, Formation Res, 38: 938–943.
of the physical neophase in terms of SARA compositions of Wang, Z., Que, G., Liang, W. and Qian, J., 1998, Chemical structure
vacuum residua, Prep-ACS, Div Pet Chem, 43: 187–189. analysis of shengli crude vacuum residuum related to thermal
Magaril, R.Z. and Aksenova, E.I., 1968, Study of the mechanism of conversion, Prep-ACS, Div Pet Chem, 43: 151 –154.
coke formation in the cracking of petroleum resins, Int Chem Weekman, V.M., 1979, Lump models and kinetics in practice, Chem
Eng, 8: 727 –729. Eng Prog Monogr Ser, 75: 3– 75.
Magaril, R.Z., Ramazaeva, L.F. and Aksenova, E.I., 1971, Kinetics of Wiehe, I.A., 1992, A solvent-resid phase diagram for tracking resid
the formation of coke in thermal processing of crude oil, Int Chem conversion, Ind Eng Chem Res, 31: 530– 536.
Eng, 11: 250–251. Wiehe, I.A., 1993, A phase-separation kinetic model for coke
Mosby, J.F., Buttke, R.D., Cox, J.A. and Nokolaides, C., 1986, formation, Ind Eng Chem Res, 32: 2447– 2454.
Process characterization of expanded-bed reactors in series, Wiehe, I.A., 1994, The pendant-core building block model of
Chem Eng Sci, 41: 989–995. petroleum residua, Energy & Fuels, 8: 536–544.
Rahmani, S., McCaffrey, W. and Gray, M.R., 2002, Kinetics of solvent Yasar, M., Trauth, D.M. and Klein, M.T., 2001, Asphaltene and resid
interactions with asphaltenes during coke formation, Energy & pyrolysis 2. The effect of reaction environment on pathways and
Fuels, 16: 148–154. selectivities, Energy & Fuels, 15: 504–509.
Rahmani, S., McCaffrey, W.C., Dettman, H.D. and Gray, M.R., 2003, Zuba, B.I., 1998, Delayed coking in the CIS, Hydro Proc, 77(10):
Coking kinetics of asphaltenes as a function of chemical structure, 132-B-132-G.
Energy & Fuels, 17: 1048– 1056.
Savage, P.E., Klein, M.T. and Kukes, S.G., 1985, Asphaltene reaction
pathways. 1. Thermolysis, Ind Eng Chem Process Des Dev, 24:
1169– 1174. ACKNOWLEDGEMENTS
Schabron, J.F., Pauli, A.T. and Rovani Jr, J.F., 2002, Residua coke
formation predictability maps, Fuel, 81: 2227–2240. Authors are grateful to PETROTECH-2005 committee for
Schabron, J.F., Pauli, A.T. and Rovani Jr, J.F., 2003, Coke predict- granting permission to publish this work. One of the authors,
ability maps, Prep-ACS, Div Pet Chem, 48: 1 –5. A.N. Sawarkar, is grateful to the University Grants Commis-
Schucker, R.C., 1983, Thermogravimetric determination of the coking
kinetics of Arabian heavy vacuum residue, Ind Eng Chem Process sion (UGC), Government of India, for providing the financial
Des Dev, 22: 615–619. support in the form of research fellowship.
Shen, H., Ding, Z. and Li, R., 1998, Thermal conversion—an efficient
way for heavy residue processing, Proc 15th World Petroleum The manuscript was received 23 September 2006 and accepted
Congress, 2: 907– 913. for publication after revision 17 January 2007.

Trans IChemE, Part A, Chemical Engineering Research and Design, 2007, 85(A4): 481–491