8960

Ind. Eng. Chem. Res. 2008, 47, 8960–8988

Petroleum Residue Upgradation via Visbreaking: A Review

Jyeshtharaj B. Joshi,*,† Aniruddha B. Pandit,† Kamal L. Kataria,† Rohit P. Kulkarni,† Ashish N.
Sawarkar,† Deepak Tandon,‡ Yad Ram,‡ and Man Mohan Kumar‡

Institute of Chemical Technology, UniVersity of Mumbai, Matunga, Mumbai-400 019, India, and Indian Institute
of Petroleum (IIP), Dehradun-248 005, India

World petroleum residue processing capacity has reached about 810 MMTPA. In the present petroleum
refining scenario, the viability of a petroleum refinery strongly depends on the flexibility of processing heavy
crudes and, in turn, heavy residues. Visbreaking is one of the major residue upgrading processes and
constitutes about 33% of the total residue processing capacity. In the present communication, the published
literature pertaining to the visbreaking process has been extensively analyzed and a state-of-the-art review
has been written that includes the following: (i) the effect of feed properties on fuel oil stability; (ii) reaction
pathways, mechanism, and kinetics; (iii) parametric sensitivity of the operating variables such as
temperature, pressure, and residence time; (iv) different visbreaker designs, viz. coil visbreaker, coil-soaker
visbreaker, soaker with internals, and high conversion soaker; (v) coking and fouling; (vi) estimation of
design parameters, viz. gas holdup in high-pressure bubble column (soaker), gas holdup in sectionalized
bubble column (soaker with internals), liquid-phase mixing and axial mixing in high-pressure bubble
column, liquid-phase mixing and axial mixing in sectionalized bubble column, and weeping; and (vii)
mathematical modeling of visbreaker, which mainly includes the coil and the soaker. An attempt has been
made to get the aforementioned aspects together in a coherent manner so that the information is available at a
glance and is expected to be useful to

researchers and practicing refiners. * To whom correspondence should be addressed: Phone: 91-22-
2414 5616. Fax: 91-22-2414 5614. E-mail: [email protected].
1. Introduction
2
University of Mumbai.
3
Indian Institute of Petroleum.
The worldwide petroleum refining capacity has reached Worldwide residue processing capacity has reached about
about 4100 million metric tons per annum (MMTPA), 1 810 MMTPA (about 20% of the processed crude) using
inclusive of India’s refining capacity of about 146 MMTPA, 2 various upgradation processes. Visbreaking, practiced since
processed in over 700 refineries. 3 Petroleum refining has the 1930s,7 is a long-time workhorse with regard to bottom-of-
entered a significant transition period as the industry has the-barrel upgradation. Visbreaking is a mild liquid-phase
moved into the 21st century. The decreasing supply of light thermal cracking process, which reduces the viscosity and pour
sweet crude oils is a matter of serious concern for the point of the residue so that, with little or no addition of lighter,
petroleum industry because the distillation of heavy crude oils valuable cutter stock, it can meet the fuel oil quality (fuel oil
leaves behind a significant amount of vacuum residue (about No. 2). Visbreaking offers an economical conversion of the
40% of the total crude processed). 4 The quantity of heavy residual feedstocks to incremental gas and distillate products
residues is expected to increase in the future in view of the while simultaneously reducing fuel oil viscosity. 7–15
progressively increasing heavier nature of the crudes. 1,5,6 In Furthermore, visbreaking can be even more attractive if the
order to satisfy the changing pattern of product demand, refiner has idle equipment available that can be modified for
significant investment in the refining conversion processes has this service. In addition, visbreaking projects can be
become inevitable to profitably utilize these heavy crude oils. implemented rapidly and a quick payout (in about 2 years) can
Although the most efficient and economical solution to this be expected. Worldwide, about 200 visbreaking units 16 are
problem depends, to a large extent, on individual country and under operation, and Europe alone accounts for about 55% 17 of
company situations, the most promising technologies involve the total visbreaking capacity. The fact that about 33% of the
the conversion of vacuum residua and extra heavy crude oil total petroleum residue processing capacity is met via
into light and middle distillate products. It has been widely visbreaking17 is testimony to the success of visbreaking
envisaged that the window of opportunity is for those process. Atmospheric residue (AR), vacuum residue (VR), and
refineries that can process low-grade crudes to produce high- slops obtained from different refinery units act as the
grade products. Another reason for giving extra attention to feedstocks for the visbreaking unit. The process benefits 18
petroleum residue upgradation is that, with the continual (Table 1) include saving of the cutter stock and incremental
increase in the crude oil prices, it has become imperative to get quantities of value-added lighter products (gas and gasoline,
the maximum out of the residual feedstocks. In the light of this bp typically <175 °C), mid-distillates, and catalytic cracker
scenario in the petroleum refining industry, petroleum residue feeds.
or bottomof-the-barrel upgradation has gained tremendous In the past, visbreakers were employed primarily to produce
importance. The processes that convert these residues into incremental gasoline and to reduce fuel oil viscosity. It may
low-boiling, valueadded products are called as bottom-of-the- also be pointed out that cutter stock reduction, not destruction
barrel upgrading processes or residue upgrading processes. of residue, was the original objective of the visbreaking
 Ind. Eng. Chem. Res., Vol. 47, No. 23, 2008 8961
process. However, current applications of visbreaking are demand from fuel oil to distillates. When the unit is intended
aimed at residue destruction to increase clean products yield. 19 to process vacuum gas oil, the process is referred to as thermal
This is due to both heavier crude slates and shifting of product cracking.7
10.1021/ie0710871 CCC: $40.75  2008 American Chemical Society Published on Web 07/04/2008
Table 1. Typical Yields from Bachaquero or Light Arabian Crude Oil (Basis ) 2.65 MMTPA Crude Oil)18
Bachaquero LightArabian
atmosphericresidue vacuumresidue atmosphericresidue vacuumresidue
Crude Unit Operation

residual flow rate, MMTPA 1.94 1.28 1.18 0.34

cutter needed (no visbreaker), MMTPA 0.52 0.92 0.19

total no. 6 fuel oil production, MMTPA 2.46 2.20 1.18 0.53
Visbreaker Operation
visbreaker yields, wt % (2) Catalytic process: residue fluidized catalytic cracking
(RFCC).
C1-C4 2.4
C5-175 °C 6.2 (3) Hydrogen-addition processes: residue hydrocracking.
175-343 °C 20.6 (4) Thermal conversion processes: visbreaking, delayed
343 °C+ 70.8 coking, fluid coking, and flexicoking.
residual flow rate, MMTPA 1.37 The solvent deasphalting process involves physical
cutter needed, MMTPA 0.17
total no. 6 fuel oil production, MMTPA 1.54
separation, and there is no chemical conversion. The
Table 2. World Residue Processing Capacity, MMTPA17 limitations of this process are high energy cost and the limited
uses of deasphalter tar. Current interest in deasphalting is
process U.S. Japan Europe restofworld total
greatest in the areas of the world where demand for motor fuel
thermal is low. This suggests
2.0 2.6 2.3
cracking/visbreaking 6.5 1 108.5 82.5 198.5 5.5 7.4 6.2
coking 93 3 31.5 61 188.5
10.5 23.5 11.8
deasphalting 13 1 0.5 5 19.5
hydroprocessing 30.5 30.25 9 49.75 119.5 82 66.5 79.7
resid FCC 31.5 12.5 10.5 37 91.5 1.05 0.78 0.27
total 174.5 47.75 160 235.25 617.5 0.66 0.13
2. Qualitative Description of Different 1.71 0.78 0.40
UpgradingProcesses
that, in the long run, solvent deasphalting, as a stand-alone
Technologies for upgrading heavy feedstocks such as heavy residue upgrading process, will have less interest worldwide.21
oil, bitumen, and residua can be broadly divided into carbon Residue fluidized catalytic cracking (RFCC) involves a
rejection and hydrogen addition processes. Carbon rejection vaporphase catalytic cracking reaction. The heavier and more
redistributes hydrogen among the various components, contaminated atmospheric and vacuum residues cannot
resulting in fractions with increased H/C atomic ratios and vaporize and eventually end up getting deposited on the
fractions with lower H/C atomic ratios. On the other hand, surface of the catalyst and tend to increase the production of
hydrogen addition processes involve reaction of heavy coke and deactivate the catalyst. Thus, residue fluidized
feedstock with an external source of hydrogen that results in catalytic cracking (RFCC) is limited in terms of its
an overall increase in H/C ratio. Within these broad ranges, all applicability to process relatively low metal and low
upgrading technologies can be subdivided as follows: asphaltene feeds.17
(1) Carbon rejection processes: visbreaking, steam
Residue hydroprocessing can process slightly heavier and
cracking, fluid catalytic cracking, and coking.
high metal content feedstocks (Conradson carbon residue
(2) Separation processes: solvent deasphalting. (CCR) up to 10 wt % and metals up to 100-150 ppm) with the
(3) Hydrogen addition processes: hydrocracking, fixed- aid of a new process such as Chevron’s Onstream Catalyst
bed catalytic hydroconversion, ebullated catalytic bed Replacement (OCR) and Shell’s HYCON unit. Thus, residue
hydroconversion, hydrovisbreaking, hydropyrolysis, and donor hydroprocessing has a distinct advantage over residue
solvent fluidized catalytic cracking as regards the processing of
processes.20 heavier feeds. However, it may be pointed out that, although
Table 2 shows the world residue processing capacity in residue hydroprocessing can produce high-quality products
different parts of the world. It can be seen that a major portion and meet the requirement of the reformulated gasoline and
of the petroleum residue upgrading (about 63%) is met via diesel in terms of low aromatic and low sulfur, hydrogen
thermal processes, viz. visbreaking and delayed coking. resource and high investment limit its application. 17 Thus,
heavy residues containing more than 10 wt % CCR and 150
There can be a brief classification of residue upgrading
ppm of metals can only be processed by using noncatalytic
processes under different headings, which have been
carbon rejection processes22 as illustrated in Figure 1.
commercially installed over the years in the refineries. The
classification can be as follows:
(1) Separation processes: solvent deasphalting.
8962 Ind. Eng. Chem. Res., Vol. 47, No. 23, 2008

3. Visbreaker Feed Properties unsuitable for sale as fuel oil. This problem is usually referred
to as fuel oil
Petroleum residua comprise a polarity continuum of instability.23,24
molecules ranging from nonpolar hydrocarbons to
intermediate polarity aromatic species to polar asphaltenes, For a given feed, the visbreaking conversion is proportional
and possibly very polar precoke materials. In the visbreaking to the cracking temperatures and residence times; however, it
process, the nature and behavior of the asphaltenes determine is limited by the induction of the instability in the residue
potential fuel oil compatibility.7 Cracking temperatures used in followed by asphaltenes flocculation, which ultimately leads to
visbreakers must be the coke formation.25 An unstable feed, defined by the
existence of n-alkane insoluble material separated from the
maltene (oily) phase, along with the formed coke leads to
fouling of the process lines and the furnace heater tubes. This
results in the incomplete combustion of the fuel oil and an
early shutdown of the units (shorter run length).
At molecular level, the stability factor, the coking tendency,
and, consequently, the permissible conversion (crackability)
are the typical characteristic features of the feed being
processed to visbreaker units.26 The feed properties vary to a
large extent depending upon (i) the crude origin and (ii) the
extent of topping. To understand the fundamental chemistry of
the cracking behavior, the induction of instability, the coke
formation, the reaction mechanism and kinetics, and the
Figure 1. Feasibility region of the commercial residue conversion product variation as a function of the feed properties, a large
processes (Phillips and Liu22).
Figure 2. Variation in feed chemical composition and physical properties.
number of AR, VR, their asphalts, and bitumen have been
processed either in continuous or batch manner. 27–39 These
carefully controlled to prevent separation of the asphaltenes feedstocks have a range of H/C ratios from 1 to 1.8, API
from the bulk of the visbroken residue (fuel oil). Asphaltenes gravity from 3 to 18, sulfur content from 0.1 to 7.5 wt %, and
are normally held in colloidal suspension by complex asphaltene content in the range of 0.6-16 wt %. Parts A-F of
peptizing agents, which occur naturally in crude oil. Under Figure 2 show the variation in the feed properties as a function
severe cracking conditions, these peptizing agents tend to be of API gravity and H/C atomic ratio of the feed. From these
broken down, which in turn leads to asphaltenes precipitation. figures, the following can be seen:
As a result of this phenomenon, the visbroken residue becomes (1) The sulfur and nitrogen content decreases with an
increase in the H/C atomic ratio (parts A and B of Figure 2),
 Ind. Eng. Chem. Res., Vol. 47, No. 23, 2008 8963
which indicates the increasing concentration of these where the CCR contents of the AR and VR were related to
heteromoieties in aromatic and unsaturated compounds. feed asphaltene content as follows:
(2) The H/C atomic ratio decreases with a decrease in the
API gravity (Figure 2C). The reduction in this ratio indicates As (wt %) ) 0.3373CCR + 0.0058CCR2 (for VR) (4)
an increase in the content of unsaturated and polycondensed
aromatic compounds. This is again confirmed by the observed As (wt %) ) 0.2854CCR + 0.0049CCR2 (for AR) (5)
trends of Conradson carbon residue (CCR) and n-C7 insoluble 49
Li et al. proposed the relationship between the relative
asphaltenes (parts D and E of Figure 2). proportions of SARA fraction on the stability of VR as
(3) H/C atomic ratio decreases linearly with an increase follows:
in CCR (Figure 2F). The linear relation 24 between CCR and
H/C ratio is given by PA
SX ) 1.36( ) + 3.11(Ar) - 1.86(S) (6) As
H ⁄ C ) 1.71 - 0.0115CCR (1) The above equations indicate that the stability improves with
Equation 1 holds within two limits; at H/C ) 1.71, the CCR ) an increase in the resins and/or aromatics content and a
0, i.e., no coke formation, while at H/C ratios ) 0.5, the CCR ) reduction in the asphaltenes and saturates in the feed. Further,
100 (which means all material gets converted to coke under Liu et al.50 studied the batch thermal cracking of about eight
test conditions). The correlation is based on 114 data points residues and their subfractions covering the wide range of
derived from 15 different crudes and has a correlation elemental (C, H, S, and N) and SARA compositions and
coefficient of 0.988. proposed cracking and coking characterization factors as
In order to ease the identification of chemical composition, follows:
heavy oil can be classified as consisting of four polarity and -
K M (7)
solubility classes of components, viz. saturates, aromatics, r ) (H ⁄ C)0.257(S % )0.158(N % )0.184( W ) 0.151
resins, and asphaltenes (SARA). Prior to chromatographic K M
separation, asphaltenes can be separated. Asphaltenes are r′ ) (Sa)0.026(Ar)0.008(PA)0.38(S)0.17( )
W -0.230 (8)
typically defined as the component that is insoluble in n- The
heptane and soluble in toluene.40,41 Using normal-phase coke
chromatography, the heptane-soluble maltene material can be
separated into saturates, aromatics, and resins using heptane, Kc ) (H ⁄ C)-3.80(MW)0.40 (9)
toluene, and toluene/methanol mixtures, respectively. The
qualitative relation between the feed characteristics and their
permissible conversion, product yield, stability, and coking Kc ) (H ⁄ C)-3.80(PA + As)0.30(MW)0.24 (10)
tendency can be summarized as follows:
Kataria et al.38 reported the batch thermal cracking of six
(1) The viscosity reduction (with high distillates
different vacuum residues and asphalts. The authors found the
production) largely increases with an increase in the extent of
distillate (VGO, LGO, Ga, G) yield to be dependent on the
conversion for a given feed with specific characteristic
severity and the initial feed concentration (SARA, n-C5
properties.31,42
insolubles, n-C7 insolubles, CCR, and S) composition as
(2) The extent of permissible conversion of a particular follows:
feed within which it is stable and has no coke yield is termed
as
“crackability”.43
(3) The stability and the crackability of the feed are
found to improve with an increase in the content of the resin- VR ) 100 As ( ) 0.02 exp [ (
-0.023 1 + n-C5 SI )] (11)
to-asphaltene ratio and with a decrease in relative aromaticity Sa CCR
between the maltene phase (oily phase) and the asphaltenic

(
phase.16,31,45,46 (4) Feeds rich in asphaltenes and CCR are found n-C
0.96
to produce higher gas, gasoline, and coke yields.31,47

)
The quantitative description between the feed characteristics
and conversion, yield, stability, and coke has also been
VGO ) 4.76(SI)0.55
proposed.38,48–50 Kuo48 reported that the extent of conversion
(crackability) is proportional to the feed aromatics and the (12)
resin content and inversely proportional to the saturates and S + CCR
the CCR content. The author proposed the following
quantitative relationship,
) 3.24(SI)0.77Sa() 0.13
Conversion (150 ° C-), % wt ) -0.7922 + 41.804(SF) (2) (13)
where SF, the stability factor of the feed, which depends on LGO
the feed composition, is given by the following, NA

SF ) (Ar + Rs) (3)

(Sa)(CCR) Ga ) 0.271(SI) ( As ) 0.19 (14) PA
8964 Ind. Eng. Chem. Res., Vol. 47, No. 23, 2008

)
(4) Asphaltenes: Asphaltenes are the heaviest and most
polar class of component in the crude oil, containing larger
(n-C5 0.45
amounts of heteroatoms than the rest of the components in the
G ) 0.43(SI) (15) crude oil.41,52 Asphaltenes comprise molecules with 4-10
CCR pericondensed aromatic rings, 4-6 alkyl carbon chains, and 20-
These correlations38,49,50 are based on batch studies at relatively 24 aliphatic carbons, with an average H/C atomic ratio of
about 1.2. IR studies have shown that about 90% of the
low severity (temperature ) 400-430 °C and residence time ) 3- hydrogen is substituted on saturated carbon, primarily in
60 min). methylene and some methyl groups. Sulfur species are
Apart from the feed characteristics, the mode of operation primarily aromatic thiophenes and aliphatic sulfides. Nitrogen
also acts as another important factor that governs the product species are primarily aromatic pyrrole and pyridinic and metal
yield and the stability. Thus, continuous improvements are chelating (Ni, V) porphyrin or pseudoporphyrin type of
being made in the thermal unit operation to achieve structures.53 The average molecular weight of an asphaltene
economical and improved conversions with desirable product component molecule is 750 g/mol, with a range of 500-1000
selectivity. g/mol.41 Groenzin and Mullins41 employed the time-resolved
fluorescence depolarization (TRFD) technique to estimate the
4. Kinetics and Mechanism of Visbreaking Reactions molecular weight of asphaltenes. TRFD can be performed on
extremely dilute solutions (10 mg/L)41 and, hence, does not
4.1. Structural and Compositional Changes during suffer from aggregation and gives the true molecular weight of
Visbreaking. Low demand for the fuel oil has forced the asphaltenes. It may also be pointed out that the mass spectral
refiners to opt for higher conversion even for feeds with low ionization techniques, viz. field ionization mass spectroscopy
crackability. In order to achieve this, the fundamental (FIMS), electrospray ionization (ESI), atmospheric pressure
understanding of the reactions involved rendering instability, photo ionization (APPI), and atmospheric pressure chemical
and the subsequent coke formation is needed. The pertinent ionization (APCI) employed for the determination of
subject has been elaborated in the following three sections: (i) molecular weight of asphaltenes corroborate the TRFD
the feed characteristics, (ii) the induction of instability and its results.54 The extent of asphaltenes aggregation depends on the
mechanistic variation, and (iii) the coke formation and fouling. type of solvent, the temperature, and the concentration. It has
4.1.1. Feed Characteristics: Salient Features. Visbreaker been found that molecular association may occur even at a
feeds comprise saturates, aromatics, resins, and asphaltenes very low concentration (150 mg/L).54 Vapor pressure
and are classified on the basis of their solubility and osmometry (VPO) and gel permeation chromatography (GPC)
chromatographic separations.40,41 In order to characterize these suffer from the fact that a high concentration of asphaltenes in
fractions, their average molecular weight (MW), average solution (about 1000 mg/L) is needed, which leads to the
molecular formula, microcarbon residue (MCR) content, and formation of aggregates or micelles of asphaltenes and not
the two important structural features (using H 1 and C13 NMR individual molecules, which in turn gives rather large
techniques), viz. aromatic and aliphatic carbon content, are estimations of asphaltenes molecular weight (nearly 10000
taken into consideration. These features fairly distinguish g/mol).41 It is known that asphaltenes are in a very delicate
subcomponents from each other and their variation within a state of dynamic equilibrium, where a minute asphaltene
class from different sources of visbreaker feeds. As thermal droplet is covered by a sequential envelopes of resins, 55
treatment progresses, the solubility parameter of the aromatics, and saturates (Figure 3) with decreasing 45,55 polarity.
components of residua fractions increases, and the apparent According to this model, asphaltenes are dispersed by the
molecular weight decreases.51 The distinguished features of surfactant-like property of resins that in turn are held in the
these subfractions can be summarized as follows: solution by aromatics. Saturates are nonsolvents for the
(1) Saturates: These components are found to have an asphaltenes. It may also be pointed out that, though the
average carbon number in the range C38-50 with relatively low asphaltenes are not soluble in petroleum without the resins, 40
heteroatom content. The structural study shows that it consists the resins might be a cosolvent. However, amphiphilic model
of long alkyl chains with few or negligible naphthenic and compounds increase the solubility of asphaltenes, and the
aromatic rings. The microcarbon residue (MCR) value (coking addition of resins reduces the asphaltene radius of gyration
tendency) reported is almost 0 wt %, which indicates that these measured by smallangle X-ray scattering.55
fractions are completely volatile and cannot directly yield 4.1.2. Mechanistic Variation and Instability. Cracking,
coke. dehydrogenation, H-abstraction, cyclization, aromatization,
(2) Aromatics: The aromatics are found to have a carbon and condensation are the major reactions that occur within the
number in the range C41-53. These are simple structures molecules during thermal processing of the feed.14 The feeds
compared to resins and asphaltenes, having low heteroatom rich in paraffin content are found to give high gas oil yield via
content and MCR value of about 3.7 wt %. (Cali-Cali, BDE ) 98 kcal mol -1) splitting reactions,31,38 while
(3) Resins: These are viscous, tacky, and volatile enough feeds rich in resin and asphaltenes contents are found to
to be distilled with hydrocarbons. Structurally, resins consist undergo
of an appreciable amount of aromatic carbon content (40-53%)
with intermediate paraffin chain length on naphthenic
structures and aromatic rings, and about 2/3 of its aromatic
carbon atoms are nonbridged. The resin fraction acts as the
dispersant for the asphaltenic component in the maltene phase.
 Ind. Eng. Chem. Res., Vol. 47, No. 23, 2008 8965
asphaltene content, and a change in the internal polarity
distribution within the asphaltenes themselves. At a particular
conversion, the thermodynamic colloidal stability of the oil
gets disturbed, which in turn leads to the separation of the

Figure 4. Idealized molecular structures for asphaltenes consistent with overall molecular size, aromatic ring systems, and chemical speciation (Groenzin
asphaltenes from the maltene phase and the feed becoming
unstable. The plausible mechanism of the asphaltene
aggregation involve (a) π-π overlap between aromatic sheets,
(b) hydrogen bonding between functional groups, and (c) other
charge-transfer interactions.55
Fuel oil stability is the most important factor that governs
the possible yield of the liquid products from the visbreaking
process. The feed residue and the visbroken residue (500 °C+)
obtained by distillation are analyzed for their stability by the
standard merit number test (IFP-3024-82, severity test for
thermal reactions by spot method and residue stability
determination). For this purpose, 2 mL of sample is diluted
with 10 mL of a mixture of o-xylene and 2,2,4-
trimethylpentane with variable o-xylene content. The above
mixture of the sample and the diluent is homogenized with the
Figure 3. Microemulsion structure of petroleum, containing solutes, As help of a glass rod. The homogeneous mixture is spotted on
(asphaltenes); dispersants, Rs (resins); solvents, Ar (aromatics); and Whatman filter paper no. 2. The spot is visually evaluated for
nonsolvents, Sa (saturates).
its homogeneity or phase separation. For homogeneous
samples, the stability is indicated by the lowest volume of o-
dealkylation (Car-Cali, BDE ) 78 kcal mol-1) to give high yields xylene added in a mixture of o-xylene and 2,2,4-
of gas, gasoline, low carbon molecules, and coke (at high trimethylpentane. The spot with a markedly darker center
severity via condensation). Typical structural variation in the (appearance of two phases) is considered as an unstable
subcomponents of the residue during thermal cracking 46,56 can sample. Samples with merit number less than 7 are considered
be summarized as follows: to be stable samples, those with merit number 7-8 are taken as
(1) There is a reduction in the paraffin chain length from borderline, and those with merit number above 7 are
C50 to C30 in saturates. considered to be unstable samples. For necessary fuel oil
(2) There is dealkylation of the aromatics and resins specifications, the merit number test is carried out for 150 or
(polar components), resulting in a decrease in the number of 350 °C+ bottoms fractions.
carbon atoms per alkyl chain (reduction in chain length) with and Mullins52).
parallel reduction in the naphthenic rings. The simultaneous
increase in the carbon number indicates their intermolecular A solubility parameter is a kind of cohesion parameter that
condensation reactions. describes the interaction between molecules in condensed
(3) Quantitatively, there is an increase in saturates materials. It could be expressed as the difference between the
content (due to thermally formed distillates), a reduction in the internal energy of the condensed material and that of an ideal
aromatic and resin content, an increase in highly polar gas of the same material at the same temperature. It may be
8966 Ind. Eng. Chem. Res., Vol. 47, No. 23, 2008
pointed out that the main purpose of the solubility parameter feed is highly dependent on (i) peptizability (aromaticity) of
concept has been to predict the precipitation of asphaltenes. 57 asphaltene core, (ii) peptizing power of the resins, (iii) relative
The complexing solubility parameter component measures the amount of the asphaltenes and the resins, and (iv) aromaticity
interaction energy that requires a specific orientation between of the maltene phase.
an atom of one molecule and a second atom of a different The physical changes and heteroatom (S, N, Ni, V)
molecule. The field force solubility parameter component distribution that occur as the consequence of the thermal
measures the interaction energy of the liquid that is not cracking include the following:
destroyed by changes in the orientation of the molecules. (1) Physical changes: The feeds with high initial
Thus, hydrogen bonding and electron donor-electron acceptor viscosity (and high sulfur content) undergo high viscosity
interactions are part of the complexing solubility parameter reduction, due to cross-linking within different components by
component, and van der Waals and dipole interactions, when weak bonds that can easily break, as compared to the feeds
significant, are part of the field force solubility parameter with lower starting viscosity.16,61 Similarly, feeds with low
component.55 It may be pointed out that the solubility of an sulfur content (<1%) resulted in a sharp increase in the
individual asphaltene moiety decreases with the increase in the asphaltene content as compared to the feeds with high sulfur
difference between solubility parameters. Also, a lower content (>1%). The pour point and density reduction were
molecular weight moiety is more soluble than the higher found to be proportional to viscosity reduction.
molecular weight molecule for a particular difference in (2) Thermal effects on heteroatom: Nitrogen in the feed
solubility parameter. It is apparent that the solubility depends exists in two forms, i.e, basic (30%) and nonbasic, and is
on both molecular weight and polarity of the particular mostly concentrated in the resins and asphaltenes of the
asphaltene molecule or associated specie. As thermal treatment residue before and after the thermal conversion. 62 The
proceeds, the solubility parameter of the components of the denitration is less susceptible to cracking and remains
residue increases and the molecular weight decreases. An concentrated in the liquid fuel, showing continuous rise in N/C
unstable system results when the depletion of the resins ratio with increasing C-rejection conversion. The sulfur
diminishes the ability of the asphaltenes/resins complexes to content and H/C ratio in the visbreakate has been observed to
increase their apparent molecular weights to autoadjust their reduce with an increase in the conversion.14,28,30,31,56 The loss in
solubility parameter to closely match the solubility parameter sulfur can be attributed to desulfurization and its distribution
of the matrix. It is at this stage that a bimodal system is formed in the gaseous phase (appears as H 2S) and distillate products.
and the onset of coking takes place. 51 Apart from the thermal The metals generally found in the heavy oils and feeds include
induction of instability, the external parameters are also found Ni and V in the form of petroporphyrins and similar structures,
to affect the asphaltene precipitation. Nielsen et al. 58 observed as well as relatively low quantities of Na and Fe.
that, with an increase in the temperature, the molar mass of the
4.1.3. Coke Formation and Fouling. Thermal cracking of
oil and the diluent/oil ratio results in a larger amount of
hydrocarbons is always accompanied by the fouling and coke
asphaltene precipitation. On the contrary, an increase in the
formation that deposit on the furnace walls and results in
pressure or the number of carbon atoms in the diluent
increased pressure drop, reduced heat transfer rates, hot spot
molecules results in a lower amount of asphaltene
formation due to uneven flow distribution, and corrosion by
precipitation. Further, the distinct features of the stable and
carbonization. The detrimental effects arising out of these
unstable feed and coke formation include the following:
phenomena are reduced run length, reduced selectivity,
(1) Structural and the characteristics variations: The production losses, and high maintenance, operating, and utility
asphaltenes and resins in an unstable feed 59 are found to have costs. The effects are even more pronounced if the residue is
low H/C ratio, high aromaticity, highly condensed aromatic heavy since heavy residues are notorious coke producers in
rings, and less alkyl and naphthenic substitution. Figure 4 cracking coils.8 The fouling process progresses through various
shows three idealized molecular structures for asphaltenes, 52 steps, viz. polymerization of unsaturated radicals,
which depict the possible interaction of the aromatic moieties dehydrogenation, subsequent specific volume contraction, and
with aliphatic chains and heteroatoms. The instability of their final conversion into coke. The prevailing mechanisms in
asphaltenes increases with increasing heteroatom content (O, this process include (1) catalytic growth and (2) radical
N, and S) and metal content (Ni and V) in the parent mechanism for carbon deposition.63
feedstock.
(2) Charge on asphaltenes and resins: Fotland et al. 60 (1) Catalytic growth: The tube metallurgy plays an
observed that a stable feed (asphaltene content of 0.3-9 wt %) important role in this kind of mechanism. During the start-up,
under the influence of dc electric voltage shows little (0.015- the exposed active metal sites (Cr, Ni, Fe, etc.) in the furnace
0.05 µS/m) or no conductivity, while an unstable feed showed coils act as a heterogeneous polyaddition catalyst, resulting in
large conductivity (0.1-0.32 µS/m). The conductivity (also initial fouling deposition by polymerization of the molecules
confirmed by the gravimetry test) was also found to be containing olefins. Once the uniform film is formed on the
proportional to the extent of instability induced by the metallic surfaces, the catalytic growth mechanism is controlled
precipitation of the asphaltenes using varying amount of by the diffusion of the monomers into the polymeric layer,
externally added paraffinic solvent. Later, it was also which is dependent on the degree of cross-linking. Thus, after
confirmed45 by the electrodeposition study (at constant voltage the initial cracking period of 15-20 days, the phenomenon of
3.6 kV) for the crude oil, fuel oil, AR, VR, and visbroken catalytic growth becomes insignificant in the visbreaking coils
residues at different temperatures. Further, the author extended due to metal surface adsorption.
the study to separated resins and asphaltenes from the residue (2) Radical and condensation mechanism: The
and found that the micelle structure consists of negatively polyaromatic (polymeric) films formed on the metal surface
charged resins surrounding a positively charged central still contain few saturated centers. The hydrogen abstraction of
asphaltene core. The study also concluded that stability of the these species by free radicals thus becomes a significant
 Ind. Eng. Chem. Res., Vol. 47, No. 23, 2008 8967
reaction. The further growth in the film occurs through the 4.1.3.5. Alonizing. Coating the surface with aluminum can
addition of an olefin or new free radical in the feed. The be an attractive remedy for preventing carburization of furnace
katacondensed aromatic layers result in the hard coke coils. The alonized surfaces show (i) little or no loss of tube
formation, which is usually difficult to remove during material, (ii) no change in the roughness, (iii) a higher
decoking of the furnace coil and transfer line exchangers. percentage of Cr, Al, Nb, and Si, and (iv) higher thermal
Adding to this, the asphaltenes from the entrapped heavy conductivity of formed coke.
boiling material mainly undergo dealkylation and 4.1.3.6. Effect of Feed Composition. The physicochemical
condensation reactions that give rise to coke precursors. Thus, properties of the feedstock are also found to affect the coking
further dehydrogenation (H/C tends to reduce <0.5-0.6), rates and their properties. Yan47 studied the coke-forming
subsequent specific volume contraction and morphological tendency of three different feeds and found that coking
modifications produce coke with unavoidable porosity. tendency follows the following order: mixed residue < Boscan
Refineries frequently use different techniques, such as crude < Arabian heavy residue. The study concluded that
additive inhibitors, surface treatment, and changes in the tube visbreaker should be operated at maximum feasible operating
metallurgy and physicochemical characteristics of feedstock, temperature with shorter residence time (which also indicates a
to combat the fouling problem in furnace coils and transfer lower degree of liquid-phase backmixing) to maximize the
line exchangers. Though most of these techniques are distillate yields and reduce coking. The physicochemical
practiced for hightemperature pyrolysis processes (such as properties of the coke formed were also found to vary
steam reforming, naphtha pyrolysis, and ethane cracking), they according to the feed type, e.g., the thermal conductivity of the
can also be extended for mild thermal cracking process such as coke produced from naphtha and gas oil cracking ranges in
visbreaking.64 The methodology can be briefly described as 3.5-5.8 W m-1 °C-1, while that of those produced from ethane or
follows:
LPG cracking ranges in 11.6-14 W m-1 °C-1.64
4.1.3.1. Additive Inhibitors. The sulfur compounds, i.e.,
4.2. Kinetics. Thermal cracking reactions have been found
dimethyl sulfide (DMS), dimethyl disulfide (DMDS),
to follow zero-order to first-order kinetics 36 and an
benzothiophene, etc., act as metal passivators and also alter the
Arrheniustype relationship:
rate of radical reactions through their interference with
Htransfer and termination reactions. Similarly, k ) A exp(-E ⁄ RT) (16)
phosphoruscontaining compounds, such as triethyl phosphite
(TEP), triphenyl phosphite (TPP), benzyl diethyl phosphite The activation energy of cracking and coke formation can
(BDP), and triphenyl phosphite sulfide (TPPS), can be used as range from 22 to 83 kcal/mol. 36 The activation energy (E) in eq
metal passivators and for facilitating easy removal of coke 16 can be assumed to be 50 kcal/mol, which is the thermal
during decoking operation (decoking time can be drastically cracking energy for C15 compound.47
reduced). The use of alkali metal compounds such as K 2CO3, The residue comprises a large number of complex organic
Na2CO3, K2O, Na2O, and Na2SO4 also promotes the components, and a detailed kinetic model involving the
gasification of coke during decoking operations. cracking behavior of each of the components is impractical.
4.1.3.2. Surface Treatment. This method reduces the Hence, the pseudo-components (also termed as lump) are
catalyzing tendency of the reactor surfaces by treatment with considered by group contribution method, based on their
hydrogen, dilute acids (HCL, H2SO4), and sulfides. Also, a physicochemical properties such as boiling point, solubility,
preoxidation step is practiced during the start-up of new coils, carbon number, etc. On the basis of the thermal property
which results in stable and adherent Cr2O3 film on the coil variation (mostly boiling point) in these components, lumped
surface. kinetic models have been proposed (Table 3). The cited
models can be categorized into (i) parallel reaction models and
4.1.3.3. Tube Metallurgy. The key considerations in (ii) parallel-consecutive reaction models.
selecting appropriate tube metallurgy are oxidation, The parallel reactions models28,31,65 consider the thermal
carburization, creep strength, high-temperature ductility, and cracking of the feed without further cracking of the products or
cost. The higher weight percentage of Ni and Fe in tube the compositional changes within the feed. Castellanos et al. 66
metallurgy accelerates the dehydrogenation reactions at divided the heavy oil into a large number of model
elevated temperatures, thereby increasing the coking rates of components based on the carbon number and considered
hydrocarbons. To overcome this, the formation of an oxide thermal cracking of each component into lower carbon number
film on the tube-wall surface is desirable to minimize the and olefinic component, while Benito et al.33 and Xiao et al.67
metal effects. The elements Cr, Ni, and Si act as effective reported parallel decomposition of residue to distillates and
barriers for carburization. Similarly, the stability of oxides coke. A detailed investigation pertaining to the product yields
formed varies in the following order: Al 2O3 > TiO > TiO2 > of the visbreaking process and their properties prediction has
SiO2 > Cr2O3 > SiO > Mn2O3 > Mn3O4 > FeO > Fe2O3 > NiO > been reported by Dente et al. 14 The authors have proposed a
CrO3. mechanistic approach that elucidates the fouling of the coils,
4.1.3.4. Polishing. The polishing of tube surfaces results in physicochemical properties of the products viz. residue
reduction in irregularities, the active metal sites (due to stability, sulfur and asphaltenes content, viscosity, specific
removal of surface irregularities), and emittance of the surface. gravity, etc.
The emittance, which influences radiation and, therefore, the The parallel-consecutive model considers sequential
temperature of reactor surface, may vary from below 0.1 for a cracking of product species as well. The lumped model by
polished surface up to 0.9 for a plain or carbonized surface. Takatsuka et al.68 predicts variation in the yields of the
For highly polished tubes, the lower emittance means lowering distillates formed, along with the compositional changes of the
of tube-wall temperature by 50-100 °C and, hence, lowering of feed. The model proposed by Del Bianco et al. 32 for predicting
the coking rates. the thermal behavior of Belaym VR showed coke formation
8968 Ind. Eng. Chem. Res., Vol. 47, No. 23, 2008
through secondary reactions of the intermediates, while tube wall becomes hotter in order to overcome the thermal
models reported by Yasar et al., 34 Xiao et al.,67 and Trauth et resistance of the coke until the maximum allowable tube-wall
al.69 deal with the interconversion of the solubility class of temperature is reached.74 The precipitated asphaltenes and the
components and their rate parameters. Recently, Kataria et al. 38 coke formed get deposited on the heater tube surface, thereby
reported the variation in the yield of pseudo-components increasing heater duties and pressure drop, which results in
(based on their boiling point), the activation energies, and the untimely shutdown of the unit.
cracking as a function of feed properties by employing a five- 5.2. Coil-Soaker Visbreaker. In coil-soaker operation, feed
lump and seven-parameter model. is allowed to pass through the coil and is kept at relatively
lower temperature and the residence time (∼ 450 °C, 2-3 min)
5. Visbreaker Designs followed by soaking is held for a longer time (15-25 min) in an
5.1. Coil Visbreaker. The coil visbreaker is operated at adiabatic hollow drum known as a “soaker”. The reduction in
high temperature and low residence time and, thus, is also the coil temperature and simultaneous increase in the residence
termed as a high-temperature, short-residence-time (HTST) time in the soaker give similar conversions as in the case of the
process. Foster Wheeler has been at the forefront of designing coil type. Thus, the coil-soaker type is described as the low-
the heaters for the coil-visbreaking process. In coil temperature, high-residence-time (LTHT) route. The
visbreaking, the visbreaker feed from the preheater enters the technology has been developed by Shell and is also known as
Shell soaker visbreaking process (SSVP). The first of the Shell
furnace at 300-330 °C and undergoes thermal cracking in the
soaker visbreakers went on stream in 1962 in Curacao,
coil in the temperature range of 460-480 °C for the duration of Netherland Antilles.74 The process flow diagram (Figure 6A)
2-5 min. Steam (∼1 wt %) is injected along with the feed to (i) for the coil-soaker type shows that the feed from the coil outlet
generate the turbulence and, hence, increase the heat transfer follows the additional soaker drum path in an upflow manner
coefficient at the wall, (ii) control the liquid-phase residence and is quenched (using appropriate cutter stock) at the soaker
time, and (iii) prevent the coking along the tube walls. The outlet before fractionation.
cracking reactions (C-C cracking) occur in the liquid phase The variation in the temperature, pressure, and conversion in
and are slightly endothermic (192 kcal kg -1 of the product the case of the coil-soaker type is shown in Figure 6B. 43 The
boiling below 204 °C).70 The cracked feed at the coil outlet is feed cracking occurs both in the coil and in the soaker. The
immediately quenched (up to 350 °C) (Figure 5A) using pressure at the soaker inlet is about 1.3 MPa and reduces
fractionator bottoms and is then distilled into different boiling
cuts.
Variations in the temperature, pressure, and conversion
along the reactor length for the coil-type visbreaker operation
are shown in Figure 5B.43 The coil-inlet pressure (∼2.5 MPa)
is maintained at higher values so as to facilitate the transport
of the feed. The pressure drop (∼1.7 MPa) along the reactor
length is mainly due to the two-phase frictional pressure drop.
The gas phase initially consists mainly of the steam. However,
along the length, the gas-phase fraction increases because of
partial vaporization and thermal cracking of the feed. The
temperature of the feed increases continuously along the coil
length; however, because of the onset of cracking and the
endothermic nature of the reactions, the rate of temperature
rise reduces slightly after a certain coil length.
The cracked products obtained vary from gas (C1-C4) to gas
oil. The products, viz. gas, gasoline, and, to a smaller extent,
gas oil, are separated in a fractionator. However, the extent of
permissible conversion and the subsequent withdrawal of
distillates are governed by the stability, CCR, viscosity, and
pour point of the visbroken residue. The insoluble content (hot
filtration test), heated tube fouling, xylene equivalent, Shell P-
value, and Merit number tests are used to ensure the stability
of the fuel oil.71–73 The nature of the cutter stock, recycle oil, or
lube-oil extract that are used to blend and meet the desired
viscosity specifications also affect, to some extent, the final
stability of the fuel oil.
In order to improve the throughput of the process, the
singlecoil-type process was modified to two-coil-type. In the
coilvisbreaking process, the heater outlet temperature (about
480 °C) is considerably higher than in the soaker visbreaking
process (about 450 °C). As a consequence, the clean tube-wall
temperature and the fluid-film temperature are
correspondingly higher. The higher tube-wall temperature
leads to the coke buildup at the wall. As a consequence, the
 Ind. Eng. Chem. Res., Vol. 47, No. 23, 2008 8969
Table 3. Summary of the Reported Work on Visbreaking Kinetics
8970 Ind. Eng. Chem. Res., Vol. 47, No. 23, 2008
phase consists of product gas, vaporized low-boiling
components, and the added

Figure 7. Soaker drum for the visbreaker process: (A) hollow drum with
high backmixing and (B) modified soaker drum with no backmixing.

turbulizing steam. The RTD and tomographic experiments 75

have shown that the average gas holdup ranges from 20-40%
with bubble size ranging from 3-15 mm. Further, the study
also revealed that the residence time of the gas phase is around
Figure 5. Coil-type visbreaking operation: (A) flow diagram and (B) 1/10 of the residence time of the liquid phase, and thus, the
operational parameters (Akbar and Geelen43). overall conversion may be assumed to be controlled by the
liquid-phase cracking only.
The advantages of the coil-soaker process over the coil
process can be summarized as follows:
(1) The soaker drum allows the heater to operate at a
lower outlet temperature by providing the residence time
required to achieve the desired reaction. Because of the lower
heater-outlet temperature, the heater duty and, in turn, the fuel
consumption is only 70% of that for the coil-visbreaking
process.
(2) The lower fuel consumption reduces the generation
of large amounts of waste heat steam. Earlier, these steam
quantities were a significant part of the visbreaking process
and quite often upset the refinery steam balance.74
(3) It is known that the activation energy required for the
cracking of the hydrocarbons is inversely proportional to their
carbon number; it is higher for low carbon number compound
and vice versa. This implies that the selectivity to the cracking
of the feed in the coil-soaker mode, due to their relatively
lower temperature vis-a`-vis the coil type, allows only the
cracking of higher molecules, and already-cracked molecules
remain unaffected.
(4) The lower temperature results in less coke formation
and less destabilization of the feed, enhancing the unit run
length typically from about 3-6 months (in the case of the coil
type) to about 18 months.
Figure 6. Coil-soaker-type visbreaking operation: (A) flow diagram and (5) The low operating temperature results in low
(B) operational parameters (Akbar and Geelen43). operating cost.
5.3. Soaker with Internals. Figure 7A shows a
slightly due to the hydrostatic head, cracking, and typical representation of a soaker. As mentioned
vaporization. The temperature in the soaker reduces axially earlier, a soaker is essentially a bubble column where
due to the endothermic nature of the cracking reactions (∼10- the volumetric coefficient increases from bottom to
20 °C). It may be pointed out that, because of the lower top. The gas phase is in the form of bubbles in a wide
temperatures and the low conversion in the heater in soaker size distribution. The column operates in the
visbreaking, the coke buildup is much slower, so that the tube- heterogeneous regime where the concentration of gas
wall temperature increase is slower by a factor of about 3, as phase varies in the radial as well as axial directions.
has been found in operating plants. 74 The soaker drum can be At this stage, it may be noted that, in the
considered similar to a bubble-column reactor in which the gas homogeneous regime, the bubble concentration is
uniform in both radial and axial directions. The
 Ind. Eng. Chem. Res., Vol. 47, No. 23, 2008 8971
extent of uniformity of gas bubbles has direct the heater tubes. These features differentiate it from
repercussions on the performance of bubble columns. the soaker used in the conventional visbreaking
For instance, the radial and axial nonuniformity in process. The liquid product (C5, 520 °C) yield of the
the gas phase (holdup and bubble-size distribution) HSC process is about 44 wt %. 82 HSC can be
results in gross liquid circulation with an upflow in integrated with the existing coilvisbreaking unit for
the central region and downflow in the near-wall higher conversions and simultaneously allows the
region. The liquid circulation velocity is given by the use of the visbroken residue (without the addition of
following the cutter stock) as a fuel oil. The authors have also
equation:76 suggested

VC ) 1.31 (17)
The circulation velocities are of the order of 0.3-3 m/s and
as a result of such an intense liquid circulation, the liquid
phase remains completely backmixed. Quantitatively, the
liquid-phase dispersion coefficient is given by

DL ) 0.31D(VC + VL) (18)

The heterogeneity of the gas phase also reduces the gas-
phase residence time as compared to that in the homogeneous
regime. The above phenomenon leads to a wide spread in the
gas- and liquid-phase residence times and affects visbroken
residue stability and mid-distillate selectivity. The direct and
obvious effect is the reduction in the intense concentration
gradients, resulting in decreased overall conversion. Also, the
overcracking of the feed due to backmixing and dead zones
causes decreased solubility of asphaltenes resulting in their
early precipitation, thereby producing an unstable fuel oil. In
some cases, decreased selectivity toward mid-distillates and
deteriorated stability have also been reported.77
The above discussion suggests that the performance of the
soaker can be improved by reducing the extent of backmixing
in the liquid phase. Since the major reason for backmixing is
the gross liquid circulation (eqs 17 and 18), in practice, the
backmixing is reduced by reducing (in an extreme case, by
arresting) the liquid circulation. This is achieved by
sectionalization (Figure 7B) with sieve plates, radial baffles,
etc. Some of the designs77–80 are shown in Figure 8. Recently,
Kumar et al.,81 using RTD experiments for a visbreaker pilot
plant with and without a sectionalized soaker drum (4L
volume) by perforated plates, concluded that the sectionalized
soaker offers several advantages, which are as follows: (i)
reduction in backmixing (by 3-5 times); (ii) increased LPG +
gasoline, gas oil, and distillate yield by 10-16%, 11-19%, and
1-17%, respectively, under an identical temperature gradient;
and (iii) better fuel oil stability.
5.4. High-Conversion Soaker Cracker. The
high-conversion soaker cracking (HSC) process
developed by Toyo Engineering Corporation (TEC)
and Mitsui Kozan Chemicals Ltd. enhances the
conversion of the VR to low-boiling products.82 The
feed for the HSC process includes the refractory
bottoms of the visbreaker unit or the vacuum-
distillation unit and involves the continuous thermal
cracking reactions with simultaneous deep steam
stripping in the countercurrent manner, i.e., the feed
moving from the soaker head and steam moving
from the bottom. Feedstock is heated to temperatures
of 440-460 °C, depending on the desired conversion
in the soaker drum. Typically, low pressure ( ∼ 0.1
MPa) and high feed velocities are chosen as the
operational and design variables to prevent coking on
8972 Ind. Eng. Chem. Res., Vol. 47, No. 23, 2008
coupling of the HSC process with the ROSE process suction due to capillary forces, and (c) the rupture of the film
(supercritical solvent extraction) for enhanced recovery of the at critical thickness.
liquid products along with the deasphalted oil. Lin et al.90 found that increasing the pressure reduces the
surface tension and increases the liquid viscosity and surface
6. Design Considerations drag, φ (the parameter defined as the velocity gradient at the
surface due to the adsorbed layer of the gas), which decreases
6.1. Estimation of Design Parameters. In the visbreaking the film-thinning velocity, thereby reducing the
soaker drum, steam, product gas (C1-C4), and low-boiling bubblecoalescence rate. Further, the reduction in maximum
hydrocarbons (∼typically boiling point (bp) < 350 °C) act stable bubble size, dB,max, occurs as a result of a decrease in the
together as the dispersed gas phase while the liquid gas-liquid surface tension and an increase in the gas density at
(∼typically bp > 350 °C) forms the continuous phase. elevated pressure. As a consequence of these phenomena,
Fractional gas holdup is an important design parameter, which higher gas holdup was observed at higher pressure. The radial
determines the effective volume of the reactor available for the gas holdup profile is given by
liquid-phase thermal cracking reactions. The single-stage flash
calculations (ASPEN Plus software version 10.2) for different G
vacuum residues under varying severity (soaker temperature ) G (r ⁄ R ′ )m }
400-440 °C, pressure ) 1.2-1.7 MPa) have revealed the
molecular weight of the vapor phase to be in the range of 20- At atmospheric pressure, the radial gas holdup profile was
100, the liquid-phase surface tension to be in the range of 5-25 found to be parabolic, i.e., holdup profile index, m, varied
mN m-1, and the liquid and vapor viscosities to be in the range from 3.6 to 2.0 with increasing VG from 20 to 180 mm/s,
of 0.1-30 Pa · s and 0.01-0.03 × 10-3 Pa · s, respectively. Such a respectively. However, with increasing the pressure from 0.1
large variation in the properties can be attributed to the to 0.7 MPa, m value increased from 3.6 to 5.4 at VG ) 20 mm s-1
physicochemical characteristics of the feedstock, the extent of and from 2.0 to 3.0 at VG ) 180 mm s-1, indicating flattening of
conversion, the wt % of the steam injected, the operating the
pressure, and the temperature. holdup profiles.95
Recently, the tomographic measurements 75 performed on a The above effects are relatively more significant for the case
1.6 MMTPA commercial soaker visbreaker have shown that of the perforated-plate sparger than that of the sintered-plate
the fractional gas holdup varies from 18% at the bottom to sparger.86 Studies with porous plates as gas distributors show a
28% at the top. The variations in the gas-liquid much smaller or no influence of pressure at low superficial gas
physicochemical properties, column pressure, and superficial velocities on the gas holdup. It has been reported that gas
gas and liquid velocities are found to affect the bubble-size bubbles formed from a porous plate show much interaction
distribution, bubble rise velocities, average gas holdup, radial with each other on the surface of the porous plate, leading to
and axial profiles of gas holdup, and liquid-phase backmixing. coalescence of the bubbles. This in turn was found to have
6.1.1. Gas Holdup in High-Pressure Bubble Column. A partial or total disguise of the effect of pressure on the
systematic investigation of gas holdup structure in bubble bubbleformation process and, thus, on the gas holdup.
columns has been reviewed by Joshi et al. 83 The studies Moreover, the magnitude of the pressure in the case of
pertaining to the effect of gas density and pressure on the noncoalescing liquids was found to be higher than that for
average gas holdup in a bubble column are summarized in coalescing liquids at elevated pressures for identical design
Table 4. On the basis of these comprehensive studies, a brief and operating conditions.
review has been given below. Wilkinson et al.,87 Reilly et al.,88 Letzel et al.,91 Krishna et
The gas holdup was found to increase with an increase in al.,96 and Yang and Fan97 have described the influence of gas
the operating pressure.84–92 This effect of pressure has been density on the regime of operation. Experimental analysis
attributed to the vapor compression leading to the increase in showed that the increased gas density stabilized the regime of
the gas density, which increases the gas-phase momentum. homogeneous bubble flow and, consequently, delayed the
Therefore, at the gas distributor, because of the higher upward transition to the churn-turbulent flow regime. Thus, the
force of the gas, the bubbles detach in their earlier growth transition superficial gas velocity was found to be a unique
stage, resulting in reduction in their size. 86 function of the gas density. Moreover, Wilkinson et al. 87
reported that the gas holdup is virtually independent of the
The Kelvin-Helmholtz stability analysis reveals that the column diameter above 0.15 m, column height-to-diameter
pressure mainly affects the stability of large bubbles, which ratio in excess of 5, and sparger hole diameter larger than 2
tend to break because of the growth of surface instability, mm.
correspondingly reducing the maximum stable bubble size, An insignificant effect of superficial liquid velocity on the
dB,max.93 As a result, the bubble-size distribution becomes gas holdup at elevated pressures has been reported by Yang
narrower and their rise velocities are reduced, resulting in the and Fan,97 indicating that the net liquid motion is negligible as
shift of the regime transition point to higher superficial compared with the bulk liquid motion and, hence, has little
transition velocities, Vtrans, of the vapor phase. effect on bubble characteristics. The influence of elevated
The variation in the interfacial fluid properties as a result of pressure on the gas holdup for low- and high-viscous systems
elevated pressure was found to affect the bubble formation, has been reported by Urseanu et al. 98 For the low-viscous
bubble coalescence, and bubble breakup rates. 89,94 The bubbles system, increasing the pressure of the system resulted in the
during their passage undergo frequent coalescence that expansion of the dense/gas phase. Therefore, overall holdup
involves (a) the approach of two bubbles to form a thin liquid was found to increase considerably. However, for high-viscous
film between them, (b) the thinning of the liquid film by the systems, the effect shown by high-pressure operation was less
drainage of the liquid under the influence of gravity and prominent, since the dense/gas phase was almost inexistent at
 Ind. Eng. Chem. Res., Vol. 47, No. 23, 2008 8973
low gas velocities, resulting in a less prominent effect of
pressure on overall gas holdup.
Gandhi et al.99 have analyzed all the published data over the
range 0.045 e D e 5.5 m, 0.0003 e VG e 0.72 m/s, 100 e P e
15600 kPa, and 284 e T e 538 K. They have developed a
predictive procedure using support vector machines.
6.1.2. Gas Holdup in Sectionalized Bubble Column. It has
been mentioned earlier that the value of εG increases along the
8974 Ind. Eng. Chem. Res., Vol. 47, No. 23, 2008

Table 4. Summary of Gas Holdup Studies in High-Pressure Bubble Column

author system geometrical details
Tarmy et al.84 air-water DC ) 0.024-0.61 m L )
N2 heptane 1.2-8.5 m
coal liquid sparger ) single and multiple
(continuous nozzle, bubble cap
)
Clark94 N2 water DC ) 0.075 m
methanol L)3m
(semibatch) sparger ) sintered metal distributor
dp ) 60 µm

Oyevaar et al.86

N2 DC ) 0.081 m
water sparger ) perforated plate, dO)
aqueous DEA solution + 0.4 mm
antifoaming agent sintered plate
(semibatch) dp ) 30 and 100 µm

Krishna et al.96
N2, He, CO2, Ar, SF6 deionized DC ) 0.16, 0.19 m L )
water, monoethylene glycol, 1.2, 4 m
n-butanol, sparger ) ring type, dO) 2 mm, n
turpentine ) 37 dO) 10 mm, n )
(semibatch) 19

Wilkinson et al.87 N2, He, CO2, Ar, SF6 n- DC ) 0.16, 0.23 m L )

heptane, 1.2 m
monoethylene glycol, sparger ) ring type, dO ) 2 mm, n )
water (semibatch) 37 dO ) 10 mm, n ) 19
Reilly et al.88 N2, He, air, Ar, CO2 DC ) 0.15, 0.23 m
ISOPAR G, H ) 1.2 m
ISOPAR M, X-type sparger
TCE, Varsol DX 3139, dO ) 0.5 mm, n ) 64
water (semibatch)

Luo et al.89 N2-Paratherm-solids

(continuous) DC ) 0.058 m L )
0.8 m
sparger ) perforated plate
dO ) 2.4 mm, n ) 37 square pitch dp )
2.1 and 3 mm

Lin et al.90 N2-Paratherm DC ) 0.0508, 0.10 m L )

(semibatch) 0.8, 1.58 m
sparger ) ring type dO )
3 mm

DC ) 0.15 m H )
1.22 m
Letzel et al.91 N2-demineralized water sparger ) perforated plate dO )
(semibatch) 0.5 mm, n ) 200
operating variables P ) 0.12-7 MPa

VG ) 0.006-0.2 m s-1 -1 T ) 25-450 °C

VL ) 0.0015-0.15 m s
 Ind. Eng. Chem. Res., Vol. 47, No. 23, 2008 8975

FL) 685-850 kg m-3 1. Increasing the pressure delayed the

σ ) 0.005-0.02 N m-1 µL ) 3 × 10-4-4 ×1 regime transition 2. Significant increase
10-4 Pa s in jεG with increase in gas density 3. At
high pressure, DL values were 2.5 times
) -0.06 m s- smaller than those observed at
10.1 MPa atmospheric conditions
T ) 25-450 °C FL) 500-1000 1. The high jεG values were
kg m-3 dp ) 45-63 µm attributed to the reduced bubble
coalescence 2. Two distinct regimes of
operation were observed, with lower half
of column in the bubbly regime while
the upper half showed churn turbulent
regime over studied VG 3. The
homogeneous regime showed 40%
higher jεG than churn turbulent region 4.
The addition of solids increased bubble
VG ) 0.01-0.10 m s-1 coalescence and reduced jεG 1. jεG
P ) 0.1-10.1 MPa increased with pressure both for water
and aqueous DEA
solution
2. Relative increase in jεG was
found smaller than with aqueous DEA
solution 3. The reduced effect of
pressure on jεG observed in the case of
porous plate was due to the bubble
coalescence in the vicinity of the
distributor surface 4. The interfacial
area, a, increased with increase in
pressure, and relative increase in a was
VG ) 0.02-0.18 m s-1, smaller at lower VG
0.03-0.33 m s-1 for DGDT 1. Increasing the pressure
P ) 0.1-2 MPa delayedthe regime transition
2. Vtrans was found to be a
unique function of gas density and liquid
properties
3. In churn turbulent regime,
VG ) 0.006-0.2 m s-1 thehigher values of jεG were attributed to
P ) 0.1-2 MPa FL ) 684-1113 bimodal bubble-size distribution 1. The
kg m-3 column dimensions had negligible effect
σ ) 0.02 -0.072 N m-1 µL 0.041 on jεG when DC g
× 10-3-1 × 10-3 Pa ) 0.15 m, L/DC > 5, dO > 2 mm
s
VG) 0.006-0.23 m s-1
P ) 0.1-2 MPa FL ) 740-1462 1. Increasing the gas density increased
kg m-3 the stability of the bubbly flow regime
σ ) 0.0235-0.0728 Nm-1 µL ) 0.575 × and Vtrans 2. In the case of the
10-3-2.433 × homogeneous regime, jεG ∝ M, and for
10-3 Pa s the heterogeneous regime, jεG ∝ M1/3,
where M ) specific gas-phase momentum
given as M ) FGVG/(1 - εG)FL
1. The average bubble size
VG ) 0.006-0.23 m s-1 decreases and the bubble-size
P ) 0.1-15.6 MPa distribution becomes narrower with
FL ) 863 kg m-3 an increase in pressure 2. The effect
σ ) 0.0285 Nm-1 µL ) 0.018 of pressure on regime transition
Pa s was found significant upto
6 MPa
1. Decrease in dB, reduced
bubble-size distribution, and bubble-rise
VG ) 0-0.3 m s-1 velocity contribute to increased εG with
P ) 0.1-21 MPa T ) 27, increasing pressure
47, 78 °C 2. An increase in gas-
phasemomentum and the decrease in the
surface tension, as a result of increase in
pressure, increased jεG while the increase
in liquid viscosity decreased jεG 3. An
increase in the temperature increased jεG
1. Increase in the pressure delayed the
regime transition and decreased the rise
velocity, VB, of large gas bubbles 2. The
VG ) 0-0.4 m s-1 rise velocity of large bubbles was found
P ) 0.1-1.3 MPa to be inversely proportional to the square
remarks root of gas density
8976 Ind. Eng. Chem. Res., Vol. 47, No. 23, 2008

author system geometrical details operating variables remarks

Yang and Fan97 N2-water DC ) 0.051, 0.10 m H ) VG ) 0-0.4 m s-1 VL ) 0- 1. Effect of liquid velocity on jεG was found to be insignificant
N2-Paratherm 0.56, 0.92 m 0.01 m s-1 2. Axial dispersion coefficient decreased with an increase in the
(continuous) sparger ) perforated plate pressure
P ) 0.1-10.3 MPa
dO ) 1.5 mm, n ) 120 FL ) 870 kg m-3
σ ) 0.029 N m-1 µL )
0.028 Pa s
Table 4. Continued
length of the column because of trays results in the redistribution
the formation of the gaseous of gas bubbles, thereby reducing
products. The gas holdup has the overall bubble size.
direct and indirect implications. Therefore, an increase in the
Directly, it decides the fraction bubble residence time, a smaller
of the liquid phase and, hence, average bubble diameter, and a
the reactor volume available for decreased bubble-coalescence
the visbreaking reactions. rate together contribute to a
Indirectly, the value of εG is not higher average gas holdup. For
uniform even in the radial the same percent free area, the
direction. Because of this, gas holdup was found to
pressure gradients are created in increase with decreasing the
the radial direction, which, in hole diameter, while for the
turn, generates intense liquid same hole diameter, increasing
circulation in the soaker. The the percent free area of the
resulting liquid circulation time sectionalizing plate was found to
is an order of magnitude lower decrease the overall gas
than the residence time, and holdup.78,100–107 The formation of
hence, the liquid phase tends to gas pockets below the trays was
be completely backmixed. Thus, also found to contribute to the
in order to reduce the intensity overall gas holdup when VG >
of the effects of the radial and 0.03 m s-1. However, this effect
the axial profiles of εG, was found to be insignificant
sectionalization is beneficial. when the column diameter
The role of sectionalization on exceeds 0.38 m.108 The
the liquid phase is discussed in partitioning-plate hole diameter
Section 6.1.4, and the effect of was found to show a more
liquid-phase mixing on the prominent effect on the gas
visbreaking reaction is described holdup than the percent free area
in Section 6.2.2. of the sieve plate. In
In the sectionalized soaker, contradictory results, Kemoun et
the overall gas holdup is al.109 found that the overall gas
affected by the geometry of the holdup remained identical for
internals, the process severity, tray and simple bubble columns;
and the mode of operation. The however, the radial density
investigations with regard to the gradients (∆εG) were found to
gas holdup in a sectionalized decrease by incorporating the
bubble column are summarized internals. The average gas
in Table 5. The research efforts holdup showed a weak
presented by different dependency on the liquid
investigators are summarized superficial velocity and the
below. geometry of the gas distributor.
In a multistage bubble 6.1.3. Liquid-Phase Mixing
column, each section acts as a and Axial Mixing in High-
spargerdominated region where Pressure Bubble Column.
the bubble size is primarily Soaker technology offers
governed by the hole diameter advantages of lower heater
of the partitioning plate. The duties, lower capital investment,
higher gas holdup observed in increased run lengths, and low
these reactors is possibly due to susceptibility to operational
the reduction in the liquid-phase upsets.43,74,110 However, the
circulation velocities within the liquid-phase backmixing
stage. In addition, higher energy occurring in the soaker often
dissipation across the perforated diminishes the process
 Ind. Eng. Chem. Res., Vol. 47, No. 23, 2008 8977

performance. The detrimental value of DL increases with

effect arising because of increasing pressure, especially
backmixing causes overcracking under high gas velocity
of the residue, which affects the conditions (VG > 100 mm/s).
fuel oil stability and reduce the Yang and Fan97 measured
selectivity toward middle liquid axial dispersion
distillates. coefficient at high pressure (up
The liquid- and gas-phase to 10.3 MPa) in 50.8 and 101.6
backmixing have been mm i.d. bubble columns. The
extensively investigated in the superficial liquid and gas
literature. However, it may be velocities were varied in the
pointed out that most of the range of 3.4-10 and 20-400
studies have been carried out at mm/s, respectively. They have
atmospheric pressure and used heat as a tracer to measure
especially for the air-water the thermal dispersion
system, whereas the coefficient. They have found
investigations under high- that the axial dispersion
pressure conditions are very coefficient decreases with an
scarce. Holcombe et al.111 increase in pressure. Further,
determined the axial liquid they have also observed that the
dispersion coefficient in a 78 distributor design does not have
mm bubble column with 1.8 m a significant effect on the liquid-
height and a pressure range 0.3- phase mixing.
0.72 MPa. The superficial gas From the foregoing
velocity was varied up to 0.6 discussion, it is clear that
m/s. They used heat as a tracer contrasting observations have
to measure the thermal been made, and these have not
dispersion coefficient. The been rationalized in a systematic
authors found that the effect of manner on the basis of well-
pressure on thermal dispersion understood fluid mechanics in
coefficient is negligible in the bubble columns. The prevailing
pressure range of their study. unsatisfactory state of the art is
Tarmy et al.84 investigated principally because of the
liquid-phase axial mixing in following reasons:
industrial coal liquefaction (1) The most common
reactors using radioactive approach to account for the
tracers. The operating pressure nonideality has been based on a
was varied up to 17 MPa. They one-dimensional (1-D) axial-
found that the values of dispersion model (ADM).
dispersion coefficient increased Because of its simplicity and
with an increase in pressure. ease of use, the ADM has
Houzelot et al.112 investigated remained popular in reactor
the axial dispersion in a 50 mm design, and numerous
i.d. column. The superficial gas correlations for the axial liquid-
velocity was varied up to 6 dispersion coefficient in bubble
mm/s, and pressure was varied columns have been developed
in the range of 0.1-0.3 MPa. over the years. However, the
They have observed an ADM makes an attempt to cover
insignificant effect of pressure all the types of nonidealities
on the axial dispersion without including the
coefficient, which is perhaps due understanding of the detailed
to the limited range of variables fluid mechanics behind each
covered in their work. nonideality.
Wilkinson et al.113 measured (2) Most of the
the axial dispersion coefficient experiments have selected air-
in a batch-type 158 mm i.d. water as the gas-liquid system.
bubble column with nitrogen- The results with organic liquids
water system at pressures at higher temperatures and
between 0.1 and 1.5 MPa using pressures have been
the electrical conductivity contradicting. For instance,
method. It was found that the Wilkinson et al.113 and Panvelkar
8978 Ind. Eng. Chem. Res., Vol. 47, No. 23, 2008

10- m s FL ) 997-1006 kg m-3 σ )

114
et al. have observed an 0.049-0.073 Nm-1
increase in axial-dispersion ) × -3- ×
coefficient (DL) with an increase
author system geometrical details
Voigt and Schugerl100 air DC ) 0.20 m L )
0.5-2% methanol 3.81 m
0.5-1% methanol perforated plates
0.5-2% glucose FA ) 0.32, 0.62, 4.82% dO )
(countercurrent) 0.5, 1, 3 mm
Pandit and Joshi78 air-water DC ) 0.385 L )
(semibatch) 1.35 m
baffle dia ) 0.28 m
HC ) 0.308 m
Kato et al.101 air-water-kerosene (cocurrent DC ) 0.066, 0.122 m L )
upflow) 2, 2.2 m
perforated plates
FA ) 6, 12.8, 28.9% dO )
6.5, 9, 1.2 mm
4 and 8 stages
Nishikawa et al.102 air, N2 tap DC ) 0.10, 0.15 m perforated
water plates
CMC dO ) 3, 5, 7.5, 8 mm
10-60% sugar sol.
(countercurrent)
Chen et al.103 air-water DC ) 0.05, 0.075 m
(cocurrent upflow) L)3m
6 mesh SS wire screen
dO ) 0.58 mm FA )
64%
37 trays
HC ) 0.05 m
Karr type trays
dO ) 12.7 mm FA )
64%
84 trays
HC ) 0.0254 m
Chen and Yang104 air DC ) 0.05, 0.075, 0.15 m
tap water L)3m
2% carboxymethyl cellulose 6 mesh SS wire screen
(cocurrent upflow) dO ) 0.58 mm FA )
64%
38 trays
HC ) 0.05 m
Vinaya105 air DC ) 0.098, 0. 154 m L )
tap water 1.1, 1.8 m
kerosene perforated trays
(countercurrent) dO ) 3, 5, 10, 12 mm
FA ) 10, 12.5, 20, 30, 38,
52%
38 trays
HC ) 0.05, 0.12, 0.2, 0.85 m
Yamada and Goto106 hydrogen-water1,2-dichloroethane, DC ) 0.02 m
carbon S. S. mesh
(cocurrent) dO ) 0.2 mm FA )
73%
15 trays
HC ) 0.015, 0.025, 0.045,
0.075 and 0.225 m
Dreher and Krishna108 air-water DC ) 0.1, 0.15, 0.38 m
(semibatch) L ) 4, 6 m
FA ) 18.6, 30.7%
dO ) 10 mm HC ) 1,
2.04 m
Kemoun et al.109 air-water DC ) 0.2 m FA )
(cocurrent) 5.16% dO )
6.35 mm
FA ) 0.14, 0.68%
operating variables
VG ) 0.03-0.25 m s-1
VG ) 0.012
VL
 Ind. Eng. Chem. Res., Vol. 47, No. 23, 2008 8979

VG ) 0.015-0.13 m s-1 VL ) 0.001-0.01 VS ) 0.115jεG-0.182

m s-1

VG ) 0.01-0.07 m s-1 VL ) 0.0- 0.03 m

s-1 HC ) 0.35, 0.4 m Homogeneous regime: jεG )
2.4(VG2/gdO)0.54(dO/
0.26 -0.22 -0.3 HC) FA (σ/σW)

VG ) 0.0-0.0822 m s-1 VL ) 0.0-0.0612 Heterogeneous regime:

m s-1 jεG ) 0.59(VG2/gdO)0.4(VL2/
0.03 -0.03 -0.41
HC ) 0.05 m
gdO) FA (σ/σW)

jεG ) VG/0.01 + 1.6VG0.8

VG ) 0-0.3 m s-1
VL ) not specified The role of gas pockets on jεG becomes
negligible when column diameter
exceeds
0.38 m

VG ) 0.0042-0.116 m s-1
VL ) 0.007-0.012 m s-1

VG ) 0.01 Vwater ) 5.3 × 10

-1.6 ×
-4 -1

10 ms
Vwater ) 5.3 × 10-5 m s-1

VG ) 0.05-0.40 m s-1

VG ) 0.016-0.167 m s-1
VL ) 0.0042-0.0142 m s-1
conclusions
Reduction in dO and % FA increased jεG

The gas holdup increased in the

presence of baffles

jεG ) VG/30 + 3.3VG(0.8)

experimental data

For Karr plates: jεG ) 0.0448VG0.81VL-

0.055 For screen plates: jεG )

0.0416VG1.23VL-0.092
8980 Ind. Eng. Chem. Res., Vol. 47, No. 23, 2008

The trayed and simple bubble columns

showed identical jεG; however, ∆εG
decreased in the presence of trays.

in pressure. However, Yang and Fan 97 reported exactly the the liquid circulation velocities are lower in the presence of
reverse. In order to rationalize the above-mentioned apparent ethanol (Figure 9). The same effect can be seen for the N 2-
discrepancies, Lorenz et al.115 performed a detailed 1butanol system. The holdup profile is flatter (C0 ) 1.99) as
computational fluid dynamics (CFD) simulation of flow
compared to that for the N2-water system.
pattern and axial mixing. A brief summary is given below:
The flow pattern depends upon the superficial gas velocity With regard to the temperature, it has been shown that, as
(VG), the column diameter (D), and the nature of the gas-liquid the temperature increases from 298 to 323 K, the liquid
system. The simulations have been carried out for three velocity and holdup increase at a constant pressure of 0.1
MPa. The CFD simulations have shown that the value of VS is
different gas-liquid (N2-water, N2-ethanol, and N2-1-butanol)
higher (C1 ) 0.319) and the fractional gas holdup (0.174) is
systems, for two different temperatures (298 and 323 K), and
lower at a temperature of 298 K than the values of VS and
for two different pressures (0.1 and 0.5 MPa). The CFD
fractional gas holdup at 323 K (0.285 and 0.195, respectively).
simulations show that, in the presence of ethanol, the values of
Similar observations have been made for the other three
VS are lower (C1 ) 0.251) and the fractional gas holdup values
different gas-liquid systems.
(0.205) are higher than the corresponding values for the N 2- Further, as the pressure increases from 0.1 to 0.5 MPa, the
water system (0.319 and 0.174). Further, the holdup profile is liquid circulation velocity decreases and the holdup increases
flatter (C0 ) at a constant temperature of 298 K. In addition, it was found
2.22) than that in the N2-water system (C0 ) 2.37). Therefore,
 Ind. Eng. Chem. Res., Vol. 47, No. 23, 2008 8981

velocity (VC) and eddy diffusivity (νt) were found to be 0.38

m/s and 0.50 × 10-3 m2/s, respectively. The corresponding
values at 0.5 MPa

Figure 9. Comparison of simulated profiles of fractional gas holdup and

axial liquid velocity for different gas-liquid systems at 0.1 MPa and 298 K
for VG ) 0.096 m/s: (1) water, (2) ethanol, and (3) 1-butanol (Lorenz et
al.115).

that the value of VS is higher (C1 ) 0.319) and the fractional gas
holdup (0.174) is lower at a pressure of 0.1 MPa than the
values of VS (0.285 and 0.284) and fractional gas holdup at 0.5
MPa. In addition, the liquid circulation velocity was found to
increase with an increase in temperature at a constant pressure.
The experimental values of axial-dispersion coefficient (DL)
have been shown in Figure 10. It can be seen that, at all the
values of VG, DL increases with an increase in pressure. In
order to understand this observation, CFD simulations of flow Figure 10. Comparison between the simulations and the experimental
pattern were performed at both pressures (0.1 and 0.5 MPa) for results of axial-dispersion coefficient (A) water, (B) ethanol, and (C) 1-
the three systems and all the superficial velocities covered in butanol: (1) 0.1 MPa, 298 K; (2) 0.1 MPA, 323 K; (3) 0.5 MPa, 298 K;
and (4) 0.5 MPa, 323 K (Lorenz et al.115).
the work. Figure 10 shows an excellent agreement between the
experimental and predicted values of DL. It may be pointed out
pressure were found to be 0.36 m/s and 0.43 × 10-3 m2/s,
that, in all the cases, the values of average liquid circulation
velocity (VC) as well as eddy diffusivity (νt) decrease with an respectively. For understanding the parametric sensitivity of
increase in pressure. However, in spite of reduction in VC and VC and νt only one parameter was varied from their base values
νt, the values of DL increase in all the cases. Authors have
made an attempt to understand the reasons for the observed of 0.38 m/s and 0.50 × 10-3 m2/s, respectively. The sensitivity
increase in DL with pressure in spite of the reduction in VC and of VC is shown in Figure 11, and that of νt is shown in Figure
νt. For this purpose, a systematic analysis has been carried out 12. From Figure 11, it can be seen that, at a constant νt, the
on the parametric sensitivity of VC and νt on the axial-
dispersion coefficient. For instance, for the case of the N 2- value of DL increases with an increase in VC. From Figure 12, it
water system, ambient temperature and pressure at VG ) 0.104 can be seen that the value of the axial-dispersion coefficient
m/s and atmospheric pressure, the average liquid circulation
8982 Ind. Eng. Chem. Res., Vol. 47, No. 23, 2008

increases with a decrease in eddy diffusivity up to 0.2 × 10-3 θmix ∝ tC (20)

2 -3 -3 2
m /s (from 0.5 × 10 to 0.2 × 10 m /s) of the base case, and a where tC is the circulation time given by
further reduction in eddy diffusivity results in a decrease in the
2HD + DC
axialdispersion coefficient.
The authors have made an attempt to understand the tC ) VC (21)
probable reasons behind the characteristic role of νt on DL as
shown in Figure 12. When the eddy diffusivity is zero, the For large HD/DC ratio, eqs 20 and 21 simplify to
spreading of a tracer is practically zero (or very low because it
occurs by bHD
θmix ) VC (22)

The other extreme occurs at high values of νt, where the

mixing is controlled by eddy diffusivity and the mixing time is
given by
cHD2
θmix ) νt (23)

Thus, in the eddy-diffusivity-controlled regime, the

denominator indicates reduction in mixing time with an
increase in νt. However, the θmix becomes proportional to HD2
as compared to the linear dependence in the circulation-
controlled regime. Thus, in bubble columns where HD/DC
ratios are high, the conversion of dependence from HD to HD2
results in an increase in θmix with an increase in νt. The above-
Figure 11. Effect of variation of convection (with 100% eddy viscosity) mentioned qualitative description has been quantitatively
on axial-dispersion coefficient (Lorenz et al.115). explained by Lorenz et al.115
The foregoing discussion provides an explanation for the
effect of pressure on DL. For instance, at a VG of 0.104 m/s,
when the pressure is increased from 0.1 to 0.5 MPa, the values
of VC and νt decrease by 3.78% and 14.59%, respectively. As a
result, from Figures 11 and 12, it can be seen that the
corresponding effect on DL is -6.89% and +15.47%,
respectively. The combined effect of VC and νt is a 16.63%
increase in DL from 0.0086 to 0.0101 m 2/s, which compares
favorably with the experimental value of 0.01 m 2/s. Similar
exercises were performed for various gas-liquid systems,
temperatures, and VG, and the results are given in Figure 10.
All these results quantitatively explain the reasons for the
increase in DL with an increase in pressure.
A comparison between the experimental observations of
Wilkinson et al.113 and the CFD predictions is shown in Figure
13. In all the cases, the value of DL was found to increase with
an increase in pressure. Figure 13 shows an excellent
Figure 12. Effect of variation of eddy viscosity (with 100% convection)
on axial-dispersion coefficient (Lorenz et al.115).
agreement between the experimental observations and the
CFD predictions. For the CFD model developed by Lorenz et
al.,115 the experimental observations of Yang and Fan 97 provide
molecular diffusivity) at any intensity of bulk mean a critical test. In this set of experiments, the values of DL
circulation. With only bulk convective motion, the added always decreased with an increase in pressure. The CFD
tracer keeps on circulating and the mixing time is very high or predictions agree fairly well with the experimental
DL is very low. By incorporating some eddy diffusion, mixing observations of Yang and Fan97 (Figure 14) that the dispersion
time reduces (and DL increases). However, the mixing time coefficient decreases with an increase in pressure.
attains a minimum at a νt value of 0.202 × 10-3 m2/s, and a The fundamental difference between the two sets of
further increase in νt results in an increase in mixing time (or a experiments (Wilkinson et al.113 and Lorenz et al.115) and the
reduction in DL). results of Yang and Fan97 is that, for the corresponding gas-
When we talk of axial dispersion, it is the combined effect liquid systems under consideration, the gas holdup (jεG)
of VC and νt in the axial direction. Thus, we can imagine two increase with respect to pressure is different. In order to
extreme situations. At low values of νt, the mixing is understand the parametric sensitivity of jεG, CFD simulations
controlled by the bulk circulation, and a typical relationship is were performed under otherwise identical conditions of VG, D,
given by HD, etc. The results are shown in Figure 15. It can be seen that
 Ind. Eng. Chem. Res., Vol. 47, No. 23, 2008 8983

an increase in jεG initially results in a decrease in DL up to a

certain critical

Figure 15. Sensitivity of liquid-phase axial-dispersion coefficient on gas

holdup (Lorenz et al.115).
Figure 13. Comparison between the simulations and the experimental data
of Wilkinson et al.113 of axial-dispersion coefficient: (1) 0.1 MPa, (2) 0.5
MPa, and (3) 1 MPa (Lorenz et al.115).
value of jεG and a minimum DL. A further increase in jεG results
in an increase in DL. The important point is the actual jεG-VG
relationship that decides the nature of the holdup profile (C0)
and the bubble-rise velocity (C1), which in turn decides the
flow pattern (VC and νt) and, hence, the DL. On the basis of
these arguments, it can be said that the actual effect of any
parameter (such as P, T, HD, D, sparger design, µ, σ, nature of
gas-liquid system) depends upon its contribution to the change
in jεG and, hence, on C0, C1, and the flow pattern. In order to
confirm this

Figure 16. Parity plot for axial-dispersion coefficient. Lorenz et al. 115
work: ], water; ∆, ethanol; 0, 1-butanol; 2, Wilkinson et al.; 113 and Yang
and Fan:97 O, 50.8 mm i.d.; b, 101.6 mm i.d. (Lorenz et al.115).

reasoning, CFD simulations were carried out over a wide

range of abovementioned parameters covered in the published
Figure 14. Comparison between the simulations and the experimental literature. The results are shown in Figure 16, and the parity
dispersion coefficients of Yang and Fan.97 (A) 50.8 mm i.d. column: (1) ],
0.1 MPa; (2) 0, 1.5 MPa; (3) ∆, 2.8 MPa; and (4) , 4.2 MPa. (B) 101.6 mm
can be seen to be excellent.
i.d. column: (1) ], 0.1 MPa; (2) ∆, 0.8 MPa; (3) 0, 1.5 MPa; and (4) , 2.8 6.1.4. Liquid-Phase Mixing and Axial Mixing in
MPa (Lorenz et al.115). Sectionalized Bubble Column. In a bubble-column reactor,
the agitating action of the rising gas bubbles generates
interactingor noninteracting-type multiple circulation cells. 116
The interstage exchange velocity, Vex, between the adjacent
cells is responsible for the net transport of the liquid from one
cell to another. The inclusion of baffle plates results in an
isolation of these circulation cells, thereby reducing the
8984 Ind. Eng. Chem. Res., Vol. 47, No. 23, 2008

exchange velocity and increasing the theoretical number of velocity and found to be independent of column diameter.
stages. Thus, the liquidphase flow in the soaker approaches Recently, Kumar et al. 81 performed radiotracer RTD studies on
plug type. The exchange velocity or the effective liquid-phase 0.1 m i.d. soaker visbreaker incorporated with and without
Peclet number was found to be a function of superficial gas perforated plates and operated at elevated pressure (1.2 MPa).
and liquid velocities, as well as percent free area and hole The radiation count data were analyzed using two-parameter
diameter of the sectionalizing plate.80,108,117 For a cascaded axial dispersion and tanks-in-series with backmixing type
bubble column, Blass and Cornelius 117 reported the following models. The study revealed that, in the presence of internals,
correlation to predict the effective number of stages: the liquid-phase backmixing can be reduced by a factor of 3.

( )
From the foregoing discussion, we summarize the
liquidphase axial mixing as follows. In a bubble column, the
eff ) 1.061 G (24) N V FA relationship between the liquid-phase dispersion coefficient
-0.19
(DL)
N VLhole2 and the number of tanks-in-series is given by

The correlation is applicable over Neff ) Pe(0.5 + f) (27a)

(
where

VGFA ) Pe ) VLHD ⁄ DL (27b)

and f is the ratio of backward flow to net flow. The value of f
can be controlled by manipulating the design of sieve plates
3 -1
used for sectionalization. The value of f decreases with a
1.3 <<10 (m s); 0.044 < FA < 0.36 (25) decrease in free area, and in a limiting case of no weeping
V condition of the sieve plates, f becomes zero. Under this
2 Lhole

Palaskar et al.80 performed RTD studies on 0.062 and 0.2 m condition, the net liquid flow across the plate is managed by
i.d. cold-flow sectionalized bubble columns, assuming it to be providing riser/downcomer depending upon the direction of
used for visbreaking application. The following conclusions net flow. For the intermediate case of finite weeping, eqs 24
can be drawn from their studies: and 26a can be used.
(1) The extent of liquid-phase backmixing can be 6.1.5. Weeping. The bubble coalescence occurring below
regulated using perforated plates. the sieve trays is a statistical process during which the gas-
liquid turbulent dispersion frequently reaches the plate and
(2) The liquid-phase dispersion coefficient was found to
transfers the liquid to the next stage. In order to satisfy the
be proportional to VL0.45-0.85 and VG0.12.
continuity, the liquid from the upper section is also transferred
(3) The introduction of gas phase by split mode of to the lower section by the weeping mechanism. The undesired
operation reduced the liquid-phase dispersion by 50-80% at phenomena of weeping in a sectionalized bubble column can
lower liquid superficial velocities. be explained on the basis of formation of a single bubble at the
(4) The liquid-phase dispersion in a large diameter orifice, wherein the buoyancy and surface tension forces act
column (0.2 m i.d.) was found to be 1.3-1.6 times higher than upon the bubble. The bubble is detached from the orifice when
that in a small diameter column (0.062 m i.d.). buoyancy just exceeds the surface tension force. As a result,
Further, the authors have correlated liquid-phase dispersion the upstream pressure reduces momentarily, and in order to
as a function of flow parameter as counterbalance this pressure, the liquid above the plate starts

( lnDL ) ( ) ) DCi -0.2 [ 2 + 0.2658{ln(FP)} -

weeping. Since the gas flow rate is continuous, the upstream
pressure is immediately recovered and, thus, the weeping
process continues over an entire cross section in a statistical
0.023{ln(FP) } manner. The formation of gas pockets are undesirable as they
VLL DC decrease gas-liquid interfacial area and reduce the effective
reactor volume. Thus, the selection of sectionalizing plate hole
0.816(26a)
diameter becomes an important criterion for designing sieve
where flow parameter FP is defined as follows: plates.
For a commercial-scale coal and residue hydrogenation
(VG)(% Free Area) process, Bakopoulos118 suggested the optimum range of hole
FP ) (26b) diameter and free area to be 50-80 mm and 1-4%, respectively.
VL Such a large hole diameter facilitates unidirectional flow of
gas bubbles without gas-phase separation below the sieve
The applicability of this correlation is largely limited by the
plates for superficial velocities greater than 0.08 m s -1. Further,
regime of operation and lower values of gas and liquid
for successful experimental simulation of a high-pressure
superficial velocities. Dreher and Krishna108 carried out
slurry
liquidphase dispersion studies in three different sizes of bubble
bubble column, the author has recommended the fluid dynamic
columns (0.1, 0.15, and 0.38 m i.d.) operated in semibatch
similarity between the following dimensionless groups for
mode. The axial-dispersion coefficient was found to increase
coldflow and industrial process conditions,
with increasing column diameter and superficial gas velocity.
The exchange velocity, Vex, was found to increase with
increasing percent free area of sieve plate and superficial gas F G VG g(FSL- FG)VG(ΩµSL)4
 Ind. Eng. Chem. Res., Vol. 47, No. 23, 2008 8985

FSL;VBM; MoSL ) FSL2σ3 (28)

where MoSL is the modified Morton number.119

The influence of solids concentration on macroscopic slurry
viscosity is given as follows:

ln Ω ) 4.6εS{5.7εS0.58 sinh[-0.71 exp(-5.8εS)ln(Mo)0.22] + 1}

(29)

For dP < 10 µm or nonsettling slurries, the macroscopic

viscosity is estimated as follows:
Figure 17. Five-lump, seven-rate parameter series-parallel kinetic model

( )
(Kataria et al.38).

µSL)µL 1 + S FL LεS (30) F + F

distillates. For an optimum performance, θmix should be
6.1.6. Methodology for Predicting Reactor Performance. comparable with τmean, and this can be achieved using suitable
It may be pointed out that scarce information is available with internals.81
regard to the adequate quantification of gas holdup and mixing 6.2. Mathematical Modeling. 6.2.1. Introduction. The coil
time in industrial-scale visbreakers. In the absence of any and the soaker units of a visbreaker can be broadly classified
design data, the preliminary estimation of gas holdup and as G-L systems wherein the generated gas, vaporized
liquid mixing time in industrial visbreakers can be performed lowboiling hydrocarbons, and superheated steam together act
by adopting the following methodology. as a dispersed phase, while the visbroken residue acts as a
(1) For a given capacity, the information related to the continuous (liquid) phase. Thus, the former can be represented
product yields can be made available from the respective as a twophase pipe-flow reactor, while the latter can be
refinery. represented as a bubble-column reactor. It is well-known that
(2) The assay data of each of the psuedo-boiling the visbreaking reactions proceed via free radical mechanism
fractions (TBP curve, specific gravity, etc.) and their and preferentially take place in the liquid phase. The extent of
respective yields including the steam wt %, operating conversion occurring in the coil as well as in the soaker drum
temperature, and pressure are given as an input to the mainly depends on the process severity and the quality of the
equilibrium flash module in the ASPEN or similar simulation feedstock. The process temperature influences the cracking-
engine. reaction rate constants, whereas the feed flow rate decides the
(3) The flash calculations can be performed using liquid-phase residence time in the coil and the soaker. The
property method such as Peng-Robinson or SRK equations of soaker-outlet pressure has an indirect effect on the liquid-
state (EOS), while for the prediction of transport properties, phase residence time. Increasing the pressure suppresses the
the respective methods have been reported by Filho and vaporization of low-boiling fractions, thus increasing the
Sugaya.70 (4) From the obtained volumetric flow rates, the liquid-phase holdup and, consequently, the liquid-phase
mean superficial gas and liquid velocities are calculated and residence time (τL). In industrial visbreakers, the superficial
correspondingly can be used in eqs 22-27 for calculation of gas velocity (VG) is typically 2-5 times that of the superficial
average gas holdup in the soaker. In the case of coil, the liquid velocity (VL), and this proportion may change with the
average gas holdup can be estimated using the correlation of extent of residue conversion, the operating pressure, and the
Hughmark.120 Thus, mean liquid-phase residence time or the quantity of steam injected in the coil. This leads to a relative
effective volumes of the coil and the soaker can be evaluated. slip condition between the liquid and vapor phases in the coil
(5) The liquid-phase dispersion coefficient, DL, can be as well as in the soaker. Because of the associated slip, the
estimated using eqs 17 and 18 with VB∞ ) 0.09 m s-1.84 From the thermodynamically calculated gas holdup significantly differs
knowledge of DL, the mixing time can be calculated as from the one calculated on the basis of the slip velocity (VS).
follows: Thus, the regime of operation may change with the extent of
the residue conversion, i.e., in the case of the coil, the regime
0.5HD2 may change from dispersed bubble to slug or annular type,
while for the soaker, it may change from homogeneous to
θmix ) DL (31) transition or heterogeneous type. Therefore, the model should
take into account these effects so as to predict the liquid-phase
The knowledge of mixing time gives us an idea of soaker residence time precisely.
behavior at respective operating conditions. If θmix , τmean, then The earlier studies on heavy oil processing included the
the soaker can be said to be fully backmixed and the fuel oil sevenlump kinetic scheme for residue hydrotreating proposed
produced will be of inferior quality. Conversely, if θmix > τmean, by Mosby et al.121 in which each lump was an industrially
then the reaction is said to be partially completed, thereby important boiling fraction, viz. gas, naphtha, mid-distillates,
reducing the overall conversions and selectivity toward the gas oil, and hard-to-crack and easy-to-crack vacuum residues.
mid- For the evaluation of kinetic constants, the authors assumed
that each hydrotreating unit behaved as a completely
backmixed reactor. The model predictions were found to be in
8986 Ind. Eng. Chem. Res., Vol. 47, No. 23, 2008

close agreement with the experimental yields. Further, a more of the pilot-plant data and predicted product concentrations for
detailed kinetic model for an industrial Dubb’s unit was each of the feeds. These modified rate constants can then be
proposed by Castellanos et al.66 In their approach, the feed and used for the estimation of
the product hydrocarbons were represented by a set of discrete industrial visbreaker product yields.
pseudo-components, characterized by their true boiling point, 6.2.2. Mathematical Model Development. The coil and the
API gravity, and molecular weight. The pseudo-components soaker can be schematically represented as a PFR coupled with
were categorized according to the number of carbon atoms j equal volume CSTRs in series superimposed with interstage
present in these compounds. The coil was considered as a backmixing (Figure 18). The backmixing coefficient was
plug-flow reactor, and the end effects were neglected. The defined as the fraction of the liquid-phase residue recycled
thermophysical properties, heat transfer coefficients, and feed back and is given as f ) m˙b/m˙f. The backmixing ratio was
and product enthalpies were also accounted for in the model. assumed to be constant for all the stages.
The activation energies and the frequency factors were The preheater and the reaction coil were heated externally
correlated as a function of molecular weight of the reacting by immersing them in the molten salt bath maintained at the
and product species. The model parameters were optimized desired temperature. The batch reactor studies 38 have shown
using Powell’s method to match the industrial unit data. that the cracking reactions initiate above 400 °C; hence, only
Though the model accurately predicted the refinery yields, the the coils (no. 4, 5, 6) in the reaction zone were considered in
complexity of the model restricted its application to the coil- the model development. To compensate for the heat losses
visbreaking units only. In a more realistic approach for heavy from the surface as well as from the endothermic nature of the
oil hydroconversion in the slurry reactor, Carbonell and cracking reactions, heaters were provided on the outer surface
Guirardello122 combined the zero-order turbulence model with of the soaker vessel, unlike industrial soakers, which are
Mosby’s lumped-reaction scheme. The simulations were adiabatic reactors. The soaker inlet and outlet temperatures
performed in two steps: in the first step, the hydrodynamics were controlled by changing the electrical power input
was established by calculating the pressure drop in the bed, the delivered to these heaters. Since the information regarding the
radial gas and slurry holdup distribution, the effective eddy exact power delivered to the soaker was unknown, the heat
viscosity, and the liquid-phase velocity profile. In the second added to the soaker was estimated from the inlet and outlet
step, the thermal cracking reactions were superimposed by enthalpies (calculated on the basis of temperature differential
accounting the radial variations in these transport properties so across the soaker and the respective conversion) and the heat
as to predict the heavy oil conversion. Their study revealed of reaction. The heat losses to the surroundings were
that the liquid recirculatory patterns (backmixing) strongly compensated in the heat flux added to the soaker. Kulkarni 125
affect the product yields. However, the validation of the has described the pertinent subject in detail.
experimental data was not performed using these models. In The generalized correlation describing the variation in the
another approach, the coil-soaker was modeled as a plug-flow liquid-phase volumetric flow rate is given as follows:
reactor (PFR) coupled with continuously stirred tank reactors
(CSTRs) in series, and the rate constants were tuned to match
the industrial visbreaker yields.123 The furnace pressure drop

( )
was estimated using the Beggs and Brill 124 correlation. The
model was able to predict the coil pressure drop, the coil- )
soaker temperature profiles, and the residue conversion within QL m˙FLf {a1Ta2} exp(-a3x150 °C-) (32)
a narrow range of accuracy. However, the model was found to

( ){
be feed-specific and accounted for only a single cracking
reaction of VR to gas and gasoline. QG ) FLf b1+ b2(Tx150 °C-) + b3(Tx150°C-)2} (33)
Kulkarni125 extended a similar approach for modeling a
pilotscale coil-soaker visbreaker. The primary objective of this m˙
study was to eliminate the discrepancy between the pilot-plant FLε¯L)(F0- c1T) exp((c2T - c3)x150 °C-) (34)
yields and those predicted using the kinetic constants
estimated from the batch reactor data. This was achieved by The regressed values of the density-holdup correlation
multiplying the kinetic constants with appropriate scaling coefficients for the three feedstocks are listed in Table 6. The
coefficients (R). The coil was considered as a dual plug-flow steadystate dimensionless material balance equation
reactor, while the soaker was modeled as CSTRs in series with representing the material balance for each species can be
interstage backmixing. The model employed five-lump series- written as
parallel kinetic scheme (Figure 17), the radiotracer RTD
information gathered for the soaker equipped with/without dyVR ) - (35)
internals, and the obtained pilotplant visbreaking data. The R FLεLACL dz m˙f(yVR+ yVGO)k1234yVR
values were estimated by performing the least-squares analysis
dyVGO FLεLACL
) (k4yVR- k567yVGO) (36)
dz m˙f(yVR+ yVGO)
dyLGO FLεLACL
) (k3yVR+ k5yVGO) (37)
dz m˙f(yVR+ yVGO)
dyGa FLεLACL
) (k2yVR+ k6yVGO)
dz m˙f(yVR+ yVGO) (38)
Figure 18. Schematic representation of the coil and the soaker as PFR coupled with j equal volumes of CSTRs in series superimposed with interstage
backmixing.
 Ind. Eng. Chem. Res., Vol. 47, No. 23, 2008 8987

dyC FLεLACL temperature drop occurring along the coil length can be
m˙ y y attributed to the endothermic nature of the reaction and the
) (k1yVR+ k7yVGO) (39) dz f( VR+ VGO)
heat loss to the surroundings. Since the coil-outlet and soaker-
The energy balance equation to calculate the dimensionless inlet temperatures were known for this section, a uniform
temperature profile for the coil is given as follows: linear temperature gradient was assumed. The initial
conditions to solve eqs 35-41 are as follows:

[
At z ) 0, yVR ) yVR0, yVGO ) yVGO0, yLGO ) 0, yGa ) 0, yGas )

dψ ) 1 UiAPL - 0, ψ ) 1.0(42)
The coil inside heat transfer coefficient was estimated using

(
the procedure130 described below. For laminar flow, the
following correlations are recommended to compute the heat
-
transfer coefficient.

)
For Gz < 100,

ACLFLεL∆HR -

( )
dz G m˙fTm˙fT0

(Nu)lm ) 3.66 + 1 +0.085Gz 2⁄3 µµwb 0.14 (43)

yiCpi ) For Gz > 100,

0.047Gz

µ 0.14
∑
G dyi
) 1.086G1Z⁄31 µwb ( )
]
(44)
(Nu)am
ψ Cpi dz (40)
For the case of turbulent and transition regime (NRe > 2100)
i)VR

The dimensionless temperature is given as ψ ) T/T0 where T0 is

the coil inlet temperature. The rate constants are written as
k1234 ) k1 + k2 + k3 + k4 and k567 ) k5 + k6 + k7, and ki )
Nu ) 0.0116(ReM2⁄3 - 125) PrM2⁄3 1 [ ( )] +
DcH
(45) d
RiAi exp(-E/RT) and) m˙sCpsteam/m˙f. 2⁄3

The heat of reaction term is given as follows: The above equations are for a straight tube and are applicable
Table 6. Correlation Coefficients
- 3
∆HR ) [(k1+k2)yVR+ (k6+ k7)yVGO]∆H*R (41) feed F O (kgm ) c1 c2 c3

The heat of reaction of 800 kJ/kg for the products boiling BHVR 1019.9 0.807 -11.1 × 10-3 0.023

below 204 °C was assumed.70 The liquid and vapor viscosities

- 3
were averaged over the studied temperature range. The AMVR 1023.2 0.828 -6.7 × 1010- 3 3.0333.033 VB feed
average specific heats were obtained from Kern 126 and are 1017.6 0.814 -8.5 ×
listed in Table 7. The vapor- and liquid-phase viscosities were
estimated using correlations proposed by Dean and Stiel 127 and
Twu.128 The liquid-phase thermal conductivity values were for a helical coil with the substitution of DH/dC for L/D terms in
obtained from the Graetz number. Thus, the Graetz number is written as
Mallan et al.129 follows:
The reaction coil was divided into two zones, viz. fired d 0.5
section and unfired section. In the first zone, heat transfer
occurs from molten salt bath to the coil to raise and maintain

( )
the temperature of the reaction mixture to a desired value. The
bath temperature was assumed to be 20 °C higher than the Gz
) ReMPrM DCH (46)
coil-outlet temperature. The temperature rise in this zone was
nonlinear and is given by eq 40. The second zone included a The soaker was considered as an axially dispersed reactor
transfer line connecting the furnace outlet to the soaker. and modeled using a series of equal volume tanks with
Because there was no additional heat input to this section, the interstage backmixing. Each section was considered as a

( )| |
f
yVR |j- 1 - 1 + + y + r VR j +
(1 f)(yVR + yVGO ) VR j
fyVR
 |
( ) | |
- f
yVGO |
j- 1 1+ yVGO + r VGO +
(1 + f)(yVR + yVGO ) j j
8988 Ind. Eng. Chem. Res., Vol. 47, No. 23, 2008
fyVGO
)
CSTR, and the exchange mass flow rate (m˙b) was assumed to
be constant. The information about the number of theoretical
stages (N) and the backmixing coefficient (f) for a soaker
(1 + f)(yVR+ yVGO) | j+1 0(54)
incorporated with or without internals can be obtained from yLGO|j-1- yLGO|j + rVGO|j ) 0 (55)
the actual plant RTD studies or from the published literature
yGa|j-1- yGa|j + rGa|j ) 0 (56)
discussed in Section 6.1.3. For species VR and VGO, the
model was represented by tanks-in-series with backmixing, yG|j-1- yG|j + rG|j ) 0 (57)

)|
while for LGO, Ga, and G fractions, it turns out to be a simple

[
CSTR-in-series type model; because they are in the vapor
phase, the vapor-phase backmixing was found to be yiCpi ψ j-1 -
insignificant. The liquid-phase holdup in each tank was
determined by the drift flux model using the reported values of
C0, Vb∞, and m.84 The material and energy balance equations
for various species are given as follows. For the first tank, G

|
]|
(47)
yVR|0 - (1 + f)(yVR- rVR)|1 + yVR+ yVGO )0
fy
y | - (1 + f)(y -r )| + VGO
)
i pi
(48)

VGO 0 VGO VGO 1 yVR+ yVGO | 2 0 yC

yLGO|0- (1 + f)(yLGO- rLGO)|1 ) 0 (49)
VR yiCpi yVR+yVGO ψ j -
yGa|0- (1 + f)(yGa- rGa)|1 ) 0 (50)
VGO

yG|0- (1 + f)(yG- rG)|1 ) 0 (51)

fyVR

f ∑ yiCpi |
∆Hr|j + yVRi)VR+ yVGOψ j+1 ) 0(58)

For the Nth tank,

( ) | |
G

(1 + f)yVR|N-1 - 1 + (yVR+fyVGO) yVR N + rVR N ) 0 (59)

yiCpi)ψ | -yiCpi ]| | ψ - ∆Hr +

(1 + f)yVGO|N-1 - ( 1 + (yVR+fyVGO) yVGO ) | N + rVGO N ) | 0

i)VR 0 VR 1 1
(60)
VGO (1 + f)yLGO|N-1- yLGO|N + rLGO|N ) 0 (61)
(1 + f)yGa|N-1- yGa|N + rGa|N ) 0 (62)

|
(1 + f)yG|N-1- yG|N + rG|N ) 0 (63)

f ∑ yiCpi
)|
[
yiCpi ψ -
i)VR
yVR+ yVGOψ 2 ) 0(52) N-1

For the jth tank, where 2 e j < (N - 1),

 Ind. Eng. Chem. Res., Vol. 47, No. 23, 2008 8989

] ||
coil length fired section 1000 mm
coil length unfired section 100 mm
soaker i.d. 100 mm
soaker equivalent length 500 mm
2. Average specific heat (kJ/(kg K))
G ipi VR 3.22
VGO 3.46
LGO 3.59
yC gasoline 3.85
gas 3.85
steam 2.06
yiCpi + i)VR
Table 8. Arrhenius Parameters, E/R (K) and A (s-1) Used for Model
i)VR yVR+ yVGO ψ N - ∆Hr N ) 0(64) Simulation
BHVR,AMVR VBfeed
The reaction rate expressions are given as follows: pathway A E /R A E /R
VR f G 1.54 × 10 2.51 38.98
20
3.36 × 10 10
23.65
HCε¯Lk1234yVR 37.94 22.64
VR f Ga
× 1019 6.25 × 30.91
9
7.54 × 10 9.92 16.66
rVR ) - VL (65) VR f LGO
15 15 31.24 × 106 16.52
VR f VGO 10 8.52 × 10
44.68 34.44
VGO f LGO 6.67 × 1023 1.68 53.20 2.85 × 106 51.39
HCε¯L
VGO f Ga 55.53 50.42
rVGO ) VL(k4yVR- k567yVGO) (66) × 1028 4.96 × 1016 1.17
VGO f G
1.29 × 1030 × 1027

HCε¯L 6.18 × 1026

81
to be 2. This high value of the backmixing coefficient
rLGO ) VL(k3yVR+ k5yVGO) (67) suggests that the soaker behavior is closer to that of a
backmixed reactor, and thus, R values were estimated
HCε¯L assuming a case of a single CSTR. The best test for any kinetic
model is the prediction of experiments that were not used for
rGa ) VL(k2yVR+ k6yVGO) (68) the parameter estimation. Hence, using these R values, the
yields were predicted for the case of a soaker incorporated
HCε¯L with internals, and the equivalent number of tanks representing
the observed yields was found out.
rG ) VL(k1yVR+ k7yVGO) (69) 6.2.3. Algorithm for Parameter Estimation.
(1) The number of tanks (N), feed mass flow rate (m˙f),
coilinlet temperature (T0), coil-outlet temperature (COT),
q’πDCHC ∆Hr*
soakerinlet temperature (TS,in), residue and product
∆Hr ) - m˙fT0 + T0 (rGa+ rG) (70) concentrations (yi), and dimensionless soaker-outlet
temperature (Ψ) were read as an input data file in the main
The average gas holdup in each tank can be obtained using the program. Also, the other data input included geometrical
drift flux model of Zuber and Findlay 131 as follows: details of the coil and the soaker, specific heats, coil-inlet
concentrations, and activation energy and frequency factors
(Tables 7 and 8).
VG )C0(VG+ VL) + Vb∞(1 - ε¯G)m (71) (2) The appropriate guess values of scaling coefficients
were provided to initialize the calculation procedure in the
ε¯G
main program.
ε¯L ) 1 - ε¯G (72) (3) Equations 35-40 were integrated numerically using
the fourth-order Runge-Kutta method with the step size equal
where C0 ) 1.0, Vb∞ ) 0.1, and m ) 1.75 to 0.001.
Since the model contains a nonlinear temperature term in the (4) The estimated coil-outlet concentrations and the
form of the Arrhenius relationship, the solution of eqs 47-64 temperature were fed as an input to the soaker. Also, these
becomes highly dependent on the initial guess value. Thus, it values along with the experimental conversion (150 °C-) and
was thought to resolve the model into a simple tanks-in-series
the soakeroutlet temperature provided the heat flux (q′) from
model with backmixing term (f) set equal to zero. For the case
the heating elements to the soaker. The heat energy added per
of a soaker without internals, the backmixing coefficient
stage was assumed to be constant.
varied in the range of 3.74-5.56 and the number of tanks was
found (5) The product concentration data for each experimental
Table 7. Input Data for Model Simulation run served itself as an initial guess for starting the iterations.
(6) The algebraic equations 53-64 were iteratively solved
1. Geometrical details (mm) in the backward direction, i.e., from j) N to j ) 2. Care has been
taken to avoid non-negative values of the variables, and the
coil i.d. 6 mm material balance was checked during each calculation over
helix dia. 17 mm each tank. Similarly, for each iteration, the converged values
8990 Ind. Eng. Chem. Res., Vol. 47, No. 23, 2008

of liquid-phase holdup were obtained using the Newton- (2) The physicochemical properties of the products of
Raphson method. visbreaking, extent of variation in the gas and distillate yield,
Table 9. Optimized Values of Scaling Coefficients coking propensity, and crackability, are highly dependent on
scalingcoefficients BHVR AMVR VBfeed the physicochemical properties of the feedstocks.
R1 0.34 0.55 0.37 (3) The study pertaining to the characteristic property of
R2 0.61 0.89 0.59 a large number of feeds shows that (i) H/C ratio increases with
R3 0.33 0.57 0.24 increasing API gravity, (ii) sulfur and nitrogen contents
R4 0.54 0.44 0.89 decrease
R5 0.36 0.35 0.36 Table 10. Comparison between Experimental and Predicted
R6 0.13 0.13 0.14 Concentrations and Soaker-Outlet Temperatures for BHVR, AMVR,
R7 0.17 0.18 0.15 and VB Feed
SSE 0.06 0.06 0.043
internals VR,wt% VGO,wt% LGO,wt% gas
(7) The convergence criteria defined by eqs 47-52 was
checked. If the absolute value of f(x) defined by eqs 47-52 runno. %FA( L /D C ) exp. pred. exp. pred. exp. pred. exp
were less than 0.0001, then the solutions were said to be Case: BHVR
converged. 14 no internals 92.90 93.19 3.44
(8) If the conditions defined in step 6 were not satisfied, 3.34 2.63 2.50 0.60
0.57 0.43 0.41 400
then the new input values of concentration and temperature
401
were obtained using the Newton-Raphson method. 15 88.83 88.32 5.27 5.60
(9) Steps 5-7 were repeated until the converged profiles 3.99 4.20 1.13 1.07
of concentration and temperature were obtained. 0.79 0.80 408 408
16 81.19 79.98 8.94 9.29
(10) The evaluated product concentrations and the 6.69 7.12 1.75 2.04
dimensionless soaker-outlet temperature were compared with 1.43 1.57 419 417
the pilotplant data. 24 16 (0.5) 91.46 92.85 4.74
(11) Steps 3-10 are repeated for each experimental run, 3.51 2.59 2.61 0.52
0.60 0.68 0.44 395
and the sum of squares of errors function (SSE) defined by the 396
following equation was minimized using the modified 25 84.71 86.97 8.27 6.23
steepestdescent method so as to obtain the optimum values of 4.80 4.67 1.05 1.22
the scaling coefficients. 1.16 0.91 404 405
26 77.46 76.90 11.13 10.70
ψy -y 7.33 8.15 1.91 2.42
2.17 1.85 416 415
SSE )
19 12 (0.5) 91.14 95.08 4.67
(73) 2.44 2.88 1.82 0.69
)1 i)VR ijexp
0.38 0.62 0.27 400
394
The optimized values of scaling coefficients for the three 20 87.79 91.70 6.01 4.06
studied feedstocks, viz. Bombay high vacuum residue 4.18 3.03 0.86 0.70
(BHVR), Arabian mix vacuum residue (AMVR), and VB 1.16 0.51 411 401
Feed, are shown in Table 9. The scaling coefficients were 21 81.43 84.21 8.19 7.50
estimated for soakeroutlet temperature ranging from 395 to 6.77 5.63 1.76 1.51
1.85 1.14 406 411
420 °C. The scaling coefficients were found to be less than 1
23 73.22 73.48 10.12 12.05 10.90
for all three feedstocks. Table 10 gives the comparison 9.25 2.53 2.80 3.23 2.16
between the experimental and predicted concentrations and 416 419
soaker-outlet temperatures for Case: AMVR
BHVR, AMVR, and VB Feed. The model predicted the 18 no internals 89.53 89.25 5.03 5.61 3.99
concentrations and the soaker-outlet temperatures in the 3.72 0.82 0.78 0.64 0.64 400
400
desired proximity.
20 82.24 82.39 7.90 7.75 6.96
6.81 1.65 1.66 1.25 1.38
7. Conclusions and Suggestions for Future Work 410 410
58 75.70 74.41 10.26 10.00 9.02
The overall conclusions and the suggestions for future work 10.33 2.57 2.80 2.45 2.37
may be summarized as follows: 418 417
(1) The trend of processing heavy crudes132 in the 37 16 (0.5) 87.26 83.21 7.05
refineries has brought about a renewed interest in bottom-of- 6.24 3.86 4.61 0.86
the-barrel upgradation. Visbreaking, because of its low capital 1.02 0.97 0.83 401
cost, low operational upsets, and ease in handling residue of 401
38 80.83 80.26 9.63 8.42
any quality, is one of the important residue upgradation
6.09 7.75 1.67 1.95
processes and contributes about 33% 17 of the total residue 1.78 1.62 410 409
processing capacity. Visbreaking provides low-cost 39 71.57 72.36 12.06 10.65
conversion, especially when idle equipment can be modified 10.89 11.09 2.52 3.20
for visbreaking service. Furthermore, visbreaking projects can 2.96 2.73 420 419
be implemented rapidly and a quick payout can be expected. 40 12 (0.5) 87.04 86.78 6.85
6.40 4.10 4.38 0.91
 Ind. Eng. Chem. Res., Vol. 47, No. 23, 2008 8991

1.08 1.10 0.89 402 thus, explain the coke (SDTC) process, which runs
402 formation. The activation at much higher severity
41 79.83 79.33 9.55 8.70 energy of cracking and levels. Distillates are
6.94 8.17 1.78 2.08
1.89 1.74 412 411
coke-formation reactions maximized by incorporating
42 67.74 68.22 14.10 11.90 can range from 22 to 83 Shell Visbreaking Vacuum
11.67 13.01 3.11 3.71 kcal/mol. The large Flashing.21
3.38 3.16 421 420 variation in these values is (12) The production of
Case: VB Feed mainly due to variation in liquid fuels and chemicals
45 no internals 86.26 85.96 8.50 8.29 3.58 the feed characteristics and
46 81.63 81.14 10.08 10.39 5.43 out of synthesis gas is
different lumping making rapid strides
17 72.88 74.40 13.19 13.16 9.12 approaches (different because of the
combinations of everdiminishing crude oil
43 16 (0.5) 83.38 85.11 10.58 8.68 4.08 compounds) used. reserves. The gasification of
44 76.00 79.83 13.12 10.97 7.50 (9) Lumped kinetic visbroken residue to
52 69.96 72.39 15.78 14.02 8.70 models are often used to synthesis gas has been a
describe visbreaking promising route. About
48 12 (0.5) 85.18 84.90 8.25 8.77 4.67 reactions. The reaction rate 18% of the total gasification
49 78.19 79.26 11.18 11.20 6.97 parameters such as capacity worldwide rests
activation energy and with visbroken residue as
55 67.87 71.20 15.41 14.50 10.75
frequency factor determined the feedstock for the
3.99 0.98 1.06 0.68 0.69 402 402 5.62 1.72 1.71 1.14 1.13 412 412 7.85 on the basis of large boiling production of electricity.133
2.95 2.72 1.86 1.87 420 422 4.28 0.93 1.16 1.03 0.77 400 401 6.06 1.68 cuts or SARA analysis The visbroken residue can
1.87 1.70 1.25 408 403 8.48 3.00 3.03 2.56 2.08 420 429 4.35 1.04 1.20 alone fail to reflect the also be utilized as a
0.86 0.78 400 404 6.23 2.02 1.98 1.63 1.32 410 412 effect of feed properties. feedstock for a partial
8.88 3.11 3.18 2.86 2.19 420 424 Thus, the detailed hybrid oxidation unit for hydrogen
models using narrow production.74
with increasing H/C ratio, maltene phase as a second boiling range or carbon
and (iii) n-C7 asphaltenes phase (also known as (13) Various forms of
number are highly desirable
and CCR contents decrease mesophase). In the visbreaking, viz.
over which thermodynamic
with increasing API gravity mesophase, asphaltenes hydrovisbreaking, hydrogen
properties such as critical
of the feed. undergo a high rate of donor solvent
temperature or pressure,
(4) The feed stability condensation, visbreaking,134 and
mean molecular weight,
is proportional to the aromatization, and aquaconversion,135 have
accentric factor, and
aromatic and resin content dealkylation to give coke. come into reckoning so as
solubility parameter do not
while being inversely (7) Fouling to increase the selectivity
vary appreciably. Though
proportional to the phenomena is the chronic toward the distillates.
such models will be too
asphaltenes and saturates effect of secondary However, it may be pointed
complex, they are expected
content. reactions, viz. out that these processes use
to determine the close
(5) On the basis of the polymerization and high hydrogen pressure and
dependency between the
reported observation in condensation. Fouling catalytic additives such as
two. The rate constants so
distillates yield and limits the maximum molybdenum and are
estimated can then be used
interclass conversions, a achievable skin temperature intended to operate at
for accurate prediction of
comprehensive mechanism and on-stream time and severe conditions. The
commercial visbreaker
involving primary and increases the pressure drop. reaction chemistry and fluid
yields.
secondary reactions has (8) The kinetics properties in these
(10) Single-coil, two- hydroconversion processes
been proposed. The primary reported is based on the coil, coil-soaker-type, and
reactions involve Cali-Cali psuedo-components (based are fundamentally different
sectionalizedsoaker from the conventional
splitting and Car-Cali on the solubility, boiling- visbreaking processes have
dearomatization to give point cuts, and carbon visbreaking regime.
shown increasing degree of
distillates formation, while number) by making the use performance in terms of (14) The soaker may
secondary reactions (via H- of parallel and parallel- conversion and selectivity. operate in transition or
abstraction, cyclization, consecutive models. The (11) Of late, there is an heterogeneous regime
condensation, and parallel model explains only increasing trend of getting depending upon the feed
aromatization) give the thermal cracking of the more of distillates out of type and operating
interclass conversions that feed to distillates and does visbreaking in addition to conditions. The
follow the order: saturate > not consider further conventional benefits in comprehensive gas holdup
aromatics > resins > cracking of the already- terms of viscosity and pour- and bubble-size distribution
asphaltenes. cracked products. However, point reduction of the measurements over a wide
(6) There exists a parallel-consecutive models residual feedstocks. Shell range of physicochemical
threshold concentration consider the cracking of the International Oil Products properties and gas-liquid
beyond which asphaltenes products and the secondary has developed the Shell superficial velocities under
precipitate out of the reactions paths as well and, Deep thermal conversion moderate-to-high pressure
8992 Ind. Eng. Chem. Res., Vol. 47, No. 23, 2008

need to be performed. The erroneous results when catalyst surface, and (vi) onal
data so obtained can be tested for an unknown feed desorption of hydrogenated area
used to estimate the and operational variables species into the bulk. It is
(m2)
transition velocity (Vtrans) outside the range covered. desirable to study the effect
and can be subsequently The detailed two- and three- of temperature, H2/steam AMVR ) Arabian mix
incorporated in the dimensional process models partial pressure, type of vacuum residue
correlations to predict gas that take into account the catalyst and its loading, AP ) surface area (m2)
and liquid holdups for an kinetics and equilibrium particle size (for supported AR ) atmospheric residue (wt
unknown G-L system. flash calculations still need catalyst), and speed of %)
(15) The diagnostic to be developed. agitation on product yields Ar ) aromatics (wt %)
tools such as noninvasive (17) Soaker internals to understand the reaction
As ) asphaltenes (n-heptane
gamma ray tomography and are designed on the basis of mechanism and determine
insolubles) (wt %)
radiotracer residence time the hydrodynamic data the rate-controlling step.
Further, the compatibility of B)
distribution studies can be obtained on cold-flow
additives (sulfur, oil- const
fruitful in throwing light on laboratory setup or pilot-
miscible metal compounds) ant
the hydrodynamics of the scale units in terms of
reactor. These techniques liquid-phase mixing time, with the feed properties b)
have been successfully residencetime distribution, needs to be explored. constant
tested and standardized on differential pressure drop, used in eq
laboratory cold-flow and radial gas-holdup profiles, Acknowledgment 22 b1, b2,
pilot-plant units. The and product yields. Such
K.L.K. and R.P.K. are b3 )
information gathered from experiments are often
the above studies can be timeconsuming and require grateful to the Centre for regression
used in assessing the a considerable capital High Technology (CHT), coefficient
performance of existing expenditure. This can be New Delhi, India, for
s in eq 33
coil-soaker visbreaker units avoided by using the providing the research
fellowship. A.N.S. is BDE )
and incorporated in the commercial CFD software
powerful CFD codes to packages. The grateful to the University bond
predict internally consistent hydrodynamic parameters Grants Commission (UGC), dissociatio
velocity, eddy viscosity, such as velocity, holdup, Government of India, for n energy
and gas holdup profiles. pressure, and eddy viscosity providing the research
(kcal mol-
can be estimated for fellowship. The research
(16) In the black-box 1
) BHVR )
different set of internals fellowship of United
approach, the output yields
using Euler-Euler type Phosphorus Limited (UPL), Bombay
are regressed as a function
models. The profiles so India, is also gratefully high
of input variables such as
obtained can be used very acknowledged.
temperature, LHSV, feed vacuum
density, asphaltenes, and well to predict the RTD or
residue c )
CCR content using best-fit mixing time for a given Nomenclature
percent free area, hole constant
correlations. The approach A)
of preliminary models (in diameter, and pitch of the used in eq
which coilsoaker is baffle plate. frequ 23 C )
assumed as a plug-flow (18) Hydrovisbreaking ency concentrati
reactor coupled with n- and aquaconversion being facto on
CSTRs in series or a one- multiphase reactions, the
r (s-1) Cali ) aliphatic carbon
dimensional axially mass transfer and chemical
dispersed reactor) includes reactions are of significant a1, Car )
iterative solution of the importance. The possible a2, a3 aromati
equations of continuity, sequence of steps through c
)
momentum, and energy to which these reactions occur carbon
predict axial pressure, can be given as (i) mass regre
C0, C1 )
temperature, and product transfer of steam/H2 to ssion
profiles. The axial pressure catalyst surface, (ii) drift
coeff
drop is calculated using reforming reaction (in the flux
icient
friction factor defined on case of aquaconversion) to constan
pseudo-phase properties produce molecular s in
t c1, c2,
(single-phase flow), while hydrogen, (iii) adsorption of eq 32
the model parameters hydrogen on the catalyst c3 )
AC )
(apparent activation energy surface, (iv) diffusion of regress
and frequency factor) are hydrocarbon species on the cross ion
tuned to match the plant catalyst surface, (v) - coeffici
data. However, these hydrogenation reactions secti ents in
methods may give (radical saturation) on the
 Ind. Eng. Chem. Res., Vol. 47, No. 23, 2008 8993

eq 34 ) h F
CCR ) r
d (
Conrad
Q m )
son
′ )
carbon F
residue ) E r
(wt %) o
d ) u
Cp )
i d
averag a e
f
e f c
specifi e t n
c heat r i u
e v m
(kJ kg-1 b
n a
K-1) t t e
Cp,Feed ) feed i i r
a o g ) acceleration due to
average specific
l n
heat (kJ kg-1 K-1) gravity (m s-2) G ) gas

Cp,steam ) steam h e fraction (C1-C5) (wt %)

e n Ga ) gasoline (C5-150 °C) (wt
average specific
a e %)
heat (kJ kg-1 K-1) t r
Gao ) initial concentration of
dB ) bubble g
gasoline at zero residence
i y
diameter (m) time
n (wt %)
dB,max ) maximum p (
Gm ) mass velocity (kg m-2 s-
stable bubble u k 1
)
t c
diameter (m) dC )
a Go ) initial concentration of
coil diameter (m) l gas component at zero
(
DC ) column W residence time
diameter (m) ) m (wt %)
o Gz ) Graetz number
DCi ) reference column l
d h ) film thickness (m)
diameter (m) -
z
DH ) helix diameter (m) 1 H/C ) molar hydrogen-to-
carbon ratio (mol mol-1)
DL ) liquid-phase axial ) )
dispersion coefficient (m2 s-1) f ) backmixing coefficient HC ) cell height (m)
d i FA ) free area H
n
O F C
c
P L
) r
e
)
h m )
o e
f
l n
l c
e t
o
a l
w
l e
d
i p a
a c
a r
m o
r
e i
a
t l l
m
e e i
r l
t q
e
e
( n u
r
m g i
t
8994 Ind. Eng. Chem. Res., Vol. 47, No. 23, 2008

d u b
m i a
h b n c
e e d k
i r e m
g k ) first-order x i
h reaction rate x
constant (s-1)
t m e
KcKc′ )
characterization ˙ d
( factors for
m coking reactions f
)
) kH, kA ) first-order rate
l
constants of the n-C7
soluble and asphaltenes, m o
H respectively (s-1) a w
D ki ) first-order s
rate constant for s r
) -1
ith species (s ) a
KrKr′ ) f t
d characterization l e
i factors for o
s coking reactions w (
p k1-7 ) rate k
e -1 g
constants (s ) r
r
k1234 ) k1 + k2 + a
s s
k3 + k4 (s-1) k567 ) t
i -
k5 + k6 + k7 (s-1) e
o 1
L ) column or coil length (m)
n
LGO ) light gas oil (150-350 ( )
°C) (wt %) k
h LGOo ) initial concentration g m
e of LGO at zero residence
˙
i time (wt
%) s f
g
m -
h
) 1 )
t
)
h f
(
o m e
m
l ˙ e
)
d b d
u
j
p ) m
a
)
p l s
r i s
t
o q
a
f u f
n
i i l
k
l d o
e w
n
 Ind. Eng. Chem. Res., Vol. 47, No. 23, 2008 8995

′
1

r ) n
a MMTPA ) million metric ) -
t tons per annum C
e Mo ) Morton number s 5

MoSL ) slurry Morton number c

MW )
( a
) l
k
g mo e n
lec - -
ula u p
s
- r p e
n
1
we
i t
) igh
n a
t
d n
m (kg
e e
˙ km x
s ol- i
1
) N n
) N) s
A
act o
s ) l
ual
t u
nu
e n b
mb
a a l
er
m p e
of s
h
CS
m t
a TR (
h
s s e w
s n n t
) i
f c %
n )
l
o u a
m r n
w
b o -
e m C
r
r a 7
a
t
t
o i )
e
f c
s n
(
h -
k
o ( h
g l e
w
e t p
s s t
-
% a
n ) n
8996 Ind. Eng. Chem. Res., Vol. 47, No. 23, 2008

e f M )
f
) h
i
e
n ) t a
s w t
o e o
l - f
f
p l
u f h u
b e a x
l c s
e t e (
W
s i
P
v r m
( e a -

w n 2

d )
t n
t Q ) net volumetric flowrate
u l (m3 s-1)
% m
Q1 ) total flow rate given as
) b n
(Q + q), (m3 s-1)
u
e QL ) liquid-phase volumetric
m
n r b flowrate (m3 s-1)
d e QG ) vapor-
a o r phase
t f volumetric
q
flow rate (m3
) C ) s-1) R ) gas
S constant (cal
n T b
a gmol-1 K-1) R′
u R
c ) column
m s k radius (m)
b Nu ) Nusselt number w R
e PA ) polar aromatics (wt %) a
Pe ) Peclet number r d
r
d
P
r )
o f
f l
) r
o
w a
d P r
r d
a a
a t i
t
n e u
a d
s
t
(
p l m
3 o
o
n f
i u s
n m -

t b 1 d
e i
s )
r
s
q k
N P
′
e r
 Ind. Eng. Chem. Res., Vol. 47, No. 23, 2008 8997

( n d te
m u e m
x
) m p
b T e
R e r
e r ) at
Rs ) resins (wt %) u
r
) r ) radial e r
distance (m) a e
ri ) rate of c (
R
t
e reaction for K
i
y the ith o )
n species (mol n T
-1
o s ) rG ) B
t
l reaction rate )
e
d for gas (mol m b
s p at
s-1) rGa )
e h
reaction rate r
n a te
for gasoline
u t m
(mol s-1) rLGO
u p
m ) reaction r
b e
rate for LGO e
e r
(mol s-1) rVGO
r ( at
) reaction
K u
rate for VGO )
R r
(mol s-1) rVR ) t
e e
reaction rate )
(
M
for VR (mol r
K
s-1) e
)
) a
S ) sulfur (wt %) tC
Sa ) saturates (wt %) ct
)
t SARA ) saturates, aromatics, i
ci
w resins, and asphaltenes o
r
o SF ) stability factor n
c
- SI )severity index (s-1) ti
u
p SSE ) sum of square of errors m
la
h S e
X ti
a (s
o
s )
) n
e T
ti
s 0
t m
R )
a e
e c
b (s
y i o
)
n l il
i T
o i
t I
l y n
)
d le
t
s i t
n o
8998 Ind. Eng. Chem. Res., Vol. 47, No. 23, 2008

l VL(0) ) j p
u liquid ) ε
e centerlin ε fractional average holdup L

n e ε )
e velocity G fr
i (m s ) -1 ) a
n VL,hole ) fr c
s liquid- a ti
o phase c o
l ti n
hole
u o a
velocity
b n l
(m s-1)
le a li
VR ) vacuum residue (wt %)
( l q
VRo ) initial concentration of
c VR at zero residence time (wt g u
o %) a i
k VS ) slip velocity (m s-1) s d
e VSL ) superficial slurry h h
) velocity (m s-1)
o o
( Vsmall ) unhindered rise l l
w velocity of small bubbles d d
t (m s-1) Vtrans ) transition u u
% velocity (m s-1) x150- ) p p
) fractional conversion to j j
TN ) soaker outlet temperature gas-gasoline fraction ε ε
(K) (<236) yi ) concentration G L
Ui ) overall inside heat of ith species yG ) ) )
transfer coefficient (W m-2 K-
1 dimensionless a a
)
3
concentration for gas on v v
V ) tank volume (m )
wt basis yGa ) e e
VB∞ ) terminal bubble rise
velocity (m s-1) dimensionless r r
VBM ) rise velocity of concentration for gasoline a a
maximum stable bubble (m s- on wt basis yLGO ) g g
1
) dimensionless e e
VC ) liquid circulation velocity concentration for LGO on fr fr
(m s-1)
wt basis yVGO ) a a
Vex ) interstage exchange
dimensionless c c
velocity (m s-1)
concentration for VGO on ti ti
VG ) superficial gas velocity
(m s-1) wt basis yVR ) o o
VGO ) vacuum gas oil (350- dimensionless n n
500 °C) (wt %) concentration for VR on a a
VGOo ) initial concentration wt basis yiN ) l l
of VGO at zero residence concentration at soaker g li
time (wt
outlet for ith species yiO ) a q
%)
concentration at coil inlet s u
VL )
for ith species z ) reactor h i
superfici
length o d
al liquid
Greek Letters l h
velocity
d o
(m s-1)
u l
 Ind. Eng. Chem. Res., Vol. 47, No. 23, 2008 8999

d p s a
u ∆ c g
a
p ε
l
ε G i (
) n
S
m
) r g

fr a c s
a d o -

c i e
1
f
ti a
f )
o l i
n d c
a e i Ω
e
l n
n
s si t )
o t
li y s
)
d g l
h r c u
o a o
r
n
l d r
s
d i t y
u e a
p n n
t c
j t -
∆H ) enthalpy change (kJ kg o
εt µ 1
) r
ra L
∆HR ) heat of reaction r
ns ) calculated on basis of liquid
) li feed conversion e

r q (kJ kg-1) c
e u ∆Hr* ) heat of reaction (kJ kg - t
1
g i ) i
i d ∆HR* ) specific heat of o
reaction for products boiling
m v n
below 204
e is °C (kJ kg-1)
tr c φ f
a o a
n si c
)
si t t
ti y
s o
o (
u r
n P
r σ ) surface tension (N m-1)
g a
f ψ
a ·
s s a )
h ) c
o R e t
e
l m
)
d d p
u r e
9000 Ind. Eng. Chem. Res., Vol. 47, No. 23, 2008

r e L
a )
t ( )
u k s
r i ) s
e n l
e τ u
ν m m r
t a e r
t a y
) i n
c v
e ) i
d v s
d i m c
y s e o
c a s
d o n i
i s t
f i r y
f t e
u y s (
s i P
i ( d a
v m e
2
i n ·
t c
y s e s
-
)
1
( t
m ) i F
2
m
τ e )
s L
-
( d
1
) s e
) ) n
l s
µ i θ i
q m t
) u i y
i x

v d (
i - ) k
s p g
c h m
o a i m
s s x -

i e i 3

t n )
y r g FG ) gas density (kg m-3)
e
( s FGatm ) gas density at
t
P i atmospheric pressure (kg m-3)
i
a d m FL ) liquid density (kg m-3)
e e F0 )
· n
standard
c (
s e s density
) ) at 15 °C
t
(kg m-3)
ν i µ
m FSL )
L S
 Ind. Eng. Chem. Res., Vol. 47, No. 23, 2008 9001

slurry y (3)
) Tr B
density en od
(kg m-3) ds us
m : zy
e T ns
Subscripts a he ki,
a n Pr M
es .
m n
en M
d t .
) a an C
t d o
a F m
r ) ut po
ur sit
i
e. io
t n H n
h u yd of
m m ro Pe
e b ca tr
t e rb ol
i r o eu
n m
c
P R
o ro es
m f ce id
e ss ua
a d . .
n a 20 P
05 re
t
, pr
b a 8 .A
4 m
) p (9 .
o ), C
b i 35 he
n . m
u
(2) .
l t
A S
k s no oc
G, L ) gas, liquid n. .,
w In D
i ) species G, Ga, LGO, iV
di
VGO, VR a’ .
)
j ) run number s P
E et
l w ne .
m a C
rg
l y he
) l O m
pt .
l io 20
Literature Cited ns 02
o
. ,
g (1) 4
C
a S 7
he
r w
m (4
i at ),
.
t y, 32
In
T. 9.
h d.
E. (4
m Gl
D
)
i ig
ob S
c .
al pe
20
R ig
07
m ef ht,
,
e in J.
in Ja
G.
a n,
g H
n In 7
a
du 0.
n
M str d
9002 Ind. Eng. Chem. Res., Vol. 47, No. 23, 2008

b (12) Hennico, A.; Peries, J. (25) Rogel, E. Theoretical Visbreaking. Ind. Eng. Chem.
o P.; Laurent, J.; Espeillac, M. Approach to the Stability of Res. 2004, 43, 1373.
ok Maximum Conversion of Heavy Visbroken Residues. Energy (39) Sawarkar, A. N.;
of Hydrocarbons. Erdoel Kohle, Fuels 1998, 12, 875. Pandit, A. B.; Samant, S. D.;
P Erdgas, Petrochem. 1992, 45, (26) Beuther, H.; Joshi, J. B. Petroleum Residue
et 19. Goldthwait, R. G.; Offutt, W. C. Upgrading via Delayed Coking:
ro (13) Iscovici, R. S. Kinetic Thermal Visbreaking of Heavy A Review. Can. J. Chem. Eng.
le Model Solves Visbreaker ResiduessA Modern Application 2007, 85 (1), 1.
u Constraint Control Problem. of Thermal Cracking. Oil Gas J. (40) Speight, J. G. The
m Hydrocarbon Process. 1994, 73 1959, 57, 151. Chemistry and Technology of
A (5), 109. (27) Al-Soufi, H. H.; Petroleum; Marcel Dekker, Inc.:
n (14) Dente, M.; Bozzano, Shanshool, J.; Savaya, Z. F. New York, 1991.
al G.; Bussani, G. A Soaker Visbreaking of some (41) Groenzin, H.; Mullins,
ys Comprehensive Program for Iraqi Residues. Fuel Sci. O. C. Asphaltene Molecular Size
is; Visbreaking Simulation: Product Technol. Int. 1987, 5, 543. and Structure. J. Phys. Chem. A
Jo Amounts and their Properties (28) Krishna, R.; Kuchhal, 1999, 103, 11237.
hn Prediction. Comput. Chem. Eng. Y. K.; Sarna, G. S.; Singh, I. D. (42) Rhoe, A.; de
W 1997, 21, 1125. Visbreaking Studies on Aghajari Blignieres, C. Visbreaking: A
ile (15) Omole, O.; Olieh, M. Long Residue. Fuel 1988, 67, Flexible Process. Hydrocarbon
y N.; Osinowo, T. Thermal 379. Process. 1979, 58 (1), 131.
an Visbreaking of Heavy Oil from (29) Thomas, M.; Fixari, (43) Akbar, M.; Geelen, H.
d the Nigerian Tar Sand. Fuel B.; Le Perchec, P.; Princic, Y.; Visbreaker uses Soaker Drum.
S 1999, 78, 1489. Lena, L. Visbreaking of Safaniya Hydrocarbon Process. 1981, 60
on (16) Brauch, R.; Fainberg, Vacuum Residue in the Presence (5), 81.
s, V.; Kalchouck, H.; Hetsroni, G. of Additives. Fuel 1989, 68, 318. (44) Savaya, Z. F.; Al-
In Correlations between Properties (30) Singh, I. D.; Kothiyal, Soufi, H. H.; Al-Azawi, I.;
c.: of Various Feedstocks and V.; Ramaswamy, V.; Krishna, R. Mohammed, H. K. Effect of
N Products of Visbreaking. Fuel Characteristic Changes of Visbreaking Conditions on the
e Sci. Technol. Int. 1996, 14 (6), Asphaltenes during Visbreaking Stability of Some Iraqi Long
w 753. of North Gujarat Short Residue. Residues. Pet. Sci. Technol.
Y Fuel 1990, 69, 289. 1988, 355.
(17) Shen, H.; Ding, Z.; Li,
or (31) Di Carlo, S.; Janis, B. (45) Taylor, S. E. The
R. Thermal ConversionsAn
k, Composition and Visbreakability Electrodeposition of Asphaltenes
Efficient way for Heavy Residue
20 of Petroleum Residues. Chem. and Implications for Asphaltene
Processing. Proc. 15th World
01 Eng. Sci. 1992, 47 (9-11), 2695. Structure and Stability in Crude
Pet. Congr. 1998, 907.
. (32) Del Bianco, A.; and Residual Oils. Fuel 1998, 77,
(18) Yepsen, G. L.;
(5) Henderson, R.; Panariti, N.; Beltrame, P. L.; 821.
Jenkins, J. H. Visbreak by Crude
Rodwell, M.; Harji, A. Consider Carniti, P. Thermal Cracking of (46) Fainberg, V.;
Unit Revamp. Hydrocarbon
Modifying your Refinery to Petroleum Residues. 1. Kinetic Podorozhansky, M.; Hetsroni,
Process. 1981, 60 (9), 117.
Handle Heavy Opportunity Analysis of the Reaction. Fuel G.; Brauch, R.; Kalchouck, H.
Crude Oils. Hydrocarbon (19) Marzin, R.; Pereira,
P.; Rivas, L.; Mcgrath, M.; 1993, 72, 75. Changes in the Composition and
Process. 2005, 84 (9), 54. (33) Benito, A. M.; Properties of the Vacuum
Thompson, G. J. Resid
(6) Gembicki, V. A.; Martinez, M. T.; Fernandez, I.; Residues as a Result of
Cowan, T. M.; Brierley, G. R. Conversion through Visbreaking
Technology. World Refin. 1999, Miranda, J. L. Visbreaking of an Visbreaking. Fuel Sci. Technol.
Upgrade Processing Operations Asphaltenic Coal Residue. Fuel Int. 1996, 14 (6), 839.
to Handle Heavy Feedstocks. NoV/ Dec, 16.
1995, 74, 922. (47) Yan, Y. T.
Hydrocarbon Process. 2007, 86 (20) Speight, J. G. The
Desulfurization of HeaVy Oils (34) Yasar, M.; Trauth, D.; Characterization of Visbreaker
(2), 41. Klein, M. T. Asphaltene and Feeds. Fuel 1990, 69, 1062.
(7) Allan, D. E.; and Residua; Marcel Dekker,
Inc.: New York, 2000. Resid Pyrolysis. 2. Effect of (48) Kuo, C. J. Effects of
Martinez, C. H.; Eng, C. C.; Reaction Environment on Crude Types on Visbreaker
Barton, W. J. Visbreaking Gains (21) Christman, R. D.
Pathways and Selectivities. Conversion. Oil Gas J. 1984,
Renewed Interest. Chem. Eng. Carbon-Out Processes for Heavy
Energy Fuels 2001, 15, 504. Sep 24, 100.
Prog. 1983, Jan, 89. Resids. World Refin. 1999,
NoV/Dec, 22. (35) Rahmani, S.; (49) Li, S.; Liu, Z.; Liu, C.;
(8) Allen, J. G.; Little, D. McCaffrey, W.; Gray, M. R. Que, G.; Liang, W.; Zhu, Y.
M.; Waddill, P. M. Visbreaking (22) Phillips, G.; Liu, F.
Kinetics of Solvent Interactions Formation of the Physical
High-Vacuum Residua. Pet. Advances in Resid Upgradation
with Asphaltene during Coke Neophase in terms of SARA
Process. 1951, June 612. Technologies Offer Refiners
Formation. Energy Fuels 2002, Compositions of Vacuum
(9) Notarbartolo, M.; Cost-Effective Options for Zero
16, 148. Residua. Prepr. Am. Chem. Soc.,
Menegazzo, C.; Kuhn, J. Fuel Oil Production. Proceedings
of European Refining (36) Schabron, J. F.; Pauli, DiV. Pet. Chem. 1998, 43 (1),
Visbreaking: Low Cost Next A. T.; Rovani, J. F., Jr. Residua 187.
Step. Hydrocarbon Process. Technology Conference, Paris,
2002. Coke Formation Predictability (50) Liu, C.; Zhu, C.; Jin,
1979, 58 (9), 114. Maps. Fuel 2002, 81, 2227. L.; Shen, R.; Liang, W. Step by
(10) Hournac, R.; Kuhn, J.; (23) Giavarini, C. Stability
of Bitumens Produced by (37) Singh, J.; Kumar, M. Step Modeling for Thermal
Notarbartolo, M. Visbreaking: M.; Saxena, A. K.; Kumar, S. Reactivities and Chemical
More Feed for FCC. Thermal Processes. Fuel 1981,
60, 401. Studies on Thermal Cracking Compositions of Vacuum
Hydrocarbon Process. 1979, 58 Behavior of Residual Feedstocks Residues and their SFEF
(12), 97. (24) Altgelt, K. H.;
in a Batch Reactor. Chem. Eng. Asphalt. Fuel Process. Technol.
(11) Stolfa, F. New Roles Boduszynski, M. M.
Sci. 2004, 59, 4505. 1999, 59, 51.
for Thermal Cracking. Composition and Analysis of
HeaVy Petroleum Fractions; (38) Kataria, K. L.; (51) Schabron, J. F.; Pauli,
Hydrocarbon Process. 1980, 59 Kulkarni, R. P.; Pandit, A. B.; A. T.; Rovani, J. F., Jr.
(5), 101. Marcel Dekker: New York,
1994. Joshi, J. B.; Kumar, M. M. Molecular Weight Polarity Map
Kinetics of Low Severity
 Ind. Eng. Chem. Res., Vol. 47, No. 23, 2008 9003

for Residua Pyrolysis. Fuel 2001, and Combustion Processes. (76) Joshi, J. B. Axial Bubble Columns. AIChE J. 1992,
80, 529. Appl. Therm. Eng. 2002, 22, 919. Mixing in Multiphase 38 (4), 544.
(52) Groenzin, H.; Mullins, (64) Towfighi, J.; ContactorssA Unified (88) Reilly, I. G.; Scott, D.
O. C. Molecular Size and Sadrameli, M.; Niaei, A. Coke Correlation. Trans. Inst. Chem. S.; De Bruijn, T. J. W.;
Structure of Asphaltenes from Formation Mechanism and Coke Eng. 1980, 58, 155. MacIntyre, D. The Role of Gas
Various Sources. Energy Fuels Inhibiting Methods in Pyrolysis (77) Fersing, M.; Gouzien, Phase Momentum in
2000, 14, 677. Furnaces. J. Chem. Eng. Jpn. L.; Mouchot, E.; Bourley, G. Determining the Gas Holdup and
(53) Mullins, O. C.; Sheu, 2002, 35 (10), 923. Processes for Visbreaking Heavy Hydrodynamic Flow Regimes in
E. Y.; Hammami, A.; Marshall, (65) Al-Soufi, H. H.; Hydrocarbons Feedstocks. U.S. Bubble Column Operations.
A. G. Asphaltenes, HeaVy Oils Savaya, Z. F.; Mohammed, H. Patent 5925236, 1999. Can. J. Chem. Eng. 1994, 72, 3.
and Petroleomics; Springer: New K.; Al-Azawi, I. A. Thermal (78) Pandit, A. B.; Joshi, J. (89) Luo, X.; Jiang, P.;
York, 2007. Conversion (Visbreaking) of B. Mixing in Mechanically Fan, L. S. High-Pressure Three-
(54) Mullins, O. C. Heavy Iraqi Residue. Fuel 1988, Agitated GasLiquid Contactors, Phase Fluidization:
Rebuttal to Comment by 67, 1714. Bubble Columns and Modified Hydrodynamics and Heat
Professors Herod, Kandiyoti, and (66) Castellanos, J.; Cano, Bubble Columns. Chem. Eng. Transfer. AIChE J. 1997, 43,
Bartle on “Molecular Size and J. L.; del Rosal, R.; Briones, V. Sci. 1983, 38, 1189. 2432.
Weight of Asphaltene and M.; Mancilla, R. L. Kinetic (79) Blauwhoff, P. M. M.; (90) Lin, T. J.; Tsuchiya,
Asphaltene Solubility Fractions Model Predicts Visbreaker Cornelissen, A. E. Continuous K.; Fan, L. S. Bubble Flow
from Coals, Crude Oils and Yields. Oil Gas J. 1991, Mar 18, Thermal Cracking Process. U.S. Characteristics in Bubble
Bitumen”. Fuel 2007, 86, 309. 76. Patent 4551233, 1985. Columns at Elevated Pressure
(55) Wiehe, I. A.; Liang, (67) Xiao, J.; Wang, L.; (80) Palaskar, S. N.; De, J. and Temperature. AIChE J.
K. S. Asphaltenes, Resins, and Chen, Q.; Wang, D. Modeling K.; Pandit, A. B. Liquid Phase 1998, 44, 545.
other Petroleum for Product Distribution in RTD Studies in Sectionalized (91) Letzel, H. M.;
Macromolecules. Fluid Phase Thermal Conversion of Heavy Bubble Column. Chem. Eng. Schouten, J. C.; Krishna, R.; van
Equilib. 1996, 117, 201. Oil. Pet. Sci. Technol. 2002, 20 Technol. 2000, 23, 61. den Bleek, C. M. Gas Hold-up
(56) Singh, I. D.; Kothiyal, (5-6), 605. (81) Kumar, M. M.; and Mass Transfer in Bubble
V.; Aloopwan, M. K. S.; (68) Takatsuka, T.; Tandon, D.; Yadram; Kulkarni, Column Operated at Elevated
Ramaswamy, V. Compositional Kajiyama, R.; Hashimoto, H.; R. P.; Kataria, K. L.; Pandit, A. Pressure. Chem. Eng. Sci. 1999,
and Structural Changes in Short Matsuo, I.; Miwa, S. A. Practical B.; Joshi, J. B. Residence Time 54, 2237.
Residues during Visbreaking. Model of Thermal Cracking of Distribution Studies for the (92) Han, L.; Al-Dahhan,
Erdoel Kohle, Erdgas, Residual Oil. J. Chem. Eng. Jpn. Development of “Advanced M. H. Gas-Liquid Mass Transfer
Petrochem. 1993, 46 (10), 386. 1989, 22 (3), 304. Soaker Visbreaking in a High Pressure Bubble
(57) Redelius, P. G. Solubility (69) Trauth, D. M.; Yasar, Technology”. In Proceedings of Column Reactor with Different
Parameters and Bitumen. Fuel M.; Neurock, M.; Nasigam, A.; XIIth Refinery Technology Sparger Designs. Chem. Eng.
2000, 79, 27. Klein, M.; Kukes, S. G. Meeting, Goa, India, 2004; p Sci. 2007, 62, 131.
(58) Nielsen, B. B.; Asphaltene and Resid Pyrolysis: 390. (93) Wilkinson, P. M.;
Svrcek, W. Y.; Mehrotra, A. K. Effect of Reaction Environment. (82) Washimi, K.; Limmer, Dierendonck, L. V. Pressure and
Effects of Temperature and Fuel. Sci. Technol. Int. 1992, 10 H. A New Unit to Thermal Crack Gas Density Effects on Bubble
Pressure on Asphaltene Particles (7), 1161. Resid. Hydrocarbon Process. Break-Up and Gas Hold-Up in
Size Distribution in Crude Oils (70) Filho, R. M.; Sugaya, 1989, 68 (9), 69. Bubble Columns. Chem. Eng.
Diluted with Pentane. Ind. Eng. M. F. A Computer Aided Tool (83) Joshi, J. B.; Parasu Sci. 1990, 45, 2309.
Chem. Res. 1994, 33, 1324. for Heavy Oil Thermal Cracking Veera, U.; Prasad, C. V.; (94) Clark, K. N. The
(59) Leon, O.; Rogel, E.; Process Simulation. Comput. Phanikumar, D. V.; Deshpande, Effect of High Pressure and
Espidel, J.; Torres, G. Chem. Eng. 2001, 25, 683. N. S.; Thakare, S. S.; Thorat, B. Temperature on Phase
Asphaltenes: Structural (71) Ben-Asher, J.; Por, N. N. Gas Hold-up Structures in Distributions in a Bubble
Characterization, Self- Prediction of Residual Fuel Oil Bubble Column Reactors. Proc. Column. Chem. Eng. Sci. 1990,
Association, and Stability Stability during Storage, Indian Natl. Sci. Acad. 64 1998, 45, 2301.
Behavior. Energy Fuels 2000, Handling, and Use-Assessment 441. (95) Kemoun, A.; Ong, B.
14, 6. of Test Methods. In Proceedings (84) Tarmy, B. L.; Chang, C.; Gupta, P.; Al-Dahhan, M. H.;
(60) Fotland, P.; of 4th International Conference M.; Coulaloglou, C. A.; Ponzi, P. Dudukovic, M. P. Gas Hold-up
Anfindsen, H.; Hadnes, F. on Stability and Handling of R. The Three Phase in Bubble Column At Elevated
Detection of Asphaltene Liquid Fuels, Orlando, FL, 1991; Hydrodynamic Characteristics of Pressure via Computed
Precipitation and Amounts p 643. the EDS Coal Liquefaction Tomography. Int. J. Multiphase
Precipitated by Measurement of (72) Zerlia, T.; Pinelli, G. Reactors. Inst. Chem. Eng. Symp. Flow 2001, 27, 929.
Electrical Conductivity. Fluid Stability Tests on Visbreaking Ser. 1984, 87, 303. (96) Krishna, R.;
Phase Equilib. 1993, 82, 15. Residues obtained at Increasing (85) Idogawa, K.; Ikeda, Wilkinson, P. M.; Dierendonck,
(61) Rahimi, P.; Gentzis, Severity. Fuel 1993, 72 (8), K.; Fukuda, F.; Morooka, S. L. L. A Model for Gas Holdup in
T.; Fairbridge, C.; Khulbe, C. 1109. Formation and Flow of Gas Bubble Columns Incorporating
Chemistry of Petroleum (73) Maxa, D.; Prochaska, Bubbles in a Bubble Column the Influence of Gas Density on
Residues in the Presence of H- F.; Sebor, G.; Blazek, J.; with a Single Orifice or Nozzle Flow Regime Transitions. Chem.
Donors. Prepr. Am. Chem. Soc., Kroufek, J. Comparison of Gas Distributor. Chem. Eng. Eng. Sci. 1991, 46, 2491.
DiV. Pet. Chem. 1998, 43 (4), Vacuum Residue Visbreaking Commun. 1987, 59, 201. (97) Yang, G. Q.; Fan, L.
634. and Hydrovisbreaking. Pet. Coal (86) Oyevaar, M. H.; Bos, S. Axial Liquid Mixing in High
(62) Wen, P.; Yu, D.; Que, 2002, 44 (1-2), 36. R.; Westerterp, K. R. Interfacial Pressure Bubble Columns.
G.; Wang, Z. Research on (74) Hus, M. Visbreaking areas and Gas Hold-ups in Gas- AIChE J. 2003, 49 (8), 1995.
Nitrogen Distribution in Thermal Process has Strong Revival. Oil Liquid Contactors at Elevated (98) Urseanu, M. I.; Guit,
Conversion of Vacuum Residues. Gas J. 1981, Apr 13, 109. Pressures from 0.1 to 8 MPa. R. P. M.; Stankiewicz, A.; van
Prepr. Am. Chem. Soc., DiV. (75) Kataria, K. L. Studies Chem. Eng. Sci. 1991, 46, 1217. Kranenburg, G.; Lommen, J. H.
Pet. Chem. 2002, 47 (1), 11. in Multiphase Reactions and (87) Wilkinson, P. M.; G. M. Influence of Operating
(63) Bozzano, G.; Dente, Reactors. Ph.D. Thesis, Spek, A. P.; Dierendonck, L. V. Pressure on the Gas Holdup in
M.; Faravelli, T.; Ranzi, E. University of Mumbai, Mumbai, Design Parameters Estimation Bubble Columns for High
Fouling Phenomena in Pyrolysis India, 2003. for Scale-up of High Pressure
9004 Ind. Eng. Chem. Res., Vol. 47, No. 23, 2008

Viscous Media. Chem. Eng. Sci. (111) Holcombe, N. T.; Visbreaker. In Proceedings of
2003, 58, 697. Smith, D. S.; Knickle, H. N.; Third International Petroleum
(99) Gandhi, A. B.; Joshi, O’Dowd, W. Thermal Dispersion Conference and Exploration;
J. B.; Jayaraman, V. K.; and Heat Transfer in Non PETROTECH-99: New Delhi,
Kulkarni, B. D. Development of Isothermal Bubble Columns. India, 1999; p 349.
Support Vector Regression Chem. Eng. Commun. 1983, 21, (124) Beggs, H. D.; Brill, J.
(SVR)-Based Correlation for 135. P. A Study of Two Phase Flow in
Prediction of Overall Gas Hold- (112) Houzelot, J. L.; Inclined Pipes. J. Pet. Technol.
Up in Bubble Column Reactors Thiebaut, M. F.; Charpentier, J. 1973, 17, 607.
for Various Gas-Liquid Systems. C.; Schiber, J. Contribution to (125) Kulkarni, R. P.
Chem. Eng. Sci. 2007, 62, 7078. the Hydrodynamic Study of Modelling of Multiphase
(100) Voigt, J.; Schugerl, K. Bubble Columns. Int. Chem. Reactors: Visbreaking. Ph.D.
Absorption of Oxygen in Eng. 1985, 25, 645. Thesis, University of Mumbai,
Countercurrent Multistage (113) Wilkinson, P. M.; Mumbai, India, 2005.
Bubble Columns. I. Aqueous Haringa, H.; Stockman, F. P. A.; (126) Kern, D. Q. Process
Solutions with Low Viscosity. Dierendonck, L. V. Liquid Heat Transfer; McGraw-Hill
Chem. Eng. Sci. 1979, 34, 1221. Mixing in Bubble Column under Kogakusha Ltd.: Tokyo, 1950.
(101) Kato, Y.; Kago, T.; Pressure. Chem. Eng. Sci. 1993, (127) Dean, D. E.; Stiel, L.
Morooka, S. Longitudinal 48 (10), 1785. I. The Viscosity of Nonpolar Gas
Concentration Distribution of (114) Panvelkar, P.; Mixtures at Moderate and High
Droplets in Multi-Stage Bubble Tierney, J. W.; Shah, Y. T.; Pressures. AIChE J. 1965, 11,
Columns for Gas-Liquid Rhodes, D. F. Backmixing in a 526.
Systems. J. Chem. Eng. Jpn. SRC Dissolver. Chem. Eng. Sci. (128) Twu, C. H. Internally
1984, 17, 429. 1982, 37, 1582. (115) Lorenz, Consistent Correlation for
(102) Nishikawa, M.; Shino, O.; Schumpe, A.; Ekambara, K.; Predicting Liquid Viscosities of
K.; Kayama, T.; Nishioka, S.; Joshi, J. B. Liquid Phase Axial Petroleum Fractions. Ind. Eng.
Hashimoto, K. Gas Absorption in Mixing in Bubble Columns Chem. Process Des. DeV. 1985,
Multistage Gas-Liquid Spouted Operated at High Pressures. 24, 1287.
Vessel. J. Chem. Eng. Jpn. 1985, Chem. Eng. Sci. 2005, 60, 3573. (129) Mallan, G. M.;
18, 496. (116) Joshi, J. B.; Sharma, Michaelian, M. S.; Lockhart, F.
(103) Chen, B. H.; Yang, N. M. M. A Circulation Cell Model J. Liquid Thermal Conductivities
S.; Mc Millan, A. F. Gas Hold- for Bubble Columns. Trans. Inst. of Organic Compounds and
up and Pressure Drop for Air- Chem. Eng. 1979, 57, 244. Petroleum Fractions. J. Chem.
Water Flow through Plate (117) Blass, E.; Cornelius, Eng. Data 1972, 17, 412.
Bubble Column. J. Chem. Eng. W. Residence Time Distribution (130) Rihani, D. N.;
Jpn. 1986, 64, 387. of Solid and Liquid in Multistage Nandapurkar, P. J. Prediction of
(104) Chen, B. H.; Yang, N. Bubble Columns in Cocurrent Yields and Properties of
S. Characteristics of Cocurrent Flow of Gas, Liquid and Visbreaker Products; Report No.
Multistage Bubble Column. Ind. Suspended Solid. Int. J. 54(80); Engineers India Limited:
Eng. Chem. Res. 1989, 28, 1405. Multiphase Flow 1977, 3, 459. New Delhi, India, 1980.
(105) Vinaya, M. Multistage (118) Bakopoulos, A. Fluid (131) Zuber, N.; Findlay, J.
Bubble Column and Liquid Dynamics and Mixing in Three- A. Average Volumetric
Pulsed Column. D. Sc. Thesis, Phase Coal and Oil Residue Concentration in Two Phase
Indian Institute of Technology, Hydrogenation Sieve Cascade Flow Systems. Trans. ASME J.
Madras, India, 1994. Reactors. Chem. Eng. Sci. 2001, Heat Transfer 1969, 87, 453.
(106) Yamada, H.; Goto, S. 56, 5131. (132) Elliott, J. D.
Gas and Liquid Holdups in (119) Fan, L. S.; Yang, G. Maximize Distillate Liquid
Multi-Stage Bubble Columns for Q.; Lee, D. J.; Tsuchiya, K.; Luo, Products. Hydrocarbon Process.
Gas-Liquid-Liquid-Solid Four- X. Some Aspects of High- 1992, 71 (1), 75.
Phase System. J. Chem. Eng. Pressure Phenomena of Bubbles (133) Minchener, A. J. Coal
Jpn. 1998, 31 (5), 813. in Liquids and LiquidSolid Gasification for Advanced Power
(107) Doshi, Y. K.; Pandit, Suspensions. Chem. Eng. Sci. Generation. Fuel 2005, 84, 2222.
A. B. Effect of Internals and 1999, 54, 4681. (134) Bakshi, A. S.; Lutz, I.
Sparger Design on Mixing (120) Hughmark, G. A. H. Adding Hydrogen Donor to
Behavior in Sectionalized Hold-up in Gas-Liquid Flow. Visbreaking Improves Distillates
Bubble Column. Chem. Eng. J. Chem. Eng. Prog. 1962, 58 (4), Yields. Oil Gas J. 1987, July 13,
2005, 112, 117. 62. 84.
(108) Dreher, A. J.; Krishna, (121) Mosby, J. F.; Buttke, (135) Roger, M.; Pereira, P.;
R. Liquid Phase Backmixing in R. D.; Cox, J. A.; Nokolaides, C. Mcgrath, M.; Feintuch, H.;
Bubble Columns Structured by Process Characterization of Thompson, G.; Houde, E. The
Introduction of Partition Plates. Expanded-Bed Reactors in Aquaconversion Process: A New
Catal. Today 2001, 69, 165. Series. Chem. Eng. Sci. 1986, 41, Approach to Residue Processing.
(109) Kemoun, A.; Rados, 989. Presented at Asia Pacific
N.; Li, F.; Al-Dahhan, M. H.; (122) Carbonell, M. M.; Technology Conference, 1999.
Dudukovic, M. P.; Mills, P. L.; Guirardello, R. The Effect of
Leib, T. M.; Lerou, J. J. Gas Bubbling Regime on Gas and ReceiVed for reView August 9,
Hold-up in Trayed ColdFlow Liquid Phase Mixing in Bubble 2007
Bubble Column. Chem. Eng. Sci. Column Reactors. Chem. Eng.
2001, 56, 1197. (110) Singh, V. ReVised manuscript receiVed
Sci. 1997, 52, 4179.
D. Visbreaking Technology. (123) Mittal, S.; Maiti, R.
December 12, 2007
Erdoel Kohle, Erdgas, N.; Lahiri, R. N.; Ram Babu, D. Accepted January 7, 2008
Petrochem. 1986, 39, 19. Modelling and Simulation of IE0710871