Chemical Engineering Science 59 (2004) 5637 – 5645

www.elsevier.com/locate/ces

Ebullated bed reactor modeling for residue conversion

Jean-Marc Schweitzer∗ , Stéphane Kressmann
Chemical Engineering Department, Institut Français du Pétrole, Echangeur de Solaize B.P. 3, 69390 Vernaison, France

Received 27 February 2004

Abstract
Ebullated bed reactors are suitable for deeply converting heavy feedstocks such as residues. In this study, a dynamic reactor model is
developed taking into account hydrodynamics, catalytic and thermal kinetics. First, a validation of the hydrodynamic model is performed
on a bench scale reactor using a radioactive tracer test for residence time distribution measurements under industrial conditions. In a second
step, the complete dynamic model including kinetics is validated on the same reactor for various operating conditions. The model gives
good predictions of conversions and yields. An improvement is required to assess this model for a larger range of operating conditions.
䉷 2004 Elsevier Ltd. All rights reserved.

Keywords: Reactor; Ebullated; Modeling; Hydrodynamics; Kinetics; Residue

1. Introduction there is no possibility to add a catalyst continuously. In or-

der to extrapolate the result, a complete reactor model is
Owing to the great increase of energy consumption in the needed to simulate commercial yields and product qualities.
world, reserves of light crude oils are being depleted fast. This paper deals with this methodology of data extrapola-
In the future, there will be two main orientations for oil tion. A radioactive tracer test was conducted on this unit in
companies: order to characterize the hydrodynamics of liquid and gas
phases in industrial conditions (reaction). From these exper-
• Valorization of natural gas, iments, a complete hydrodynamic model was developed and
• Upgrading of heavy crude oils. validated describing both liquid and gas flow patterns and
gas–liquid mass transfer phenomena. Then, a complete dy-
Residue of heavy crude oils can be upgraded through various namic reactor model was developed including the hydrody-
existing processes. Among these processes, the catalytic hy- namic model and a lumped kinetic reaction scheme. A first
droconversion ebullated bed process (H-Oil) has the unique validation of this model was performed with a set of bench
feature of deeply converting heavy feedstocks with high unit data.
impurity contents while operating with a long cycle time
(Colyar and Wisdom, 1992, 1994, 1997). In this work, we
focus on the modeling of an ebullated bed reactor in which 2. Description of the ebullated bed reactor
the residue is converted into lighter cuts (fuel gas, etc.) and
a low sulfur fuel oil. For the development of ebullated bed A schematic of the ebullated bed reactor is shown in Fig. 1.
process, a new bench scale unit has been constructed. This It is a three-phase system in which the catalyst is kept in a flu-
unit operates on commercial operating conditions except that idized state through the upward lift of liquid reactants (feed,
oil and recycle) and gas (hydrogen). The height of the ebul-
lated bed catalyst is controlled by the rate of liquid recycle.
∗ Corresponding author. Tel.: +33-4-78-02-25-98; fax: +33-4-78-02-20- The main advantage of this reactor is the good mixing of cat-
09. alyst particles and the liquid phase which leads to better ther-
E-mail address: j-marc.schweitzer@ifp.fr (J.-M. Schweitzer). mal control due to recirculating flow (low thermal gradient).
0009-2509/$ - see front matter 䉷 2004 Elsevier Ltd. All rights reserved.
doi:10.1016/j.ces.2004.08.018
5638 J.-M. Schweitzer, S. Kressmann / Chemical Engineering Science 59 (2004) 5637 – 5645

D8

two phases D7
zone

D6

three phases
zone

D5
Fig. 1. Ebullated bed reactor.

Moreover, the bed plugging and the channeling due to coke

formation are eliminated which allows to operate the unit D4
for a long period (several years) because of constant and low D2
pressure drops. The catalyst used for ebullated bed is typi- D3
cally a 0.8 mm diameter extrudate with Nickel–Molybdenum
active metals (Morel et al., 1997). Due to the low reactor di-
ameter in the bench scale unit, wall–particle friction forces
are high compared to particle–particle friction forces. Then Ar D1
a high recycle ratio (about 15) with respect to the feed flow-
rate is necessary in order to fluidize the bed. However, for a
commercial unit, the reactor diameter is high (at least 5 m)
and particle–wall interactions become negligible compared Fig. 2. Experimental setup.
to particle–particle interactions. Then, a low recycle ratio of
about 3 is enough to fluidize the bed. Because of the high re-
cycle ratio in the bench scale reactor, the thermal gradient is
low, and therefore isothermal conditions may be considered.
In commercial reactor, the temperature gradient observed is Table 1
less than 5 ◦ C. Reactor operating conditions

Pressure (bar) 100–200

Temperature (◦ C) 380–430
Recycle ratio with respect to feed flow-rate (L/L) 10–20
Hydrogen flow-rate with respect to feed flow-rate (NL/L) 650
3. Hydrodynamic characterization of the ebullated bed Liquid hourly space velocity LHSV (L/L/h) 0.74
bench scale reactor

A residence time distribution study was conducted in the

bench scale reactor injecting Argon 41 as radioactive tracer
(Kressmann et al., 2000). Experiments were performed un-
der industrial conditions in terms of pressure, temperature 3.1. Experimental setup
and reaction. The aim of this study is to characterize the hy-
drodynamics of the reactor. The tracer was injected at the Fig. 2 shows the location of detectors along the reactor.
reactor bottom and several radioactive detectors were placed Only one of them (D7) is located in the two-phase zone while
along the column in order to follow the response of a pulse others are located in the ebullated bed zone. From detector
injection. Then, a model describing liquid and gas flow pat- D1, it is possible to have a characterization of the signal
terns and gas–liquid mass transfer was developed in order just before the reactor inlet. Table 1 summarizes operating
to predict detector responses. conditions of the reactor.
 J.-M. Schweitzer, S. Kressmann / Chemical Engineering Science 59 (2004) 5637 – 5645 5639

120 Table 2
Operating conditions for RTD measurements
100 Test LHSV H2 flow-rate Temp Recycle ratio
number (h-1) (NL/L) (◦ C) (L/L)
%vol of Ar in liquid

80
1 0,74 733 T 8.9
2 0,74 533 T 8.9
60 Soave
3 0,74 533 T 8.9
Peng-Robinson 4 0,74 1151 T 8.9
40 5 0,74 1151 T + 16 8.9
6 0,74 783 T + 16 8.9
20 7 0,74 783 T + 16 8.9
8 0,74 783 T + 16 23.6
9 0,74 783 T + 16 23.6
0
10 0,74 783 T + 16 23.6
0 5 10 15 20 25 30 11 0,74 480 T + 16 23.6
bench unit typical ratio recycle ratio 12 0,74 1151 T + 16 23.6
13 0,74 1151 T + 16 23.6
Fig. 3. Argon solubility. 14 0,74 360 T + 16 23.6
15 0,74 480 T + 16 19.7
16 0,74 480 T + 16 19.7

3.2. Tracer solubility

Argon 41 was used as radioactive tracer. Its period is

110 min and the gamma-ray energy is 1.29 MeV. Argon is 6000 Detector D1
gaseous at reactor operating conditions. However, when Ar- Detector D2
gon is introduced in the reactor, a part of it is dissolved in the 5000
Detector D3
liquid phase. The solubility of Argon depends on the liquid Detector D4
recycle ratio. Indeed, when this ratio increases, more Argon 4000
photon flux

Detector D5
is absorbed because of the liquid flow-rate increase. Fig. 3 Detector D6
3000
shows the evolution of Argon solubility as a function of the Detector D7
liquid recycle ratio. These gas–liquid equilibrium calcula- 2000 Detector D8
tions were performed using two thermodynamic models for
Ki estimations: Soave–Redlich–Kwong, Peng–Robinson. It 1000
clearly appears that for a typical recycle ratio in the bench
unit (10.5), 55% of Argon is dissolved in the liquid phase. 0
This means that both phases (gas and liquid) were traced 0 200 400 600 800 1000 1200 1400
simultaneously. This partial dissolving of Argon leads to a Time (s)

more complicated interpretation of RTD experiments. In- Fig. 4. Typical responses collected at different levels after 1 injection.
deed, the hydrodynamic model should take into account
gas and liquid flow patterns and also the gas–liquid mass
transfer of Argon. Unfortunately, no other tracers, which are
completely liquid and stable under industrial conditions, are
available. 3.4. Hydrodynamic model

3.3. RTD experiments The reactor is divided into two zones: the catalytic zone
and the bubble column. Liquid flow in both zones is modeled
Table 2 summarizes operating conditions of 16 tracer in- with a dispersed-plug flow model (Froment and Bischoff,
jections. The tracer test was carried out at two temperatures 1990). The dispersion is closed between the bubble column
(where T is in the range of 350–450 ◦ C), 3 recycle ratios and the ebullated bed. The gas flow is assumed to be plug-
and 6 conditions of hydrogen flow-rate. The liquid hourly flow. The gas–liquid mass transfer of Argon is taken into
space velocity (LHSV) was maintained constant. account (tracer in liquid and gas phases is not in equilibrium)
Fig. 4 shows responses of detectors after one tracer injec- (Froment and Bischoff, 1990). The liquid recycle flow is
tion. It can be observed from this figure that there is a dis- assumed to be plug-flow. Model bases are summarized in
persion of the tracer along the column. Moreover, since no Fig. 5.
gas is present in the recycle line, the second peak confirms Material balances of Argon in the liquid phase, in both
the partial absorption of Argon in the liquid phase. This is ebullated bed and bubble column, are given by Eqs. (1) and
in agreement with gas–liquid equilibrium calculations. (2), respectively.
5640 J.-M. Schweitzer, S. Kressmann / Chemical Engineering Science 59 (2004) 5637 – 5645

kl.a liquid
εs = 0
PAr *
C Ar ε1 = 1 εg2
gas
εg2 CAr

gas = plug flow Liquid = dispersed plug flow

2
Dax
kl.a

PAr
*
C Ar

CAr
Qr gas
liquid + catalyst

εg1
εs
ε1 = 1 εg1 εs

gas = plugflow Liquid = dispersed plug flow

1
D ax

Fig. 5. Hydrodynamic model basis.

Ebullated bed: The material balance in the recycle line is given by

Eq. (3).
2  
* CAr *CAr PAr
Dax
1
− ul1 − kl a CAr − *CAr *CAr
HAr − ur = .

*z 
*z 

2 (3)
 
dispersion convection gas–liquid mass transfer

*z  *t 


convection accumulation
*CAr
= . (1) Material balances of Argon in the gas phase, in both ebul-
*t 

 lated bed and bubble column, are given by Eqs. (4) and (5).
accumulation
Ebullated bed:
 
Bubble column: *PAr (1 − εg1 − εs ) PAr
− ug1 + kl aRT CAr −
εg1 HAr
2  
*z 
 
2 * CAr *CAr PAr
Dax − ul2 − k l a CAr − convection gas–liquid mass transfer
HAr

*z 
*z 

2
  *PAr
gas–liquid mass transfer
= . (4)
dispersion convection
*t 


*CAr
= . (2) accumulation
*t 

 Bubble column:
accumulation
 
*PAr (1 − εg2 ) PAr
−ug2 + kl aRT CAr −
εg2 HAr

 *z 

The liquid velocity is considered to be constant in each zone
of the reactor (ebullated bed and bubble column). Indeed, in  
convection gas–liquid mass transfer
each part of the column the gas holdup does not drastically
change due to the low gas consumption. Then, the liquid *PAr
= . (5)
holdup is assumed to be constant and, therefore, the liquid *t 


velocity. accumulation
 J.-M. Schweitzer, S. Kressmann / Chemical Engineering Science 59 (2004) 5637 – 5645 5641

Detector 4 Detector 5
1.2 1.2

1 1

0.8 0.8 exp

C norm
C norm

exp simul
0.6 simul 0.6

0.4 0.4

0.2 0.2

0 0
0 500 1000 1500 2000 2500 0 500 1000 1500 2000 2500
time s time s

Detector 6 Detector 7
1 1
0.9 0.9
0.8 0.8
0.7 0.7

C norm
C norm

0.6 exp
0.6 exp
0.5 0.5 simul
simul
0.4 0.4
0.3 0.3
0.2 0.2
0.1 0.1
0 0
0 500 1000 1500 2000 2500 0 500 1000 1500 2000 2500
time s time s

Fig. 6. Comparison model/exp. for 1 injection.

The gas velocity is considered to be constant in each zone Table 3

of the reactor. For the same reason as for the liquid velocity, Parameter estimates
this assumption is also reasonable. Parameters Optimized value Confidence interval at 95% t-value
The Henry coefficient of Argon H =0.15 bar m3 /mol (de-
εg1 9.3960E-1 8.9330E-2 9.8590E-2 40.6
termined by a gas–liquid equilibrium calculation). This co- ε g1 1.1833E-2 −2.8384E-3 2.6504E-2 1.6
efficient does not change in the range of operating condi- Dax1 (m 2 /s) 2.5979E-3 2.2967E-3 2.8990E-3 17.3
tions used for this study. Only 5 hydrodynamic parameters Dax2 (m 2 /s) 1.4184E-4 4.4459E-4 2.3921E-3 2.9
have to be estimated, which are kla (s−1 ) 3.9842E-2 3.8003E-2 4.1681E-2 43.3

• axial dispersion coefficients of the liquid phase in three

and two-phase zones Dax1 , D2 ,
ax
• gas holdup in three- and two-phase zones εg1 , εg2 , persion and mass transfer coefficients) were modeled as a
• gas–liquid mass transfer coefficient kla. function of operating conditions. The axial dispersion coef-
ficient could be predicted by the following correlation:

Dax
1
= aεl + bεg + cv sl + dv sg + eεs . (6)
3.5. Experimental results and model validation
Fig. 7shows the good agreement in the three-phase zone be-
Fig. 6 shows a comparison between the model and ex- tween this correlation and experimental data. Nevertheless,
periments for 1 tracer injection (test number 5). A good the agreement in the two-phase zone is less good due to the
agreement was obtained between the model and experimen- poor significance of hydrodynamic parameters in this zone.
tal results. Table 3 gives optimized values of parameters for A model for gas holdup was developed based on approach
one condition. A statistical analysis of parameters was done of Krishna et al. (1997). The three-phase zone works in
in order to evaluate the significance of each of them. It is the churn-turbulent regime for gas holdup point of view. In
clearly shown that parameters relative to the bubble column this region the gas holdup is determined by the following
are less significant than those in the three-phase zone. This equation:
is due to the lack of experimental information. Indeed, only
1 detector was located in the bubble column region that is εg = εgtrans (1 − εb ) + εb with
not enough for a good hydrodynamic characterization of this  
2
1
zone. The same procedure was used for the other injections εb =
1 (vsg − Vsmall εgtrans )
2 . (7)
and hydrodynamic and mass transfer parameters (axial dis- D
5642 J.-M. Schweitzer, S. Kressmann / Chemical Engineering Science 59 (2004) 5637 – 5645

2.5E-03 2.0E-03

1.8E-03
Catalytic zone
Bubble column
2.0E-03 1.6E-03

1.4E-03
Dax correlation (m2/s)

Dax correlation (m2/s)

1.5E-03 1.2E-03

1.0E-03

1.0E-03 8.0E-04

6.0E-04

5.0E-04 4.0E-04

2.0E-04

0.0E+00 0.0E+00
0.0E+00 5.0E-04 1.0E-03 1.5E-03 2.0E-03 2.5E-03 0.0E+00 5.0E-04 1.0E-03 1.5E-03 2.0E-03 2.5E-03
2
Dax model (m /s) Dax model (m2/s)

Fig. 7. Correlation validation.

The two-phase zone works in the homogenous regime from Table 4

the point of view of gas holdup. In this region the gas holdup Pseudo-component definition
is determined by the following equation: Pseudo- Temperature Molecular Specific Weight
component range weight gravity fraction
vsg
εg = . (8) (◦ C) (g/mol) (kg/m3 )
Vsmall
A 524 + 903 1073 64.78
The gas–liquid mass transfer coefficient did not change for B 343–524 393 975 31.18
all tracer injections. C 177–343 210 888 4.04
D 80–177 113 799 0

3.6. Conclusions on the hydrodynamic part

Catalytic cracking B+H2 Thermal cracking B
A complete hydrodynamic model was developed and val-
AB BC AB BC
idated with experimental data. A good agreement was ob-
tained between experiments and simulated data. From this A+H2 AC
C+H2 BD A AC
BD
C
radioactive tracer test, a modeling of hydrodynamic param-
CD CD
eters for various operating conditions was performed. Now,
we are able to represent the hydrodynamic in the bench scale AD AD
reactor for classical operating conditions used. D D

Fig. 8. Kinetic reaction schemes.

4. Reactor model
Table 5
4.1. Representation of the feedstock
Kinetic expressions

The feedstock which enters the reactor is character- Reaction Catalytic kinetic expression Thermal kinetic expression
ized by a True Boiling Point curve called TBP. From this c = k c C C 0.5 t = kt C
i→j rij ij i H2 rij ij i
curve, 4 pseudo-components were defined using a break-
down method. Each pseudo-component is characterized
by a temperature range, an average molecular weight, an
average specific gravity and a weight fraction in the feed-
stock. The feedstock we used in our test is a Canadian lumped kinetic schemes were used to describe the hydro-
heavy crude oil. The decomposition of the TBP curve into cracking and thermal cracking (Jacob et al., 1976). Fig. 8
pseudo-components is given in Table 4. shows these two schemes.
Each pseudo-component could react with or without hy-
4.2. Kinetics drogen to form lighter pseudo-components. The reaction
takes place at the solid–liquid interface (catalytic process)
Due to high process temperature (350–450 ◦ C), catalytic and in the liquid interface (thermal process). Kinetic ex-
and thermal cracking occurs simultaneously. Then, two pressions used for each reaction are given in Table 5 where
 J.-M. Schweitzer, S. Kressmann / Chemical Engineering Science 59 (2004) 5637 – 5645 5643

i = A, B, C, D, j = A, B, C, D; i is heavier than j and Table 6

k = ko e−E/RT (Arrhenius law). Operating conditions for kinetic study
The formation rate of each pseudo-component can be writ- Test LHSV H2 flow-rate Temp Recycle
ten as follows: number (h-1) (NL/L) (◦ C) ratio

1 0,752 665 T 18.4

rAc = (−rAB
c c
− rAC c
− rAD ) rAt = (−rAB
t t
− rAC t
− rAD ), 2 0,752 533 T − 10 11.5

 
  3 0,752 557 T − 25 11.5
catalytic process thermal process
rBc = (rAB
c c
− rBC c
− rBD ) rBt = (rAB
t t
− rBC t
− rBD ),
c c c c t t t t Two-phase zone:
rC = (rAC + rBC − rCD ) rC = (rAC + rBC − rCD ),
rDc c
= (rAD c
+ rBD c
+ rCD ) rDt
= (rADt t
+ rBD t
+ rCD ),  
εg2 Hi *Cil
c c
rH2 = −(rAB + rAC + rAD c c c
+ rBC + rBD + rCD ). c
(9) 1+
(1 − εg2 )RT *t
* Cil *Cil
2
Assumptions. The reaction rate follows a power law. The = Dax
2
− u l
stoichiometric coefficient of hydrogen for the whole set of *z 2 2
*z
catalytic reactions is 3. A heavier cut could be converted εg2 *(ug2 Hi Cil )
only into lighter cuts. Reaction orders are the same for all ± rit − . (13)
(1 − εg2 )RT *z
reactions. The HDO, HDS, HDN, HDM reactions are not
included in this first scheme. The gas velocity decreases along the reactor due to the gas
consumption. Another equation is necessary to calculate the
gas velocity profile. This equation is obtained summing ma-
4.3. Material balances terial balances of each specie and assuming the total pres-
sure constant. In both zones this leads to:
The gas–liquid equilibrium is assumed for hydrogen Three-phase zone:
and pseudo-components. Henry coefficients of each specie
 l 2 l 
which characterizes the partition between gas- and liquid- * i Ci 1 * i Ci * i Cil
phases is defined as follows: = Dax − ul1
*t *z2 *z
 s ε s
c
± ri
Pi yi Pt Pt 1 − εg1 − εs
Hi = =  l = Ki  l . (10) i
Cil xi i C i i Ci  εg1 Ct *ug1
± rit − . (14)
(1 − εg1 − εs ) *z
i
Partition coefficient Ki is calculated for various operating
conditions and feed composition using a Peng–Robinson Two-phase zone:
equation of state. It was shown that the composition does  2 
not strongly affect Ki . Then, these coefficients are correlated * i Cil 2 *
l
i Ci * i Cil
= Dax − ul2
(in an analog form than the Antoine law) as a function of *t *z 2 *z
pressure and temperature.  ε g C t * ug2
± rit − 2
. (15)
(1 − εg2 ) *z
i
ai
log10 (Pt Ki ) = + bi . (11)
T 4.4. Experimental results and model validation

Kinetics is included into the hydrodynamic model. The ma- Kinetic parameters were estimated using experimental
terial balance of each pseudo-component i and hydrogen data coming from the bench scale unit. Table 6 shows operat-
(with i = A, B, C, D, H2 ) in both zones may be written as ing conditions used for the model validation. The same total
follows: pressure was used for these three tests. In a first approach,
Three-phase zone: kinetic parameters of thermal cracking were not taken into
account. These parameters will be estimated in a further
 
εg1 Hi *Cil work, with more experimental data. Table 7 shows catalytic
1+ kinetic parameters at reference temperature T . A statisti-
(1 − εg1 − εs )RT *t
cal analysis will be done in a further work when more ex-
* Cil *Cil
2
s ε s
= Dax
1
− u l ± ric perimental data will be available. However, from these first
*z 2 1
*z 1 − εg1 − εs results, it may be noticed that kinetic constants are higher
εg1 *(ug1 Hi Cil ) when a heavy pseudo-component cracks instead of a lighter
± rit − . (12) one. This is in agreement with the literature data (Weitkamp,
(1 − εg1 − εs )RT *z
1975; Luzarraga and Voorhies, 1973; Haynes et al., 1983).
5644 J.-M. Schweitzer, S. Kressmann / Chemical Engineering Science 59 (2004) 5637 – 5645

Table 7 Partial pressures profiles

Kinetic parameters at reference temperature Pt-6 bars
3.5 151.6
Kinetic parameters at T Optimized value 151.4

Hydrogen partial pressure

3

A,B,C,D partial pressure

151.2
c
kAB 8 × 108 2.5 151
c
kAC 2 × 108 A 150.8
c
kAD 2 × 108 2
B
c 150.6
kBC 3 × 108 1.5 C
c 150.4
kBD 3 × 108 D
c 1 150.2
kCD 2 × 108 H2
150
Activation energy E (J/mol) 200000 0.5 Pt-8
149.8bars
0 9.6
0 0.5 1 1.5 2 2.5 3 3.5 4
80 reactor height (m)

70 Fig. 11. Partial pressures profiles.

60
% wt simulations

the three-phase zone, pressures do not change because no

50
thermal cracking was taken into account.
40 conversion
Yield of A
30
yield of B
5. Conclusions and perspectives
20 yield of C
A complete experimental study was carried out on a bench
10 yield of D
scale unit. A hydrodynamic model was developed and vali-
dated using the radioactive tracer test. Then, chemical reac-
0
0 20 40 60 80
tion was introduced in this model in order to predict reac-
% wt experiments
tor performances. A first validation was done without taking
into account thermal cracking. A good agreement is obtained
Fig. 9. Comparison model/experiments (conversion and yields). between experimental results and calculations.
More experimental data should be collected for various
operating conditions in order to assess the kinetic model. In
Yields profiles particular, the ratio between the three- and the two-phase
0.7 zones should be varied in order to estimate kinetic parame-
0.6
ters for the thermal cracking. Even though kinetics and hy-
drodynamics are very complex, the model gives promising
0.5 results.
A
yields % wt

0.4 B
0.3 C
D Notation
0.2
0.1 Cil , CAr liquid concentrations of component i and Argon,
mol/m3
0
D reactor diameter, m
0 5000 10000 15000 20000 25000
time (s)
Dax
1 , D2
ax liquid axial dispersion coefficient in three- and
two-phase zones, m2 /s
Fig. 10. Outlet time profiles of yields. E activation energy, J/mol
εg1 , εg2 gas holdups in three- and two-phase zones, di-
mensionless
Fig. 9 shows a comparison in conversion and yields of all εs solid holdup, dimensionless
species between experimental data and calculations at 3 op- εgtrans transition gas holdup between homogenous and
erating conditions. A good agreement is obtained between churn turbulent regime, dimensionless
model and experiments. Fig. 10 shows the time evolution Hi , HAr Henry coefficients of component i and Argon,
of yields at the reactor outlet. Steady-state conditions are bar m3 /mol
reached after 5 h. Fig. 11 shows steady-state partial pres- kijc ,kijt kinetic constants in catalytic and thermal pro-
sures profiles along the reactor. It can be noticed that after cesses, dimensionless
 J.-M. Schweitzer, S. Kressmann / Chemical Engineering Science 59 (2004) 5637 – 5645 5645

kla Mass transfer coefficient, s−1 Colyar, J.J., Wisdom, L.I., 1997. The H-oil process: a worldwide leader
in vacuum residue processing. 1997 National Petroleum Refiners
Ki gas–liquid partition coefficient of component i,
Association Annual Meeting, San Antonio, TX.
dimensionless Froment, G.F., Bischoff, K.B., 1990. Chemical Reactor Analysis and
Pi , PAr partial pressures of component i and Argon, bar Design, second ed., Wiley Series in Chemical Engineering. Wiley, New
or Pa York, pp. 259, 446.
Pt total pressure, bar or Pa Haynes, H.W., Parcher, J.F., Helmer, N.E., 1983. Hydrocracking of
ric , rit kinetic formation rate of component i in cat- polycyclic hydrocarbons over a dual-functional zeolite (faujasite)-based
catalyst. Industrial Engineering Chemical Process and Development 22
alytic and thermal processes (3), 401–409.
rijc , rijt kinetic reaction rate in catalytic and thermal pro- Jacob, S.M., Gross, B., Voltz, S.E., Weekman, V.W., 1976. A lumping
cesses, mol/kg/s mol/m3 /s and reaction scheme for catalytic cracking. A.I.Ch.E. Journal 22 (4),
R ideal gas constant, J/mol/K 701–713.
Kressmann, S., Boyer, C., Colyar, J.J., Schweitzer, J.M., Viguié,
T temperature, K
J.C., 2000. Improvements of ebullated-bed technology for upgrading
ug1 , ug2 gas velocities in three- and two-phase zones, m/s Heavy oils. Oil & Gas Science and Technology-Revue IFP 55 (4),
ul1 , ul2 liquid velocities in three- and two-phase zones, 397–406.
m/s Krishna, R., De Swart, J.W.A., Martina, G.B., Maretto, C., 1997. Gas
ur recycle velocity, m/s holdup in slurry bubble columns: effect of column diameter and slurry
concentrations. A.I.Ch.E. Journal 43 (2), 311.
vsg , vsl superficial gas and liquid velocities, m/s
Luzarraga, M.G., Voorhies, A., 1973. Hydroisomerization and hydro-
Vsmall velocity of small bubbles, m/s cracking of cyclohexane in the presence of a palladium–hydrogen–
faujasite-catalyst. Industrial Engineering Chemical Product Research
and Development 12 (3), 194–198.
References Morel, F., Kressmann, S., Harlé, V., Kasztelan, S., 1997. Processes and
catalysts for hydrocracking of heavy oil and residues. Studies in Surface
Colyar, J.J., Wisdom, L.I., 1992. Upgrading vacuum residue from Mexican Science and Catalysis 106, 1.
crudes for petroleos mexicanos hydrodesulfurization residue complex Weitkamp, J., 1975. The influence of chain length in hydrocracking
miguel hidalgo refinery. Japan Petroleum Institute, Tokyo, Japan. and hydroisomerization of n-alkanes. In: Ward, J., Qader, S.A. (Eds.)
Colyar, J.J., Wisdom, L.I., 1994. Second generation ebullated-bed Hydrocracking and Hydrotreating A.C.S. Symposium Series, vol. 20.
technology. JPI Petroleum Refining Conference, Tokyo, Japan. pp. 1–27.