Ind. Eng. Chem. Res. 1995,34, 3737-3748
3737
CATCRAK: A Process Simulator for an Integrated FCC-Regenerator System Sanjay Kumar, Ajay Chadha, Rajan Gupta, and Raj Sharma* Department of Chemical Engineering, Malaviya Regional Engineering College, Jaipur 302 01 7, India
A comprehensive integrated simulator has been developed to simulate the riser and the regenerator of a fluid catalytic cracking (FCC) process. The gas-oil feed sent to the reactor of a n FCC unit is cracked to yield gasoline and light fractions. The simulator predicts the yield of
+
C-lump ( C I - C ~ Hz),gasoline, light fuel oil, heavy fuel oil, regenerator temperature, coke burnoff, and the flue gas composition for different feed characteristics and operating conditions (temperature, pressure, catalyst circulation rate). It can also be used for trouble-shooting, optimization, and control of the process, so a s to obtain the maximum yield of cracked products. The simulator integrates Weekman’s 10-lump kinetic model and plug flow hydrodynamics for the riser reactor with Errazu et. al’s grid model for the regenerator. A reactor-regenerator heat balance has been incorporated in the simulator to calculate the required catalyst to oil ratio. On testing the simulator over a wide range of operating conditions the predicted C-lump, gasoline, light fuel oil, and heavy fuel oil yields show similar results when compared with the existing limited information in the literature.
Introduction The importance of petroleum fuels in meeting man’s energy requirements needs no elaboration. Petroleum crude oil is refined into several useful lighter products of which gasoline and light distillates are considered to be the most important. With an increasing population, a “shrinking world”, and development of rapid mass transit systems-both individual and public-there is an ever growing demand for transportation fuels. There is a natural limit to which these light fuels can be obtained from crude oil by fractional distillation alone. Newer processes and newer catalysts have been developed, and are being developed, t o meet this growing demand for low molecular weight liquids. Fluid catalytic cracking is one such process which converts high molecular weight fractions into gasoline and light distillates. Fluid catalytic cracking (FCC) has undergone several stages of development for enhanced performance and improved efficiency since the first FCC unit was commissioned during World War I1 to meet the then war-time demand for transportation fuels. Basically, an FCCU consists of a reactor with a regenerator. Cracking, or breaking down of long-chain hydrocarbons present in virgin or processed feedstocks into smaller chain hydrocarbons, takes place in the reactor with coke deposition on the surface of the zeolite catalyst leading to a loss of catalyst activity. Therefore, t o maintain catalyst activity, it is continuously regenerated in the regenerator by burning off the carbon from its surface with a stream of air. An FCCU is generally operated to maximize gasoline production and minimize coke formation to make it economically attractive. With increasing energy costs and depleting petroleum resources, it is imperative that operations of an FCCU are optimized to yield products that satisfy current demand patterns. Optimization of an FCCU by “trial-and-error”on-line is undesirable, is very expensive, and could lead t o loss of production and consequently affect profits. Process simulation, because of access to high-speed computers, provides an alternative tool for “optimization” of an FCCU that eliminates the inherent risks of the “trial-and-error” procedure.
* Author t o whom correspondence should be addressed. 0888-588519512634-3737$09.00/0
Several process models to predict the conversion of charge stock have been suggested by various researchers (Blanding, 1953; Davidson and Harrison, 1963; Kunii and Levenspiel, 1969; Voltz, 1971; Weekman, 1979). The gasoline yield, coke deposition on catalyst, light distillate yields, regenerator temperature, coke burn-off, and flue gas composition can be calculated using these models. In this work, an integrated computer simulator, based on the choice of models existing in the literature, has been developed for a reactor-regenerator system that can examine various alternatives for optimizing the cracking process and product yields. An energy balance between the reactor and the regenerator that allows calculation of the catalyst circulation rate has also been incorporated in the simulator. A graphics panel has been developed that shows the process flow of an FCCU. This simulator can also be used for monitoring, troubleshooting, and control of the fluid catalytic cracking unit.
The Fluidized Catalytic Cracking Process Kumar (1987) has traced the history of the development of the fluid catalytic cracking process as summarized in Table 1. Figure 1presents a typical modern fluid catalytic cracking process. The system represents an advanced design, side-by-side reactor and regenerator combination. This arrangement provides unsurpassed operability and a highly selective yield pattern. The reactor section features a short contact time “riser” equipped with a rapid catalyst and hydrocarbon vapor separation system to make the best use of the highly active zeolite catalyst. The feed is contacted with the hot regenerated catalyst at the base of the riser. The cracking reactions take place “instantly”,and the catalyst-vapor mixture rises in near plug flow through the riser. This catalyst-vapor mixture is then separated by passing it through a cyclone. After disengagement, the catalyst is passed through a baffled stripping zone t o displace the hydrogen-rich hydrocarbon vapors from the cooked catalyst. The catalyst regenerator provides an intimate air-catalyst contact. The operation of the regenerator is such that a low catalyst loading is
0 1995 American Chemical Society
3738 Ind. Eng. Chem. Res., Vol. 34, No. 11, 1995 Table 1. Some Important Events Relating to Catalytic Cracking (Kumar, 1987) event
year
catalytic cracking
1936
first commercial silicdalumina produced
1940
World War I1
1941- 1945
original up flow FCC thermofor catalytic cracking (TCC) synthesis and commercialization zeolite X zeolite Y zeolite cracking catalyst short contact time riser FCC
1941 1942
controlled combustion zeolite cracking catalyst
1975
1954 1959 1962 1971
Ir
greatly improved efficiency with use of natural clays (montmorillonite) as catalyst synthetic si02/A1203 with superior properties became dominant cracking catalyst great demand for aviation gasoline and other liquid fields; spurred acceleration of refining process development model I, pioneer fluid cracker first moving bed catalyst cracking process future moving bed cracking catalyst future component of zeolite cracking catalyst revolutionized catalytic cracking maximum utilization of active zeolite catalysts, minimum overcracking and coke formation promoted zeolite cracking catalysts capable of converting CO to CO2; improved yields, selectivity; lower CO stack emissions
TO FRACTIONATOR
-3
TO HEAT AND POWER
significance
SEPARATOR
/\
RECOVERY
STRIPPER
-
REGENE RAT01
STRIPPING STEAM
SPENT CATALYST TRANSFER LINE
yq
REACT 0R
-
COMBUSTION
CATALYST
-
AIR
COMBINED GAS-OIL
FEED
Figure 1. A typical modern FCC unit.
provided to the regenerator cyclones. The regenerated catalyst is sent to the reactor and the cycle is repeated.
The FCCU-Regenerator Simulation The Choice of Models. The riser reactor and the regenerator models have been integrated with the reactor-regenerator energy balance to simulate the FCC system. Several models have been proposed by various researchers for the kinetics of the cracking reactions and the hydrodynamics of the reactorlregenerator. Kumar (1987)has discussed these in detail, with their inherent limitations, leading to the choice of models used in this work. The models used for this simulator are briefly discussed below. Detailed discussion on the justifications of the choice of models and the models themselves is beyond the scope of this article, and the reader is referred t o Kumar (1987).
Figure 2. Lumped kinetic scheme. PI = wt % paraffinic molecules, 430-650 "F. N1 = wt % naphthenic molecules, 430-650 OF. CAL = wt % carbon atoms in aromatic rings, 430-650 "F. A1 = wt % aromatic substituent groups, 430-650 OF. P h = wt % paraffinic molecules, 650 OF+. Nh = wt % naphthenic molecules, 650 OF+. C f i = wt % carbon atoms in aromatic rings, 650 OF+.A h = w t % aromatic substituent groups, 650 OF+. G = G lump (Cj 430 OF). C = C lump (CIto Cq + coke). Cu = PI NI + AI = LFO, 430-650 "F. Cfi = P h + Nh f Ah = HFO, 650 OF+. Adapted nomenclature for rate constants is detailed in the figure for the paraffinic molecules. Similar rules apply for the other reaction steps.
+
Reaction Kinetics. 1. Cracking Kinetics. Weekman (1979) proposed the 10-lump cracking kinetics model. The cracking pattern is illustrated in Figure 2. Weekman has not given any values for the various kinetic constants except for their relative orders. The major difficulty encountered in this work has been the lack of data in the literature on kinetic constants for Weekman's model. Therefore, as a first approximation, the rate constants for these reactions were estimated on the basis of Greensfelder's study of cracking of pure hydrocarbons (1945). The estimated rate constants are presented in Table 2. The model equations as presented in Table 3, which have been derived from the continuity equation, are used to calculate the final product yield after the cracking reaction is complete. The coke deposited on the catalyst surface during the cracking reaction is estimated using Amoco's model (Wollaston (1975); in particular, eq 7, Table 3). The deactivated catalyst is then regenerated by burning off the carbon in a stream of air. 2. Combustion Kinetics. The coke deposited on the catalyst is burnt off in the regenerator. The kinetics of coke burning on cracking catalyst is relatively simple. Weisz and Goodwin (1966) studied the reaction and found that the kinetics is controlled by the rate of pore
Ind. Eng. Chem. Res., Vol. 34,No. 11, 1995 3739 HYDROCARBON PRODUCTS
ye
DEACTIVATED CATALYST
1
I
=h
PLUG FLOW
Variable
Range
Temperature,%
480 to 560
Pressure, am
2 to 3 20 10 40 It04
Riser length, m Riser Diameter, m
r“
8 9
10
Nh’C &-Ai Ah-G Ah-c
Cas Oil Feed
150
400 300 140 250
:
F, yo
: F,,
To, Cco
Figure 4. Schematic of the fluidized bed including the jet region, the bubbles, and the emulsion phase.
Table 2. Rate Constants Estimated Using Greensfelder’s Data for Silica-Alumina Catalyst (Greensfelder and Vage, 1945) rate constant rate constant no. reaction (lh) (k) no. reaction (lh) (k) 11 CAh-CAl 300 1 Ph-P] 400 12 CAh-c 100 250 2 Ph’G 13 P i - G 100 3 Ph-c 150 14 P i - C 80 380 4 Nh-Ni 15 N1-G 270 5 Nh-G 380 7
cSolid
Figure 3. Schematic of the riser.
6
Emulsion
Y”
Air
I 1
Regenerated Catalyst
y-
16 17
18 19 20
Ni-C Ai-G Ai-C CA1-c G-C
150 250 250 100 20
diffusion of oxygen to the coke surface. They also found that the reaction kinetics could be best described by the first order rate equation for the combustion of coke in an FCC unit. Table 4 presents the equations used for the regenerator kinetics. Hydrodynamics. 1. Reactor Hydrodynamics. Figure 3 presents the schematic of a riser reactor. The riser reactor, with the plug flow hydrodynamics, is a long channel where the mixture of catalyst an oil flows upward. The slip ratio, catalyst holdup, and catalyst circulation rate are important factors for reactor hydrodynamics. These can be calculated from equations presented in Table 3. Lower slip velocities must be maintained to obtain high conversions. 2. Regenerator Hydrodynamics. The grid model proposed by Errazu et al. (1979) has been selected to simulate the regenerator and is presented in Figure 4. This model incorporates thermal effects. It was demonstrated that a simple CSTR model without bypass of gas feed entering the bed provides a good model for representing the fluidized bed, including the grid region (Errazu et al., 1979). The correlations used for calculating various bed parameters for the regenerator are given in Table 4.
Simulator Architecture. The simulator algorithm is divided into three parts: (1)reactor simulation which provides the yield of gasoline, coke, composition of light fuel, and heavy fuel oil; (2) regenerator simulation which determines the regenerator temperature and the flue gas composition coming out of the regenerator; (3)the reactor-regenerator heat balance for calculating the catalyst circulation rate, catalyst to oil ratio and the heat required by the’reaction. Each of the above aspects is discussed below. 1. Reactor Simulation. The reactor is simulated to observe the effect of the following parameters on the conversion of gas-oil feed: feed composition; reactor temperature; weight hourly space velocity; catalyst residence time. At specified reactor and feed conditions heavy fuel oil conversion is calculated by a stepwise solution of the reactor model equation (Errazu et al., 1979) given in Table 3. The weight percent conversion of the feed stock is given by the sum of G-lump, C-lump, and light fuel oil. The weight percent coke deposited on the catalyst surface is calculated from eq 7 of Table 3. Figure 5 summarizes the steps involved in the development of the “software”. 2. Regenerator Simulation. The regenerator is simulated to observe the effect of the following parameters on coke burn-off and regenerator temperature: air flow rate; catalyst circulation rate. For an assumed regenerator temperature, the coke conversion is calculated using eq 19, Table 4. The actual regenerator temperature is then calculated using the coke conversion. This is then compared with the assumed value, and their difference is minimized. Finally, at this regenerator temperature the flue gas composition is calculated from the material balance equations shown in Table 5. The steps followed in the computer programming are presented in Figure 6. 3. Reactor-Regenerator Heat Balance. Continuous catalyst circulation is necessary for the stable operation of an FCCU. The catalyst circulation plays two functions: firstly, it brings the deactivated catalyst into the regenerator t o restore its activity and, secondly, it transfers the heat produced in the regenerator to the
3740 Ind. Eng. Chem. Res., Vol. 34, No. 11, 1995 Table 3. Model Equations for Reactor Kinetics and Hydrodynamics (Jacob et al., 1976; Voltz et al., 1971; Weekman et al., 1979) no. equations assumptions Kinetic Equations catalytic sites are nonselective
continuity equation in terms of ordinary differential equation:
first order reactions
dAJ A+( dt,) -dx (1+ Khc,h)wHsv @(tJ= exp(-at,) for ANR a = 32(ANR)0.1177
a = 31.53(ANR)'
1686 for
> 2.0
0.7 < ANR < 2.0
a = 35.98(ANR)O5515 for ANR
paraffinic > aromatic The effect of the nature of charge stock on gasoline yield is due to the individual cracking behavior of naphthenic,
paraffinic, and aromatic molecules (Greensfelder and Voge, 1945). In the absence of actual rate constants, it is reasonable to compare the trends of cracking behavior. As is evident from Figure 9, these trends are similar to the trends observed in the literature. The negative bias that is observed in the predicted values is possibly because of the inaccuracies in the rate constants that have been used. Effect of Feedstock Composition and Space Velocity on C-Lump (Cl-C* H2,Coke) Yield. The effect of the nature of charge stock on the C-lump yield as a function of space velocity is presented in Table 9 and Figure 10. As is evident, the C-lump (Cl-c4 Hz, coke) yield decreases in the following order of the nature of the charge stocks:
+
+
aromatic > paraffinic > naphthenic The aromatic charge stocks give the highest yield of C-lump due to cracking of aromatic rings into C-lump. Naphthenes crack in both the rings and side chain into fragments of three or more carbon atoms. Paraffins also crack in fragments of three or more carbon atoms. Here again a negative bias in the predicted values is observed relative to the literature values. Variation of Product Composition with Conversion. The variations in the yield of heavy fuel oil and light fuel oil constituents at different conversion levels are presented in Tables 10 and 11 and Figures 11 and
Table 7. Gasoline Yields for Three Different Charge Stocks with Varying Weight Hourly Space Velocity gasoline yield (wt %) aromatic charge stock (PA331 naphthenic charge stock (N3) paraffinic charge stock (P3) WHSV 1it.b calcd dev 1it.b calcd dev 1it.b calcd dev -8.8 40.0 31.2 43.6 -8.4 52.0 5 48.0 38.3 -9.7 -8.1 30.0 21.9 -9.3 42.0 32.7 -11.7 10 40.0 28.3 16.9 -7.1 -7.8 24.0 34.0 26.2 -7.5 15 30.0 22.5 13.8 -6.2 -6.1 20.0 21.9 -7.3 28.0 20 26.0 18.7 a
Catalyst residence time = 5 min; reactor temperature = 482.5 "C. Jacob et al., 1976.
Ind. Eng. Chem. Res., Vol. 34, No. 11, 1995 3743 Table 8. Molecular Composition of Charge Stocks (Voltz et al., 1971) charge stock
P1 P2 P3 N1 N2 N3 PN33 PA33 PA331 PA37 AA45
paraffins (wt %)
naphthenes
aromatic
(wt %)
(wt %)
aromatic ring (wt %)
51.9 40.9 46.2 11.3 8.6 9.8 27.8 31.3 17.7 30.2 11.0
33.7 36.5 35.1 68.8 59.4 64.0 49.9 30.4 26.2 23.7 14.2
5.6 15.2 10.4 13.8 21.5 17.2 12.8 22.4 33.5 11.0 43.5
8.8 7.4 8.2 6.1 10.9 9.1 9.7 15.9 22.6 35.1 31.3
+
Table 9. C-Lump (C1-G H2,Coke) Yields for Two Different Charge Stocks with Varying Weight Hourly Space Velocity"
Coke adsorption Regen. Catalyst @ Regen. Bed T
Fresh feed and
WHSV 5 10 15 20
-c
1
Heat uf reaction
Spent catalyst
@RxnT
Recycle vapor @ Rxn T
Stripping and
injection steam
@ supply T
--c Radiation &
Mix. loss (C)
REGENERATOR HEAT BALANCE
naphthenic charge stock (N2)
lit.b
calcd
dev
lit.*
calcd
dev
18 12 8 6
25.9 17.1 13.0 10.6
7.9 5.1 5 4.6
16 10 6 5
25.6 16.7 12.7 10.3
9.6 6.7 6.7 5.3
Catalyst residence time = 5 min; reactor temperature = 482.5 "C. Jacob et al., 1976. a
+-Flue gas
Spent catalyst @ Rxn T
Combustion of
-+Regen. cat @ Bed T
coke @ Regen. T Radiation & Misc. loss Combustion air @ Blower Discharge T
12. Different conversion levels are obtained by varying the weight hourly space velocity from 5 t o 30 kg of oil! (kg of catalyst-h). The weight percent composition of paraffinic (Pd, naphthenic (Nh), and aromatic substituent group (Ah) decreases with an increase in conversion. However, the concentration of the aromatic rings ( C h ) decreases slowly in comparison to others. This is expected because the aromatic rings (Cfi) are cracked to a lesser extent. Cfi cracks to yield lighter end ( C d and then further to C-lump, without forming the gasoline molecule. On the other hand light naphthenic group (NM)first attains a maximum at about 20%(by weight) conversion and then decreases continuously with an increase in conversion. However, the concentration of the light aromatic rings (CAI)increases with an increase in conversion. This may be ascribed t o the formation of more light aromatics due to the shearing off of the substituent group from the heavy (Ah) and light (AI)aromatic lumps. Product Distribution. Variation in gasoline yield, light fuel oil yield, C-lump yield, and heavy fuel oil yield are plotted in Figure 13. The trends obtained are as expected. The gasoline and C-lump yields decrease with the increase in space velocity while the light fuel oil yield first increases and then decreases to a constant value at higher space velocities. This curve would be very useful in product optimization.
Desorption of coke
't
Figure 7. Schematic diagram of reactor-regenerator heat balance.
Reactor Temperature Effects on Gasoline and C-Lumps ( C I - C ~ Hz,Coke) Yield. To observe the effect of temperature on gasoline and C-lump, lumped activation energies are assigned to groups of reactions according to the Arrhenius relation. The various lumped activation energies used are given in Table 12. Gasoline and C-lump yields as a function of space velocity for a paraffinic charge stock are presented in Table 13 and Figure 14. At temperature 482.5 OF and 453 "C and catalyst residence time of 5 min, the gasoline yield is higher than the C-lump yield. The yield of gasoline and C-lump decreases a t higher space velocities. The gasoline yield decreases as the reactor temperature increases. At lower temperatures the cracking conditions are severe enough to crack the material from the feedstock into the gasoline-boiling-range components but not severe enough to crack the material out of the gasoline boiling range into the lighter hydrocarbons. Coke Burn-Off from the Catalyst. Coke burn-off as a function of air flow rate and catalyst circulation rate is presented in Figures 15 and 16. It is observed that, with the increase in air flow rate, coke burn-off
+
Table 10. Light Fuel Oil Conversions with Varying Weight Hourly Space Velocity" light fuel oil composition (wt %) light aromatics (AI) light naphthenes (Ni) wt % cOnv of cOnv light paraffins (Pi) calcd litb dev lit.b dev calcd lit.* dev charge (WHSV) (wt %) calcd 4.0 0.08 6.03 3.4 2.63 4.08 10.0 -0.4 5 59.36 9.6 5.25 4.2 1.05 7.0 1.77 10 39.61 12.13 13.5 -1.37 8.77 1.03 5.13 4.1 -1.29 9.41 8.4 1.01 12.91 14.2 15 29.61 0.84 0.57 4.84 4.0 9.57 9.0 20 24.17 13.26 14.24 -0.98 4.56 3.8 0.76 9.5 0.08 25 20.3 13.46 14.3 -0.84 9.58 4.32 3.6 0.72 9.54 0.0 30 17.52 13.58 14.35 -0.77 9.54 a
Syn Crude vapor @ Rxn T
L'
recycle @ combined feed T
C-lump yield (wt %) paraffinic charge stock (P3)
n-
REACTOR HEAT BALANCE
Catalyst residence time = 5 min; reactor temperature 482.5 "C. Jacob et al., 1976.
light aromatic rings (CAI) calcd lit.* dev 10.97 10.5 0.47 9.0 0.38 9.38 7.99 7.8 0.19 6.9 0.11 7.01 -1.44 4.56 6.0 3.5 2.27 5.77
3744 Ind. Eng. Chem. Res., Vol. 34, No. 11, 1995
+
C-LUMP YIELD 30
L
-
--.
%)
(Wt.
& I I
Calculate flue gas compositio U
0
5
15
10
25
20
SPACE VELOCITY ( K g oil /Kg c a t ) / h r &
P3 (LIT 1
-4-
P3 (CALC)
++
N2 ( L I T )
-E-
N2 ( C A L C )
Figure 10. C-lump yield for different charge stocks as a function of space velocity. Catalyst residence time = 5 min; reactor temperature = 482.5 "C.
Print Results
COMPOSITION ( w t % ) 16 I
Yes
Figure 8. Logical diagram for integrated reactor-regenerator simulation.
1 4 1 \,, 12
-
10 -
0-
64 -
U
01 0
I
5
1
I
I
20
25
I
10
15
SPACE VELOCITY ( K g oil / Kg. cat)/hr P3 ( L I T ) 8- N3 ( C A L C )
+ P3
*
(CALC)
PA33 ( L I T )
++ N3 ( L I T ) PA33 ( C A L C !
Figure 9. Gasoline yield with different charge stocks as a function of space velocity. Catalyst residence time = 5 min; reactor temperature = 482.5 "C.
from the catalyst also increases due to the increased oxygen supply. The rate of increase of coke burn-off decreases as the air flow rate is increased beyond 30 m3/s. On the other hand, with an increase in catalyst circulation rate (700 kg/s), the coke burn-off is only 2540%, which shows the dominance of catalyst circulation rate on the regenerator operation. Low conversion of coke on catalyst will obviously result in only a partially regenerated catalyst and the conversion of the charged oil in the reactor will be affected. Regenerator Temperature. Regenerator temperature as a function of catalyst circulation rate for varying coke on catalyst and air flow rates is presented
15
20
25
30
35
40
45
50
55
60
CONVERSION ( w t % ) c
Paraffinic (Heavy)
-It A r o m a t i c s ( H e a v y )
+ Naphthenic -6-
(Heavy)
Aromatic Ring ( H e a v y )
Figure 11. Heavy fuel conversion with varying weight hourly space velocity. Catalyst residence time = 5 min; reactor temperature = 482.5 "C.
in Figures 17 and 18. As the catalyst circulation rate increases, regenerator temperature decreases significantly. A moderate temperature drop is observed for low coke concentration on catalyst (Cco= 0.0066) with an increase in catalyst circulation rate (Figure 18). It increases with an increase in air flow rate as is evident from Figure 18. This increase is due to the increased oxygen supply which causes more coke burn-off. For low catalyst circulationrate (100-200 kg/s), regenerator temperature increases with an increase in catalyst circulation rate. A linear temperature drop is observed (45-50 "C) as catalyst circulation rate is increased in steps of 100 kg/s, from 200 to 700 kg/s. The calculated values show a similar trend as observed in the literature
Ind. Eng. Chem. Res., Vol. 34, No. 11, 1995 3746 Table 11. Heavy Fuel Oil Conversions with Varying Weight Hourly Space Velocity heavy fuel oil composition (wt %) wt % conv of charge (WHSV) 5 10 15 20 25 30 a
(wt %)
heavy paraffins ( P h ) calcd 1it.b dev
heavy naphthenes (Nh) calcd lit.b dev
heavy aromatics (Ah) calcd lit.b dev
heavy aromatic rings ( C h ) calcd lit.6 dev
59.36 39.61 29.95 24.17 20.3 17.52
2.5 6.39 8.69 10.17 11.19 11.95
1.57 5.43 8.18 10.08 11.46 12.51
1.92 6.57 9.87 12.15 13.8 15.05
3.2 6.21 7.24 8.66 9.28 9.72
conv
0.7 0.69 0.89 1.17 0.69 0.64
1.8 5.7 7.8 9.0 10.5 12.0
1.27 4.9 7.8 9.2 10.3 11.4
0.3 0.53 0.38 0.88 1.16 1.11
1.5 5.6 8.0 10.8 13.0 18.0
0.42 0.87 1.87 1.35 0.8 -2.95
8.0 9.4 10.0 10.5 10.5 11.0
-4.8 -3.19 -2.76 -1.8 -1.22 -1.2
Catalyst residence time = 5 min; reactor temperature = 482.5 "C. Jacob et al., 1976.
Table 12. Activation Energy for Lumped Reactions (Jacob et al., 19'76) activation energy (kcaUmo1) no. lump 5 1 gasoline formation reaction from paraffins and naphthenes 9 2 C-lump (c1-c4 Hz,coke) formation reactions from paraffins and naphthenes 14 3 gasoline formation reactions from aromatic heavy and light compounds 20 4 C-lump formation reaction from gasoline
COMPOSITION ( W T % J
I +
50
2 t
0'
15
I
I
I
I
1
1
I
,
20
25
30
35
40
45
50
55
60
YIELD (wt.%)
40
CONVERSION ( W T % ) z
PARAFFINIC ( L I G H T )
+ NAPTHENES
++
AROMATIC ( L I G H T )
-8 AROM RING ( L I G H T )
(LIGHT)
30
Figure 12. Light fuel oil yield variation with conversion. Catalyst residence time = 5 min; reactor temperature = 482.5 "C. 20 YIELDS ( w t %)
60 10
50
0
40
30
0
5
15
10
S P A C E VELOCITY (Kg oil/Kg
-
Gasoline 402 5 C
+
C-lump 4825 C
+
20
25
cat ) / h r
Gasoline 543 C C-lump 543 C
Figure 14. Effect of temperature on yield for a charge stock (P3). Catalyst residence time = 5 min.
20
10
0 0
+
5
10
15
20
25
30
35
SPACE VELOCITY (Kg oil/Kg cat )/hr Heavy fuel oil
+
C-lump
-8 Gasoline
Light fuel oil
Figure 13. Yields of paraffinic charge as a function of space velocity. Catalyst residence time = 5 min; reactor temperature = 482.5 "C.
(Kumar, 1987). The temperature values compare well for the high catalyst circulation rate, and a deviation of 5-10 "Cis observed. A maximum error of 5% in the regenerator temperature is observed for low catalyst circulation rate which might be due t o the inaccuracy
in the calculation of total catalyst inventory in the regenerator. Influence of Change in Catalyst Circulation Rate on the Overall Coke Conversion. Overall coke conversion as a function of initial coke concentration is shown in Figure 19. It is observed that for a fxed catalyst circulation rate overall coke burn-off initially increases with increase in coke concentration and then it finally remains constant. Thus, it can be shown that two characteristic regions exits (de Lasa et al., 1981; Errazu et al., 1979). These are (a) a zone where coke conversion varies with the initial coke concentration and (b) a zone where coke conversion depends strongly on the oxygen supply. The calculated values show a similar trend and compare very well with the observed data (Errazu et al., 1979). Negligible errors were obtained in this case.
3746 Ind. Eng. Chem. Res., Vol. 34, No. 11, 1995 Table 13. Temperature Effecta in Charge Stock P3 gasoline yield 482.5 "C 543.0 "C WHSV lit.b calcd dev lit.b calcd 48.0 40.0 30.0 26.0
5 10 15 20
40.5 31.0 25.0 21.0
-7.5 -9.0 -5.0 -5.0
33.0 28.0 24.0 22.0
39.6 30.0 23.8 20.0
C-lump yield 482.5 "C
543.0 "C
dev
lit.b
calcd
dev
lit.
calcd
dev
6.6 2 -0.2 -2.0
12.0 8.0 7.0 6.0
28.6 19.13 14.7 12.1
16.6 11.13 7.7 6.1
26.0 16.0 13.0 10.0
27.4 18.2 13.9 11.4
1.4 2.2 0.9 1.4
Catalyst residence time = 5 min. *Jacob et al., 1976. C O K E B U R N OFF ( w t % )
REGENERATOR TEMPERATURE ( K )
100
1100
1000
900
1
40
201 15
20
,
1
25
30
1
35
40
45
800 0
AIR FLOW RATE ( C u rnt /s)
100
200
300
400
500
600
700
800
CATALYST CIRCULATION RATE ( K g l s )
-
C a t Circ Rate
300 K g / S e c
+
Figure 15. Coke burnoff as a function of air flow rate (effect of catalyst circulation rate). Coke on catalyst = 0.0104 kg of cokekg of catalyst; feed temperature = 472.4 "C
1050
02
0126
++ 0104
0066
-E-
REGENERATOR TEMPERATURE ( K )
1000
r
I 60
INITIAL C O K E C O N C
Figure 17. Regenerator temperature as a function of catalyst circulation rate (effect of wt % coke on catalyst). Air flow rate = 18 m3/s; feed temperature = 472.4 "C.
C O K E B U R N OFF ( w t Z )
100
6o
-
400 K g / S e c
950
i
900
401
850
1
20 0
100
200
300
400
500
600
700
800
0
-
0078
+
0104
-
0126
Figure 16. Coke burnoff as a function of catalyst circulation rate (effect of wt % coke on catalyst). Air flow rate = 25 m3/s; feed temperature = 472.4 "C.
this also shows that model equations are correctly chosen for the regenerator simulation.
Conclusions Simulator Highlights. The simulator can predict the conversion of charge stock to C-lump (CI-C~,H2),
200
300
400
500
600
700
800
CATALYST CIRCULATION RATE ( c u rnt /sec )
CATALYST CIRCULATION RATE ( C u rnt / s ) INITIAL C O K E C O N C
100
-
AIR FLOW RATE Q=40 cu mt /sec
++ Q=30 c u mt /sec
+ 8-25 c u mt /sec -8-
Q=l8 cu mt /sec
Figure 18. Regenerator temperature as a function of catalyst circulation rate (effect of air flow rate). Coke on catalyst = 0.0104 kg of cokekg of catalyst; feed temperature = 472.4 "C.
gasoline, light fuel oil, and heavy fuel oil for different operating conditions and feed characteristics. The final compositions of the heavy fuel oil and light fuel oil can be calculated. The regenerator temperature, the coke burn-off, and the flue gas composition from the regenerator can also be calculated.
Ind. Eng. Chem. Res., Vol. 34, No. 11, 1995 3747 COKE B U R N OFF (Kg coke/Kg c a t ) X 100
*2
i
I o.8
t
i
2
I
7
I
1
0
I
2
1
1
3
4
I 5
INITIAL C O K E CONC (Kg coke/Kg c a t ) X 100
-
CATALYST ClRC RATE 277 Kg/sec
++ 694
Kg/sec
+
416 Kg/sec
-0-
1382 Kg/sec
Figure 19. Influence on changing the catalyst circulation rate on the overall coke transformation. Air flow rate = 28.14 m3/s; feed temperature = 474.1 “C.
Simulator Performance. The performance of the simulator was evaluated for varying run conditions (space velocity, reactor temperature, catalyst circulation rate, and air flow rate) and was compared with the existing limited literature information. The calculated results show a good agreement with the observed values. However, a general negative deviation is observed in the case of the gasoline and light gas yields, which may be due to the inaccuracy in the rate constants used and different reactor configuration. The regenerator temperature and coke burn-off are found to be close enough to the reported value. The maximum deviation in regenerator temperature is 5% and 20% in coke burn-off calculations. Simulator Graphics. The graphics panel on computer indicates a well-illustrated working model of FCCU,with moving streams to give a clear picture of the process a t hand. Simulator Applications. The simulator can be used for preliminary calculations for designing, monitoring, and optimization of a fluid catalytic cracking unit. Suggestions for Future Work The work presented in this report lays the foundation for future development in this area. The following are some of the suggestions for carrying out further work. Better estimates of the rate constants need to be obtained for improving the accuracy of simulator predictions. Actual pilot-plant refinery data at varying run conditions would be required to “fine-tune”the simulator. A non-isothermal simulator needs to be developed to account for any variations in temperature along the length of the riser. The applicability of the simulator should be examined for use as a part of the on-line control system of the process. Studies need to be carried out for identifylng optimum product yields that satisfy demand patterns. Effect of volume expansion, as a result of cracking of higher molecular weight hydrocarbon, on conversion
needs to be incorporated in the simulator for better predictions. An economic analysis can be done to study the feasibility of any change brought in parameters.
Nomenclature a = group defined by eq 22 (Table 4) aj = specific area of the jet (l/m2) A =jet cross-sectional area (m2) A N R = aromatic to naphthenic ratio A1 = group defined by eq 5 (Table 3) AJ = concentration lump j (mol of j/g of gas) Ah = weight percent aromatic substituent group in HFO A1 = weight percent aromatic substituent group in LFO b = group defined by eq 22 (Table 4) c = group defined by eq 23 (Table 4) Cc = coke concentration in the regenerator feed (kg of cokekg of catalyst) C,. = initial coke concentration on catalyst (kg cokekg catalyst) C h = weight percent aromatic rings in HFO CCR = catalyst circulation rate C/O = catalyst to oil ratio CA = weight percent aromatic rings in LFO C p ,=~ specific heat of air (kJ/(kgK)) CP,s= specific heat of cracking catalyst (kJ/(kgK)) CP,oz= specific heat of oxygen gas (kJ/(kgK)) CP,co= specific heat of carbon monoxide (kJ/(kgK)) Cp,+O = specific heat of water vapor gas (kJ/(kgK)) C p , ~=z specific heat of nitrogen gas (kJ/(kgK)) db = bubble equivalent diameter (m) D = riser diameter (m) DC = diffusivity coefficient (m2/s) DO = hole diameter (m) F = gas volumetric flow rate at STP (m3/s) FS = cracking catalyst mass flow (kg/s) g = gravity acceleration (m/s2) G = G-lump, wt % gasoline (CsL;430 “F) h =jet penetration (m) H = catalyst holdup (kg) 12 = kinetic constant for the reaction of coke (mol of gas/(mol of 02.s)) K1, K2 = mass transfer coefficient between jet and emulsion phase (kg/(m2.s)) KJ = rate constant of jth lump (g of catalyst4cm3 h1l-l Kh = heavy aromatic ring adsorption coefficient (wt % C,h)-’ Kbc,Kce, Kbe = interchange coefficient (l/s) L = riser length (m) m, r = catalyst deactivation constants M A = molecular weight of air Mc = molecular weight of coke Nh = weight percent naphthenic molecules in HFO Nl = weight percent naphthenic molecules in LFO NO = number of holes in distributor NT = gas molar flow feed to each hole (kmol/s) NT~ = gas molar flow required to maintain the bed at minimum fluidization (kmol/s) N T= ~ N T N ~- N T (kmol/s) ~ P = absolute pressure (atm) Ph = weight percent paraffin molecules in H F O P I = air volumetric flow rate per hole at the operating temperature and pressure (m3/s) r = coke reaction rate R = gas constant WHSV = weight hourly space velocity (kg of feed/(kg of catalystah)) Qo = air flow rate in the regenerator (m3/s) S = slip ratio t = time (s) t, = catalyst residence time (min)
3748 Ind. Eng. Chem. Res., Vol. 34,No. 11, 1995
tR = bubble residence time (s) tv = vapor residence time (s) T = reactor temperature ("C) To = feed t e m p e r a t u r e of gas and solid ("C) u b = velocity of the bubble ( d s ) U, = velocity of the j e t ( d s ) Umf= m i n i m u m fluidization velocity ( d s ) u = feed velocity ( d s ) v b = bubble volume (m3) Ve = emulsion p h a s e volume (m3) V, = velocity of vapor in the riser ( d s ) V, = velocity of catalyst in the riser ( d s ) W = total catalyst weight in the regenerator (kg) X = conversion y = oxygen molar fraction y o = initial concentration of oxygen in air (mole fraction) Ye = oxygen molar fraction at the regenerator exit Yf= oxygen molar fraction at the regenerator feed Yh = oxygen molar fraction in the jet and at level h Ytr = oxygen molar fraction in a bubble with residence t i m e tR Y = oxygen molar fraction in the emulsion p h a s e z = axial coordinate (m) Greek Letters al= mass transfer p a r a m e t e r in equation 25 of Table 4
a = catalyst deactivation c o n s t a n t 6
= bed void fraction
e = vapor density ea = air density e,
= catalyst density ps = density of s p e n t catalyst (kg/m3) u = mole COdmole CO #(t,) = catalyst decay function
de Lasa; et al. Analysis of Fluidized Bed Catalytic Cracking Regenerator Models in an Industrial Scale Unit. Can. J . Chem. Eng. 1981,59, 549. Errazu, A. F.; et al. A Fluidized bed Catalytic Cracking Regenerator Model. Can. J. Chem. Eng. 1979,57,191. Fryer, C.;Potter, 0. E. Experimental Investigation of Models For Fluidized Bed Catalytic Reactors. M C h E J. 1976,22(11, 3847. Gary, J. H.; Handwerk, G. E. Petroleum Refining; Marcel Dekker Inc.: New York, 1979. Gates, B. C.; et al. Chemistry of Catalytic Processes; McGraw Hill: New York, 1979. Greensfelder, D. S.; Voge, H. H. Catalytic Cracking of Pure Hydrocarbons. Znd. Eng. Chem. 1945,37(61, 514-520. Hano, T.; et al. The Burning Rate of Coke Deposited on Zeolite Catalyst. J. Chern. Eng. Jpn. 1975,8, 127. Jacob, S. M.; et al. A Lumping and Reaction Scheme for Catalytic Cracking. M C h E J . 1976,22(41, 701. Kumar, S. Computer Aided Simulation of a Fluid Catalytic Cracking Unit. M-Tech. Thesis, Indian Institute of Technology, Kanpur, (1987). Kunii, D.; Levenspiel, 0. Fluidization Engineering; Wiley Eastern: New York, 1969. Oblad, A. G. Molecular-Sieve Cracking Catalysts. Oil Gas J . 1972, 70 (131,84. Voltz, S. E.; et al. Application of Kinetic Model for Catalytic Cracking. Znd. Eng. Chem. Process Design Dev. 1971,10, 538. Voorhies, A., Jr. Carbon Formation in Catalytic Cracking. Znd. Eng. Chem. 1945,37 (41, 318. Weekman, V. W., Jr. Lumps, Models and Kinetics in Practice; AIChE: New York, 1979. Weisz, P. B.; Goodwin, R. B. Combustion of Carbaneous Deposits within Porous Catalyst Particles. J . Catal. 1966,6 , 227. Wollaston, E. C.; et al. FCC Model Valuable Operating Tool. Hydrocarbon Process. 1975,54 (191, 93. Yates, J . G. Fluidized Bed Reactor. J . Chem. Eng. 1975,303.
Literature Cited Bischoff, K. B.; Froment, G. F. Rate Equations For Consecutive Heterogeneous Processes. Znd. Eng. Chem. Fundam. 1962,1 , 195. Blanding, F. H. Reaction Rates in Catalytic Cracking of Petroleum. Znd. Eng. Chem. 1953,45(61, 1186-1197. Davidson, J. F.; Harrison, D. Fluidized Particles; Cambridge University Press: New York, 1963.
Received for review December 14, 1994 Accepted May 17, 1995@ IE940754K
Abstract published i n Advance A C S Abstracts, August 1, 1995. @