5288
Ind. Eng. Chem. Res. 2010, 49, 5288–5298
Computational Fluid Dynamics Simulation of Fluid Catalytic Cracking in a Rotating Fluidized Bed in a Static Geometry Waldo Rosales Trujillo* and Juray De Wilde* Unite´ IMAP, UniVersite´ catholique de LouVain, Place Sainte Barbe 2, 1348 LouVain-la-neuVe, Belgium
The application of a rotating fluidized bed in a static geometry to fluid catalytic cracking is evaluated by means of computational fluid dynamics (CFD) simulations using an Eulerian-Eulerian model and the kinetic theory of granular flow. The reactions are described by a 10-lump model. First, the reaction kinetics is based on currently allowable cracking temperature and catalyst activity. Typical reactor dimensions required are presented, and an evaluation of the process intensification potential is made, based on a comparison with riser technology. Next, the possibility of using a higher cracking temperature or a more active catalyst is evaluated. Introduction Rotating Fluidized Bed in a Static Geometry. Rotating fluidized beds in a static geometry (RFB-SG) have been recently developed.1-3 The rotational motion of the particle bed is generated by the tangential injection of the fluidization gas in the fluidization chamber via multiple slots (Figure 1). Tangential and radial fluidization of the particle bed can be combined. In the radial direction, the gas-solid drag force is counteracting the centrifugal force. The latter can be multiple times gravity, allowing operating conditions different from those in conventional fluidized beds and overcoming their limitations. In particular, small particle bed heights and high particle bed densities at high gas velocities are possible. High-G operation was also shown to allow the fluidization of smaller particles.4 A particular feature of RFB-SGs is the flexibility in the fluidization gas flow rate, the latter affecting the centrifugal force and the counteracting radial gas-solid drag force in a similar way.2,3 An experimental proof of concept of RFB-SGs was given by De Wilde and de Broqueville (2007),1 who also studied their basic hydrodynamics, stability, and heat transfer characteristics.2 Both experiments and computational fluid dynamics (CFD) simulations demonstrated the excellent particle bed mixing in RFB-SG,2,3 for example resulting in a particle bed temperature uniformity much better than in conventional fluidized beds. Preliminary CFD calculations also showed an improvement of the fluidization when using a polygonal instead of cylindrical reactor chamber, due to a reduced solid’s shear with the chamber’s outer wall. Challenges and Opportunities. In a RFB-SG, the particle bed height is of the order of centimeters, compared to the order of meters or tenths of meters in conventional fluidized beds or risers. The radial gas velocities can be comparable to the axial gas velocities in risers or much higher.1-3 However, because of the negligible radial motion of the particles in RFB-SGs, the radial gas-solid slip velocity can be much higher than in conventional fluidized beds or risers, which is advantageous for gas-solid mass and heat transfer. The combination of small bed heights and high radial gas velocities causes the gas phase residence time in RFB-SGs to be easily 1-2 orders of magnitude smaller than in conventional fluidized beds or in risers. When * To whom correspondence should be addressed. E-mail:
[email protected] (W.R.T.);
[email protected] (J.D.W.).
used as a reactor, this may be limiting the conversion. To compensate for the smaller gas phase residence times, RFBSGs offer particle bed densities significantly higher than those in risers. Furthermore, High-G fluidization was shown to improve the particle bed uniformity, bubbling being suppressed. As such, bypassing of solids (catalyst) by the gas, encountered in conventional fluidized beds, can be avoided in RFB-SGs and the gas-solid contact improved. Finally, the improved particle bed mixing and related uniformity may allow the use of higher reaction temperatures and/or a more active catalyst. To evaluate the impact of the different above-mentioned factors and the process intensification potential of RFB-SG type reactors, CFD simulations are carried out. The fluid catalytic cracking (FCC) process is simulated because it is the most important industrial process using fluidized bed, more particular, riser technology, and because of the availability of data in the literature. First, the intrinsic process intensification potential of RFBSGs is evaluated, that is, using the same catalyst and kinetics used in the riser simulations5 and a typical cracking temperature of 775 K. Next, operation at higher cracking temperatures and the use of a more active catalyst are evaluated. In contrast with
Figure 1. Schematic representation of the rotating fluidized bed in a static geometry reactor (RFB-SG).
10.1021/ie901610f 2010 American Chemical Society Published on Web 03/30/2010
Ind. Eng. Chem. Res., Vol. 49, No. 11, 2010 Table 2. Constitutive Equations
Table 1. Continuity Equations mass
∂ (ε F ) + ∇ · (εgFgb u ) ) -S˜coke ∂t g g ∂ (ε F ) + ∇ · (εsFsb V ) ) S˜coke ∂t s s
species
momentum
source terms
solid phase pressure
ps ) 2Fs(1 + ess)εs2g0,ssΘs
radial distribution
g0,ss )
3εs 1 + (1 - εs) 2(1 - εs)2
∂ (ε F w ) + ∇ · (εgFgb u wi) ) Si (for lump i) ∂t g g i
solid phase collisional viscosity
∂ (ε F w ) + ∇ · (εsFsb V wcoke) ) S˜coke ∂t s s coke
solid phase kinetic viscosity
µs,kin )
solid phase frictional viscosity
µs,fr )
gas-solid drag force
εs e 0.2:
∂ (ε F b u ) + ∇ · (εgFgb ub u ) ) -εg∇p + ∂t g g ∇ · cτg + β(V b-b u) ∂ (ε F b V ) + ∇ · (εsFsb Vb V ) ) -εs∇p ∂t s s ∇ps + ∇ · τcs + β(u b-b V)
granular temperature
5289
6-8
3 ∂ (F ε Θ) + ∇ · (Fsεsb V Θ) ) (-pscI + 2 ∂t s s cτs):∇V b + ∇ · (κΘ∇Θ) - γΘ - 3βΘ
[
]
S˜coke ) fcokeMcoke
∑R
CkνCk
4 µs,col ) εsFsdsg0,ss(1 + ess) 5
CD )
εsdsFs√Θsπ 2 1 + (1 + ess)(3ess - 1)εsg0,ss 6(3 - ess) 5
[
]
ps sin φ 2√I2D V -b u | -2.65 3 εsεgFg | b β ) CD εg 4 ds
24 [1 + 0.75(εgRes)0.687] εgRes
εs > 0.2:
β ) 150
εs2µg 2
εgds
+ 1.75
V -b u| Fgεs | b ds
k
Si ) Mi(
∑R ν
ik ik
k
k
-
∑R ) ji
j
increasing the cracking temperature, increasing the catalyst activity accelerates all reactions with the same factor. The CFD simulations of FCC in a RFB-SG also illustrate the effect of strong density variations (due to cracking) on the hydrodynamics of RFB-SGs, a crucial issue for their application as a reactor. Simulation Model Continuity Equations. The Eulerian-Eulerian two-phase model allows for the modeling of two fully penetrating and interacting phases. Mass, momentum, and species continuity equations are solved for each phase (Table 1). Gas phase density variations due to cracking are accounted for. The solid phase physical properties are calculated from the kinetic theory of granular flow,6 which requires the solution of an additional continuity equation for the granular temperature, also shown in Table 1. The constitutive equations derived from the KTGF are shown in Table 2. The restitution coefficient of particle-particle collision, ess, was given a value of 0.9. The gas-solid interactions include momentum and mass transfer. The latter accounts for coke deposition on the catalyst. Gas-solid species mass and heat transfer limitations are neglected due to the high gas-solid slip velocities and related gas-solid mass and heat transfer coefficient.2 Furthermore, heat transfer and mixing in the rotating particle bed were shown to be extremely fast.1 If the time scale of the heat transfer and particle bed mixing are assumed relatively small, that is, compared to the coking time scale and the related average residence time of the particles in the reactor, the temperature and catalyst coke content in the reactor can be assumed uniform and can be imposed in the simulations. In case the catalyst residence time is low to limit the increase in catalyst coke
content, a uniform coke content assumption is also justified. For the calculation of the gas-solid momentum transfer, the drag coefficient model of Gidaspow6 was used. A Reynolds-averaged approach was taken. Turbulence was accounted for, using the k-ε turbulence model for each phase. The grid was sufficiently refined to obtain a grid-independent solution and to calculate possible mesoscale phenomena, like bubbling or clustering. This also requires a nonstationary (transient) calculation. The mesh size is as small as 0.2 mm, and the time step is 1 × 10-5 s. Calculations were continued until a statistically stationary state was reached. Boundary conditions at solid walls are based on a no-slip behavior for the gas phase, and a partial slip behavior for the solid phase. The Johnson-Jackson model9 was adopted, using a specularity coefficient of 0.2 and a particle-wall restitution coefficient of 0.9. Reaction Mechanism and Kinetics. The ten-lump model of Jacob et al.10 was used to describe the catalytic cracking of gas oil. A scheme showing the lumps, as well as the possible reactions, is shown in Figure 2. The molar mass of the different lumps, as well as initial feed composition are shown in Table 3. The reaction rate for the formation of lump i out of lump j is calculated from Aije-(Eij/ RT ) F ε φC 1 + KArhwArh s s j (
Rij )
)
(1)
The pre-exponential factors and the activation energies for the conventional catalyst are listed in Tables 4 and 5, respectively. The effect of the adsorption of heavy aromatics is accounted for via the nondimensional term (1 + KArhwArh)-1. Catalyst deactivation is modeled as a function of the coke content on the catalyst by φ ) (1 + 69.47(100wcoke)3.8)-1. The C lump is considered as a mixture of coke and light gases (C1-C4). The CFD model associates coke with the solid phase. A given fraction fcoke of the C lump was considered to be coke,
5290
Ind. Eng. Chem. Res., Vol. 49, No. 11, 2010 Table 3. Gas Oil (GO) Composition and Molar Mass of the Different Lumpsa
Ph Nh Ash Arh G a
wi (%)
Mi
17 20 24 12 0
339 339 339 339 114
Pl Nl Asl Arl C
wi (%)
Mi
14 9 2 2 0
226 226 226 226 34
Data from the work of Jacob et al.10
The code describing the kinetic model was validated by repeating the 1D riser simulations described in the work of Froment and Bischoff.5 Simulation Cases and Setup
Figure 2. Ten-lump model for the catalytic cracking of gas oil.
10
Figure 3. (a) Periodic calculation domain and (b) 12-slot polygonal body reactor.
the remaining fraction being light gases. A value fcoke of 0.19 was used. For the calculation of the physical properties of the lump of light gases, propane was taken as the representative component.
The problem is simulated in 2D using FLUENT v.3.6.26 with user defined functions to describe the reaction kinetics. The calculation domain is shown in gray in Figure 3a and represents a periodical section of a 12-slot polygonal body reactor, as shown in Figure 3b. A section of the gas distribution chamber is included to prevent imposing the gas velocity directly at the particle bed boundary. The gas distribution chamber allows for pressure changes in the bed to be transported upstream the injection slot, reflecting the physical reality. Catalyst is continuously fed via a side inlet, nearly tangentially. When reaching the statistically stationary state, catalyst losses via the chimney compensate for the continuous feeding of catalyst. A separate catalyst outlet in the dense particle bed region could be considered and was, for example, used in the experiments of De Wilde and de Broqueville.1 The reactor dimensions and operating conditions are summarized in Table 6. First, FCC using a conventional catalyst and cracking temperature of 775 K is evaluated. Gas oil conversion, gasoline, and light gasses yield and selectivity are compared with values obtained using a 30 m tall riser reactor, taken from the work of Froment and Bischoff,5 and the process intensification potential is discussed. Next, operation at higher cracking temperatures or using a more active catalyst and the related additional process intensification potential is evaluated. Convergence was assumed to be reached when the residuals dropped below 1 × 10-6 for the energy equation and 1 × 10-3
Figure 4. Contour plot of the solids volume fraction in the reactor: (a) accounting for the effect of density variations due to cracking of hydrocarbons, (b) not accounting for density variations, solids loading as in a, and (c) not accounting for density variations, solids feeding rate as in a. For reactor characteristics and operating conditions, see Table 6.
Ind. Eng. Chem. Res., Vol. 49, No. 11, 2010 Table 5. Activation Energies From the Work of Jacob et al.
5291
10
reaction
Eij
gasoline formation reactions from Ph, Nh, Pl, and Nl gasoline formation reactions from Ash and Arh C lump formation reactions from Ph, Nh, Pl, and Nl C lump formation reactions from Ash, Arh, Asl, and Arl formation of C lump from gasoline formation of Pl, Nl, Asl, and Arl from Ph, Nh, Ash, and Arh
23045 60755 35615 73325 83800 33939
Table 6. Reactor Dimensions and Operating Conditions distribution chamber number gas of inlets 12 gas inlets width 15 gas velocity at the inlets 19.6 gas density 4.72 reactor number of slots 12 slots width 3 number of catalyst inlets 12 catalyst inlets width 2.6 reactor outer radius 0.6 reactor inner radius (chimney) 0.2 particle diameter 80 particle density 1500 catalyst velocity at the inlets 15 catalyst volume fraction at the inlets 0.3 catalyst loading 23 catalyst coke content 0.15
Figure 5. Radial profile of the tangentially averaged solids volume fraction in the RFB-SG and in a riser reactor. Profiles for the RFB-SG: (a) accounting for the effect of density variations due to cracking of hydrocarbons, (b) not accounting for density variations, solids loading as in part a, and (c) not accounting for density variations, solids feeding rate as in part a. For reactor characteristics and operating conditions, see Table 6. Riser data is from the work of Froment and Bischoff.5
for the other continuity equations. Furthermore, the overall mass balances were verified. Typically, reaching a statistically steady state required 60 s, corresponding to 600 h of calculation time on a single processor.
mm m/s kg/m3 mm mm m m µm kg/m3 m/s kg/m wt %
between the particle bed and its freeboard region less pronounced. The latter implies that it becomes more difficult to avoid particle losses via the chimney. The particle bed uniformity is, however, slightly improved. Similar conclusions can be drawn from Figure 5. The latter also suggests that the radial solids volume fraction gradient in the particle bed (slope of the curve) is affected by the solids feeding rate, whereas the density variations due to reactions cause the vertical shift in the solids volume fraction profile. The loss of a well-separated particle bed freeboard is resulting both from the density variations due to reactions and from the continuous solids feeding required to maintain a sufficiently high solids loading in the reactor. Values for the solids volume fraction in the RFB-SG range from 0.15 to 0.28, with an average of 0.225. This is about 7.5 times higher than the average value in the riser. The low solids volume fraction near the outer wall in the RFB-SG is caused by the gas injection and was confirmed experimentally.1,3 Bubbling does not occur and a stationary, rather than statistically stationary, state is achieved. Figure 6a and b shows the contour plots of the fluid phase density and the static pressure field. In Figure 6c, radial profiles of the tangentially averaged tangential and radial solid phase velocities are shown. In the latter, the effect of the variation of the density due to cracking is again illustrated and found to be limited. Additionally, Figure 6c shows the effect of solids feeding on the particle bed tangential and radial velocities. Solids being fed nearly tangentially, tangential momentum is added to the particle bed. The radial velocity is mainly affected in the freeboard region, where solids are forced to leave the reactor
Results and Discussion Conventional Cracking Catalyst and Temperature. Figure 4a shows the contour plot of the solids volume fraction in the RFB-SG. Except in the immediate vicinity of the outer reactor wall, where the effect of the polygonal design and a limited nonuniformity due to the gas and solids injection are visible, the tangential uniformity of the solids volume fraction profile is acceptable. A radial profile in the range 0.2 < R < 0.6 m of the tangentially averaged solids volume fraction in the RFBSG is shown in Figure 5 in which a comparison is made with the axial profile of the cross-sectional averaged solids volume fraction in the riser. The coordinates were normalized for comparison purposes. Figure 4b and c illustrate the effect of density variations due to cracking on the hydrodynamics in RFB-SGs. Contour plots of the solids volume fraction in the RFB-SG for the given gas flow rate, but in the absence of density variations due to reactions are shown in Figure 4b and c. In Figure 4b, the solids loading is kept the same as in Figure 4a, which required to stop the solids feeding. In Figure 4c, the solids feeding (and loss) rate is kept equal to that of Figure 4a; but then the RFB-SG is operated at higher solids loading. The relationship between the solids loading in the RFB-SG and the rate of solids losses via the chimney was demonstrated by De Wilde and de Broqueville.3 As a result of the density variations due to cracking, the particle bed is seen to become less dense and the separation Table 4. Pre-exponential Factors Aij From the Work of Jacob et al.10 Ph Ph Nh Ash Arh Pl Nl Asl Arl G C
0
Nh 0 0
Ash 0 0 0
Arh 0 0 0 0
Pl 1.28 0 0 0 0
Nl 0 1.39 0 0 0 0
Asl 0 0 1.17 0 0 0 0
Arl 0 0 3.09 3.61 × 10-1 0 0 0 0
G
C -1
6.00 × 10 9.21 × 10-1 2.78 × 10-2 0 2.6 × 10-1 7.2 × 10-1 8.18 × 10-1 0 0
6.3 × 10-1 1.20 1.12 × 10-3 4.79 × 10-2 7.6 × 10-1 6.6 × 10-1 1.18 × 10-2 3.27 × 10-1 7.65 × 10-2 0
5292
Ind. Eng. Chem. Res., Vol. 49, No. 11, 2010
Figure 6. Contour plots of (a) static pressure and (b) fluid phase density in the RFB-SG. (c) Radial profiles of the tangentially averaged tangential and radial solid phase velocities in the RFB-SG, accounting or not for the effect of density variations due to cracking on the hydrodynamics. For reactor characteristics and operating conditions, see Table 6. Table 7. Lump Mass Fractions and GO Conversion Comparison between RFB-SG and the Riser Reactor Technologya wi (%)
Ph
Nh
Ash
Arh
Pl
Nl
Asl
Arl
G
LG
coke
χGO (%)
RFB-SG riser
10.12 6.14
9.41 4.02
7.73 1.89
10.32 8.7
12.90 11.77
7.24 5.37
3.26 3.51
6.88 9.23
23.38 37.54
7.09 9.58
1.67 2.25
32.14 49.36
a For RFB-SG reactor characteristics and operating conditions, see Table 6. RFB performance with conventional catalyst and cracking temperature of 775 K. The riser data is from the work of Froment and Bischoff.5
Table 8. Gasoline Conversion, Light Gases Conversion, Gas Oil Conversion, Gasoline Selectivity, Average Coking Rate, Average Coking Rate Per Unit Mass Catalyst, Allowable Catalyst Residence Time, Corresponding Number of Turns in a RFB-SG, and Average Solids Feeding or Regeneration Rate for the Different Cracking Temperaturesa T [K]
χG [%]
χGO [%]
χG/χGO [%]
χLG [%]
〈Rc〉 [kmol/(m3 s)]
〈R′〉 × 104 [kgcoke /(kgcat s)]
τcat [s]
N [-]
m ˙ cat [kg/s]
775 800 820 840 860 880
23.38 25.12 26.16 27.13 28.38 28.83
32.14 35.85 38.55 41.38 45.02 47.63
72.75 70.07 67.84 65.55 63.03 60.53
7.09 8.68 10.03 11.53 13.46 15.21
0.0016 0.0019 0.0022 0.0025 0.0030 0.0034
0.93 1.16 1.38 1.63 1.89 2.20
0.81 0.64 0.54 0.46 0.40 0.34
4.03 3.22 2.70 2.28 2.01 1.71
181.4 222.2 256.9 295.1 344.7 389.3
a For RFB-SG reactor characteristics and operating conditions, see Table 6. A conventional cracking catalyst was used. This is based on a maximum increase of the catalyst coke content of 5%.
when solids are fed. The average radial gas velocity is 2.67 m/s, implying a gas phase residence time of 0.15 s. The mean particle bed rotation speed is 15 m/s, corresponding to a centrifugal acceleration of approximately 38G at the outer wall of the reactor. As such, the average solids rotation period is 0.19 s.
Figure 7 shows the gas oil, gasoline, and light gases mass fraction contours in the RFB-SG reactor. The gas oil mass fraction is calculated as the sum of the mass fractions of the different lumps of which gas oil is composed. The gas oil conversion χGO can be defined as 1 - wGO, the conversion into gasoline as χG ) wG and that of light gases as χLG ) wLG. The
Ind. Eng. Chem. Res., Vol. 49, No. 11, 2010
5293
Figure 7. Contour plots of the mass fraction of (a) gas oil, (b) gasoline, and (c) light gases. (d) Radial profiles of the tangentially averaged gas oil conversion, conversion into gasoline, and conversion into light gases in the RFB-SG. For reactor characteristics and operating conditions, see Table 6. The conventional catalyst and cracking temperature is 775 K.
radial profiles of the tangentially averaged gas oil conversion and conversion into gasoline and into light gases are shown in Figure 7d. A maximum gas oil conversion of 31.24% and conversion into gasoline of 23.38% is obtained in the RFBSG. Table 7 compares the mass fraction of the different lumps and the gas oil conversion obtained at the outlet of the RFBSG and the riser reactor.5 A comparison between the gas oil and gasoline conversion profiles in the RFB-SG and in the riser is shown in Figure 8. For the given RFB-SG geometry and operating conditions, results indicate somewhat lower conversions compared to the riser. Essential is, however, the process intensification factor, which accounts for the differences in gas oil flow rate and reactor volume, as discussed hereafter. If required, the conversion in the RFB-SG can be increased by increasing the gas phase residence time (increasing the particle bed height or reducing the gas flow rate), or by increasing the reaction rates (increasing the operating temperature or catalyst activity). The latter two options will be considered in more detail further in this paper. The process intensification (PI) potential can be defined as the ratio of the amount of gasoline produced per unit time and per unit volume reactor (2).
Figure 8. Gas oil conversion profile and conversion into gasoline profile: comparison between RFB-SG and riser type reactor. For RFB-SG characteristics and operating conditions, see Table 6. The conventional catalyst and cracking temperature is 775 K. Riser data is from the work of Froment and Bischoff.5
5294
Ind. Eng. Chem. Res., Vol. 49, No. 11, 2010
PI(χGO) )
XGRFB(χGO) XGriser(χGO)
where
XG(χGO) )
wG(χGO)m ˙ GO V(χGO) (2)
In (2), wG(χGO) is the gasoline mass fraction produced for a given gas oil feed rate, m ˙ GO, and gas oil conversion χGO. V(χGO) is the particle bed volume required to achieve this gas oil conversion at the given gas oil feed rate. The PI factor was calculated as a function of the gas oil conversion, χGO, and is shown in Figure 9. The PI factor increases with the gas oil conversion and values as high as 7 can be easily achieved. This is mainly due to the higher particle bed density and the improved particle bed uniformity. Finally, the high cat-to-oil ratios at which RFB-SGs can be operated should be remarked. Typical values in riser reactors are of the order of 10, maximum 20.5 In the RFB-SG simulated, the gas and catalyst motion is mainly cross-flow and the catto-oil ratio, calculated from 3, amounts to roughly 500. cat-to-oil ratio )
εsFs〈Vtang〉 εgFg〈urad〉
Figure 9. Process intensification potential of RFB-SG for the catalytic cracking of gas oil as a function of the gas oil conversion: comparison with riser technology.5 For RFB-SG reactor characteristics and operating conditions, see Table 6. RFB-SG operated with conventional catalyst and cracking temperature of 775 K.
(3)
The maximum allowable average solids residence time can be estimated in terms of the allowable coke content variation, at the given coking rate profile, Rc. If an average 5% coke content increase is admitted: τcat )
0.05wcokeεsFs Mcoke〈Rcoke〉
where
Rcoke ) fcoke
∑ν
CkRCk
k
(4) The required catalyst feed or regeneration rate can then be calculated from m ˙ cat )
εsFsVbed τcat
(5)
A contour plot of the calculated coking rate in the RFB-SG is shown Figure 10, an average value is shown in Table 8 and allows an average particle residence time of 0.81 s when using the conventional catalyst and cracking temperature of 775 K. This implies that the catalyst is to be replaced on average every 4.03 turns, which requires a catalyst feeding or regeneration rate of 181.4 kg/s per meter of reactor axial length. In the simulation, the catalyst feeding rate was 702 kg/s per meter of reactor axial length. An alternative approach for facilitating the treatment of the high solids mass flow rates and to be considered in more detail in future work is the combined catalytic cracking and catalyst regeneration, that is, in a single vessel, using a multizone concept type RFB-SG reactor. An illustration of a possible multizone concept is shown in Figure 11. Gas oil is injected in the cracking zone, whereas air is injected in the catalyst regeneration zone. The two zones can be separated from each other by additional “buffer” zones in which steam is injected. The multizone concept also has the potential of increasing the efficiency of the heat transfer between catalyst regeneration and catalytic cracking, allowing further process intensification. Conventional Catalyst, Increased Cracking Temperature. The excellent particle bed mixing properties of RFB-SGs and related particle bed uniformity2 open perspectives for increasing the cracking temperature or using a more active catalyst. First, simulations were carried out to determine the
Figure 10. Catalyst coking rate (kmol/m3 s) in the RFB-SG, operated with conventional catalyst and cracking temperature of 775 K. For RFB-SG reactor characteristics and operating conditions, see Table 6.
Figure 11. Tangential multizone concept: injection of different gases through the different gas inlet slots.
influence of a higher cracking temperature. The reactor characteristics and other operation conditions were kept identical. The main results are summarized in Table 8 and Figures 12 and 13. The tangential uniformity of the conversion profiles remains acceptable, and Figure 12 shows the radial profiles of tangentially averaged values for different cracking temperatures.
Ind. Eng. Chem. Res., Vol. 49, No. 11, 2010
5295
Figure 12. Radial profiles of the gasoline mass fraction, the light gases mass fractions, the gas oil conversion, and the gasoline selectivity, for different cracking temperatures. For RFB-SG reactor characteristics and operating conditions, see Table 6. Conventional FCC catalyst activity 1a.
Figure 14. Solids volume fraction dynamics over a 5 s period, in a point at R ) 0.32 m: comparison for two catalyst activities, 1a and 5a. Figure 13. Process intensification potential of RFB-SGs for the catalytic cracking of gas oil as a function of the gas oil conversion: comparison with riser technology.5 For RFB-SG reactor characteristics and operating conditions, see Table 6. RFB-SG operated with a conventional catalyst and at different cracking temperatures.
In the cracking temperature range studied, the gas oil conversion increases almost linearly with the cracking temperature, as shown in Figure 12c. The conversion into gasoline increases as well, but less rapidly than the gas oil conversion increases, so that the gasoline selectivity χG/χGO gradually
decreases, as shown in Figure 12d. At temperatures above 900 K, the conversion into gasoline will decrease with increasing cracking temperature, overcracking occurring and the reactions producing coke and light gases becoming dominant. In case the production of light gases is aimed at, operation at higher temperatures is advantageous as illustrated in Figure 12b showing the evolution of the mass fraction of light gases in the RFB-SG for different cracking temperatures.
5296
Ind. Eng. Chem. Res., Vol. 49, No. 11, 2010
Table 9. Gasoline Conversion, Light Gases Conversion, Gas Oil Conversion, Gasoline Selectivity, Average Coking Rate, Average Coking Rate Per Unit Mass Catalyst, Allowable Catalyst Residence Time, Corresponding Number of Turns in an RFB-SG, and Average Solids Feeding or Regeneration Rate for the Different Catalyst Activitiesa act [-]
χG [%]
χGO [%]
χG/χGO [%]
χLG [%]
〈Rc〉 [kmol/(m3 s)]
〈R′〉 × 104 [kgcoke /(kgcat s)]
τcat [s]
N [-]
m ˙ cat [kg/s]
1a 1.5a 2a 3a 4a 5a 6a
23.38 27.33 31.17 38.47 41.21 43.02 44.67
32.14 37.81 43.41 54.78 59.69 63.38 67.01
72.75 72.28 71.82 70.23 69.05 67.88 66.66
7.09 8.47 9.90 13.20 14.94 16.46 18.06
0.0016 0.0019 0.0022 0.0029 0.0033 0.0037 0.0040
0.93 1.24 1.51 1.88 2.29 2.67 3.01
0.81 0.60 0.50 0.40 0.33 0.28 0.25
4.03 2.97 2.39 2.08 1.69 1.44 1.28
181.4 218.3 253.0 337.1 383.2 422.9 464.3
a For RFB-SG reactor characteristics and operating conditions, see Table 6. A conventional cracking catalyst was used. This is based on a maximum increase of the catalyst coke content of 5%.
Figure 15. Radial profiles of the gasoline mass fraction, the light gases mass fraction, the gas oil conversion, and the gasoline selectivity for different catalyst activities. For RFB-SG reactor characteristics and operating conditions, see Table 6. A conventional cracking temperature of 775 K was used.
Figure 13 shows that by allowing higher cracking temperatures, the FCC process in a RFB-SG can be further intensified and PI factors (defined by 2) of up to 18 are achieved. At high temperatures, the average coking rate increases, requiring a more frequent regeneration of the catalyst. The latter can eventually be facilitated using a multizone reactor, as discussed before. In the current FCC technology, the heat for the endothermic cracking reactions is provided by the combustion of the coke deposited on the catalyst. Therefore, operation at higher cracking temperatures may require somewhat more coke deposition on the catalyst. This aspect was not yet accounted for in the simulations and requires a coupled crackingregeneration simulation.5
Increased Catalyst Activity, Cracking Temperature of 775 K. Simulations of catalytic cracking in the RFB-SG using different catalyst activities were carried out. The conventional catalyst activity a was increased by a factor i ) [1.5, 2, 3, 4, 5, 6] by adapting the activation energies of the different reactions. The reactor characteristics and the operating conditions were maintained (Table 6), and a cracking temperature of 775 K was kept. The results are summarized in Table 9, and Figures 14-16. Up to a catalyst activity of 3a, no fluctuations in the flow field were observed. Higher catalyst activities result in fluctuations in the radial direction which are initiated near the particle bed freeboard, that is, in the inner and more dilute part of the particle bed. Fluctuations in the more dense outer region of the particle
Ind. Eng. Chem. Res., Vol. 49, No. 11, 2010
5297
The demonstration of the intrinsic process intensification potential of rotating fluidized beds in a static geometry on the FCC process no doubt justifies a further investigation. Acknowledgment The authors thank the “Fonds de la Recherche Scientifique (FNRS)” for the financial support, under project FRFC no. 75394/2.4.597.07 F. The authors also acknowledge the “Institut de calcul intensif et de stockage de masse (CISM)” for the parallel computing infrastructure and technical support. Notation
Figure 16. Process intensification potential of RFB-SG for the catalytic cracking of gas oil as a function of the gas oil conversion: comparison with riser technology.5 For RFB-SG reactor characteristics and operating conditions, see Table 6. RFB-SG is operated at the conventional cracking temperature of 775 K and using different catalyst activities.
bed were not observed. Figure 14 illustrates the fluctuations occurring. The tangential uniformity of the composition and conversion profiles remained, however, acceptable, and the temporally and tangentially averaged radial profiles of the gasoline and light gases mass fractions are shown in Figure 15a and b, respectively. For the given reactor characteristics and operating conditions (gas oil flow rate, particle bed height, etc.), the gasoline and light gases yields and the gas oil conversion increase less than proportionally with increasing catalyst activity, as shown in Figure 15a-c, respectively. The resulting gasoline selectivity is seen to decrease with increasing catalyst activity, but less than when increasing the cracking temperature. Above a given catalyst activity, the gasoline yield is expected to decrease with increasing catalyst activity, significant overcracking causing losses of gasoline. As mentioned previously, this type of operation may be advantageous when the production of light gases is desired. Figure 16 shows that when using a more active catalyst, the FCC process in the RFB-SG can be further intensified, the PI factor 2 reaching values of up to 70. Conclusions CFD simulations using the Eulerian-Eulerian approach with the kinetic theory of granular flow and a 10-lump reaction model confirm the potential of rotating fluidized beds in static geometry for the intensification of the fluid catalytic cracking process. Using the conventional catalyst and cracking temperature, a 1 order of magnitude process intensification can be easily achieved. The gasoline or light gases selectivity can also be optimized. Rotating fluidized beds in a static geometry have excellent particle bed mixing and heat transfer properties.2 Therefore, the additional process intensification potential of applying higher cracking temperatures or of using a more active catalyst was also evaluated. Increasing the cracking temperature or catalyst activity increases the gas oil conversion in a similar, less than proportional way, but decreases the gasoline selectivity and increases the light gases selectivity. Allowing a higher cracking temperature or the use of more active catalyst, RFB-SGs may intensify the FCC process by up to a factor 70.
Aij ) pre-exponential factor of the reaction of formation of lump i out of lump j, kmol/(m3 s) Cj ) molar concentration of lump j, kmol/m3 Eij ) activation energy of the reaction of formation of lump i out of lump j, kJ/kmol ess ) solid-solid restitution coefficient, dimensionless ds ) particle diameter, m g0,ss ) radial distribution function, dimensionless h ) bed height, m I2D ) second invariant of the deviatoric stress tensor, 1/s2 KArh ) aromatic adsorption constant, dimensionless m ˙ ) mass flow rate, kg/s Mi ) molar mass of lump i, kg/kmol N ) average number of catalyst turns in the RFB-SG, dimensionless p, ps ) gas phase pressure, solid phase pressure, Pa Rij ) reaction rate of lump i out of lump j, kmol/(m3 s) Rc ) coking rate, kmol/(m3 s) R′ ) coking rate per mass catalyst, kgcoke/(kgcat s) S˜c ) mass source term in the continuity equation, kg/(m3 s) Si ) source term of lump i in the species continuity equation, kg/ (m3 s) T ) temperature, K b u, b V ) gas phase and solid phase velocity, m/s V ) reactor volume, m3 wi ) mass fraction of lump i, dimensionless Greek Letters β ) interphase momentum transfer coefficient, N/m3 γΘ ) collisional dissipation of energy, J/(s2 m2) ε ) phase volume fraction, dimensionless Θ ) granular temperature, J/kg κΘ ) granular diffusion coefficient, dimensionless µ ) viscosity, Pa s νik ) stoichiometric coefficient ratio of the formation of lump i out of lump k, dimensionless Fg, Fs ) gas phase density, solid phase particle density, kg/m3 τcat ) catalyst residence time, s cτ ) phase stress-strain tensor, N/m3 φ ) angle of internal friction, rad χGO ) gas oil conversion, dimensionless Subscripts cat ) catalyst g, s ) gas phase, solid phase GO, C, G, LG ) gas oil, C lump, gasoline, light gases h, l ) heavy lump, light lump
Literature Cited (1) De Wilde, J.; de Broqueville, A. Rotating fluidized beds in a static geometry: Experimental proof of concept. AIChE J. 2007, 53, 793–810. (2) de Broqueville, A.; De Wilde, J. Numerical investigation of gassolid heat transfer in rotating fluidized beds in a static geometry. Chem. Eng. Sci. 2009, 64, 1232–1248.
5298
Ind. Eng. Chem. Res., Vol. 49, No. 11, 2010
(3) De Wilde, J.; de Broqueville, A. Experimental study of fluidization of 1G-Geldart D-type particles in a rotating fluidized bed with a rotating chimney. AIChE J. 2008, 54, 2029–2044. (4) Watano, S.; Nakamura, H.; Hamada, K.; Wakamatsu, Y.; Tanabe, Y.; Dave, R. N.; Pfeffer, R. Fine particle coating by a novel rotating fluidized bed coater. Powder Technol. 2004, 141, 172–176. (5) Froment, G. F.; Bischoff, K. B. Chemical Reactor Analysis and Design, 2nd ed.; John Wiley and Sons (WIE): New York, 1990. (6) Gidaspow, D. Multiphase Flow and Fluidization: Continuum and Kinetic Theory Descriptions, 1st ed.; Academic Press: New York, 1994. (7) Syamlal, M.; Rogers, W.; O’Brien, T. J. MFIX Documentation: Vol. 1, Theory Guide, DOE/METC-9411004 NTIS/DE9400087; 1993.
(8) Syamlal, M. The Particle-Particle Drag Term in a Multiparticle Model of Fluidization, DOE/MC/21353-2373 NTIS/DE87006500; 1987. (9) Johnson, P. C.; Jackson, R. Frictional-collisional constitutive relations for granular materials, with application to plane shearing. J. Fluid Mech. Digital ArchiVe 1987, 176, 67–93. (10) Jacob, S. M.; Gross, B.; Voltz, S. E., Jr. V. W. W. A lumping and reaction scheme for catalytic cracking. AIChE J. 1976, 22, 701–713.
ReceiVed for reView October 15, 2009 ReVised manuscript receiVed March 10, 2010 Accepted March 16, 2010 IE901610F
