Advanced Chemical Reaction Engineering CBE9450 Project
Advanced Chemical Reaction Engineering CBE9450 Project
Advanced Chemical Reaction Engineering CBE9450 Project
A Dispersion Model for Fluid Catalytic Cracking Downer Reactor with a Six Lump Kinetic Model
Written and Illustrated by: Dawood Al-Mosuli MEng in CBE at Western University
A Dispersion Model for Fluid Catalytic Cracking Downer With a Six Lump Kinetic Model
Abstract
The fluid catalytic cracking FCC is a process and apparatus where riser is used for cracking of oil feed stocks in the presence of catalyst and regenerator is used to regenerate the spent catalyst. This process is modified by the development of the downflow riser (downer). In downer reactor where the solids and gas move downward co-currently the flow behavior is near plug flow reactor, it is possible to obtain uniform distribution of the catalyst with the feed and to reduce the contact time between them. In addition the amount of coke produced in the process is reduced. In this work, a dispersion model for downer reactor is proposed. The model will combine the six lump kinetic models, the hydrodynamic model of the downer and the dispersion mixing model to predict the fluid catalytic cracking performance. The results should be compared with that of a dispersion model for the riser reactor. This will give a prediction of the effect of back mixing on both types of FCC reactors, and if the assumption of plug flow reactor is an oversimplification for any of these reactors.
1- INTRODUCTION AND LITRATURE SURVEY. 1-1 DEVELOPMENT OF DOWNER FREACTOR. 1-2- MATHEMATICAL MODELING RELATED TO DOWNRS. 1-3-AIM AND SCOPE OF WORK. 2- MODELING RELATED TO DOWNERS 2-1- DISPERSION MODEL AND ASSUMPTIONS RELATEDTO OUR MODEL. 2-2- THE KINETIC MODEL 2-3- CONCENTRATION, TEMPERATURE, PRESSURE AND REACTION TIME PROFILES IN THE REACTOR.
3
6 6 8 9 9
9 12
13
2-4- HYDRODYNAMIC MODEL 2-5-MIXING TEMPERATURE 3- SEQUENCE OF CALCULATION STEPS 4- CONCLUSIONS AND SUDGUSTIONS FOR FURTHER WORK. REFERENCES
15 16 17
18 19
NOMENCLATURES
A Reactor cross section area (m2) Ar Archimedes number (-) Cp Heat capacity (kJ/kg.K) D Reactor diameter (m) d Particle diameter (m) Ej Activation energy (kJ/kmole) Fr Fround number H Heat enthalpy (kJ/s) Hj Heat enthalpy of jth reaction (kJ/kg) Kj Kinetic reaction rate constant of jth reaction Kuop UOP characterization factor
L Reactor Height (m) MW Molecular weight (kg/kg mole) m Mass rate (kg/s) P Reactor pressure (Pascal) R Universal ideal gas constant (atm m3/kmole K) Re Reynold number SG Feed specific gravity T Temperature (K) u Velocity (m/s) X Conversion (wt %) yi Weight fraction of ith lump z Axial position of riser height (m) Greek letters Voidage Catalyst deactivation function Density (kg/m3) Slip factor Viscosity (Pa.s) factor of VGO to gasoline reaction Subscripts cok Coke cat Catalyst ds Dispersion or Atomizing steam
5
f Feed fg Flue gas fl Feed in the liquid phase fv Feed in the vapor phase g Gas phase in Flowing in j 1,2,3,4 and 5 for the reactions VGO to GLN, VGO to C4s, C5s,etc o Superficial out Flowing out p Particle rcat Regenerated catalyst rcoke Coke on the regenerated catalyst s Steam scat Spent catalyst scok Coke on the spent catalyst t Terminal velocity
produced in the process. In order to understand the important of this development, It is important to notice that for up flow riser (see e.g., U.S. pat. No. 3565790,U.S. pat. NO. 3607126, U.S. pat.NO. 3492221), the petroleum feed should be in contact with the catalyst for a relatively long time to obtain an efficient conversion for the feed. This long contact time is necessitated by the upflowing configuration of the riser which includes acceleration of the catalyst from stationary case to feed velocity against the gravity force. This caused a lot of problems with the riser like catalyst backmixing and non-uniform catalyst distribution through the feed. Due to this relatively long contact time with the catalyst and inefficient contact between the catalyst and petroleum feed, there will be an increase in natural tendency of heavy petroleum cuts to form large amounts of coke at the expense of gasoline production. To overcome these problems, the FCC reaction vessel is provided with a transport reactor at the top of the reaction vessel. In this arrangement, the catalyst is forced to flow downward from the regenerator into the reactor. This downflow eliminates the problems of catalyst backmixing and its non-uniform distribution in through the feed. Moreover the uniform catalyst distribution in the feed stream is gained in a relatively short time. This will enable rapid separation of catalyst at the bottom of the reactor and low coke formation. So, the net result of of providing a downflow riser is decreasing coke formation, increasing gasoline selectivity and production of higher octane gasoline at same conversion. Detailed representation of downflow riser technology is given in (U.S. pat.No. 4385985, U.S. Pat.NO.7087154, US. Pat.NO. 4411773, U.S. Pat.NO.5582712, U.S. Pat.NO.4693808, U.S. Pat.NO. 4797262).
Because of the importance of this development in refining, then it is important to develop a mathematical model to simulate the dynamic behavior of the downer. In addition, this model will be an important tool to study the effect of operational parameters on the productivity and performance of the process.
influence of the operating conditions and feed properties on the yield of downer type reactor.
2-Modeling related to downer 2-1- Dispersion model and assumptions related to our model
The axial gas dispersion of a pilot-scale downer reactor was studied by Wei et al. (1995)[16]. They found that the axial Peclet number is 1-2 times larger than that of riser.
These studies in both the riser and downer indicate that the dispersion model can describe the model and that a certain extent of gas and solids mixing occurs in both the riser and downer reactors. To make the formulation in a simple way, the assumptions are needed:
1.The system is at a steady state condition. 2. No heat or mass transfer resistance between gas and solid phases due to high mixing rate. 3. Axial mixing can be described by one dimensional dispersion model. 4. Due to good mixing then the cracking reactions is controlled by chemical kinetics. 5. Temperature of chemical reaction is constant. 6. The flow in the reactor is a fully developed flow. 7. Pressure changes through the riser length is due static head.. 8. Immediate evaporation of feed at the riser inlet. 1-4- Mass and heat balance The axial dispersion model is combined with the six lump kinetic model to gas-oil catalytic cracking. The system is assumed to be close-close boundary, then the boundary conditions of Danckwerts (1953)[] could be applied to the system. The mass balance of the small element in reactor is formulated as follows:
10
In order to solve the model numerically, the riser is divided into equal sized disc like elements of thickness (dz) as in the figure below
d2 yi/dz2=Pe/L*{dyi/dz+r(i)}
(i=1,2,3,4,5,6)
(a)
With Danckwerts (1953) boundary conditions: At Z=0 dyi/dz-Pe/L*{yi (at z=0+) yi (z=0-)} At Z=L dyi/dz=0 (b) (c)
11
Where the rate of reaction r(i) is obtained from the kinetic model. The volumetric expansion due to generation of moles is taken into account by calculating the superficial gas velocity at the exit of each control volume.
Fig 3.3 representation of the six lump model For each reaction the rate r was given as a function of weight fraction yi , deactivation and kinetic constant ki. The rate constant for each of the above reactions is given as below:
12
Gas oil: r1 = -(k1 + k2 +k3 + k4 +k5)*y1^2* ---------------------------1 Gasoline: r2= (k1y1^2 k6y2 k7y2 k8y2 k9y2)* -----------2 C4s : r3 = (k2y1^2 + k6y2 k 10 y3 k11y3)* --------------------------3 C5s: r4= (k3y1^2 + k7y2 + k10y3 k12y4)* -----------------------------4 Dry gas: r5 = (k4y1^2 + k8y2 + k11y3 + k12y4)* -----------------------5 Coke: r6 = (k5y1^2 + k9y2)* --------------------------------------------------6 is the deactivation function, ki are the kinetic constants and yi are the weight fractions for each of the six lumps. The values of k is given by Arrhenius equation ki = Ai exp(-Ei/RT) -----------------------------------------------------------------7 The kinetic parameters are taken from literatures as[ 20,22,23].
2-3- CONCENTRATION, TEMPERATURE, PRESSURE AND REACTION TIME PROFILES IN THE REACTOR.
Following the procedure given by (rohani[29] ,faheem[24]), the concentration profile for each lump along the reactor length could be put in the following set of ODE:
13
For VGO lump: d2 y1/dz2 =(PE/L)*{-dy1/dz-C*{k1+k2+k3+k4}*y1^2 } ------------------8 For gasoline lump: d2 y2/dz2 =(PE/L)*{-dy2/dz+C*{k1*y1^2-(k6+k7+k8+k9)*y2}} -----9 For C4s : d2 y3/dz2 =(PE/L)*{-dy3/dz+ C*{k2*y1^2+k6*y2-k10*y3-k11*y3)} ----------10 For C5s : d2 y4/dz2 =(PE/L)*{- dy4/dz+C*{k3*y1^2+k7*y2+k10*y3-k12*y4}} -----------11 For dry gas : d2 y5/dz2 =(PE/L)*{-dy5/dz+C*{k4*y1^2+k8*y2+k11*y3+k12*y4}}------------12 For coke : d2 y6/dz2 =(PE/L)* {-dy6/dz+C*(k5*y1^2+k9*y2} } ------------------------------13 Where C =A *g* * g/mg ------------------------------------------------------14
In the same way the Danckwerts boundary conditions (equations (b) and (c)) should be defined for each component. The temperature profile along the reactor could be obtained using the following equation as in (rohani et al [29] and [24]) dT/dz= { (A *g* * g )/(mcat*cpcat+mg*cpg)}(k1 H1 +k2 H2 +k3 H3+k4 H4+k5 H5)*y1^2+(k6 H6 +k7 H7+k8 H8+k9 H9 )*y2+(k10 H10+k11 H11)*y3+k12 H12y4 -----------------------15
The catalyst residence time could be calculated using the following equation[ 24] dtc/dz=A* *cat{mcat* +*1/Mwg]*mg*(1-y6)* cat*(101325*RT/P)}-----16
The vapor mass flow rate through the reactor can obtained by the following equation[ 24]
14
mg=mf*(y1+y2+y3+y4+y5}+Mds --------------------------------------------------------17
Where the quantitiesof dispersion steam (Mds) 1% [ 8]. The vapor phase densityis calculated by ideal gas law. g=P*Mwg/(101325RT) ---------------------------------------------18 Average vapor phase molecular weight is obtained from [25] Mwg=1/{(y1/MwVGO)+(y2/Mwgasolie)+(y3/MwC4s)+(y4/MwC5s)+(y5/Mwdry gas)} ----19
dp/dz=- cat*g*(1- g)-------------------------------------------------------------------------20 The deposition of coke on the catalyst surface is represented by the catalyst activation function using this formula given by Koratiya et al.[30] =(1+51(mxcok/mcat)}^(-2.78) mxcoke is mass flowrate of carbonized catalyst, mcat is the mass flowrate of catalyst.
15
Frt=ut/
---------------------------------------------------------------------25
The super ficial velocity uo=mg/(A* g) -------------------------------------26 Average particle velocity up=mcat/( cat*A*(1- g) --------------------27 By combination of equations 23 ,26 ,27, we get the average void fraction in gas phase g g= cat*mg/( g*mcat* + cat*mg) -------------------------------------28 and gas velocity can be evaluated by ug=uo/g ---------------------------29 particle velocity up=ug/ -------------------------------------------------------30 residence time in gas phase t= z/ug ------------------------------------------31 particle terminal velocity can be calculated from [35 36 37] ut =Ret*ug/( g*dp) ------------------------------------------------------------32
16
A=(
-mf*cpf*Tf
B= (mrcat + mxcat)*cpcat+(mrcok)*cpcok+mls*cps +mf *cpf Equation 35 is the initial boundary condition for the differential equation (15)
Ug , Up, X) using equations 7 and 17-32. The calculated values represent the exit
conditions of the current volume element and at the same time the inlet conditions of the next volume element. Also initial values for the first step of the equation are given by computing results of the plug flow model ( without dispersion). With these initial values and the boundary conditions in the outlet of the reactor it is possible to compute the conditions for each incremental step. 5. Increasing the amount of z by small value to move to the next volume element. 6. Calculating the new values of ODE 8-16 depending on exit values of previous volume element (step4). 7. Steps 4- 6 should be repeated until the value of z reaches the total height of the reactor. 8- It is possible to repeat the calculation for a number of iterations till reaching a certain degree of accuracy
17
The previous calculation can be calculated using Microsoft Excel depending on Runge Kutta numerical method technique. 10. Now the output variables i.e. yield, conversion cracking efficiency, selectivity, delta coke etc can be calculated. 4- Conclusions
After my short literature survey, It seems that a lot of work is still needed to develop a mathematical representation of both downer and riser FCC reactors. Developing such a model is much better than using the empirical correlations provided by the manufacturing companies. Empirical correlations couldnt be generalized to all types of downers or risers. Non ideality of FCC reactors can have a large effect on the productivity of gasoline, for example increasing axial Peclet no. from 0.1 to1000 will increase the yield of gasoline by 11% under the same conversion[18]. Moreover, incorporating six lumps model make the model more accurate than the traditional 4 or 3 lumps model used in many previous works. Providing accurate model is an important issue especially when it is necessary to change the feed or the operating conditions or both, or it is necessary to make an optimization or control system for the FCC process. More work could be done by changing the assumptions done to simplify the model. For example by introducing any one or more of the following inside the model: radial dispersion, two dimensional flow model, heterogeneous model, no isothermal behavior, changing physical properties along the reactor, changing the deactivation function along the reactor.
18
References
1-Berg, D.; Briens, C.; Bergougnou, M. Can. J. Chem. Eng. 1989,67, 96. 2-Bolkan-Kenny, G.; Pugsley, T.; Berruti, F. Computer Simulation of the Performance of Fluid Catalytic Cracking Risers and Downers. Ind. Eng. Chem. Res. 1994, 33, 3043. 3-Cao, C.; Jin, Y.; Yu, Z.; Wang, Z. In Circulating Fluidized Bed Technology IV; Avidan, A., Ed.; AIChE: New York, 1994; p 406. 4-Danckwerts, P. V. Chem. Eng. Sci. 1953, 2, 1-13. 5-Gartside, R. QC-A New Reaction System. In Fluidization VI; Grace, J., Shemilt, L., Bergougnou, M., Eds.; Engineering Foundation, New York, 1989; p 25. 6-Gianetto, A.; Faraq, H.; Blasetti, A.; de Lasa, H. I. Fluid Catalytic Cracking Catalyst for Reformulated Gasolines Kinetic Modeling. Ind. Eng. Chem. Res. 1994, 33, 3053. 7-Kauff, D.; Bartholic, D.; Steves, C.; Keim, M. Successful Application of the MSCC Process. NPRA Annual Meeting, 1996. 8- Personal communication .Iraqi ministry of oil. Operating manual. 9-Kraemer, D.; de Lasa, H. I. Catalytic Cracking of Hydrocarbons in a Riser Simulator. Ind. End. Chem. Res. 1988, 27, 2002. 10--Li, Y. C.; Wu, P. In Circulating Fluidized Bed Technology III; Basu, P., Horio, M., Hasatani, M., Eds.; Pergamon Press: Toronto, Ontario, Canada, 1990; p 581. 11-Luo, G. H.; Yang, G. L. Axial Gas Dispersion in a Fast Fluidized Bed. FLUIDIZATION91, Science and Technology; Kwauk, M.,Hasatari, M., Eds.; Science Press: Beijing, China, 1991; p 102. 12-Shimizu, A.; Echigo, R.; Hasegawa, S.; Hishida, M. Int. J.Multiphase Flow 1978, 72, 271.
19
13-Wang, Z.; Bai, D.; Jin, Y. Hydrodynamics of Concurrent Downflow Circulating Fluidized Bed. Powder Technol. 1992, 70, 271. 14-Wei, F.; Lin, S.; Yang, G. Gas and Solids Mixing in a Commercial FCC Regenerator. Chem. Eng. Technol. 1993, 16, 109. 15-Wei, F.; Wang, Z.; Jin, Y.; Yu, Z.; Chen, W. Dispersion of Lateral and Axial Solids Mixing in a Cocurrent Downflow Circulating Fluidized Bed. Powder Technol. 1994, 81, 25-30. 16-Wei, F.; Liu, J.; Jin, Y.; Yu, Z. The Gas Mixing in CDCFB. Chem.Eng. Technol. 1995, 18, 59-62. 17-Wei, F.; Lai, Z.; Jin, Y.; Yu, Z. A CFB Reactor Model for Synthesis of Acrynitrile. Asian-Pacific Chemical Reaction Forum; Beijing,1996. 18- Wei. Fei. ;Ran Xing, Zhou Rujin, Luo Guohua A Dispersion model for fluid catalytic cracking riser and downer reactors. Ind. Chem. Res. 1997 ,36, 5049-5053. 19- Sadeghbeigi R., "Fluid Catalytic Cracking Handbook", Gulf publishing (2000). 20- Zaidoon M. Shakoor, Estimation of Kinetic Parameters of Complex Reactions byMATLAB Software, University of Technology. 21- Ancheyta J., Modeling And Simulation Of Catalytic Reactors For Petroleum Refining, Wiley Publication, (2011) 22- Souza J. A., Vargas J. V. C., Von Meien O. F., and Martignoni W. P., Modeling and Simulation of Industrial FCC Risers, Thermal Engineering, Vol. 6, No.1, June (2007).
20
23-
Riser Reactor in Industrial FCC Unit, Petroleum & Coal, 50 (2) (2008), 15-24, 24 - Fahim M.A., Al-Sahhaf T.A., Elkilani A.S., " Fundamentals of Petroleum Refining" Elsevier (2010) 25 - Erthal R. H., Negrao C.O. R., and Rossi L. F. S., Modeling the Riser of a Fluid Catalytic Cracking Unit, 17th international congress of mechanical eng., 10-14, Nov., (2003). 29-H. Ali, S. Rohani and J. P. Corriou, Modeling and Control of a Riser Type Fluid Catalytic Cracking (FCC) Unit, IChemE, Vol. 75, Part A, May (1997). 30- Koratiya, V.K and Kumar.S. (2010). Modeling, simulation and optimization of FCC downer reactor Petrloeum &Coal ISSN 1337-7027.
21