Sie sind auf Seite 1von 12

Full Paper

Coupled Single-Particle and Population


Balance Modeling for Particle Size Distribution
of Poly(propylene) Produced in Loop Reactors

Zheng-Hong Luo,* Shao-Hua Wen, De-Pan Shi, Zu-Wei Zheng

A comprehensive model was developed for the PSD of PP produced in loop reactors. The
polymeric multilayer model (PMLM) was first applied to calculate the single particle growth
rate under intraparticle transfer limitations. In order to obtain the comprehensive model, the
PMLM was solved together with a steady-state
particle population equation to predict the PSD in
the loop reactors. The simulated PSD data
obtained under steady-state polymerization con-
ditions agreed with the actual data collected from
industrial scale plant. The comprehensive model
was also used to predict the effects of some critical
factors, including the intraparticle mass and heat
transfer limitations, the feed catalyst particle size
and the catalyst deactivation, etc., on the PSD.

Introduction operational variables may affect PSD via simulation may be


useful for the understanding of the propylene polymeriza-
Poly(propylene) (PP) can be produced in various types of tion in loop reactors, especially the intraparticle mass and
reactors, such as autoclave, fluidized-bed reactor (FBR) or heat transfer limitations, the feed catalyst particle size and
loop reactor. The last one is certainly the most important at the polymerization temperature, etc. on the PSD.
present.[1] In the loop reactor, small solid catalyst particles In order to predict the effects of the intraparticle heat and
(e.g., 20100 mm) react with monomers to form polymer mass transfer, the feed catalyst particle size, etc. on the PSD,
particles in a size range of 1005 000 mm in a liquid phase material and energy balance equations for particle growth
and the polymer particles are produced as a solid have to be solved together with the particle population
suspension in the liquid stream.[1] In addition, the balance equations.[3]
polymerization rate, the cost of post-treatment after Up to now, a number of models have been proposed to
polymeric process and the polymer properties are influ- describe the particle growth in the solid-catalyzed olefin
enced by the polymer particle size distribution (PSD).[27] polymerization, namely single particle models.[8] So far,
Therefore, the polymer PSD modeling thus analyzing how four single particle models, namely, the solid core model
(SCM),[9] the polymeric flow model (PFM),[10,11] and the
Z.-H. Luo, S.-H. Wen, D.-P. Shi, Z.-W. Zheng
multigrain model (MGM),[12] have been widely used to
Department of Chemical and Biochemical Engineering, College of describe the particle growth in the solid-catalyzed olefin
Chemistry and Chemical Engineering, Xiamen University, Xiamen polymerization. In addition, the polymeric multilayer
361005, China model (PMLM) is a versatile model that can also be used
Fax: 86 592 218 7231; E-mail: luozh@xmu.edu.cn to simulate olefin polymerization.[1315]

Macromol. React. Eng. 2010, 4, 123134


2010 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim DOI: 10.1002/mren.200900040 123
Z.-H. Luo, S.-H. Wen, D.-P. Shi, Z.-W. Zheng

In olefin polymerization field, there are many publica- the model of Choi et al.[18] to account for the combined
tions on the PSD by using the particle population balance. effects of intraparticle mass and heat transfer limitations
Zacca et al.[16] presented a population balance modeling on the PSD for highly active catalyst assuming negligible
of multistage olefin polymerization processes using particle agglomeration.[20] Recently, Yiannoulakis et al.
the catalyst residence time as main coordinate. Their developed a steady-state population balance model to
work allowed the consideration of particle size selection predict the PSD in an ethylene copolymerization FBR and
effects within the process in a FBR, but the single calculate the growth rate of a single particle, for which the
particle growth was simplified and no experimental or PFM is employed.[3] However, most of these comprehensive
industrial data were associated for comparison and model models are still used for the FBR and are not compared with
validation in their work. Hatzantonis et al.[2] developed a actual industrial data. Although there is nothing difference
generalized steady-state population balance model between them when both FBR and loop reactor are treated
rigorously accounting for the combined effects of particle as a continuous stirred-tank reactor (CSTR) in simulation,
growth, attrition, elutriation, and agglomeration in a gas- they are still different in plant operation and solution
phase FBR. However, the particle growth was simplified method of the population balance equation.
and the intraparticle heat and mass transfer limitations In the present work, a comprehensive model is developed
were neglected in their model. Harshe et al.[17] developed for the PSD of the PP produced in loop reactors. The PMLM is
a comprehensive model for propylene polymerization first applied to calculate the single particle growth rate
in a FBR. The need for coupling the reaction-engineering under intraparticle transfer limitations. The PMLM is solved
model with the population balance equation was together with a steady-state particle population equation
demonstrated in their work. Their model could be used to to predict the PSD in the loop reactors. Particular attention is
predict the polymers molecular structure and morphology, paid to the effect of the intraparticle mass transfer
including the PSD, but did not provide the single resistance on the PSD and to compare the computational
particle growth and comparison of model prediction to and the actual industrial data. We also simulate the
experimental or industrial operation data. Recently, we whole process for propylene polymerization in the loop
presented a model for the PSD of PP produced in a loop reactors including the prepolymerization and the main-
reactor.[7] The model took into account the flow pattern, polymerization reactors.
the polymer particle dynamics, and the attrition.
Furthermore, the simulated PSD data were compared with
the actual data collected from industrial scale plant.
However, our model still accounted for both internal Process for Propylene Polymerization in the
and external heat and mass transfer limitations of Loop Reactors
the particle via a simplified single particle growth
model. And only one grade of data obtained from certain In the present study, Spheripol Loop technology with the
plant was used to estimate the model parameters and verify continuous tubular loop reactors as its heart is studied. To
the model. In addition, our past model ignored the illustrate the concepts discussed in this paper, a typical
agglomeration. Based on above discussion, it becomes schematic flowsheet is given in Figure 1.
clear that the early modeling efforts in this field are There are many papers related to the technology.[1,7,16]
made to account for the detailed aspect of the population Comprehensive reviews have also been reported.[21,22]
balance model along with simplifying/neglecting the Some points related to this work are listed in the following
single particle growth. Moreover, it is important that most contents. First, the loop reactors, including the prepolymer-
of the PSD models established are mainly regarding the ization reactor (R200, 0.46 m3) and the main-polymeriza-
polyolefin PSD in the FBR and are not validated with actual tion reactor (R201, 56 m3), are at the heart of the technology.
industrial data. Secondly, R201 consists of three continuous tubular
On the other hand, there are some comprehensive PSD reactors connected in sequence with each other. In practice,
models suggested, which predicted the PSD via solving the loop reactor is a closed tube as a whole, wherein the
monomer and energy balances for a growth particle reacting slurry driven by a recycling pump circulates at
together with the particle population balance equation. high-recycle rates, as depicted in Figure 1. Under these
Choi et al.[18] incorporated an isothermal simplified multi- conditions, it is reasonable to regard the loop reactor as a
grain SCM into a steady-state population balance model to CSTR with constant volume, but with various slurry
predict the effects of catalyst deactivation on the PSD in a density. The reaction slurry is assumed to be a mixture of
FBR. Khang and Lee[19] developed analytical expressions a liquid phase (monomer and hydrogen) and a solid phase
for the calculation of the PSD in a FBR, taking both the (polymer and catalyst). Finally, the catalyst is assumed to
monomer diffusion rate and catalyst deactivation in the undergo the common activation-polymerization-decay
MGM into consideration. Hatzantonis et al. extended pathway.

Macromol. React. Eng. 2010, 4, 123134


124 2010 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim DOI: 10.1002/mren.200900040
Coupled Single-Particle and Population Balance Modeling for . . .

The Polymeric Multilayer Model


(PMLM)
To simulate single polymer particle
growth, the PMLM (Figure 2)[1315] was
applied in this paper. Moreover, the
reaction slurry in the loop reactor is
driven by a recycling pump and has a
high recycle rate, no propylene concen-
tration gradient in film layer around
particle possible during polymerization
in (almost) pure propylene. In addition,
high thermal diffusivity of liquids and
turbulence in loop reactors make external
heat transfer resistance (if present) neg-
Figure 1. Himont spheripol loop process (Fujian Petrochemical Company of SINOPEC).
ligible. Accordingly, the external bound-
(A, catalyst; B, propylene; C, hydrogen; D, coolant in; E, coolant out; 18, control sites; ary transfer resistance of the polymer
9E201, heat exchanger; 10E203, heat exchanger; 11P200, pump; 12P201, pump; 13 particle is unimportant and it is reason-
Z203, mixer; 14R200, prepolymerization reactor; 15R201, main-polymerization reactor; ably to neglect it in the single particle
16D202, buffer tank). model.[8] The boundary conditions of the
model used in this study are somewhat
different from those of the single particle.
Single-Particle Growth Modeling The PMLM accounting for the intraparticle mass and heat
transfer limitations and the deactivation of active sites
Polymerization Kinetics comprises the following differential equations and bound-
ary conditions.
To describe the kinetics of propylene polymerization on the
Ziegler-Natta catalyst, a simple kinetic model is  
@Mr; t De @ 2 @M
employed. [1,2,7,16,17]
The polymerization kinetic scheme r  Rp (5)
@t r2 @r @r
comprises a series of elementary reactions, namely, site
activation, propagation, site deactivation, chain transfor-  
@T r; t Ke @ 2 @T
mation, and chain transfer reactions. Among them, the r C
p P r  Qp (6)
@t r2 @r @r
propagation reaction and the site deactivation reaction
play an important role in the particle growth rate.[7] By
where M(r,t) is the propylene concentration in the
using the simple kinetic model and the pseudo-kinetic rate
polymeric particle at time t, T(r,t) is the temperature
constant method, the following kinetic equations can be
in the polymeric particle, Rp is described by Equation
obtained.
(1)(4).
Qp is expressed as
propagation rate: Rp kp C rcat M (1)
 
Qp DHp Rp (7)
where kp is given by
  
kp k0p exp EA Rgas T (2)

Because propylene is consumed by the propagation


reaction, the polymerization rate is given by Equation (1).
Catalyst deactivation can be described by a first-order
deactivation equation:

C C0 expkd t (3)

with
  
kd k0d exp ED Rgas T (4) Figure 2. Schematic representation of PMLM.

Macromol. React. Eng. 2010, 4, 123134


2010 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.mre-journal.de 125
Z.-H. Luo, S.-H. Wen, D.-P. Shi, Z.-W. Zheng

The initial and boundary conditions of Equation (5) are Equation (1)(13) describing the monomer and temperature
profiles in a growth polymer particle.
Mr; 0 0 (8) As depicted in Figure 1, the loop polymerization reactors
used in the paper including the prepolymerization reactor
@M0; t (R200) and the main-polymerization reactor (R201), consist
0 (9)
@r of continuous tubular reactor connected each other. More-
over, any loop reactor depicted in Figure 1 can be handled as
MR;tM0 (10)
a CSTR. During the solid-catalyzed propylene polymeriza-
tion in the loop reactors, the small catalyst particles are
and for Equation (6) as follows
continuously fed into R200 at a constant rate. The catalyst
T r; 0T0 (11) fragments into a large number of small particles or layers
under mild reaction conditions in R200, accordingly, the
@T 0; t polymerization reaction is initiated. Subsequently, the
0 (12) small particles initiated are transported into R201 and react
@r
with the incoming monomers. Simultaneously, the parti-
T R; tT0 (13) cles can grow in size due to the polymerization reaction in
R201, and can be entrained by the liquid medium in R201.
The volume and boundary position of each layer must be Furthermore, it is well-known that the softening tempera-
updated after a predetermined time interval Dt. The ture of PP depends on their molecular weight. The molecular
monomer concentrations in the previous time step are weight of poly(propylene) obtained in the early period of
used for this purpose, and the equations are expressed as the polymerization is usually low and its value does not
exceed 360 K. Accordingly, some polymer particles with

0 4  0 3  0 3 low-molecular-weight may undergo particle agglomera-
Vj p rj1  rj (14) tion, if the operation temperature closes to the polymer
3
softening temperature.[2,23] For the sake of simplicity, the
" i # polymer particles are assumed to be of spherical shape and
i1 i
kp Cj Mj Mm Dtrcat
Vj Vj 1 (15) with constant density. Following the developments of
rp
Randolph and Litster, and Yiannoulakis et al.,[3,24,25] we can
 1=3 derive a population balance equation to be used to describe
i1 3 i1  i1 3 the steady-state PSD in the reactor. Corresponding equation
rj1 V rj (16)
4p j can be described as Equation (18)(20).


In these equations, the superscripts indicate time and the d Gp DNp D
i i
subscripts indicate the radial position; e.g., Vj and rj are dD
the volume of layer j and the radial position of the layer 1

during the growth, the polymeric particle in the ith time BD Fin Nin D  Fpp Npp D (18)
VR
interval, respectively. For updating the volume, the
concentration of active sites that depends on the volume  
i ZD Kag D3  l3 1=3 ; l N D3  l3 1=3 N l dl
of layers in layer j in the ith time interval, Cj , is expressed D2
B  2=3
as 2 D3  l3
0
i
Cj Vji (19)
i1
Cj i1
(17)
Vj Z 1
D N D Kag D; lN ldl (20)
i 0
The effect of deactivation on Cj is described by Equation
(3) and (4). where Np(D) and Nin(D) are the number density functions
of the particles in the reaction and in the feed stream (Fin),
respectively. Kag D; l is the agglomeration kernel for
PSD Modeling
particles of sizes D and l, Gp(D) is the corresponding
particle growth rate, which can be obtained via the PMLM.
Population Balance Equation
The computational process is as follows:
To calculate the steady-state PSD in a loop reactor, a particle First, we obtain the radial profiles of the monomer
population balance model must be solved together with concentration, the temperature, the concentrations of

Macromol. React. Eng. 2010, 4, 123134


126 2010 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim DOI: 10.1002/mren.200900040
Coupled Single-Particle and Population Balance Modeling for . . .

active sites, and the polymerization rate in the polymeric Solution Method and Particle Parameter
particle based on the PMLM. In addition, we also obtain the Values
polymerization rate of each layer in the polymeric particle
according to Equation (1). Accordingly, the total particle In the present work, both R200 and R201 as shown in
polymerization rate is given by Figure 1 are considered. The catalyst particles with an
adhering thin layer of polymer first flow into R200 and form
Xn R i V i Ci the thin polymer films. Then the polymer films flow into
pj j j
G (21) R201.
P
n
i i
j1 Vj Cj The PMLM, i.e., Equation (1)(13) are solved together with
j1
the particle population balance model to obtain the steady-
state PSD in the reactors. To calculate the monomer and
where n is the total number of layers. Simultaneously, the temperature profiles in a growth polymer particle,
 
polymeric particle size D 2  rni can be obtained Equation (1)(13) are numerically implemented as an
according to Equation (14)(16). Finally, the least square MATLAB PDEPE-function in C language and solved with a
method is used to fit the particle growth rate with D as the orthogonal collocation method.[27] Moreover, in the case of
independent variable and G as the dependent variable. with or without particle agglomeration, the adjustable
Therefore, the PMLM is incorporated into a steady-state geometry discretization method[3,25,28] is associated for
population balance equation in order to predict the effects solving the population balance equations.
of some critical factors, including the intraparticle mass In the next analysis, the simulated results depend on the
and heat transfer limitations, the feed catalyst particle size input parameter values presented in Equation (1)(23). In
and the catalyst deactivation, etc., on the PSD. the present study, the parameter values include the
values of the physical and transport properties of the
reaction mixture and the kinetic rate constants. Many
The Agglomeration Kernel researchers[14,17,2932] studied olefin polymerization
including propylene polymerization in the slurry-phase
As discussed earlier, when high active small-size polymer
loop reactor. A set of reference values of these parameters
particles with low-molecular-weight lie in these circum-
have been selected and were listed in Table 1. Unless
stances which temperature is near the polymer softening
otherwise noted, the parameters used for the simulation
temperature, they will interact each other. Accordingly, the
are those showed in Table 1. In addition, some nominal
agglomeration is initiated, which leads to the formation of
operating conditions are also needed and directly obtained
particle agglomerates with unexpected different morpho-
from certain plant. They are also listed in Table 1.
logical properties.
Though there are lots of publications on the agglomera-
tion phenomena, it is still difficult to describe accurately the
particle agglomeration in olefin polymerization. Here, an
empirical approach is used for the identification of the Results and Discussion
agglomeration kernel, Kag, appearing in Equation (19)(20).
Following Yiannoulakiss and Arastoopours models,[2,26] Model Verification
the agglomeration kernel for two particles in sizes D1 and
Two plant poly(propylene) grades, namely T30S and T36F,
D2, is given by the following equation
are used in this work. The plant operation conditions of
   T30S and T36F are shown in Table 2. The catalyst used is an
Kag D1 ; D2 K0 D41 D42 D3 3
1 D2 (22) advanced fourth generation of the Ziegler-Natta catalyst,
and are sampled and characterized by means of microscope
where K0 is an agglomeration rate constant. In the present to obtain the particle accumulation curve, and then the
work, it is assumed that K0 can be expressed by particle density as the function of catalyst size is obtained
by derivation and is shown in Figure 3, where the catalyst
   size data are the input data for the model. To calculate the
K2 Ts;1 Ts;2
K0 K1 exp (23) PSD in the reactor, the initial catalyst size distribution is
T sf divided into a number of size cuts and the overall PSD in
the reactor is obtained as a weighted sum of the PSDs
where Ts,1,Ts,2 denote the surface temperatures of particles resulting from the individual catalyst cuts.[2] Accordingly,
of size D1 and D2, respectively. T sf is the average polymer the polymer PSD can be simulated based on above model.
softening temperature of particle D1 and D2, K1 and K2 are The polymer powders obtained from R201 are analyzed by
proportionality constants. using a sizer to obtain PSD.

Macromol. React. Eng. 2010, 4, 123134


2010 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.mre-journal.de 127
Z.-H. Luo, S.-H. Wen, D.-P. Shi, Z.-W. Zheng

Table 1. Reactor conditions and parameters for simulation. Table 2. Plant parameters.[1] qm represent the mass flows, which
are presented directly from the controlling computer at the plant
Parameter Value Reference site.

Parameter of R200 Sample T qm


3
M0/(mol  m ) 9 700 plant data K kg  h1
T0/K 293 plant data
Fpp/(kg  h1) 250a) plant data Propylene Catalyst H2 AlR3
VR/m3 0.46 plant data T30S 343.0 25 670 0.9 0.8 3.8
Parameters of R201 T36F 343.0 21 220 0.5 0.5 3.2
M0/(mol  m3) 9 700 plant data
T0/K 343 plant data
1 a)
Figure 4 shows that the comparison between the plant
Fpp/(kg  h ) 20 00028 000 plant data PSD and the simulated PSD under the steady-state
VR/m3 56 plant data polymerizations in R201. The results turn to be good as a
Properties and kinetics parameters whole, although the model slightly overestimates the
Ke/(W  m1  K1) 0.15 [31] amount of the fine powers, less large particles, and wider
1
Cp/(J  kg  K) 1 400 [31] distribution than those of the plant PSD. The reason for
these differences may be due to the negligence of the
DHP/(J  mol1) 85 830 [31]
electrostatic effects during the experimental analyses of
De/(m2  s1) 0.5  1010 [31] the polymer sample in sizers, which may alter the shape
1
EA/(J  mol ) 5  104 [31] of the PSD. In addition, the assumption of well-mixed in
ED/(J  mol1) 5  104 [31] tubular loop reactor also takes effects on the prediction
R0/m 20100 plant data errors. Furthermore, another physical explanation can be
Mm/(kg  mol1) 0.042 [31] the simultaneous occurrence of agglomeration of small
3 particles, and attrition/break-up of large particles. In
rcat /(kg  m ) 2 840 [17]
general, the model developed here can be used to predict
rpp /(kg  m3) 910 [17] the polymer PSD in a steady-state tubular loop reactor.
C0 /(mol  kg1) 0.2 [12]
K1/(kg  s  m1) 2.5  10 13
[3]
K2 2.5 [3] Single-Particle Growth

Kp0 /(m3  mol1  s1) 1.2  104 b)


plant data In order to describe accurately the effects of the intraparticle
mass and heat transfer resistances on the growth of the
a)
Presented directly from the controlling computer at the plant
site; b)Kp0 can be obtained from Equation (1), (2).[29]

Figure 4. Comparison of the simulated PSD with the plant PSD.


(T0 343 K, M0 9 700 mol  m3, Kd0 1.6  104, De 1011 m2  s1,
Fpp 28 000 kg  h1, Rec 100, K1 2.5  1013 kg  m1  s1,
Figure 3. PSD of the plant catalysts. K2 2.5).

Macromol. React. Eng. 2010, 4, 123134


128 2010 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim DOI: 10.1002/mren.200900040
Coupled Single-Particle and Population Balance Modeling for . . .

Figure 5. Effect of De on the radial profile of propylene concen- Figure 7. Effect of catalyst size on the temperature profile in
tration in polymer particle. (D0 50 mm, M0 9 700 mol  m3, polymer particle. (De 1011 m2  s1, M0 9 700 mol  m3,
T0 343 K, t 0.1 h). T0 343 K, t 0.01 h).

polymer particle, the monomer, and temperature profiles in polymerization stage, the profile of propylene concentra-
the growth particle are simulated with an assumption of tion in the polymer particle is very steep, which means
the catalyst feed consisted of uniform size particles for the that the effect of the mass transfer resistance is significant.
model input. With the polymerization proceeding, the propylene con-
Figure 5 illustrates the effect of the effective diffusion centration in the inner layer increases and the profile
coefficients on the radial profile of the propylene concen- becomes flat. After an hour polymerization, the propylene
tration in the polymer particle. Figure 5 indicates that the concentration in the center of the polymer particle is
propylene concentration in the inner layer is lower than almost the same as that on the particle surface. Above
that in the outside layer because of the mass transfer intraparticle mass transfer resistance decreases is due to
resistance, and the profile of the propylene concentration in low propylene consumption rate due to catalyst deactiva-
the polymer particle is steep, especially when the value of tion after longer polymerization time.
De is small. It implies that the effect of the intraparticle mass Figure 7 depicts the effect of the initial catalyst size on the
resistance is important for the propylene polymerization in temperature profile in the polymer particle. Figure 7 shows
the loop reactors. that the highest temperature position appears in the lay
The effect of the polymerization time on the radial profile near the particle surface. As to the highest temperature, it is
of the propylene concentration in the polymer particle is also weak. Namely, there is no serious temperature gradient
shown in Figure 6, which shows that at the early in the polymer particle. In fact, according to Figure 7, the
biggest temperature rise is only 0.008 T/T0 (about 2.4 K),
thus the effect of the intraparticle heat resistance is not
important for the propylene polymerization in the loop
reactors. Therefore, although the effect of the intraparticle
heat resistance is considered in the PMLM, no further
discussion on heat transfer within particle will be made in
the following.
The effect of the initial catalyst particle size on the
particle growth rate is shown in Figure 8. Based on the
figure, the polymer particle grows to 2025 times in
diameter after 100 min from initial catalyst. Yet, Figure 8
also shows that the growing rate of the polymer particle
polymerizing from the smaller catalyst particle is faster
than that from the bigger catalyst particle. In fact, the
propylene concentration in the small particles is higher
than that in the big particle under the effect of the
Figure 6. Effect of polymerization time on the radial profile of intraparticle mass transfer resistance. Accordingly, the
propylene concentration in polymer particle. (De 1011 m2  s1, increase rate of the small particle is faster than that of
D0 50 mm, M0 9 700 mol  m3, T0 343 K). the big particle via polymerization.

Macromol. React. Eng. 2010, 4, 123134


2010 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.mre-journal.de 129
Z.-H. Luo, S.-H. Wen, D.-P. Shi, Z.-W. Zheng

Figure 8. Particle growth rate with the different initial catalyst


size. (De 1011 m2  s1, M0 9 700 mol  m3, T0 343 K). Figure 10. Effect of De on the PSD in R201. (T0 343 K, M0
9 700 mol  m3, Kd0 1.6  104, D0 50 mm, Fpp 28 000 kg  h1,
Rec 100).

PSD in the Loop Reactors


in R200 is more important than that in R201. In practice, the
Single-Size Catalyst Feed shorter diffusion length represents the less internal mass
transfer resistance. Corresponding higher intraparticle
The Effect of the Internal Mass Transfer Limitation on average monomer concentration can be obtained. Accord-
the PSD ingly, higher volumetric polymerization rate in smaller
particles can be obtained.
Figure 9, 10 show the effect of the intraparticle mass
transfer limitation on the calculated PSDs in R200 and R201, The Effect of the Polymerization Temperature on the PSD
respectively. Figure 9, 10 show that both the PSDs in R200
Figure 11, 12 show that the effect of the bulk phase
and R210 become wider and the mean particle sizes
temperature on the calculated PSDs in R200 and R201,
increase with the increase in De, In addition, the comparison
respectively. It can be found that both the PSD widths in
between the two figures show that the effect of De on the
R200 and R201 become wider and the mean particle sizes
PSD in R201 is less than that in R200. Furthermore, Figure 6
change larger with the increase in the temperature because
shows that the intraparticle propylene concentration is
the polymerization rate is a function of temperature as
significantly influenced by the mass transfer resistance at
expressed in Equation (1)(4). It can be found that the
the early polymerization stage. So the effect of De on the PSD

Figure 11. The effect of temperature on the PSD in R200.


Figure 9. The effect of De on the PSD in R200. (T0 293 K, (M0 9 700 mol  m3, Kd0 1.6  104, D0 50 mm, De 1011
M0 9 700 mol  m3, Kd0 1.6  104, D0 50 mm). m2  s1).

Macromol. React. Eng. 2010, 4, 123134


130 2010 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim DOI: 10.1002/mren.200900040
Coupled Single-Particle and Population Balance Modeling for . . .

Figure 14. Effect of Kd0 on the PSD in R201. (T0 343 K,


Figure 12. Effect of temperature on the PSD in R201.
M0 9 700 mol  m3, D0 50 mm, De 1011 m2  s1, Fpp
(M0 9 700 mol  m3, Kd0 1.6  104, D0 50 mm, De 1011
28 000 kg  h1, Rec 100).
m2  s1, Fpp 28 000 kg  h1, Rec 100).

polymerization rate increases and the catalyst deactivation sizes become smaller with the increase in k0d . In practice, the
changes quicker with the increase in the temperature. particles remaining in the reactors longer than on average
Accordingly, the polymerization rate increases with the finally grow slower than on average due to catalyst
increase in the temperature as indicated in Figure 11, 12. deactivation. Accordingly, the PSD widths become marrow.
However, in certain actual plant, it is a problem to remove In addition, with the increase in k0d , the catalyst deactivates
much polymerization heat produced in the polymerization more quickly and the polymerization rate becomes smaller,
with the increase in the polymerization temperature. and the mean particle size becomes smaller in both reactors.
Moreover, the effect of k0d on the PSD in R200 is not as
The Effect of the Catalyst Deactivation on the PSD
significant as that in R201 because the residence time of the
Since the catalyst deactivation is proportional positively to catalyst in the R200 is less than that in R201.
k0d according to Equation (3), (4), the effect of the catalyst
The Effect of the Catalyst Size on the PSD
deactivation on the PSD can be simulated via the change of
k0d . Figure 13, 14 show the simulated results of such changes. Figure 15, 16 show the effect of the catalyst size on the PSD
According to Figure 13, 14, we find that both the PSD widths in both reactors. Simulated results show that the mean
in R200 and R201 become narrow and their mean particle polymer particle size increases with the increase in the

Figure 13. Effect of Kd0 on the PSD in R200. (T0 293 K, Figure 15. Effect of D0 on the PSD in R200. (T0 293 K,
M0 9 700 mol  m3, D0 50 mm, De 1011 m2  s1). M0 9 700 mol  m3, Kd0 1.6  104, De 1011 m2  s1).

Macromol. React. Eng. 2010, 4, 123134


2010 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.mre-journal.de 131
Z.-H. Luo, S.-H. Wen, D.-P. Shi, Z.-W. Zheng

Figure 18. Effect of Rec on the PSD in R201. (T0 343 K,


Figure 16. Effect of D0 on the PSD in R201. (T0 343 K,
M0 9 700 mol  m3, Kd0 1.6  104, D0 50 mm, De 1011
M0 9 700 mol  m3, Kd0 1.6  104, De 1011 m2  s1, Fpp
m2  s1, Fpp 28 000 kg  h1).
28 000 kg  h1, Rec 100).

catalyst size due to more catalyst active sites in the bigger Because the reactor residence time is affected by Fpp and
catalyst particle than that in the smaller catalyst particle. Rec, with the the decrease in Fpp or the increase in Rec, the
Figure 15, 16 also show that the catalyst size has significant residence time of the particle in the reactors increases,
effect on the PSD in both R200 and R201. so as to the particle size.
The Effects of the Propylene Flow Rate and the Recycle The Effect of the Agglomeration on the PSD in R201
Ratio on the PSD
To predict the effect of particle agglomeration on the PSD in
Propylene flow rate and recycle ratio (Rec) are important the loop reactors, the population balance model is also
operation variables in loop reactors, where Rec is defined as numerically solved together with the empirical particle
Rec Fec =Fout . agglomeration Equation (22), (23). Figure 19 shows
Figure 17, 18 illustrate that the effects of Fpp and Rec on the comparison of the simulated PSDs in the cases of
the PSD in R201. According to Figure 17, 18, we find in the presence of the agglomeration and without
that the decrease in Fpp or the increase in Rec result in the the agglomeration. Figure 19 illustrates that as the
increase in volume fraction of the larger particle.

Figure 19. Effect of agglomeration on the PSD in R201.


Figure 17. Effect of Fpp on the PSD in R201. (T0 343 K, (T0 343 K, M0 9 700 mol  m3, Kd0 1.6  104, D0 50 mm,
M0 9 700 mol  m3, Kd0 1.6  104, D0 50 mm, De 1011 De 1011 m2  s1, Fpp 28 000 kg  h1, Rec 100, K1 2.5  1013
m2  s1, Rec 100). (kg  s1)  m1, K2 2.5).

Macromol. React. Eng. 2010, 4, 123134


132 2010 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim DOI: 10.1002/mren.200900040
Coupled Single-Particle and Population Balance Modeling for . . .

Figure 20. Two initial catalyst PSD. Figure 21. Effect of the initial catalyst size distribution on the
PSD in R201. (T0 343 K, M0 9 700 mol  m3, Kd0 1.6  104,
D0 50 mm, De 1011 m2  s1, Ke 0.16 W  (m  K)1, Fpp
30 000 kg  h1, Rec 100, K1 2.5  1013 (kg  s)  m1, K2 2.5).

agglomeration appears, the PSD becomes broader, while its


peak value is shifted toward larger particles, for small
particles interact into the bigger particles. On the other limitations, the PMLM is first employed and solved
hand, the long tail of the PSD obtained is due to the together with the steady-state particle population model.
agglomeration of particles. Data for the steady-state polymerization collected from
industrial scale plant were applied to verify the model.
Multi-Size Catalyst Feed
Although the model predictions have more fine powders
In the propylene polymerization processes, the initial and less large particles than the plant data, the results
catalyst size distribution significantly affects the polymer are fundamentally found to be in agreement with the
PSD in the reactors. In fact, the initial catalyst PSD is divided plant data.
into a number of size cuts and the corresponding polymer The effects of the internal mass and heat transfer
PSD can obtained according to the above model. Further- limitations on the particle growth, the intraparticle
more, the overall PSD in R201 is obtained as a sum of PSDs monomer and temperature profiles are obtained by using
resulted from the individual catalyst cuts.[2,3] the suggested model. The simulated results show that the
Figure 20 shows that two catalysts feed are different in effect of the intraparticle mass resistance is important and
initial PSD. The two corresponding polymer PSDs are the intraparticle temperature gradients is not severe for the
obtained and illustrated in Figure 21. In addition, we must propylene polymerization in the loop reactors. Moreover,
point out the two catalysts are divided into seven size cuts, the effects of critical factors including the internal mass
namely, 020, 2040, 4060, 6080, 80100, 100120, and transfer resistance, feed catalyst particle size, its distribu-
120140 mm, in order to obtain the over polymer PSDs. The tion and catalyst deactivation, polymerization temperature
mean values of corresponding cuts are inputed into our on the PSD in the loop reactors are also investigated in this
model to calculate the polymer PSDs. According to Figure 21, work. The results show that the internal mass transfer
we can know that the shape of the overall polymer PSD resistance, as well as the initial catalyst size, has a strong
directly depends on the initial catalyst shape. With the impact on the PSD in the prepolymerization loop reactor. It
increase in catalyst size, the polymer particle size increases is also shown that the PSDs in both the prepolymerization
in the outlet polymer PSD and their distribution changes reactor and the main-polymerization reactor is greatly
wide. influenced by the feed catalyst particle size and the bulk
phase polymerization temperature. In addition, the results
show that the PSD shifts to large particle size with the
Conclusion decrease in the propylene flow rate or the increase in the
recycle ratio. The effect of the agglomeration on the PSD is
A comprehensive steady-state population balance model also simulated in this article. The results show that as the
was developed for the PSD of the PP produced in loop agglomeration exists, the PSD becomes broader while its
reactors. To calculate the growth rate of a single particle peak value is shifted toward larger particle sizes with a
under the internal and external heat and mass transfer long tail.

Macromol. React. Eng. 2010, 4, 123134


2010 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.mre-journal.de 133
Z.-H. Luo, S.-H. Wen, D.-P. Shi, Z.-W. Zheng

Nomenclature Received: June 26, 2009; Revised: August 18, 2009; Published
online: November 4, 2009; DOI: 10.1002/mren.200900040

Cp heat capacity of the polymer particle (J  kg1  K1) Keywords: loop reactor; particle population balance; particle size
C active site concentration (mol  kg1) distribution; polymeric multilayer model; poly(propylene)
C0 initial active site concentration (mol  kg1)
D particle size (m)
De effective diffusion coefficient (m2  s1)
EA activation energy of propylene polymerization
(J  mol1)
ED activation energy of catalyst deactivation reaction [1] Z. H. Luo, Y. Zheng, Z. K. Cao, S. H. Wen, Polym. Eng. Sci. 2007,
(J  mol1) 47, 1643.
[2] H. Hatzantonis, A. Goulas, C. Kiparissides, Chem. Eng. Sci.
Fec volume recycle rate (kg  h1)
1998, 53, 3251.
Fin volume feed rate (kg  h1) [3] H. Yiannoulakis, A. Yiagopoulos, C. Kiparissides, Chem. Eng.
Fout volume withdrawal rate (kg  h1) Sci. 2001, 56, 917.
Fpp Propylene flow rate (kg  h1) [4] F. A. N. Fernandes, L. M. F. Lona, J. Appl. Polym. Sci. 2001, 81,
G(D) particle growth rate (m  s1) 321.
kd catalyst deactivation rate constant (s1) [5] M. Caracotsios, Chem. Eng. Sci. 1992, 47, 2594.
[6] B. Z. Yang, W. Jiang, J. D. Wang, Y. R. Yang, J. Chem. Eng.
k0d frequency factor of catalyst deactivation reaction Chinese Univ. 2005, 19, 461 (in Chinese).
(W  m1  K1) [7] Z. H. Luo, P. L. Su, D. P. Shi, Z. W. Zheng, Chem. Eng. J. 2009, 146,
Ke effective thermal conductivity coefficient of poly- 466.
mer particle (W  m1  K1) [8] T. F. McKenna, J. B. P. Soares, Chem. Eng. Sci. 2001, 56, 3931.
kp propagation rate constant (W  m1  K1) [9] W. R. Schmeal, J. R. Street, AIChE J. 1971, 17, 1189.
[10] R. Galvan, M. Tirrell, Comput. Chem. Eng. 1986, 10, 77.
k0p frequency factor of propagation reaction [11] R. Galvan, M. Tirrell, Chem. Eng. Sci. 1986, 41, 2385.
(W  m1  K1) [12] R. A. Hutchinson, C. M. Chen, W. H. Ray, J. Appl. Polym. Sci.
K1 proportionality constant (kg  s  m1) 1992, 44, 1389.
K2 proportionality constant [13] Z. H. Luo, S. H. Wen, Z. W. Zheng, J. Chem. Eng. Jpn. 2009, 42,
576.
K0 agglomeration rate constant (kg  s  m1)
[14] J. B. P. Soares, A. E. Hamielec, Polymer 1996, 37, 4607.
M propylene concentration (mol  m3) [15] J. Z. Sun, C. Eberstein, K. H. Reichert, J. Appl. Polym. Sci. 1997,
Mm molecular weight of monomer (kg  mol1) 64, 203.
M0 bulk poly(propylene) concentration (mol  m3) [16] J. J. Zacca, J. A. Debling, W. H. Ray, Chem. Eng. Sci. 1996, 51,
N size distribution in the reactor (m1) 4859.
Nin size distribution of the feed catalyst particles (m1) [17] Y. M. Harshe, R. P. Utikar, V. V. Ranade, Chem. Eng. Sci. 2004,
59, 5145.
T temperature (K) [18] K. Y. Choi, X. Zhao, S. Tang, J. Appl. Polym. Sci. 1994, 53, 1589.
T0 bulk temperature (K) [19] D. Y. Khang, H. H. Lee, Chem. Eng. Sci. 1997, 52, 421.
T sf average polymer softening temperature (K) [20] H. Hatzantonis, A. Yiagopoulos, H. Yiannoulakis, C. Kiparissides,
Qp total polymerization heat (W  m3) Spring AIChE Meeting, Houston TX, March 14-18, 1999.
t time (s) [21] K. Y. Choi, W. H. Ray, J. Macromol. Sci. R. M. C. 1985, 25, 1.
[22] K. Y. Choi, W. H. Ray, J. Macromol. Sci. R. M. C. 1985, 25, 57.
R radius of polymer particle (m) [23] G. Dompazis, V. Kanellopoulos, C. Kiparissides, Comput. Aided
Rgas gas constant (8.314 m3  Pa  K  mol1) Chem. Eng. 2005, 21, 345.
Rp polymerization rate (mol  s1  m3) [24] A. D. Randolph, M. A. Larson, Theory of particulate processes:
r radial position (m) Analysis and techniques of continuous crystallization, Aca-
demic Press, New York 1971.
VR reactor volume (m3)
[25] J. D. Litster, D. J. Smit, M. J. Hounslow, AIChE J. 1995, 41, 591.
rp polymer particle density (kg  m3) [26] H. Arastoopour, C. S. Huang, S. A. Weil, Chem. Eng. Sci. 1988,
rcat catalyst density (kg  m3) 43, 3063.
DHp heat of polymerization (J  mol1) [27] J. Villadsen, M. L. Michelsen, Solution of Differential Equation
Models by Polynomial Approximation, Prentice-Hall, Engle-
wood Cliffs 1978.
[28] D. Ramkrishna, Population Balances: Theory and Applications
to Particulate Systems in Engineering, Academic Press, San
Diego 2000.
Acknowledgements: The authors thank the National Natural [29] U. P. Veera, Chem. Eng. Sci. 2003, 58, 1765.
Science Foundation of China (no. 20406016) for financial support [30] Y. Chen, X. G. Liu, Polymer 2005, 46, 9434.
and Fujian Petrochemical Company of SINOPEC for plant data and [31] J. J. Zacca, W. H. Ray, Chem. Eng. Sci. 1993, 48, 3743.
permission to publish it. [32] A. S. Reginato, J. J. Zacca, A. R. Secchi, AIChE J. 2003, 49, 2642.

Macromol. React. Eng. 2010, 4, 123134


134 2010 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim DOI: 10.1002/mren.200900040

Das könnte Ihnen auch gefallen