Professional Documents
Culture Documents
CHAPTER 2
3.1 METHOD
Developing an accurate simulation of propylene polymerization in fluidized-
bed reactors using Aspen Polymers requires special expertise. Predicting
reliable physical and thermodynamic properties of a chemical system is
important in process simulation for simulating chemical unit operations
such as reactors and separators. Unfortunately, creating thermophysical
properties for polymers, as well as polymerization reaction models, is
especially challenging and requires attention to detail. Thermodynamic
phase equilibrium models are categorized as either activity-coefficient or
equation-of-state (EOS) based models. In the present work, EOS models
were selected for the polypropylene process due to the nature of the
species and high operating pressure involved. EOS models are suitable for
systems at moderate to high pressures and reactor pressure for this gas-
phase polypropylene process ranges from 20 to 30 bar [5]. According to
Gross and Sadowski [20], a model based on the perturbed-chain statistical
associating fluid theory (PC- SAFT EOS), is the best for describing physical
and thermodynamic properties of a polymeric systems. Therefore, PC-
SAFT EOS was used to model the physical and thermodynamic properties
for the polypropylene process. The PC-SAFT model is an extension of the
prestigious SAFT EOS model. The main difference between these
models is that the PC-SAFT model considers hard sphere chains while
SAFT model accounts for hard spheres. This allows PC-SAFT EOS to
include the connectivity of segments that comprise the chains when
considering the attractions between species. This improvement shows a
more realistic description of the thermodynamic behavior of chainlike
molecules [5, 21, 22]. The PC-SAFT EOS also shows superior prediction
for vapor-liquid equilibrium and binary mixtures of small molecules
compared to SAFT and Peng-Robinson models, respectively [5, 6].
Table 5 lists the components included in the model. The applied pure-
component parameters for the PC-SAFT equation of state model (EOS) are
listed in Table 6 [6]. The data listed in these tables were integrated into the
Aspen Polymers simulator associated with the built-in PC-SAFT EOS
model.
Species Function
1. Propylene C 3 H 6 Monomer
2. Polypropylene(PP) Polymer
3. Propylene segment C 3 H 6 R Polymer segment
4. Hydrogen H 2 Chain transfer agent
5. Titanium Tetrachloride (CAT) Catalyst
6. Triethyl-Aluminium (COCAT) Co-catalyst
7. Nitrogen N 2 Inert
8. Water H 2 O Cooling water
No Component A U K K
B r mol g
1. Hydrogen H 2 0.8285 2.973 12.53 -
2. Propylene C 3 H 6 1.960 3.3536 207.2 -
3. Polypropylene(PP) - 4.147 298.6 0.0253
4. Catalyst 25 2.668 198.8 -
5. Co-catalyst 25 2.668 198.8 -
Based on the thermodynamic model and the component parameters that
are suitable for the reaction conditions involved in the present process [5,
6], a RCSTR module within Aspen Polymers was selected to model the
gas-phase fluidized polymerization reactor. Furthermore, the
characterization of the ZieglerNatta catalyst and a set of two-site kinetic
schemes with reaction rate constants were incorporated into the model.
The emulsion phase can be simply presented by a single RCSTR model
which takes gaseous feed that includes the propylene gas, hydrogen gas
and the nitrogen gas as an inert and 88% of the solid feed which is Ziegler-
Natta catalyst [14]. For modeling the bubble phase, which accounts for the
gaseous feed and 12% of the solid feed, there are two options in Aspen
Polymers: a single RPlug or a series of RCSTR. Since the behaviour of a
plug-flow reactor can be approximated by that of a multistage CSTR [23], a
series of four equal- volume RCSTR modules was chosen to represent the
bubble phase due to an impractical low polypropylene production using an
equivalent RPlug module.
A major problem faced during the process simulation was the convergence
of the recycle stream which was not solved in a previous work [6]. In the
present study, according to the calculation procedure of the process flow
sheet, the problem was addressed and solved with some difficulty. From
this angle, this work can be regarded as the first of its kind to be able to
estimate the real gas-phase fluidized bed polymerization system which
considers both the emulsion phase and bubble phase for predicting the
polymer production rate, weight average molecular weight, number
average molecular weight, PDI and MFI with Aspen Polymers. The flow
sheet for the complete gas phase catalytic propylene polymerization
fluidized-bed reactor is shown in Figure2.
Several other models have been proposed in the literature to describe the
kinetic reaction for a heterogeneous Ziegler-Natta catalyst [1, 12-15]. The
Ziegler-Natta catalyst, considering multiple active sites of two (2) or four (4)
and as well as heat- and mass-transfer resistance, tends to produce a
polymer with a broad molecular weight distribution. The effect of multiple
site types is more
significant than the effect of heat- and mass-transfer resistances. The
single-site kinetic model is incapable of describing the kinetic behavior,
production rate and molecular weight distribution of propylene
homopolymerization. The added complexity of using more sites may limit
the models use for simulation purposes, and the estimation of the large
number of kinetic parameters may be restrictive [12]. Therefore, a two-type
active site was considered in the present study to describe polymer
properties for both models. Using similar methodology to McAuley et al. [1],
Carvalho de et al. [12] and Kissin [13], the kinetic model was developed for
ZieglerNatta catalysts containing multiple active sites to describe the
homopolymer production rate, molecular weight and its distribution.
The reaction rate constants used in this work were taken from different
published works on similar reactive systems [1, 6, 15]. This is mainly due to
the lack of a unique source that covers all the kinetic parameters for
propylene polymerization. Characterization of polymer properties was
modeled using population balances and method of moments. The kinetic
scheme comprising of a series of elementary reactions and the rate
parameters, used for the two phase model and Aspen Polymers simulator,
are listed in Tables 1 and 2, respectively. Aspen Polymers and the two-
phase model were standardized to use the same components in the input
flow and kinetic parameters
The emulsion phase is not at the minimum fluidization condition and the
bubble phase contains solid particles.
There is negligible resistance to heat and mass transfer between the gas
and polymer particles due to small catalyst particles, low to moderate
catalyst activity or polymerization rates [19].
3.2 MODELLING
In the fluidized bed reactor calculations, the first step taking in resolving
problem is to calculate Minimium Fluidization Velocity which is the velocity
at which the fluidized bed reactor starts to fluidized. The Minimium
Fluidization Velocity can be calculated from an equation below which is a
form of Ergun equation for Geldart particles of type B (Lucas et al, 1986)
Re mf { 29.5 0.0357 Ar} 2 29.5 (3.1)
2 1
d 3p g polymer g g
Remf (3.2)
g2
Re mf g
U mf (3.3)
d pg
polymer Bulk
polymer density, g
g gravitational acceleration, g gas density
diameter of the bubble formed initially just above the distributor plate, and
d bm is the maximum bubble diameter attained if all the bubbles in any
horizontal plane coalesce to form a single bubble (as they will do if the bed
is high enough). The maximum bubble diameter, d bm has been observed to
follow relationship
d bm 0.652[ Ac U 0 U mf ]0.4 (3.6)
For all beds, while the initial bubble diameter depends upon the type of
distributor plate. For porous plates, the relationship
d bo 0.00376U 0 U mf 2
(3.7)
d min imium Is observed, and for perforated plates, the relationship
0.4
d bo 0.347 Ac U 0 U mf 1 (3.8)
nd
this equation appears to be valid, in which nd is the number of perforations.
For beds with diameters between 30 and 130cm, these relations appear to
predict bubble diameters with an accuracy of about 50%; for beds with
diameters between 7 and 30cm, the accuracy of prediction appears to be
approximately +100%, -60% of the observed values.
Werther developed the following correlation based on a statistical
coalescence model: 11
d b =0.8531+0.272
U O U mf 3.1 0.0684
The bubble size predicted by this model is close to that predicted by Mori
and Wen for large diameter bed(2m) and smaller than that suggested by
Mori and Wen for small diameter beds (0.1m)
From experiments with single bubbles, Davidson and Harrison found that
the velocity of rise of a single bubble could be related to the bubble size by
U br 0.71 gd b 2 (3.9)
1
When many bubbles are present, this velocity would be affected by other
factors. The more bubbles that are present, the less drag there would be on
an individual bubble; the bubble would carry each other up through the bed.
The greater number of bubbles would result from large amounts of gas
passing through the bed (i.e., a large value of U 0 ). Therefore, the larger the
value of U 0 )., the faster should be the velocity of a gas bubble as it rises
through the bed.
Other factors that should affect this term are viscosity of the gas and the
size and density of the solid particle that make up the bed. Both of these
terms also affect the Minimium Fluidization Velocity and so this term might
well appear in any relationship for the velocity of bubble rise; the higher the
Minimium Fluidization Velocity, the lower the velocity of the rising bubble.
Adopting an expression used in gas-liquid systems, Davidson and Harrison
proposed that the rate of bubble rise in a fluidized bed could be
represented by simply adding and substracting these terms: for situation
where many bubbles occur as in the case of a fluidized bed reactor. In a
fluidized state the bubble velocity U b
U b U br U 0 U mf (3.10)
U 0 Superficialvelocity
U mf velocityat the minimium fluidization
Substituting the velocity of single bubble
U b U 0 U mf 0.71 gd b 2 (3.11)
1
The bed fraction in the emulsion phase (which includes the cloud) is
1 . Letting Ac and c representing the cross sectional area of the
bed and the density of the solid particles, respectively, a material balance
on the solids give:
Solid flowing downward in emulsion=Solid flowing upward in wakes
Ac c 1 U s U b c Ac (3.12)
U b
Us (3.13)
1
A material balance on the gas flows gives
AcU 0 Ac U b Ac mf U b Ac mf 1 U c (3.14)
(Total gas flow rate)=(gas flow in bubbles)+(gas flow in wakes)+(gas flow in
emulsion)
AcU 0 Ac U b Ac mf U b Ac mf 1 U c (3.15)
AcU c Total gas flow
Ac U b Gas flow in bubble
Ac mf U b Gas flow in wakes
Ac mf 1 U e Gas flow in the emulsion
U mf
The velocity of the emulsion phase Ue Us
mf
Many recent studies show that the emulsion phase does not stay at
minimum fluidization but it may contain more gas at higher gas velocity
( Abramhamson et al., 1980, Chaouki et al., 2000). It is expressed as :
U mf
The velocity of the emulsion phase Ue U s -------------(3.19)
mf
The relative velocity between the solid particle and the gas used in
calculating the Reynolds number will be taken as
In the emulsion phase, the equation would be one that applied to fixed-bed
operation with a porosity in the bed equal to and a velocity of . the equation
recommended by Kunii and Levenspiel:
Where and are the concentration of A in the bubble and cloud phase
respectively, and represents the number of moles of A transferred from
the bubble to the cloud per unit time unit volume of bubble (). The concept
of basing all mass transfer (and later, all reaction) on the bubble volume
proves to simplify the calculations markedly. For the products, (e.g., B in A-
B) the rate of transfer into the bubble from the cloud is given by a similar
equation
Where
Using Davidsons expression for gas transfer between the bubble and the
cloud, and then basing it on the volume of the bubble, Kunii and Levenspiel
obtained this equation for evaluating
Where is the moles of A transferred from the cloud to the emulsion per unit
time per unit volume of bubble. Note thet even though this mass transfer
does not involve the bubble directly, it is still based on the bubble volume.
Using Higbies penetration theory and his analogy for mass transfer from a
bubble to a liquid, Kunii and Levenspiel developed an equation for
evaluating
Where U is the velocity of bubble rise in cm/s and the other symbols. And
the order of magnitude of this coefficient is equal to 1s-1. K can also be
thought of as the exchange volume between the cloud and the emulsion.
With the knowledge of mass transfer coefficients, the amount of gas
interchange between the phases of a fluidized bed can be calculated and
combined to predict the overall mass transfer behavior or reaction
behavior of fluidized bed process.
In which the reaction rate is defind per unit volume of bubble. In the cloud
phase,
Similarly in the emulsion,
Where are the specific reaction rates in the emulsion, cloud and bubble
respectively. In the latter two equations, the reaction rate is also defined per
unit volume of bubble.
3.2.14.2
3.2.15