Professional Documents
Culture Documents
Steady State and Transient Behaviour of Continuous PET Condensation
Steady State and Transient Behaviour of Continuous PET Condensation
Synopsis
The steady-state and transient behavior of a continuous stirred-tank reactor for melt transesterification of dimethyl terephthalate with ethylene glycol in the presence of metal acetate
catalyst is presented. The kinetic model includes the main transesterification reactions and side
reactions leading to diethylene glycol and carboxylic acid end groups. The effect of various reactor
operating parameters such as [EG]/[DMT] mole ratio and feed catalyst concentration on the
product distribution under steady-state reactor operating conditions is analyzed. The dynamic
process model has also been solved and the reactor transients to step changes in various reactor
parameters are reported.
INTRODUCTION
Polyethylene terephthalate (PET) is one of the fastest growing thermoplastic polyesters and used widely for fibers, films, bottles, and injection-molded
parts. PET is manufactured by step-growth melt polymerization process
which consists of three reaction steps: (1) melt transesterification of dimethyl
terephthalate (DMT) with ethylene glycol (EG), (2) prepolymerization a t
250-300C and
30 mm Hg, (3) finishing polymerization at 280-300C and
< 1 mmHg. Due to the stringent industrial specifications that today's polymers must meet, the need for enhanced quality control through more effective
reactor operation is now more acute than ever. For instance, in order to
produce PET resins offering a combination of high strength, stiffness, dimensional stability, chemical and heat resistance, and good electrical properties,
fundamental polymer properties such as molecular weight, molecular weight
distribution, and unwanted side product concentrations must be controlled
within limits that would not damage industrial end use requirements. The
presence of diethylene glycol (DEG), even in small concentration (1-2 wt %),
in final polymers causes decreased crystallinity of polymers, resulting in
lowered melting point, reduced heat resistance, decreased thermal oxidative,
and ultraviolet (w)light stability.
Thus, it is important to understand the relations between reactor operating
conditions and polymer properties for precise polymer quality control. The
ester interchange between.DMT and an excess of EG represents the first step
708
MODEL DEVELOPMENT
Consider the process flow diagram for a continuous melt transesterification
process shown in Figure 1. Molten DMT is fed to the ester interchange reactor
which is kept under an inert atmosphere within the temperature range
180-200C. Ethylene glycol is passed in a similar fashion to the reactor from
the EG storage tank. The metal acetate catalyst (e.g., zinc acetate) is injected
into the EG feed stream from the catalyst storage tank. The resultant vapors
(mostly methanol, EG, and trace amounts of other volatile side products)
from the combination of DMT and EG to produce bis-hydroxyethyl terephthalate (BHET) are rectified in the distillation column; thus, EG is refluxed
back into the reactor. The more volatile components in the vapor phase are
condensed in the condensor and their amounts are used to measure the extent
of reaction in the liquid phase. The transesterification product is transferred
t o the prepolymerization reactor which is maintained between 10-30 mmHg
and 280-30O0C. The polymer melt is pumped to the finishing polymerization
reactor, usually an extruder type or a wiped film reactor.
709
DMT
Distillation column
EG
Cat.
EG
TRANSESTERIFICATION
PREPOLY MERlZATlON
REACTOR
.
T o finisliing reactor
R,
+ G P PI + M .
k2
ki
The chemical formulae for the symbols are shown in Table I. Since the vapor
pressure of methanol is very high at the reaction temperature, it is assumed
that methanol leaves the reaction mixture as soon as it is formed. When the
continuous removal of liberated methanol (M) is efficiently performed, the
reactions (1)and (2) can be approximated as irreversible reactions. In addition
710
to main reactions (1)and (2), important side reactions lead to the formation of
DEG, which deteriorates the properties of the final PET produced. Metal
acetate catalysts, commonly used in PET production, promote side reactions
which lead to the formation of DEG, acetaldehyde, acid end groups, vinyl end
groups, and water. Distillation is not likely to remove DEG immediately upon
formation due to its rather high boiling point, 245OC. Therefore, the side
reactions considered in this work deal directly with the formation of DEG;
Here, T, and Q, species contain carboxylic end groups which are formed
through the reaction between EG and DEG end groups in R, and PI,
respectively. Acid end groups, formed in the same reaction as DEG, can
participate in polymer chain growth as long as the concentration of hydroxyl
end groups is high compared to acid end groups. If the acid end group
formation rate is very high, then the final degree of polymerization decreases.
Controlling the acid content in the polymer is very important for the thermal
711
dD
Vdt'
Doqo - Dql
dG
Vdt'
klVCcQtDG
(5)
k3)VcCQtRlG
- '3"CQtPiG
(6)
712
Q-D
4
Rl
DO
PI
xq = -
xg =
0,
(TO
GD
x3 = DO
X6 =
""%
TI
x7 =
Go - G
GO
x, = -
Ccat, - Ccat
x9 =
Q1
DO
(T -
Tk
cca,
- Tk)r
Yl =
Tk)r
Y3 =
Go%
Tk
Y4
B,
h/P
AHlG0r
=
PcpTk
Greek letters:
e=- V
Qo
Here, perfect backmixing of the reacting mixture has been assumed. Note that
vM
where
denotes the molar volume of methanol.
It is convenient to put the modeling Eqs. (5)-(14) into dimensionless form.
For this purpose, the dimensionless variables arid parameters defined in Table
I1 were substituted into the material and energy balances resulting in the
713
+ q ( l -XI)
+z,Dur,(l
h,
- - -1
dt
- q)(l
+ q ( l - x , ) + Dur,(l
- x,)x,(l
+Dur,(l
+Du(l
- = 2 3 - qx3
dt
-= Z,
dt
- x,)exp
+ z,Du(l
dt
- x,)exp
1 + x9/r
I 1 + x9/r
+ x,)(l
(16)
- x8)exp
-z,Du(l
- x2)x4(1
- x,)exp
- z5 - q x ,
+ z2Dur,(l
- x,)x,(l
-z,Du(l
[
[
1 :Z:/r]
1
1
x9
- x,)exp
:z:/r]
x9
- x2)(x4
-z,Dur,(l - x,)x,(l
h 5
-
(15)
- x,)(l - x,)(l
- x,)exp
- x2)x4(1- x,)exp
+ Du(1 - xz)x5(l
- x,)exp
- x,)(l
1 :~:/r]
1 :Z:/r]
x9
1 + x9/r
- x,)exp
- x z ) x 5 ( l - x,)exp
1
[
-?::/r]
+%r
+ xg/r
xg
(17)
714
h
-=
dt
-B,Dur,(l
-B3Du(l
- x2)x4(1 - x,)exp
- x2)x,(1
1 ?z:/r]
- x,)exp
- (B"+ Yd"9 + Y 7 ) )
where
1 - y3Dur1(l- ",)(I
- x,)(l - x,)exp
[ %]
1
STEADY-STATE ANALYSIS
The steady-state model follows directly from the dynamic modeling equations once the left-hand side of the model equations is set to zero. The
resulting nine nonlinear equations may be reduced to the following five after
715
-1
O = -1
+Dur,(l
+Du(l
X:/r]
- x 2 ) x 4 ( 1 - x,)exp
- x2)x4(1 - x,)exp
+ D u ( l - x,)x,(l
O = z4 - qx,
- x2)(1 - x8)exp
[ +xx9g/r]
[ ;lg/r]
1
- x,)exp 1
+ z,Dar,(l - x , ) ( l
- x 2 ) ( 1 - x,)exp
-B,Darl(l
- x,)(l
- x,)(1
-B,Dur2(1 - x,)x4(1
- x,)exp
- x,)exp
+Dur,(l
:l:;/r]
1 :~:/r]
- x 2 ) x 4 ( 1-
x8)exp 1it
(25)
716
where
z5
+ z,Dur,(l
x5 =
q
x6 = 2,q-l
+ z,Du(l
- xz)x4(1- x8)exp
- x,)(1 - x8)exp
+ z,Du(l
[ :i:/r]
1
X9
1 + x9/r
- x 2 ) x 5 ( l - x,)exp
(31)
x9
+ xg/r]q-l
717
TABLE I11
Numerical Values of Physical Constants and Parameters
A,
8.8513
lo9
-42
6.0531 X
lo9
1.3020 X 10
574.0
0.245
12.60
14340.0
14784.0
29800.0
1400.0
1400.0
0.0
0.0
8936.0
450.0
298.15
10.0
0.05363
P
1.11
P1
1.11
PD
1.088
L2
mo12 h
L2
mo12 h
L2
mo12 h
cm2
cal
gK
cal
mol K
cal
mol
Cal
~
mol
Cal
mol
cal
mol
Cal
mol
cal
mol
Cal
mol
Cal
mol
K
K
Cal
cm2 h K
cm
I!
cm3
718
5.0
1.0
2.4
2 .o
4 .O
- nu n
.v
s!
X
.
-
3 .O
0.6
2.0
0.4
1.0
0.2
0 .o I - I
1.0 I
1.6
0.8
2
0
v)
a
W
>
0.6
0.4
0.2
0
0
W
0.0
180
0
0
W-
5a:
170
a
2
W
I-
i60
0
RESIDENCE TIME, HR
Fig. 2.
comparison of our simulation results with experimental data was not possible.
However, the observed DMT conversion versus residence time curves are
qualitatively similar to that of semibatch transesterification process.'*'2 Note
that [EG]/[DMT] mole ratio significantly affects the formation of DEG, in
particular for large residence time. For instance, 20% increase in [EG]/[DMT]
ratio almost doubles the DEG concentration for residence time of 4 h. Even
for long residence times with [EG]/[DMT] = 2.4, considerable amounts of
unreacted methyl ester groups are present in the reaction mixture. The
concentrations of MTPA and HTPA represent the acid end groups. The sum
of the concentrations of MTPA and HTPA equals the concentration of DEG
719
9 .o
6.0
'
.
X
3.0
0
w
D
' 0.0
1.6
4 .O
1.2
3 .O
0.8
2.0
0.4
1.0
0.0
RESIDENCE T I M E , H R
720
0.9
0.6
d[
[[CcatIJDMT]
C c a t 1 DMT 1 0
0.3
0.0
3.E-4
l.E-3
2.0
1.0
0.0
RESIDENCE TIME, H R
residence times, methanol is continuously produced vigorously since the material is processed so rapidly. Thus,much heat is lost due to evaporative cooling,
causing the reactor temperature to drop significantly below that of the
heating medium [cf. Fig. 2(c)]. But, as residence time approaches 5 h, the four
chemical reactions are allowed to proceed to near completion, thus reducing
the steady-state methanol flow rate and allowing the reaction temperature to
very closely approach the heating medium temperature.
The effect of catalyst concentration is illustrated in Figures 3(a)-(f). Increasing the feed catalyst concentration has the predictable effect of increas-
4 .O
1.2
3 .O
0.8
2.0
721
I
-m
6W
0.4
1.0
0 .o
0.0
"
z
X
I-
s1
2.0
1.0
0 .o
0
RESIDENCE TIME, H R
ing both DMT and EG conversion. With increased catalyst feed concentration
(i.e. [Cc,,]o/[DMT]ovalue from 3.0 X
to 1.0 X lop3),the concentrations
of DEG and carboxylic acid end groups at long residence times increase by
about 80%.
TRANSIENT BEHAVIOR
An understanding of the transient behavior of the reactor is important in
designing the process control system for the precise control of polymer
722
l o p 4 to
1.0
Z
ccn
0.9
(r
!2
cn
U
>
>
0
0
t-
0.8
0
0
a
W
0.7
0.25
c
I0
0.75
t
0
0.20 -
s?
.-----a
- 0.67
U
U
'
U
G
0.15
- 0.58
3
tW
IU
0.10
0.50
3.0
2.5
2.0
1.5
8.5
4.0
c
Q
0
U
0.0
3.0
7.5
2.0
!2
I,"
c
TIME, H R
Fig. 4. Reactor transients to a step change in the feed [EG]/[DMT] mole ratio from 1.6 to 2.4.
723
0.90
0.85
---_
__-
- 0.9
(I)
>
>
z
0
0
+
-
0.8
0.75
o.ao[
0.70
0.7
0.25
0
Z
ga
0.75
9
c
(7
W
6c
0
4
0.67
a
U
c
%
c
I-
0.20
,-
0.58
I
m
I
c
0
X
0
t-
o
c
LT
U
+
>
2
fi
w
1.0
i
0
20
10
30
40
50
60
*n
d l
-2
0
c
0
Q
U
U
I-
2I-
7.0j
0
l
10
20
30
.
40
50
60
>
TIME. HR
Fig. 5.
0.001.
724
0.90
0
u3
a
w
>
Z
0
0
t-
aj-
'
0.85
0.9
0.75
0
0
(7
LU
0.70
9
18
27
36
45
0.25
0
k-
0
c
0
4
<
K
3
>
0.8
1
2
v)
0.80
0.225
U.
t
W
0.20
0
?8
27
36
10.50
45
0
X
0
t
0
4
a
U
20
1.0
I
0
10
20
30
40
8.0
2.5
P
X
0
X
7.0
2.0
L\
0
c
0
4
a
U
6.0
1.5
t
4:
0
t
5.0 I
0
'0
70
30
40
50
1.0
a
c
I
I
TIME, H R
Fig. 6. Reactor transients to a step change in the reactor residence time from 3 to 2 h.
725
- 1.0
0.90
0.05 -
cn
,--- - -1 0.9
U
W
>
. 0.8
Z
0
0
0.75 -
I-
- 0.7
0
U
W
>
Z
0
0
I0
Q
$
IW
z?
18
27
0.15
0
0
4.0 I
36
45
1 1.1
z
0
I0
Q
3 .O
(r
2.0
E
0
1.0
II
2.0
20
60
6.0
_----
&
40
1.5
I-
- 3.5
a
a
LL
1.0
F
2I-
c
X
G
0.5
0
20
40
60
2
I-
TIME, HR
Fig. 7. Reactor transients to a step change in the heating medium temperature from 180 to
200OC.
726
Figure 6 illustrates the effect of step decrease in reactor residence time from
3 to 2 h by increasing the reactant feedrate. As expected, the concentrations of
all the species except MHET decrease with an increase in the feedrate. Note
that a 50% reduction in the reactor residence time results in a 5% decrease in
DMT conversion and as much as a 25% decrease in DEG concentration.
The effect of step increase in the heating medium temperature from 180C
to 200C is shown in Figure 7. Note that an approximate 5%increase in both
DMT conversion and EG conversion is induced by the 20C increase in the
heating medium temperature. However, the DEG and carboxylic acid end
group concentrations in MTPA and HTPA increase by more than 150% over
the original steady-state value. The increased heating medium temperature
also gives rise to the rapid increase in the methanol flow rate, which then
decreases to a new steady-state value.
CONCLUDING REMARKS
In this work, we have presented the results of both steady-state and
transient behavior analysis for the metal acetate-catalyzed melt transesterification in a continuous stirred tank reactor. The kinetic scheme used in our
modeling includes the main transesterification reactions and side reactions to
form DEG and carboxylic acid end groups. The reactor variables studied in
this work (e.g., [EG]/[DMT] mole ratio, catalyst concentration, reactor temperature, and residence time) have been found to have a significant influence
on the overall efficiency of the transesterification process. In particular, the
DEG concentration in the reaction mixture was very sensitive to the variation
in the feed [EG]/[DMT] mole ratio. Excessive evaporation of methanol has
been found to occur when the feed catalyst concentration was increased. It
was also observed that the response of the DEG concentration to process
disturbances was much slower than that of other system variables.
This work was supported by the National Science Foundation (CBT-8504874) and in part by
the Systems Research Center at the University of Maryland.
APPENDIX
Notation
frequency factor for reaction 1, L2 mo12/h
frequency factor for reaction 2, L2 mo12/h
frequency factor for reactions 3 and 4, L2 mo12/h
heat transfer area, cm2
dimensionless adiabatic temperature rise for reaction i
dimensionless heat of vaporization for methanol
reactor catalyst concentration, mol/L
feed catalyst concentration, mol/L
heat capacity and reaction mixture, cal/g/K
molar heat capacity of methanol gas, cal/g mol/K
reactor DM T concentration, mol/L
feed DM T concentration, mol/L
Damkohler number
activation energy for reaction 1, cal/g mol
activation energy for reaction 2, cal/g mol
91
727
MHET concentration
BHET concentration
HTPA concentration
MTPA concentration
Greek Letters
dimensionless heat transfer capability
feed catalyst/DMT mole ratio
reaction mass density, g/L
feed density, g/L
exit flow density, g/L
dimensionless activation energy
ratio of activation energies E , and E3
ratio of activation energies E, and E3
residence time. h
References
1. K. Ravindranath and R. A. Mashelkar, J. Appl. Polym. Sci., 27, 471 (1982).
2. K. Ravindranath and R. A. Mashelkar, Polym. Eng. Sci., 22(10), 610 (1982).
3. K. Ravindranath and R. A. Mashelkar, J. Appl. Polym. Sci., 26, 3179 (1981).
4. K. Ravindranath and R. A. Mashelkar, J . Appl. Polym. Sci., 29, 437 (1984).
5. A. Kumar, V. K. Sukthankar, C. P. Vaz, and S. K. Gupta, Polym. E r g . Sci., 24(3), 185
(1984).
6. A. Khan, M.S. thesis, University of Maryland, 1987.
7. K. V. Datye and H. M. Raje, J. Appl. Polym. Sci., 30, 205 (1985).
8. R. H. Perry and D. W. Green, Chemical Engineers Handbook, McGraw-Hill, New York,
1984.
9. G. Challa, Makromol. Chem., 38, 105 (1960).
10. Du Pont Co. DMT Material Safety and Data Sheet, E-77976 (1985).
11. J. L. Kuster and J. M. Mize, Optimization Techniques with Fortran, McGraw-Hill, New
York, 1973.
12. K. Y. Choi, Polym. Eng. Sci., 27(22), 1703 (1987).