Pergamon Chemical Er~gineerin# Science, Vol. 50, No. 4, pp.

715 736, 1995

Copyright © 1995 Elsevier Science Ltd
Printed in Great Britain. All rights reserved
0009-2509/95 $9.50 + 0.00
0009-2509 (94) 00456-0

THE DYNAMIC BEHAVIOR OF CONTINUOUS SOLUTION

POLYMERIZATION REACTORS--VII. EXPERIMENTAL
STUDY OF A COPOLYMERIZATION REACTOR

J. C. PINTO* and W. H. RAY t

Department of Chemical Engineering, University of Wisconsin, 1415 Johnson Drive, Madison, WI 53706,
U.S.A.

(Received 2 November 1993; accepted in revised form 16 September 1994)

Abstract--The dynamic behavior of continuous free-radical solution copolymerization reactors was

studied, using VA (vinyl acetate) and MMA (methyl methacrylate) as comonomers. Due to the lack of data
in the literature, a mathematical model was built and some kinetic parameters were determined for different
operation conditions. This model was used to build a set of bifurcation diagrams, where oscillatory and
ignition-extinction phenomena could be predicted as a function of operation conditions. Additional
experiments validated the model predictions and showed that the dynamic behavior of these reactors may
be extremely sensitive to changes in the feed composition.

1. I N T R O D U C T I O N ization reactors. The closest results are those pre-

The first work to study the detailed dynamic structure sented by Keane (1972) where it is shown how impor-
of solution polymerization reactors was by Jaisin- tant the stability analysis of solution copolymeriz-
ghani and Ray (1977). Based on literature data, part of ation reactors may be for proper control and safety
the bifurcation structure of isothermal bulk M M A design. Even the theoretical results seem to be of very
and styrene homopolymerization was determined in limited scope, in spite of the industrial importance of
order to predict when multiple steady states and oscil- such systems. For instance, Shastry and Fan (1973)
latory behavior could be expected in certain ranges of and Balaraman et al. (1982, 1986) showed that
the parameter space. This first model was extended by copolymerization reactors could present multiple
Schmidt and Ray (1981) to allow different solvent steady states in certain ranges of operation condi-
concentrations in the feed stream and multiple tions, while Kuchanov et al. (1986) showed that self-
steady-state responses were then presented both the- sustained oscillations could occur in continuous bulk
oretically and experimentally for homopolymeriz- copolymerization processes. However, the only bifur-
ation. Hamer et al. (1981) extended the model to allow cation analysis was that of Hamer et al. (1981) and this
nonisothermal and copolymerization reactions. The depended on rough estimates of some important
monomers studied were VA and MMA and it was kinetic parameters not available in the literature.
shown that the dynamics of VA homopolymerization The main objective of the present work was to build
was much richer than that of M M A homopolymeriz- a detailed mathematical model to describe a con-
ation. It was also predicted that the presence of small tinuous solution copolymerization reactor and vali-
amounts of M M A could change the bifurcation struc- date the model with experimental results. The sys-
ture of VA homopolymerization dramatically. Later, tem chosen for experimentation was VA/MMA
Schmidt et al. (1984) predicted the existence of iso- copolymerization in tert-butanol (TB), due to its
lated branches of steady-state solutions and con- industrial importance. Homopolymer VA and
firmed the results experimentally. Most recently, copolymers with up to 2% M M A are made commer-
Teymour and Ray (1989, 1991, 1992a, b) improved the cially under similar conditions in solution CSTRs as
mathematical model and extended the bifurcation the first stage of polyvinyl alcohol production. Be-
diagrams for homopolymerization and sucessfully cause of the lack of data in the literature, some kinetic
confirmed many of the predictions experimentally. parameters had to be evaluated from some experi-
There are no reported experimental results on the mental runs. After evaluating the parameters neces-
dynamic behavior of continuous solution copolymer- sary for simulation, the model was used to predict
conditions where VA homopolymerization and
VA/MMA copolymerization would present qualitat-
ively different dynamic characteristics. Additional ex-
periments were then carried out to confirm the model
*Present address: Programa de Engenharia Quimica/
COPPE Universidade Federal do Rio de Janeiro, Cidade predictions and show that the stability of such systems
Universit~iria-CP: 68502, Rio de Janeiro-RJ 21945 - Brazil. may be extremely sensitive to small changes in the
tAuthor to whom correspondence should be addressed. feed composition.
715
716 J. C. PINTO and W. H. RAY
2. THE EXPERIMENTAL REACTOR SYSTEM Initiation:
The basic experimental system and process opera-
kd
tion have been described by Teymour and Ray (1989, I ,2R
1992a) so that we emphasize only important points
and constraints related to the experimental proced-
ure. Figure 1 shows a diagram of the copolymeriz- R+M1 kr'' Plo
ation process and general equipment. It is important
to note that air was used as coolant medium, that the
R + M 2 k'2,Qo 1.
entire experimental setup was kept under inert (nitro-
gen) atmosphere to avoid oxygen inhibition, that the
Propagation:
feed conditions were properly controlled and that
reactor temperature and conversion were measured Pij "1-Mx kPll~Pi+l,j
on-line. In the feed section two different containers
were used to store a mixture of solvent (TB) and
initiator (2,2'-azobis-2-methyl-propionitrile--AIBN) Qij + M1 kp2~jPi+l,j
and the comonomer mixture (VA/MMA). The in-
itiator was dissolved in the solvent alone to avoid
polymerization in the storage tanks. The monomers
Pij + M2 kpl2)Qi,j+l
were premixed because of the low M M A feed concen-
trations used in the experiments (below 2% in volume, Qij + M2 kp22 Qid+ l"
as shown later), which made it difficult to control the
feed composition otherwise.
Termination:
In order to monitor conversion, an on-line process
refractometer was used (model SSR-72, from Electron kti 1
Pij q- P,,,. ~Ai+n,j+m
Machine Corporation). However, these dynamic con-
version data were used for qualitative model analysis
k~12 A
only. Whenever precise conversion data were needed, Pij + Qnrn ~"H+n,j+m
as in the analysis of steady-state conditions, where the
reaction and heat transfer rates are uncoupled (from
a parameter evaluation point of view), samples were Qo + Qnm kt22 Ai+n.j+m.
taken from the output line and gravimetrical analysis
was performed. The technique consisted of taking
It was assumed that transfer reactions could be
samples of 50 ml, cooling the samples down to 15°C
neglected because our main interest was the dynamic
with cold water, weighting the samples in precision
behavior of global variables, such as reactor temper-
scale and drying them to constant weight under low
ature and conversion. It is known (Ray, 1972) that
temperature and vacuum conditions.
these additional reactions are either uncoupled or
As in the earlier homopolymerization experiments,
very weakly coupled with the dynamics of these global
the experimental setup was designed for operation at
variables.
atmospheric pressure, so that the reactor temperature
In order to describe the cross-propagation and
was not allowed to rise above the boiling temperature
cross-termination kinetic constants, new variables are
of the reaction mixture. This is an important con-
normally defined. The first ones are the reactivity
straint to be included into the mathematical model.
ratios, that measure how much slower the cross-
Regarding the sources of the chemicals used, VA
propagation is when compared to the the homo-
was bought from Pfaltz and Bauer, Inc., M M A
propagation. The reactivity ratios are defined as
and TB (99.5% + pure) from Aldrich Chemical Co.,
AIBN (99.0% + pure) from Eastman Kodak Co. and kpii
rij = - - . (1)
nitrogen (99.996% + pure) from Liquid Carbonic. kpij
VA and M M A were distilled under vacuum and inert
atmosphere one day before the experiment was car- The other additional variable is the cross-termina-
ried out and were stored in a refrigerator at 10-15°C. tion constant, which measures how much faster the
The other chemicals were used as received. cross-termination is when compared to an ideal
cross-termination value. It is defined as
kto
3. MATHEMATICAL M O D E L I N G ~b0 = ~ (2)
The mathematical model developed was based on
the classical kinetic mechanism for free-radical solu- Although this is a quite standard kinetic mecha-
tion copolymerization reactions, where the propaga- nism for free-radical copolymerization, it has been
tion and termination steps are extended to allow criticized lately by different authors based on different
reactions between different types of monomer and arguments. The first type of argument is based on the
living macro-radicals. The reaction mechanism is: fact that qJ depends strongly on the feed c o m p o s i t i o n
 MonomerLine
SolventLine
F

w
z rar~
m
t~
'If
I
I

i ~.
>
~n

Thermoeouples L ...... JJ

r . . . . . . . . .
I

Thermocouples
I

ii
j~
t
f

~ n
~Q

i,

LIL 'iiA--SaOlOeaa uot.lez!aatU£lod uo.llnlos snonu!luoo jo aO!neqoq O.ttUeU,((I

718 J. C. PINTOand W. H. RAY
in many common copolymerization systems, includ- monomer 1 mass balance:
ing the system VA/MMA (Brandrup and Immergut,
1989). According to Atherton and North (1962) and dv,_p,fvxfdz
Pl v~(q°+oRatel)kq,
more recently Prochfizka and Kratochvil (1983) the
termination step is not chemically controlled, but dT d(1/p~)
diffusion controlled. Thus ~b should be a function of + Pl vl dz dT (5)
feed composition, so that for the termination step kt
should have the form monomer 2 mass balance:
k, = ~ xik,i~ (3)
i dr2
dz P2fV2f
P2 V2(q°-~-
\qi O Rate2)
where x~ is the mole fraction of species i in the polymer
chain. dT d(1/p2)
+ PEr2 (6)
Having a similar point of view but following a dif- dr dT
ferent approach, Hamielec and coworkers (Jones
Initiator mass balance:
et al., 1986; Bhattacharya and Hamielec, 1986; Yaras-
kavitch et al., 1987) emphasized the importance of the
gel effect to describe copolymerization reactions. dr = ciy - cl + kaO . (7)
Briefly, the gel effect is the decrease of the termination
rate observed when the reaction medium contains energy balance:
high polymer concentrations. This leads to an auto- dT Pz(Tz- T) 0
catalytic reaction rate due to increasing concentra- (1 + e) dz t- - - (AHll v l p a k p l l P
P P%
tions of free-radical species. Eventually, at very high
polymer concentrations, the rates of all different reac- + AHIE vzp2kpl2P + AH21 vlplkp21Q
tions involving macro-radicals also decrease. The gel
effect is caused by the diffusion limitations that can be + AH22v2P2kp22Q ) - ~ ( U A ( T - Tc)
expected in a network of highly entangled polymer PCv
chains. + c t U A ( T - T~) + Qlat) (8)
Fukuda et al. 0985) provided another point of view
where qo/qi is the ratio between the output and input
and suggested that eq. (2) is correct, but that the
volumetric flow rates, given by
propagation step is incorrect. According to the
authors, polymer chain composition and sequence qo = p~fv, f + p l f V l f +. p2fv2f
length distribution obtained experimentally exhibit qi Ps Pl P2
significant dcMations from the theoretical results ob-
tained with the classical kinetic model. However, the + O(Ratel vl(pt/pp - 1) + Rate2 v2(P2/Pp - 1))
authors showed that experimental data for dTf d(1/ps) d(1/p, )
MMA/styrene copolymerization could be fitted ex- +~p.v.~+p,v, d---T--
tremely well by a model incorporating a penultimate
effect, even when qJ was considered to be constant and d(l/p2) d(llpp)'~
+ p2v2--~-- + P.V.---d~; • (9)
equal to one.
It is clear that agreement has yet to be reached on
how to describe copolymerization reactions properly. The normalized rates of monomer consumption are
However, it is almost impossible to develop given by
a modeling study based on any of the alternative Ratel = k p l l P + kp21Q + Rckr (10)
kinetic schemes described in the last paragraphs with-
out carrying out detailed kinetic studies, as additional Rate2 = kp12P + kp22Q + Rckr (11)
kinetic constants and semi-empirical parameters not where R~k, is the normalized rate of initiation, given
found in the literature are necessary to describe the by
system. Thus in our analysis here, we choose to use
the standard kinetic mechanism and take most of the 2kdCi
Rckr -- (12)
parameters necessary to describe the system from the M~/fl + M2/f2
literature and to determine the remaining ones experi- where the denominator accounts for changes in the
mentally.
initiation efficiency due to changes in the comonomer
Assuming that the reactive mixture occupies a con-
composition.
stant volume (imposed experimentally), that volume The concentration of species P and Q can be written
additivity holds and that the quasi-steady-state ap-
as
proximation is valid for radicals R, P and Q, the
following equations can be derived: Rckr(Ml -t- M2)
solvent mass balance: Q = ~ / k t l l k 2 + 2kt12k~ + kt22 (13)
dv~ = psfVsf Vs qo + psv~ d T d(1/p,) (4)
dz p~ qi dz dT P = ksQ (14)
 Dynamic behavior of continuous solution polymerization reactors--V[l. 719

where internal walls of the reactor, before reaching the con-

denser. The theoretical solution for my is not available
ks = (kp2zM1)/(kpl2M2). (15)
either without additional specification of the heat
The termination and propagation constants in eqs transfer areas and of the nitrogen flow through the
(8)-(15) include the gel effect and so have the general condenser. The contribution of latent heat to the
form energy balance is not important except close to
boiling conditions. Thus the effects were represented
kas = kajgaj
o
(16) by
0
kpi j = kpijgpi j
Ql,t = 0, T<Tb
where k°s and k°o are the kinetic constants at zero (24)
dT dTb
polymer concentration and #tij and gpo are the gel - T>~Tb
effect correlations that take into account the effects of dr dr '
increasing polymer concentration. where Tb is the boiling temperature of the reacting
Some global variables in the energy balance are solution and may be easily computed with a proper
computed as thermodynamic correlation. In order to avoid discon-
tinuities in the model, which might not allow the use
p = Vsp s + r i p t + v2p 2 + Vppp (17)
of continuation algorithms, eq. (24) was rewritten as
pCp = VsPsCps + VlPlCpl + V2P2Cp2
0, T < Tb
"k VpppCpp (18) (25)
Q,,t= 2 ( T - Tb) 2, T>~ rb
where where 2 is a numerical parameter that controls the
pp = Wppl + (1 - w)pp2 (19) strength of the boiling constraint. This same numer-
ical approach was used successfully by Teymour
Cpr = W%pl + (1 -- w)Cpp 2 (20) (1989).
and w is the polymer composition (mass fraction of In order to calculate the boiling temperature, the
structures of type 1). In the energy balance, e is an classical Flory-Huggins equation for solvent activity
additional term which accounts for extraneous heat in polymer solutions was used as
capacity due to the reactor walls and additional
equipment. The dynamic copolymer composition ln(~)=ln(vi)+Vp+ Ziv~ (26)
may be determined from
with the additional assumption that the mixture of
dw vl Pl
d--~-= (1 -- w)Ratel 0 solvent and monomers behaved ideally.
Vppp
4. PARAMETER EVALUATION
-- W Rate2 v_2p2®. (21)
Vppp 4.1. Parameters related to the heat transfer
In eqs (8) and (25) four different parameters related
However, for most of our model calculations, to the heat transfer mechanism are defined. Three of
simpler expressions were used because polymer com- these (e, UA, ctUA) could be evaluated from indepen-
position does not vary much in continuous reactors dent heat transfer experiments, while the fourth one
and these variations have a negligible effect on the (2) had to be fitted based on numerical experiments.
process dynamics. Thus copolymer composition was 2 should neither be too large, in order to allow the use
calculated by an approximate mass balance, based on of continuation algorithms for the computation of
the monomer compositions in the inlet and outlet steady states, nor too small, in order to simulate the
streams these computations are exact at steady-state boiling constraint properly.
conditions). However, the total conversion (whether Figure 2 shows a typical temperature profile ob-
including or not the solvent) can be calculated exactly tained from heat transfer experiments. A total of eight
as different heat transfer experiments were carried out
for polymer solutions of different composition. In
X = Vppp (22) these experiments initiator was not added to the sys-
Vsps + Vlpl "t- v2p 2 q- Vppp
tem and the monomers were inhibited with 5 ppm of
and, as shown later, this is the important variable for hydroquinone to avoid spontaneous thermal polym-
the gel effect correlation. erization. The overall results obtained after parameter
Regarding the latent heat of vaporization in eq. (8), estimation are shown in Table 1. It is important to
one could write emphasize that these numbers did not depend on the
solution composition as the heat transfer to the cool-
Qlat "-" my AHv (23)
ant air jacket was the controlling step in the heat
where mr is the rate of condensation and AHv is the transfer mechanism. Additionally, no significant vari-
latent heat of condensation of the vapor phase. How- ation was observed in e, although UA and ctUA
ever, it is extremely difficult to evaluate m~ experi- seemed to spread within a range of +_ 10%, probably
mentally because part of the vapor condenses on the due to unavoidable changes in the arrangement of the
720 J.C. PINTOand W. H. RAY
360.0 --
Calculated Reactor Temperature
.... Experimental Reactor Temperature
350.0_ ~ 4 t ~ .
Experimental Air Temperature
f
/
/

,//
N 340.0- i
i
/
/ s"''''-
pt
330.0 -- /
/
/
i
/
320.0 --
\
\
310.0 -- \

300.0

0.0 2000.0 4000.0 6000.0 8000.0

Time ( Seconds )

Fig. 2. Results from a heat transfer experiment (results obtained for a mixture of 60% VA and 40% TB in
volume, with 5ppm of hydroquinone, e=0.55, UA=0.858J/kgm2K, ctUA=0.0746J/kgm2 K,
V= 500ml, Ta = 298K).

Table 1. Parameters describing heat trans- 4.3. Parameters for M M A homopolymerization

fer phenomena Although a large collection of kinetic data is avail-
able for MMA homopolymerization (Brandrup and
Parameter Value
Immergut, 1975, 1989), two problems had to be faced
UA 0.836J/K s in order to define the set of kinetic equations a priori.
ctUA 0.07315 J/K s The first one was related to the kinetic constants.
e 0.55 Basically, two sets of parameters have been used re-
V 500ml
cently in the literature. The first set (called SET I from
now on) seems to have originated from kinetic data
presented by different authors and listed in Brandrup
experimental setup that occurred from experiment to and Immergut (1989) for MMA polymerization in-
experiment. itiated by AIBN. This set of parameters was used by
Hamer et al. (1981) and more recerltly by Choi and
Butala (1991). The second set of parameters (called
4.2. Parameters for VA homopolymerization
SET 2 from now on) seems to have originated by
Table 2 presents all parameters necessary to de-
kinetic data presented by Mahabadi and O'Driscoll
scribe the VA homopolymerization. As it can be seen,
(1977) and was used by Schmidt and Ray (1981) and
most of the data are the same as used by Teymour
more recently by Maschio and Moutier (1989). Both
(1989). One important change, however, regards the
sets of parameters (presented in Table 3) have been
definition of the individual kinetic constants kp and k ,
successful in representing experimental data, although
necessary for copolymerization studies. The gel effect
it may be hard to understand how the reactivity ratios
correlation was also taken from Teymour (1989), and
for the system VA/MMA can be constant, as usually
is of the form
reported in the literature, if SET 2 is correct, since the
Opll = 1 activation energy of the propagation constant for
(27) MMA is very different from the activation energy of
0,xl = exp( - 0.4407x - 6.7530x 2 - 0.3495x 3)
the propagation constant for VA.
where x is the total conversion (including solvent) as A second important issue is the gel effect correla-
defined in eq. (22). Parameter fitting was not necessary tion. The gel effect is extremely important for the
for VA homopolymerization. description of MMA homopolymerization (Hamer
 Dynamic behavior of continuous solution polymerization reactors VII. 721

Table 2. Parameters describing VA homopolymerization

Parameter Reference

74.12
Teymour (1989)
Ps 60.21 + 0.116 T g/ml
Pl = 0.9584-1.3276 x 10 -3 ( T - 273.15)g/ml Teymour (1989)
pp = 1.211 - 8.496 x 10-" ( T - 273.15)g/ml Teymour (1989)
cp~ = 0.716cal/g K Teymour (1989)
cpl = 0.4479 + 5.625 x 10 -4 ( T - 273.15)cal/g K Kroschwitz (1986)
cpp= 0.3453 + 9.55 x 104 ( T - 298.15)cal/gK Teymour (1989)
2658.29 "~
Il s**' = exp 16.8548 - - - 95.5) m m a g
T Reid et al. (1986)

II~' = exp(16.1003 2744.68 "~mmH Reid et al. (1986)

r-:3~.3~: g
AH1 x = 87.78 kJ/mole Kroschwitz (1986)
kpxl = 3.2 x 107exp( - 6300/RT)l/mols Kroschwitz (1986)
ktl x = 3.70 x 109 exp ( - 3200/RT ) l/mol s Kroschwitz (1986)
ka = 1.58 x 101s exp ( - 30800/R T ) I/S AIBN
A =0.8 Mean value from various references in Brandrup
and Immergut (1975)
011 = empirical correlation Eq. (27)
gs = 0.5 For lack of data, as suggested by Allcock
gl = 0.5 and Lampe (1981)

Table 3. Kinetic constants for MMA homopolymerization 1, v: > v:p~

(29)
g~,2z = 7.1 x lO-Sexp(171.53 v:), v: <~ v:p~
kp22 = 7.0 x l0 s exp ( - 6300/R T)l/tool s
Set 1 ktz2 = 1.76 x 109 exp ( - 2800/R T) I/mol s where
A = 0.6
v:,c = 0.1856 - 2.965 x 10 -4 ( T - - 273.15) (30)
Set 2 kp2 2 = 4.92 x l0 s exp ( - 4353/R T )l/mol s
k,2 2 = 9.80 x l0 Texp ( - 701/RT)l/mol s v:~ = 0.05 (31)
A =0.5
a n d the total free volume vy, is given by

/): = 1)fl I; 1 "~- l)f2 V 2 -~ UfsV s "~- I)fpVp (32)

et al., 1981) a n d n o kinetic d a t a are available for where
M M A polymerization in TB. This lack of d a t a is
easily u n d e r s t o o d if one realizes t h a t P M M A is not v:, = 0.025 + a , ( T - T,,) (33)
soluble in T B even at low polymer c o n c e n t r a t i o n a n d a n d ~q and Tgi are presented in Table 4 for some
m o d e r a t e t e m p e r a t u r e s ( a r o u n d 50°C), a l t h o u g h the chemical species. It was assumed that eqs (28)-(33)
presence of M M A enhances the polymer solubility. would be valid for polymerization in T B a n d t h a t arB
As T B is extremely beneficial for VA polymerization would be equal to the universal value 0.001. TgxB was
(Teymour, 1989), it was decided to adjust the gel effect then treated as a n adjustable p a r a m e t e r t h a t could be
correlation for M M A h o m o p o l y m e r i z a t i o n in TB, equal to either 181 K (no difference from kinetics in
using the well k n o w n Ross a n d Laurence e q u a t i o n ethyl acetate) or 2 1 3 K [as predicted from Soh a n d
(Ross a n d Laurence, 1976) as a basis. Thus, the gel Sundberg (1982)].
effect correlation for M M A can be written as:

= ~'0.10575exp(17.15Vy -- 0.01715 ( T - 273.15)), vf > v:,c (28)

0r22 (2.3 x 10-6 exp (75v:), v: <~v:tc

Table 4. Parameters for MMA gel effect correlations

Chemical species

Parameter VA MMA PMMA EA

~t 0.001 0.001 0.00048 0.001

Tg (K) 173 167 387 181
722 J. C. PINTO and W. H. RAY
Figures 3-5 show experimental and simulation re- of parameters necessary to describe the system is
sults for M M A homopolymerization in TB at steady- presented in Table 5.
state conditions when the SET 1 of kinetic constants
and TgTa equal to 181 K were used for simulation. 4.4. Parameters for VA/ M M A copolymerization
These parameters were much better at predicting In our study here, we will be operating with low
a wide large range of experiments for MMA (Pinto, M M A content in the polymer and this influences our
1991) than other parameter choices. Thus these were choice of model parameters. The additional para-
used in all model calculations. The complete set meters needed to describe the copolymerization reac-

350.0

= 340.0-

E
g
330.0 -

320.0 -

310.0
I I t I I I I I
0.0 1000.0 2000.0 3000.0 4000.0 5000.0 6000.0 7000.0 8000.0 9000.0

Residence Times ( Seconds )

1.00
e., Air Temperature = 323 K
o
Experimental Results:
>
C
0.80 /k MR4
o
O MR5
• MR6
0.60 • MR7

0.40

o /
f
0.20
/ A

0.00
I I I I I I I I
0.0 I000.0 2000.0 3000.0 4000.0 5000.0 6000.0 7000.0 8000.0 9000.0

Residence Times ( Seconds )

Fig. 3. Steady-state c o n d i t i o n s for M M A h o m o p o l y m e r i z a t i o n at Tc = 50°C (v2: = 0.40, v~: = 0 . 6 0 ,

c i y = 0.04 gmol/l, Ta = 298 K, kinetic p a r a m e t e r s from SET 1 a n d Tgs = 181 K).
 Dynamic behavior of continuous solution polymerization reactors VII. 723

350.0

340.0
~D
e-,

330.0 6
t l / - ' - "

320.0

310.0
I I 1 I I I I I
0.0 1000.0 2000.0 3000.0 4000.0 5000.0 6000.0 7000.0 8000.0 9000.0
Residence Times ( Seconds )

1.00
Air Temperature = 333 K

e- Experimental Results:
.o o.so _
A MR4
>
c-
O
O MR5
~J
• MR6
0.60 -
• MR7

0.40 _

0.20 --

0.00
I I 1 I I I I I
0.0 I000.0 2O00.0 30OO.0 4OOO.0 S000.0 ~00.0 70O0.0 80O0.0 9000.0
Residence T i m e s ( Seconds )

Fig. 4. Steady-state conditions for MMA homopolymerization at Tc = 60°C (v2f = 0.40, v~s = 0.60,
Cis = 0.04 gmol/l, To = 298 K, kinetic parameters from SET I and Tg, = 181 K).

tions are shown in Table 6, except for the ones related tinuous polymerization. In the second case, ~ is ex-
to the cross-termination step. pected to assume different values at the low M M A
Table 7 shows typical values of ~ reported in the concentrations used in this study. For these reasons,
literature. It seems that ~b has been always reported experiments were designed to allow the evaluation of
for batch reactors and high amounts of MMA. In the ~, from dynamic and steady-state experiments for dif-
first case, due to the composition drift, ~ is obtained ferent M M A feed concentrations.
as an average for the batch and not pointwise as Before evaluating ~, the proper gel effect correla-
a function of M M A composition as needed for con- tion had to be defined. Two different assumptions
724 J. C. PINTO and W. H. ,RAY
350.0

,,/

340.0

E
[-,

330.0

320.0

3 lO,O

0.0 1000,0 2000.0 3oo0.0 4000.0 sooo.o ~00.0 7000.0 8000.0 9000.0
Residence Times ( Seconds )

1.00

Experimental Results: Air Temperature = 343 K

A MR4
'~ 0.80 © MR5
• MR6

0.60

0.40

0.20

0.00
I I, I I I I I
0.0 I~.0 20~.0 3~.0 400o.0 s~.o ~,~0.0 7~.0 8m~.o 9~x~.o

Residence Times ( Seconds )

Fig. 5. Steady-state conditions for MMA homopolymerization at Tc = 70°C (v2/= 0.40, V,y = 0.60,
cly = 0.04 gmol/l, T, ffi 298 K, kinetic parameters from SET I and T~ -- 181 K).

seemed to be reasonable. The first one would extend produced is mostly composed of VA molecules. So,
the definition of k, z2 to the gel effect correlation, so &ts should describe the diffusion limitations properly
that gtz2 would be defined as: for all k~l~.
Table 8 and Figs 6-9 show the results obtained
gz12 ----~ t 2 2 " (34) when the gel effect was computed as in eq. (34). The
theoretical profiles were obtained assuming that the
The second assumption was based on the fact that heat transfer coetficients were subject to small vari-
when the MMA feed concentration is low the polymer ations within the range _ 10%.
 Dynamic behavior of continuous solution polymerization reactors--VII. 725

Table 5. Parameters describing MMA homopolymerization

Parameter Reference

74.12
Ps Teymour (1989)
60.21 + 0.116 T ~/ml-'
P2 = 0.9654 -- 0.00109 (T - 273.15) - 9.7 x 10 -7 (T - 273.15)2 g/ml Schmidt et al (1984)
p2
Schmidt et al. (1984)
PP = 0.754 - 9.0 lO-* (T - 343.15) g/ml
c~, = 0.716 cal/g K Teymour (1989)
%2 = 0.490 cal/g K Schmidt et al (1984)
cp~,= 0.339 + 9.55 × 10-4(T - 298.15)cal/gK Schmidt et al. (1984)
2658.29 \
l-I~= = exp 16.8548 ~--~-.5)mmHg Reid et al (1986)

5441.04 \
l-I~=t = exp (19.8567 7z+ 3-~.32)m m H g From experimental data in Brandrup and Im-
mergut (1975)
AH22 = 57.7 K J/mol Kroschwitz (1986)
kp22 = 7.0 x 106 exp ( - 6300/R T ) I/tool s Brandrup and Immergut (1975)
kt22 = 1.76 x 109 exp ( - 2 8 0 0 / R T ) l/mol s Brandrup and Immergut (1975)
kd = 1.58 x 101Sexp( -- 30800/RT)I/s AIBN
f2 = 0.6 .... Mean value from various references in Bran-
drup and Immergut (1975)
gz2 = free-volume theory Hamer et al. (1981)
Z~ = 0.5 For lack of data, as suggested by Allcock
X2 = 0 . 5 and Lampe (1981)

Table 6. Additional parameters describing VA/MMA copolymerization

Parameter Reference

AHt2 = AH21 = 72.0 kJ/moi Geometric mean of AHt t and AH22

r12 -- 0.015 Kroschwitz (1986)
r21 = 2 0 Kroschwitz (1986)
~p Table 8
gtt2 Geometric mean of g,t t and gt22

Table 7. W as rePorted by different authors

V Source Remarks

120 Choi and Butala (1991) Batch reaction with high MMA concentration (0.10-0.50mole fraction).
49°C < T < 65°C
150 Kuo et al. (1976a) Batch reaction with 30% MMA (molar basis). T = 60°C
170 Kuo et al. (1976b) Batch reaction with high MMA concentration (0.30-0.50 mole fraction).
T = 60°C
400 Chen et al. (1981) Batch reaction with 30% MMA (molar basis). T = 60°C
400 Burnett and Gersmann (1958) Batch reaction with high MMA concentration (0.10--0.30 mole fraction).
400C < T < 60°C

Table 8. q' Evaluated in this work It can be seen from Table 8 that ~, increases steadily
when M M A feed concentration increases and that it
% MMA % Q* % MMA* may eventually reach the range of values shown in
monomer feed solution polymer Table 7 for higher M M A feed compositions. N o t e that
volume basis molar basis molar basis ~F
tends to the ideal value of 1 as the M M A feed
0.00 0.00 0.00 1.00 t composition tends to zero.
0.25 17.0 0.26 1.45 Regarding the gel effect, using the VA value, g,t ~, as
0.375 22.0 0.40 1.95 a global correlation for all compositions always led to
0.50 32.0 0.54 5.25 p o o r results. Thus, it was decided to use eq. (34) in all
2.00 73.0 2.50 3O.O
our model computations.
*From simulation. Although ~, must be a function of the internal
tExtrapolated from data. reactor composition, no a t t e m p t was m a d e to model
726 J. C. PINTOand W. H. RAY
74 I I

......... Experiment
Simulation

ft
72
v

70

68

66

i
s
64

624 ! i

Time (min)

Fig. 6. Dynamic temperature profile following a perturbation in Tc ( h : = 0.3990, v2: = 0.0010,

v,: = 0.6000, c~: = 0.04 gmol/l, To = 298 K, ® = 90 min, UA = 1.00(UA)o, ~ = 1.45, T~changed from 53 to
48°C at steady-state conditions).

76- -x I

.-, 74-
I
Experiment
v
Simulation

72-
E
7O-

68-

66-
03
Time (min)

Fig. 7. Dynamic temperature profile following a perturbation in Tc (vtf=0.3985, v2:=0.0015,

vs: = 0.6000, c~z = 0.04 gmol/l, To = 298 K, ® = 90 min, UA = 0.95(UA)o, ~b = 1.95, Tc changed from 55 to
48°C in a series of small step changes).

~k. In our calculations here, q/was taken as the func- operation conditions. In order to show how sensitive
tion of M M A comonomer feed composition, shown in these structures are to changes in the heat transfer
Table 8. mechanism, Figs 11-14 show how the bifurcation
diagrams change when the coolant air temperature
changes by _ 2°C or the heat transfer coefficients
5. BIFURCATION ANALYSIS change by + 10%.
Using the mathematical model and parameters de- Figures 10-14 also indicate that V A / M M A
scribed in the previous sections, a series of bifurcation copolymerization dynamics may be extremely sensi-
diagrams was built with the capabilities of the con- tive to small changes in the feed composition. Parti-
tinuation routines of A U T O (Doedel, 1986). More cularly, it may be seen that:
details of this analysis will be presented in a com-
panion paper (Pinto and Ray, 1995). The bifurcation (1) At the experimental conditions analyzed, VA
diagrams are shown in Fig. I0 for different homopolymerization is always stable in the whole
comonomer feed compositions at certain standard range of residence times.
 Dynamic behavior of continuous solution polymerization reactors VII. 727

72- -~

-" .j

¢D
E-
< /

66-
Experiment
Simulation
64-
3
; & ,&
Time (min)

Fig. 8. Dynamic temperature profile following a perturbation in Tc (vl: = 0.3980, .v2j. = 0.0020,
vs: = 0.6000, c~: = 0.04 gmol/l, Ta = 298 K, O = 120 min, UA = 1.00(UA)o, ~b = 5.20, T~ changed from 57
to 52°C at steady-state conditions).

•00o

g 350.

340.
/
E
/
320.
/
310,
,o
1(~30. 3(303. 5(330. 7O0O. 9000.
Residence Time (s)

1.(3(
0.9(
I I I I
0.8C
0.7f
"--1 • Experiment
Simulation
I

/
.o 0.60
/
0.50 /
0 0.4(,
0.3(" .,w"
0.2C
O.IC
]
O.DC __.~J
l(D0. 3(XD. 5003. 7030. 9000.
R e s i d e n c e T i m e (s)

Fig. 9. Steady-state conditions for VA/MMA copolymerization at Tc = 90°C (vl: = 0.3920, v2: = 0.0080,
vs: = 0.6000, ci: = 0.04 gmol/1, To = 298 K, UA = 1.00(UA)o, ~, = 30).
728 J.C. PINTO and W. H. RAY
360

v
•e o e o o e e t - ....
.o ~, , " "%-- ". . ..
35~
~A, -. . ~ ". .
e~
' '," " "•. " ~ - . "11 .
340

n.,.~
I LLI[;
V ^~ -' , ~ ", • " ,~~ t ~~ /
11 • o e e e e o Oc- ~ ...... --~ ....
330
",,~ / ~ •
0.375% MMA • • • ~, • • • e

320 _

0.50% MMA
0.25% MMA
310
I I I I I I I I
0 2000 4000 ~000 80~
100(~ 3000 ~000 70~0 9000

Residence Time (s)

Fig. 10. Bifurcation diagrams for different MMA comonomer feed concentrations (v,s = 0.6000,
c~y = 0.04 gmol/l, T, = 298 K, T~ = 48°C, UA = 1.00 (UA)o).

360

00000000000%

E
m
350_
,, \ , , ". -.. ~ . •

340 _

330
Pure VA) ~, ', _ 2 ,,o • o o o o o o , o o o v

320
j//.) 0.50% MMA
r
0.25% MMA
310
I I I I I I I I
0 2000 4000 6000 0000
1000 3000 5000 7000 9000

Residence Time (s)

Fig. 11. Bifurcation diagrams for different MMA comonomer feed concentrations (v,s = 0.6000,
c~y = 0.04 gmol/1, Ta = 298 K, Tc = 50°C, UA = 1.00 (UA)o).

(2) V A / M M A copolymerization with comonomer (4) V A / M M A copolymerization with comonomer

feed composition of 0.25% in volume of M M A must feed composition of 0.50% in volume of M M A shows
always be stable, unless the heat transfer to the sur- that the upper steady state for VA homopolymeriz-
roundings is larger than at the standard conditions. In ation disappears, so that reaction extinction may be
this case, oscillations may develop around residence expected; also at residence times around 120rain os-
times of 90 min. cillatory behaviour may be expected. However, if heat
(3) V A / M A copolymerization with c o m o n o m e r transfer to the surroundings is higher than at standard
feed composition of 0.375% in volume of M M A must conditions, the self-sustained oscillations may disap-
lead to unstable operation at residence times around pear.
90 min. If the heat transfer to the surroundings is
lower than expected, however, the oscillations may Figures 15-19 show experimental results that con-
disappear. firm the theoretical predictions. Figure 15 shows
 Dynamic behavior of continuous solution polymerization reactors--VII. 729

360
v

= 350- ,,;-,,,,.
\?,~,.._ ,, i;' ~"'.'~.
. ,; • •
--_, ,, •

E
~- 3 4 0 - ",,,, ",,,,/-,,,
Y, -.--
a-'~"--m "-~-~ "- ".._
,, -. "... -,, -._ - - ~
"~ "',% " ,., ,,.,. OOo~i 00 ID O u oO lad •
330 -
Pure V A ) '~ "', ~ IDO00 'o o u o o o o a

320.

310,
I I I I I I I I
O. 2000. 4000. 6000. 8000.
i000. 3000. 5000. 7000. 9000.
Residence T i m e ( s )

Fig. 12. Bifurcation diagrams for different MMA comonomer feed concentrations (vsi=0.60~,
cu- = 0.04 gmol/1, T, = 298 K, Tc = 46°C, UA = 1.00 (UA)o).

360.

Ouu~OOOOO099

"~ 350.
k "'.3 " ~ OOo,,.
34o_
"; ",..'-,. I
330 -
PureVAi \, "", ~ "°'°''" ,,~o"~ --

/ ] 1"~0.375%MMA / "o
0.50% M M A
320.

310.
I I I I I I I I
O. 2000. 4000. 6000. 8000.
i000. 3000. 5000. 7000. 9000.
Residence T i m e (s)

Fig. 13. Bifurcation diagrams for different MMA comonomer feed concentrations (hi =0.6000,
c u . = 0.04 gmol/l, Ta = 298 K, T,, = 48°C, UA = 1.10 (UA)o).

a typical dynamic pattern that developed when the Figures 16 and 17 show the development of self-
comonomer feed composition was changed from pure sustained oscillations when the comonomer feed com-
VA to 99.75% VA and 0.25% M M A in volume, at position was initially equal to 99.625% VA and
steady-state conditions. The oscillations are damped 0.375% M M A in volume and that fast convergence to
and eventually lead to a new steady state or to a very stable operation occurs when the feed composition
shallow limit cycle. was changed to pure VA.
CES 50:4-L
730 J. C. PINTO and W. H. RAy

360
• • OIIIIIIIIIIlellJllll
P ,r--- -,.,_%.
3 350-
%)
e~
r-
D
~" 3 4 0 _

330 --
Pure V A , ', ~ ",, _ _ •,
/ ~ \ 0.375% M M A ; • • o o ~ e o o l e o • - ~ ,

320-
~ 0 50% M M A ~

310,
1 I I I I I I I
O. 2000. 4000. 6000. 8000.
i000. 3000. 5000. 7000. 9000.
Residence T i m e (s)

Fig. 14. Bifurcation diagrams for different MMA comonomer feed concentrations (v~s = 0.6000,
c~I = 0.04 gmol/l, Ta = 298 K, Tc = 48°C, UA = 0.90 (UA)o).

Changing Feed Composition

83.0-:i
L) : I Reactor Temperaturel
73.0... r ~-- - ~ ~ ,,~ ~ ~1
p
_=
60.0- ~i
50.0- ' Air Coolant Temperature I

43.0-
2):) 39 ( 30 E )0 1{}[30

ICO .(3-

llll==lhll.=~
=
O
83.0-
~
O
c-

%)
43.0-
23.0-
It
0.0-
)3 Do 8 1( 73
Time (min)

Fig. 15. Dynamic temperature profile following a step change in the feed composition (vsl = 0.6000,
c~/= 0.04 gmol/l, To = 298 K, Tc = 48°C, ® = 90 min, v2I changed from 0.0000 to 0.0010 after steady state).

Figure 18 shows a dying reaction. The extinction 99.50% VA and 0.50% M M A in volume. The oscilla-
occurred when the comonomer feed composition was tions disappeared very fast when the feed composition
changed from pure VA to 99.50% VA and 0.50% was changed to pure VA.
M M A in volume at the steady state obtained at resi- Figures 20 and 21 show that the theoretical predic-
dence time of 90 min. tions are not only good from a qualitative point of
Finally, Fig. 19 shows the development of self- view, but are also quantitatively correct. These figures
sustained oscillations at a residence time of 120min show theoretical and experimental profiles for experi-
when the comonomer feed composition was equal to ments where self-sustained oscillations were present.
 Dynamic behavior of continuous solution polymerization reactors--VII. 731
Changing Feed Composition
80.0- I
\ ,~.,_ . R.eactor Temperature.
% 70.O-
.xt,,% - -
k] ,

C~ Air Coolant Tern erature

E 5O.O-
40.0-

1c0.c- I
.o-I
= 63.0--
.2
40.9-
c-

~J
O 29.9-
0.9-
00
Time (min)
Fig. 16. Dynamic temperature profile following a step change in the feed composition (Vss= 0.6000,
Cis=0.04gmol/l, T~=298K, Tc=48°C, O=90min, V2s changed from 0.0015 to 0.0000 during
oscillatory state).

Changing Feed Composition

80.0
'i~ IReactor Temperaturel
G " I ^ I /NI /N
o 70.0
,J
60 .C
~/ I Air Coolant Temperature I
-= 5O .0
'i..........
l,....]
.....,........................
#
40.0
2OO 400 600 800 i000 1200 1400

i00.0-

8oo-
o= [," - - - - ~ ..,.,.,,~ ,.,.,.,.~ ...,., ~ ",,J~
• .~ 6o.o-
40.0-
2O .O-

0.0- f
0 200 400 600 800 i000 1200 1400
Time (min)
Fig. 17. Dynamic temperature profile following a step change in the feed composition (v,s = 0.6000,
cii=0.04gmol/l, Ta=298K, T~=51°C, ®=90min, Vzl changed from 0.0015 to 0.0000 during
oscillatory state).

It can be seen that the temperature profiles agree very 6. DISCUSSION

well. As discussed before, the experimental conversion From previous theoretical and experimental results
profiles evaluated by the on-line refractometer some- it seems clear that the dynamics of solution VA
times presented deviations to lower values, although homopolymerization is extremely sensitive to addi-
they might be reasonably good as in Fig. 21. tion of small amounts of MMA. In a more general
732 J. C. PINTO and W. H. RAY
Changing Feed Com )osition
,..., 80.0- ' I
if" ' ' ~ ~ l Reactor Temperature]
.~ 73.0-
2
6).0- ,! !
i l l a i r Coolant Temperature I
E
m.0- ' ...... ............. ) ............. +.....
40.0- / /

)(3 230 5)0

103.0-
80.0-,
o= 83.0-
40.0-
20.0-
0.0-
i) ]D
Time (rain)

Fig. 18. Dynamic temperature profile following a step change in the feed composition (v,i = 0.6000,
c+s = 0.04 gmol/l, To = 298 K, Tc = 48°C, ® = 90 min, v2/changed from 0.0000 to 0.0020 after steady state).

Changing Feed Composition

80.0 - - I
A ] Reactor TemperatureJ
P
v
70.0 - -

=1
]+ 60.0 - -

E 50.0 - -
0
I-
[ Air Coolant Temperature [
40.0 - -
I t I
800 1000 1200 1400 1600 1800 2000

100.0 -

~e 80.0 -

8 so.0-
40.0 -
8
o 20.0 -

0.0 --

800 1000 1200 1400 1600 1800 2000

Time (min)

Fig. 19. Dynamic temperature profile following a step change in the feed composition (V,s = 0.6000,
cii=0.04gmol/l, T , = 2 9 8 K , T<=50°C, O = 9 0 m i n , V2s changed from 0.0020 to 0.0000 during
oscillatory state).

sense, the stability of continuous polymerization reac- feed concentration and that it increases steadily as the
tors may be very sensitive to changes in the feed M M A concentration increases. As the rate of cross-
composition. termination is proportional to @, an increase in the
In order to analyze the source of instabilities in the M M A feed concentration causes an increase in the
V A / M M A copolymerization, one should recall that rate of termination and a consequent decrease in the
Table 8 shows that @ depends strongly on the M M A concentration of free-radical species. The net result is
 Dynamic behavior of continuous solution polymerization reactors--VII. 733
80.0-

o Reactor Temperature
pr%I

70.0-

E V\~ !

60.0-

50.0-
"""'1 tm ~ q * " q

Air Coolant Temperature

40.0-
400 600 800 i000 1200 1400
Time (rain)

i00. O-

: 80.0-

e-
,2
o 'db~'Q~l~, ~__3dd~ t~Q~b..
cJ 6 0 . 0 -

40.0-
Experiment
Simulation
20.0-

0.0-
400 600 800 i000 1200 1400

Time (rran)
Fig. 20. Theoretical dynamic temperature profile (conditions as in Fig. 16, UA = 1.01 (UA)o).

a decrease in the rate of reaction. So, from this point of example, it was shown that changes in the M M A
view, M M A molecules inhibit the VA homopolymer- comonomer feed concentration in the range
ization. Therefore, similar results may be expected in 0.00-0.50% in volume could change the stability of
other types of free-radical copolymerization where VA homopolymerization, leading to self-sustained os-
@ depends strongly on the feed composition, for cillations and extinction. A mathematical model was
example MMA/styrene and VA/styrene. developed to describe the dynamic behavior of con-
tinuous free-radical solution copolymerization reac-
tors and, after preliminary evaluation of some kinetic
7. CONCLUSION parameters, it was able to predict changes in the
The stability of free-radical solution copolymeriz- stability of VA homopolymerization when very small
ation may be extremely sensitive to changes in the amounts of M M A are added to the system. The theor-
feed composition. Using the system VA/MMA as an etical results were confirmed experimentally.
734 J.C. PINTO and W. H. RAY
80.0-

0 Reactor Temperature I ~ l
._J
. i

2 70.0-

60.0-
Q A
/ \2
" ' i"
//\
I /', ...__

50. O-

E,~ir Coolant Temperature I

40.0-
800 I000 120C 1400 1600 1800 2000
Time (min)

i00. O--

.o 80.0--

60.0--

40.0--

[ Simulation J
20.0--

0.0--
8 O0 1040 1280 1520 1760 2000
Time (min)

Fig. 21. Theoretical dynamic temperature profile (conditions as in Fig. 19, UA = 0.95 (UA)o).

Acknowledgements--We sincerely thank David Bahr and gpij gel effect correlation for kplj
James Schneider for their priceless help. We also thank gtij gel effect correlation for k,j
Conseiho National de Desenvolvimento Cientifico e Tec- I initiator
nol6gico (CNPq-Brazil) for supporting J. C. Pinto's stay in
Madison, WI. We are grateful to the industrial sponsors of kd kinetic constant for initiator decomposition
the University of Wisconsin Polymer Reaction Laboratory kpij kinetic constant for propagation of free rad-
(UWPREL) and to the National Science Foundation for ical i with monomer j
support of this research. kri kinetic constant for initiation of free-radical
i
k, kinetic 'equilibrium constant', as defined in
NOTATION
eq. (15)
c~ molar concentration of initiator k, kinetic constant for termination, as defined
Cp heat capacity in eq. (3)
f~ initiator efficiency for initiation of ktij kinetic constant for termination of free rad-
monomer i ical i with free radical j
 Dynamic behavior of continuous solution polymerization reactors VII. 735

mo rate of vapor condensation REFERENCES

Mi m o n o m e r i, molar concentration of mono- Alcock, H. R. and Lampe, F. W., 1981, Contemporary Poly-
mer i mer Chemistry. Prentice-Hall, Englewood Cliffs, NJ.
total molar concentration of free radicals Pi~ Atherton, J. N. and North, A. M., 1962, Trans. Faraday Soc.
P
58, 2049-2057.
P~j molar concentration of free-radical 1 (hav- Balaraman, K. S., Kulkarni, B. D. and Mashelkar, R. A..
ing m o n o m e r 1 at the active site), containing 1982, Chem. Engn 0 Commun. 16, 349-360.
i units of monomer 1 a n d j units of m o n o m e r Balaraman, K. S., Kulkarni, B. D., Mashelkar, R. A. and
2 Madhavan, K. P., 1986, J. appl. Polym. Sci. 31, 885-900.
Bhattacharya, D. and Hamielec, A. E., 1986, Polym. 27.
qo/qi volume shrinkage, as in eq. (9)
611~18.
Q total molar concentration of free-radicals Qi~ Brandrup, J. and Immergut, E. H., 1975, Polymer Handbook,
molar concentration of free-radical 2 (hav- 2nd Edition. Wiley, New York.
ing m o n o m e r 2 at the active site), containing Brandrup, J. and Immergut, E. H., 1989, Polymer Handbook,
i units of m o n o m e r 1 a n d j units of m o n o m e r 3rd Edition. Wiley, New York.
Burnett, G. M. and Gersmann, H. R., 1958, J. Polym. Sci. 28,
2 655-670.
Qlat rate of heat removal due to condensation Chen, C. Y., Luo, C. W. and Kuo, J. F., 1981, J. Chin. Inst.
rij reactivity ratio, as defined in eq. (1) Chem. Enors 12, 183-191.
fl initiator fragment Choi, K. Y. and Butala, D. N., 1991, Polym. Engn9 Sci 31,
normalized rate of reaction of monomer i 353-364.
Rate~
Doedel, E. J., 1986, AUTO: Software for Continuation and
Rekr normalized rate of initiation Bifurcation Problems in Ordinary Differential Equations.
T temperature California Institute of Technology, Pasadena.
T. ambient temperature Fukuda, T., Ma, Y. D. and Inagaki, H.. 1985. Macromole-
Tb boiling temperature cules 18, 17-26.
Hamer, J. W., Akramov, T. A. and Ray, W. H., 1981, Chem.
T~ coolant air temperature Engno Sci 36, 1897-1914.
T,, glass transition temperature of species i Jaisinghani, R. and Ray, W. H., 1977, Chem. Engng Sci 32,
UA global heat transfer coefficient to coolant 811-825.
jacket Jones, K. M., Bhattacharya, D., Brash, J. L. and Hamielec,
free volume A. E., 1986, Polym. 27, 602~10.
vl
Keane, T.R., 1972, Single-phase polymerization reactors in
critical free volume for propagation Chemical Reaction Engineerin9, Proceedings of the 2nd
Vftc critical free volume for termination International Symposium on Chemical Reaction Engin-
V reactor volume eering. Elsevier, Amsterdam.
X total m o n o m e r conversion Kroschwitz, J. J., 1986, Encyclopedia of Polymer Science and
Enoineerin9. Wiley, New York.
Xi molar fraction of species i Kuchanov, S. J., Efremov, V. A. and Slin'ko, M. G., 1986,
W polymer composition (mass fraction of spe- Proc. Acad. Soc. USSR: Phys. Chem. 283(2), 686~i90.
cies 1) Kuo, J. F., Chen, C. Y. and Lee, C. S., 1976a J. Chin. Inst.
Chem. Enors 7, 59-67.
Kuo, J. F., Lee, C. S. and Chen, C. Y., 1976b, J. Chin. Inst.
Greek letters Chem. Engrs 7, 75-81.
~ti expansion coefficient of species i, for free Mahabadi, H. K. and O'Driscoll, K.F., 1977, J. Macromol.
Sci. Chem. All, 967-976.
volume calculation Maschio, G. and Moutier, C., 1989, J. appl. Polym. Sci. 37,
ctUA global heat transfer coefficient to the ambi- 825-840.
ent Pinto, J. C., 1991, Anhlise da Dindmica de Sistemas de
AHij heat of reaction for propagation kpij Polimeriza~o Pela Teoria de Bifurcaf6es. Ph.D. thesis,
COPPE - Universidade Federal do Rio de Janeiro, Rio de
AHv latent heat of vaporization
Janeiro (in Portuguese).
e external capacitance Pinto, J. C. and Ray, W. H., 1995, Chem. Engn9 Sei (in
O residence time press).
2 numerical constant in eq. (25) Prochfizka, O. and Kratochvil, P., 1983, J. Polym. Sci.:
vi volume fraction of species i Polym. Chem. 21, 3269-3279.
Ray, W. H., 1972, J. Macromol. Sci. - - Revs. Macro. Chem.
Hi partial pressure of species i
C 8 , 1-56.
FIT"t saturation pressure of species i Reid. R. C., Prausnitz, J. M. and Polina, B. E., 1986, The
p density Properties of Gases and Liquids, 4th Edition. McGraw-
r dimensions time (frO) Hill, New York.
Zi interaction parameter for species i Ross, R. T. and Laurence, R. L., 1976, A.I.Ch.E. Syrup. Ser.
72, 74-79.
~Oij cross-termination constant, as defined in eq. Schmidt, A. D. and Ray, W. H., 1981, Chem. Engn 0 Sci. 36,
(2) 1401-1410.
Schmidt, A. D., Clinch, A. B. and Ray, W. H., 1984, Chem.
Engn9 Sci, 39, 419-432.
Special subscripts Shastry, J. S. and Fan, L.T., 1973, Chem. Engn9 J. 6, 129-143.
1 monomer 1 Sob, S. K. and Sundberg, D. C., 1982, J. Polym. Sci.: Polym.
2 monomer 2 Chem. 20, 1331-1344.
Teymour, F., 1989, The dynamic behavior of free radical
f feed solution polymerization reactions in a continuous stirred
p polymer tank reactor. Ph.D. thesis, University of Wisconsin,
s solvent Madison.
736 J. C. PINTO an0 W. H. RAY

Teymour, F. and Ray, W. H., 1989, Chem. Engng Sci. 44, Teymour, F. and Ray, W. H., 1992b, Chem.'Engng Sci. 47,
1967-1982. 4133-4140.
Teymour, F. and Ray, W. H., 1991, Chaos, Solitons and Yaraskavitch, J. M., Brash, J. L. and Hamielec, A. E., 1987,
Fractals l, 295-315. Polym. 28, 489-496.
Teymour, F. and Ray, W. H., 1992a, Chem. Engng Sci. 47,
4121-4132.