Académique Documents
Professionnel Documents
Culture Documents
by
A Thesis
Doclor of Philosophy.
@ Sc?tember, 1986
..
LATEX PRODliCTIO:" TECH:"OLOGY - REACTOR DESIG:" CO:"SIDERATIO:"S
..,.-
....
~
llanaOonou.\ou
,
DOCTOR OF PHILOSOPHY ~c:'o1ASTER U:"IVERSITY
Production Technology,
".-
'. .,
Thessaloniki: Greece)
- "
II
--
ABSTRACT
vinyl chloride). sustained oscillations (limit cycles! OCcur in conversion and all the late" and
pol~'mer properties. This creates severe problems with the commercial application of
continuous reactors. As shown by Pollock (1984) advanced control theory is inadequate for
controlling these reactors. However, it was shown that the limit cycle behaviour could be
eliminated through redesign of the reactor train configuration. The redesigned production
trIli.'1 was shown via simulation to be free of oscillations and to oITer greatly increased
In this thesis the mathematical model of Kiparissides 1I97S) and Pollock 1I9S~1.
based on a particle age distribution analysis and incorporating detailed chemistry and
physics of the polymeri:ation phenomena involved, is modified and is used to simulate the
dynamic behaviour of these reactol';'1bf different modes of process operation tbatch, semi-
the redesigned reactor system are carried out in a pilot plant stainless steel reactor train with
onli..'1e monitoring of monomer conversion with a set of on- line densitometers. Control of the
-- .
final1atex pa.~icle si;:e and conversion is achieved through manipulation of the initiator feed
rate to the first particle nucleating reactor, and the split of the monomer'and water fiow rates
tll
.----..
. .
ACKNOWLEDGEMENTS
I wish to thank Drs. J.F. MacGregor and A.E. Ha'l'ielec for their guidance,
'"'Patience, confidence
. and encouragement throughout this work. Thanks are also due to
Dr. G.A. Irons for his service in the dissertation committee, and to the staff of the Engineering
_
.. -
Word Processing Centre for t~lling this manuscript. :\1y gratitude is extended to Drs. O.
.Chiantore, H.A. ~1eyer and N.T. Ali for their lively dis<:ussions during the numerous aspects
of this research. I particularly ....i sh to thank the :\1d1aster Institute for Polymer Production- "
Tec~ology(~lIPPT) for pro~-iding facilities for this project. To Pavlos and Vina Kanaroglou I
wish to e:q>ress m~' appreciation for their help and friendship through many years. To the
for pro,-iding finll.llcia! support. Finally my everlasti.'1g love goes to my parents and sister, for
..
. .
\
TABLE OF CO"TE),'TS
Page
DESCRIPTIVE "OTE ii
ABSTRACT III
ACK"OWLEDGE~IE:\"TS l\'
TABLE OF CO"TE)'"TS v
LIST OF FIGURES
LIST OF TABLES "
:'(\"1
"OME"CLATURE ::tVll
2.1 Introduction 5
2.2
2.2.1
~lodels for Emulsion PolymeriUltion
t:ses oDlodels
....,,I .9
6
v.
..
Page
"
.-
'3.6
S.6.1
3.7 -I
. Particle Si:e Determination: Electron :'1icroscopy
. Experimental --
75
I Particle S::e DetermiI1lltion: Other :'leasurement :'lethods. 78
,,
.J
3.8 :'101ar :'lass :'leasurements 78
3.8.1 Light Scattering 79
3.5.2 Low Angle Laser Light Scattering lL.-\LLSPl 83
3.8.3 On-line SEC-LALLSP 84
3.9 Conclusions 85
Introduction 86
...~
4.1
~ Batch Reactor :'lodel 87
4.2.i Xumerical Values of:'lodel Parameters and 88
Rate Coefficient Expressions
..."
.~~
"t ...... :'lodel Parameters and Parameter Estimation 89
, Application oOladel Equations 94
. 4.3.1 Comparison oOlodel Predictions with Literature Data 94
Comparison oOladel Predictions with Experimental 96
.
.....
4.5
Results :'leasured Herein
Discussion of Experimental Run'
ConClusions
98
1::l5
\',.",
CHAPTERS: SE:'lI-BATCH C~l'ROL POLICIES FOR THE PARTICLE SIZE 1::l6
.DlSTRlBLl'IOX
vi
Page
.-- 7.3.1
-,
".&f
i.5
7.6
The Effect of Process Control Variables
S>:perimental Set-up
Discussion oi"the E:cperi:nental Results
Conclusions
lSI
IS3
IS6
201
Page
CHAPTER 10:
9.9 Conclusions
t
FINAL CONSIDERATIONS CONCER:s'ING THE ON-LINE
CONTROL OF Dll;'LSION POL Y~lERIZATION REACTORS,
262
10.1 lnt..~uc:tion
10.2 Part 1: On-line Control ofE:nulsion Processes 265
10.2.1 Gene:-al 265
10.2.:! Kaloa.'"'l Fnte:-s for E:=t:.lsio:1 Polyme~~tion Syste::1s 266
10.2.3 Application of Ka1:::1a."'l FUters to the Two Reactor. 2'71
Syste=.
10.2.4 The ?o::.ential for On-1L'"'le ?;l~cle Sl::e:'1easu:"e~ent 2'75
10.2.5 Fi..."'l:ll Ste?~, E=ulsion Rea~o!." Con::-ol ~77
10.3 ?:l:"'t 2: <Y:..he:- Genc:"2.1 Ft:.tu:-e Recc:=':::1enc.a::ons 279
:0.4 p~ 3: CO:'lduc..:..~g Rc-"1"ks 282
\~.,
\
Page
/
/
--
:.x
-::.---
.' .
LIST OF FIGURES
Figure Page
(3-12) Specific Turbidity I."ersus Conl."ersion for Runs BR, and BR5 72
(4-2) (a) Average Particle Diameter and (b) Conversion versus Reaction 99
time for a Soap-free Case
(4-3) Batch Emulsion VAc: Conversion vs. Time for Two Different
101
Emulsifier Concentrations
(';"41 Blitch Emulsion VAc: Conversion vs. Time for Two Different 102
Initiator Concentrations
(';"61. . Batch Emulsion VAc: Diameter vs. Conversion for Two Different 107
Emulsifier Concentrations
t';" 7) Batch Emulsion 'lAc: ;:-';umber and Weight Average :Molecular 108
Weights vs. Conversion
(';"S) Schematic Representation of the Experimental Set-up for the Batch Runs 109
(';"9) (a) Run BR2 and (b) Run BR3: On-line Densitometer Conversion
\s. Time
112-113.
. .. ,
(';"10) On-line Densitometry: (a) Run BR4 (b) Runs BR2, BR4 and BR5. 115
. Induction Time Effects
(';"12) Particle Size vs. Conversion Histories for Runs BR7, BRS, BRIO and BR11 120
.- (';"13) Weight Average :\lo1ecular Weight vs, Conversion'Histories for Runs 122
BR6. BRIO. BR11, BR1-'., BR21 and BR22
(';"1-'.) Effect of Agitation Rate on Conversion for the Batch Emulsion 123
Polymerization ofVAc
(5-1) '"'umber of Particles vs. Reaction Time for a Typical Batch 130
Emulsion 'lAc Run
(5-2) ,",umber of Particles vs. Reaction Time for a'Typical Batch Homogeneol:s . 132
Emulsion 'lAc Run (IF = 0.005 gmoleSl1it water)
xi
-
Figure . Page
(5-4) Number of Particles vs. Time Corresponding to the Am(t) Variation 136
of Figure 5-3
(5-5) Comparison ofNp(t) Between the Usual arid the Extended Case 138
(IF = 0.005 gmolesllit water: SF = 0.0437 gmolesllit water for the
Usual Case) . ,
(5-6) Total Polymer Particle Surface Area Corresponding to the Figure 5-3 139
Variations .
(5-7) Total Emulsifier Level in the Reactor Corresponding to Figue 5-,3 140
(7-1) Free Micellar Soap and Number.of Polymer Particles Variations for 169
a T)"pi~1 Continuous Emulsion Polymerization Case in a Single CSTR
(7-2) Conversion and Number of Particles versus Number of Residence Times' . 171
for a Single CSTR. Conditions: 8:; IS00 sec, 1= 0.01 gmoleJJit, .
S=O.OI gmolellit.
(7-3) Average Number of Branch Points per Polymer Molecule and Average 172
Particle Diameter versus Number of Residence Times for a Single CSTR.
Same Conditions as in Figure 7-2.
(7-4) Conversion and Number of Particles versus Number of Residence times 179
for the New Configuration. Conditions: 8 = IS00 sec, 1=.0.01 gmoleJJit,
S= 0.01 gmoleJJit. VI =75 ml, split= 0.4
(7-5) Average :"umber of Branch Points per Polymer ~Iolecule and Average 'lS0
Particle Diameter versus :"umber of Residence Times for the :"ew , '
Configuration. Same Conditions as in Figure 7-5
(7-8) Run CRl, Parts A and B: R2 Conversion versus Dimensionless Time IS9
(7-10) .' Run CR1, Parts A, B and C: R2 Conversion "ersus (tJ82) 192
o
- .'
.
,
Figure Page
(7-13) Run CR2, Parts A and B: Conversion versus Dimensionless Time 199
(7-14) Run CR2, Parts A and B: On-line Conversion versus (tl62) 200
. (7-15)' Run CR2, Parts A and B: Average Particle Diameter versus 202
Dimensionless Time
Conversion vs. Dimensionless Time: Unseeded 'S. Seeded PVC CSTR 219
Number Average Diameter vs. Dimensionless Time: Cnseeded vs. Seeded 221
PVCCSTR
..c. Se.eded
Overflow CSTR. Conversion vs. Dimensionless Time: Effect 228
of Process Variables (See also Table 8-2)
xiii
.~ ...
Figure Page
(9'3) Simulation Results: Batch Reactor Operation, MSI Level of 100 ppm. 243
Effect of'KMI
(94) Run IM1: Conversion vs. Time Data and Model Predictions 250
(9-5) Conversion vs. Time Data from Runs IM1 and 1M2 252
(96) Conversion vs. Time Data from Runs IM1 and IM3 253
(9i) Conversion-Time Histocies for Runs IM3,11\14', IM5 and IM6 255
(9-9) Evolution of Particle Size with Time for Runs [M1 and [:\12 258
(910) Runs [M1, [:\13 and IM6: Diameter vs. Time 259
(
(9-11) Average Particle Size vs. Batch Time for Runs [M6-[M6-IM4 260
(101) :\lodel Prediction and Filtered Estimate for Conversion and Average 274
Size from the Kalman Filter Analysis
.. :. ,
(102) Computer Control of Continuous Latex Reactors 280
, ..
(VI) Schematic Diagram of the KM..,{6 LALLSP 35i ,.
(V2) Mw-Re Relationship 358
(VI-l) Effect of the Ratio ("'m/a.p) on Conversion and Number ofParticles 3i6
xiv
Figure Page
(X.l) Run CRl, Parts A and B: R2 and Rl Conversioll versus (tl82) 406
.'
xv
(
LIST OF TABLES
Table Page
8-1 Kinetic Parameters and Other Physical Constants Used in the PVC 210
- Computer Program
-'.
XI-I
.
Physical Property and Hazard Data for :\lajor Experim!lntal Chemicals 413-414
L~' .
, . .
XI-2 :\lortality Data for :'lajor Chemi~ls L sed 414
-. .-.
. .'
XVI
--
--
NOMENCLATURE
Bs(t) average number oflg,ng-chain branch points per polymer molecule. 111-
latex _ "
Cm ratio ofkfm to k p
Cp ratio of kfp to k p
"
Cw concentration ofradica:ls in the water phase. gmolel1latex
dp(tr,ti) particle diameter of a particle born at time ti. now being at time tr. dm
Da.ve some average particle diameter. dm
xvii
J'
~Dp diffusion coefficient ofnomomeric radicals in polymer particles, dm2fsec
E(t) exit age distribution (residence time distribution) function for an ideal
CSTR,sec- 1
I denotes initiator
I w,
.'
initiator concentration in the reactor at time t= (}, gmOlefl-latex\
\
J
XUI