See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/37467588

Scale-Up of Polymerization Process: A Practical Example

Article in Organic Process Research & Development · May 2003

DOI: 10.1021/op025605p · Source: OAI

CITATIONS READS

28 5,018

1 author:

Thierry Meyer
École Polytechnique Fédérale de Lausanne
130 PUBLICATIONS 1,355 CITATIONS

SEE PROFILE

All content following this page was uploaded by Thierry Meyer on 01 May 2021.

The user has requested enhancement of the downloaded file.

Scale-Up of Polymerization Process: A Practical Example
Thierry Meyer*
Swiss Federal Institute of Technology (EPFL), Institute of Chemical and Biological Process Science (SB-ISP-UPRE),
CH-1015 Lausanne, Switzerland

Abstract: hydrodynamics, and product properties. As the economic

The scale-up/-down of polymerization reactors has to deal with factor becomes more and more important, other aspects such
large viscosity changes during the process, addressing mass- as product quality, EHS, and productivity should not be
and heat-transfer issues. A practical example on scale-up of forgotten.
styrene and methyl methacrylate free radical bulk and solution Depending on the chemical reaction involved, the scale-
polymerization is presented. The main critical parameters are up methodology should be adapted to the process charac-
mixing at molecular level (micromixing) and heat removal teristics. Often the process development is an iterative
capacity. The operating parameters being kept constant are: procedure combining scale-up and scale-down studies and
reaction conditions (temperature, pressure, chemistry) and thus trying to optimize factors such as: EHS, quality, process,
the reaction time. A pilot plant issued from scale-down of design, control, costs, and constraints. Bulk and solution free
possible industrial sizes was developed to represent, at best, radical polymerization will be used as an illustrative example.
larger scales. The main parameter being scaled-up is the heat
removal capacity, which has to be maintained constant among Polymerization Reaction
the different sizes. New concepts are adapted to dissipate the
Polymerization reactions can be classified as either step-
mixing energy where it is the most suitable, and the final step
growth or chain-growth reactions. It is important to note that
is the scale-up/-down strategy. Another issue addressed in this
this is a classification of reaction mechanisms, not of the
contribution is the need for in-line analytics that could operate
repeating unit structure, since many polymers can be
at different plant scales and thus give important information
synthesized either by step-growth or chain-growth polym-
for process control. Scale-up/-down strategy must include the
erization. Free radical polymerization belongs to the second
whole process, not only the reaction stage but also what happens
class.
before, after, and simultaneously, i.e., upstream, downstream,
In chain-growth polymerization, polymer molecules gen-
and peripheral operations. Finally the measurable success of a
scale-up/-down analysis could only be proved at industrial scale, erally grow to full size in a time-scale, which is much smaller
where a good agreement between pilot- and larger scale should than the time required for high conversion of monomer to
be observed. The concepts in terms of transfer phenomena, polymer. The lifetime of a growing polymer molecule may
analytics, separation, product properties, feasibility, and eco- be less than a few seconds for a free radical polymerization,
nomics should also be included in this analysis. which is a typical example of chain-growth polymerization,
while a typical polymerization time to obtain high monomer
conversion may be several minutes or hours such as that for
Introduction the step-growth polymerization. Chain-growth polymeriza-
During the R&D of a chemical process, plant design tions require an active center, which may be a free radical,
requires special attention. Scale-up or scale-down procedures cation, or anion. Once an active center is created, the polymer
are part of the integrated process development. The challenge chain grows extremely rapidly, and when the growing chain
is to find the optimum between the chemistry, the design, is deactivated by a termination reaction, the polymer chain
the EHS (environment, hygiene, and safety) compliance, and is “dead” and no longer takes part as a reactant. With free
economic factors. This iterative process requires concept radical polymerization, however, the so-called dead polymer
discrimination at the early stage of the process development. chain is not always truly dead, because under certain
A direct transfer from laboratory scale to the industrial scale circumstances it may itself react with radicals. The active
is rarely feasible. Generally, one or several scales between center may initiate the growth of several polymer chains.
the lab and the industrial process may be used. Thus, we For reactions evolving in complex media, most operating
have expressions such as bench scale, mini-pilot plant, pilot parameters change with time or size scale. In fact bulk or
plant, quarter scale, half scale, demonstration unit, and so high solid content polymerizations start in water-like viscos-
on that often vary from one company to another.1 ity media (∼1 mPa‚s) ending in high viscosity range (10-
A process development requires a good understanding of 1000 Pa‚s). Hence, scale-up procedures are not trivial, and
the chemical system in terms of kinetics, thermodynamics, special attention has to be paid to this viscosity increase
* To whom correspondence should be addressed. E-mail: Thierry.meyer@ during polymerization, decreasing the heat- and mass-transfer
epfl.ch. rates. One has to describe correctly the transfer phenomena,
(1) Euzen, J.-P.; Trambouze, P.; Wauquier, J.-P. Scale-up Methodology for
Chemical Processes; Gulf Pub. Co.: Houston, TX, Paris, EÄ ditions Technip, the kinetics, and the media composition in evolving and
1993. rheologically complex fluids.
Figure 1. MMA polymerization modeling for two tempera-
tures (initiator DTPB 50 mol‚m-3, 10% w/w ethyl benzene) 2.
Experimental value: dots; simulation: lines.

Figure 2. Examples of high-viscosity reactors.4

High solid content polymerization of methyl methacrylate
(MMA) with free radical initiator is the chosen model system
for the investigation of diffusion-controlled bimolecular industrial scale. One of the major advantages is the reaction
termination and propagation and the decrease in initiator volume being less than that in classical stirred tank reactors
efficiency at high monomer conversions (cage effect). It has and the control of temperature, which is greatly improved.
the largest Trommsdorff-Norrish effect because most of the The bulk/solution polymerization of styrene and methyl
polymer chains are produced by bimolecular termination and methacrylate were carried out continuously with remarkable
the higher-average molecular masses increase dramatically product properties (Mw ≈ 150 000, Mn ≈ 90 000). The pilot
in the absence of a chain-transfer agent. The MMA polym- reactor worked continuously for several weeks without
erization reaction, measured by differential scanning calo- plugging or fouling and delivered a constant product quality.
rimetry at two different temperatures and modeled according The pilot plant used in this work, with a maximum
to Fleury et al. and Zeilmann,2,3 is presented in Figure 1. production capacity of 200 kg per day, is schematically
The gel or Trommsdorff-Norrish effect (being accepted as depicted in Figure 3. This pilot installation consists mainly
a polymerization rate auto-acceleration) is clearly observable of four parts indicated with the letters A-D. It consists of a
at 135 °C, starting after 250 s and being less present at 155 tubular loop reactor (A) followed by a plug flow reactor
°C. The reaction time to complete the MMA polymerization (pptube). The preheating (B) and devolatilization (C) parts
is about 14 and 30 min at 155 and 135 °C, respectively. guarantee the continuous separation of polymer from volatiles
The correlation between the experiments and the model is for the production of low-residual-content polymer. The
really good, indicating that the parameters and the model monomer-feed stage (D) ensures a controlled composition
describe well the observed measurements. This suggests that and monomer feed into the reactor.
this model could be used for the scale-up/-down study.
Therefore it is necessary to have a reliable kinetic model Polymerization Modeling
before accepting too many compromises that could lead to Since 1980, modeling of polymerization reactors has
wrong development and expectations. become more comprehensive. To develop a predictive model,
account must be taken of the chemistry and physics of all
Polymerization Reactors the microscopic processes that occur in the polymerization
As the bulk- or solution-polymerization reaction evolves process.5 The polymer reactor model is now becoming
in high-viscosity media, special reactors could be used as accepted as a valuable tool whose use contributes signifi-
presented in Figure 2. These kinds of reactors are generally cantly to all aspects of process technology for polymer
suitable for high-viscosity processes, especially for poly- manufacture. This includes process design, optimization, state
condensation processes. Tubular-based loop reactors could estimation, and control. Polymers with a unique and desirable
be used instead for bulk or solution free radical polymeri- combination of properties can often be obtained through
zation. process design. Process parameters such as residence time
A special reactor composed of a tubular loop reactor distribution (RTD) are usually not considered by polymer
followed by a tubular part, all being completely filled with synthesis chemists, although RTD can influence chemical
motionless mixers (Sulzer SMXL), was developed in col- composition distribution, molecular mass distribution, long-
laboration with the manufacturer (Figure 3). The loop reactor chain branching, and gel/sol ratios. In the early days of the
can act as a continuous stirred tank as well as a plug flow polymer industry, the chemist played the major role in
reactor, depending on the recycle ratio used. This reactor product and process development and scale-up. This has
concept was proved to be efficient for polymerization up to changed, with the process engineer now playing a significant
role in all phases of commercialization of new and improved
(2) Fleury, P.-A.; Meyer, Th.; Renken, A. In 4th International Workshop on
Polymer Reaction Engineering; Kreysa, G., Reichert, K.-H., Eds.; VCH:
polymer products. His broad experience with process fun-
Weinheim, 1992; Vol. 127, p 31.
(3) Zeilmann, T. Continuous methymethacrylate polymerization process, Swiss (4) Ullman’s Encyclopedia of Industrial Chemistry; VCH: Weinheim, 1992.
Federal Institute of Technology (EPFL), No. 2554, 2002. (5) Wilks, E. S. Industrial Polymers Handbook; Wiley-VCH: Weinheim, 2001.
Figure 3. Polymerization pilot plant.

Figure 5. Dynamic monomer conversion profile for a MMA

polymerization at 145 °C 3, same conditions as for Figure 4.

Figure 6. Scale-up of intensive mixing zone in loop reactors.

Figure 4. Dynamic temperature profile in a tubular reactor be observed that depending on the heat removal capacity,
for a MMA solution polymerization (12% solvent mass fraction, the temperature profile is strongly different going from a
Twall: 418 K, Tinlet: 417 K, Cinitiator: 50 mol/m3) 3.
hot-spot presence to a quasi-isothermal polymerization. The
damentals and computer modeling is essential to obtain high- corresponding monomer conversion profile is presented in
quality products, safely, respecting EHS constraints, and in Figure 5 where the strong gel effect is observable since 20%
an economic manner.4 conversion. Stability and control of polymer reactors should
Dynamic reactor models can be used in a variety of ways. be considered at the design stage and control problems
An example is given (Figure 4) for the MMA polymerization minimized then, rather than corrective action taken after the
by dynamic modeling in tubular reactor, showing the effect plant is built. Complex interactions which are involved in
of heat-transfer coefficient on the temperature profile. It can polymerization (highly nonlinear temperature and concentra-
Figure 7. Schematic drawing of an industrial polystyrene plant, copyright Sulzer,9 reprinted with permission.

tion effects) preclude optimal design based on experimenta- case of loop reactors, special attention has to be paid to the
tion alone because the cost would be prohibitive. Models degree of mixing (a better understanding of the degree of
can be used to identify potential sources of product variability mixing in high-viscosity media is of crucial interest with
and strategies to minimize their effects. Models can also be regard to the product quality during a polymerization
used to store information on process technology in a concise reaction), energy dissipation, heat transfer, and required heat
and readily retrievable and modifiable form. capacity removal (∼1 MW/m3).
Process models can be used to train chemists, chemical Sulzer static mixers were used in our polymerization pilot
engineers, and plant operators and give them an awareness plant. The choice, among several types (Kenics, Koch, and
of the dynamics of the polymerization process. The most so forth), without including an economic factor at this stage
expensive aspect of model development is experimental of process development, was imposed by the need of high
estimation of model parameters; highly instrumented bench-, micromixing efficiency and heat transfer on one hand and
pilot-scale, and plant-scale reactors are required. Statistically by the scale-up possibilities on the other hand. Other mixer
designed experiments should be performed to permit efficient types could be used when heat capacity removal is lower or
parameter estimation and model development. Modeling is when mixing is not the critical issue.
an iterative process, and the very act of developing a Concerning mass-transfer phenomena, micromixing as-
deterministic model permits a greater understanding of the pects should be taken into consideration. The reactor design
relevant microscopic processes which occur during polym- could also be improved to dissipate the mixing energy at
erization or polymer modification. As additional data (plant, locations where intensive mixing is required (in our case one
pilot-plant, and bench-scale) become available, model struc- location is situated at the monomer inlet feed). This concept
ture and parameters can be updated. should be feasible in scale-up/-down procedures. In loop
reactors it is possible to dissipate 80-90% of the total
Scale-Up Aspects of Polymerizations mechanical energy (achieved by Sulzer SMX-type mixer)
Very little publicly available information could be found only in the zone where intensive mixing is required 7,8
on the scale-up of polymerization reaction and processes.6 (mixing of low-viscosity fluid, i.e., a monomer mixed with
Several questions have to be answered during the scale-up/- the high-viscosity reactive polymer solution) as presented
down process, such as: What should be kept constant in the in Figure 6 at pilot and industrial scale (dark zone ) intensive
process? Which factors should be scaled-up/-down? What mixing). The remainder of the dissipated energy used in the
are the important parameters? Which design concept is rest of the reactor maintains mixing and heat transfer. This
better? What will the process control strategies be? Is the concept is the result of the scale-up study that could be
process EHS-compliant? applied either at pilot or industrial scale, enforcing the scale-
To focus on crucial parameters in scale-up studies, it is up/-down compatibility.
important to define what has to be kept constant. In the case The heat transfer or more precisely, the heat-transfer
of solution and bulk polymerization, reaction time, heat coefficient per volume unit should be kept constant among
removal capacity, and the degree of homogeneity should be the pilot and the industrial scale. Two different types of static
kept constant. To maintain constant reaction time means that mixers have to be employed to fulfill the above-mentioned
operating parameters such as chemical composition, tem- requirements. At industrial scale the Sulzer SMR mixer could
perature, pressure, and mixing quality must also be kept afford a global heat-transfer coefficient of 26 kW‚m-3.K-1
constant. Assuming that the chemical recipe is identical, one
has to concentrate on heat- and mass-transfer aspects. (6) Meyer, Th. Curr. Opin. Drug DiscoVery DeV. 2002, 5, 960.
Before there is any scale-up/-down design, one has to (7) Meyer, T.; Renken, A. In Proceedings of the Third Berlin International
Workshop on Polymer Reaction Engineering; Reichert, K.-H., Geiseler,
evaluate the following: Where are the limitations are? How W., Eds.; VCH: Weinheim, 1988; p 285.
can they be faced? And what are the consequences? In the (8) Meyer, T.; Renken, A. Chem. Eng. Sci. 1990, 45, 2793.
Figure 8. Parity plot for off-line (sampling and gas chromatography analysis) and in-line measurements (UPV).

Figure 10. Scale-up/-down of UPV in-line control.

the polymerization reaction in the reactor was developed.10-12

Figure 9. MMA monomer conversion evolution in the dynamic The sensors can be directly inserted into the tubular reactor
state. (Figure 3) and do not require a sampling circuit. This allows
in-line monitoring of the instantaneous monomer conversion
from the size DN100 to DN1500. The scale-down procedure in high-viscosity media as presented in Figures 8 and 9 for
will indicate the suitable size of SMXL mixers that corre- the styrene and MMA polymerization. It could also be
sponds to the same coefficient. The result is that a SMXL adapted to pilot scale as well as industrial scale as presented
DN 20 will give a global heat-transfer coefficient of 28 in Figure 10. The same approach used in pilot scale could
kW‚m-3.K-1. The pilot-scale reactor must be constructed at be applied to larger scale without the need of tedious scale-
this size; other configurations will lead to a nonrepresentative up studies and technical developments.
pilot-industrial scale-up/-down procedure.
The industrial-size plant drawing is presented in Figure Conclusions
7, where the abovementioned scale-up/-down results are The choice of the right scale for the pilot plant is of crucial
integrated for the mass- and heat transfer. relevance towards the industrial size to be addressed. The
main parameters as well as the kinetics, thermodynamics,
and technologies should be clearly defined. For the studied
Scale-Up Strategy Applied to In-Line Control
polymerization system, reaction parameters such as chem-
Once a process has been correctly described, an adequate istry, temperatures, pressures, and reaction time were kept
strategy for process control has to be established in terms of constant among the scale-up process. All the efforts were
reactor behavior and product quality by the means of sensors, concentrated on heat- and mass transfer, these being defined
models and processes. The useful operating measurements as crucial parameters. The limiting factors such as mixing
(temperature, pressure, flow rate, monomer conversion, and efficiency and heat-removal capacity were addressed to be
viscosity) have to be scale-up/-down compliant. This seems comparable at pilot and industrial scale. The scale-up process
trivial for temperature and pressure but could be difficult (9) Sulzer-Chemtech, AG. Misch- und Reaktionstechnik; No. 23.27.06.20-X.94-
for viscosity, conversion, and product properties. The ap- 100.
(10) Cavin, L.; Meyer, Th.; Renken, A. Polym. React. Eng. 2000, 8, 201.
plication of the UPV (ultrasound propagation velocity) (11) Cavin, L.; Renken, A.; Meyer, Th. Polym. React. Eng. 2000, 8, 225.
technique to follow the media composition evolution during (12) Zeilmann, T.; Lavanchy, F.; Meyer, Th. Chimia 2001, 55, 249.
 was an iteration between scale-up and scale-down. It has and engineering. All the concepts being developed should
been demonstrated that not every pilot scale is the right be employed at pilot scale as well as industrial size. The
industrial scaled-down size. scale-up studies should be able to predict the reactor behavior
The scale-up process should play an important role in the from pilot plant to full scale.
design procedure. Special attention has to be paid to the
technical limitations and when possible take advantage of Acknowledgment
scale-up similitude as it was the case with static mixers being
totally different at both scales (Sulzer SMXL for pilot and The Swiss “Office Fédéral de l’Education et de la Science
Sulzer SMR for industrial size). Control and measurements, (OFES)” and the industrials partners, Sulzer Chemtech AG,
especially in- or on-line ones, should be associated with BASF AG, and DSM are gratefully acknowledged for their
scale-up/-down procedures. In- or on-line sensors are pre- financial and material support.
ferred and assessed in the scale-up feasibility.
Scale-up/-down studies are part of the iterative process
Received for review October 24, 2002.
development and should be closely integrated in a multidis-
ciplinary approach encompassing several branches of science OP025605P

View publication stats