(494f) The Transfer Of Semi-Batch Polymerization Processes Into Continuous Plug Flow Reactors

Introduction

Copolymers of the same family with different compositions and characteristics are usually manufactured in multiproduct batch plants. Recently, the idea of producing different products in modular continuous plants has been investigated intensely, e.g. in the European research project F3 Fast Flexible Future Factory [BUC09].  Such plants should combine the advantages of continuous production, such as product consistency and simplified operation, with the flexibility of a batch plant. Moreover, better heat transfer characteristics enable higher reaction rates, e.g. making higher solids contents in solution polymerizations feasible. While the transfer of batch recipes from a batch plant to a plug-flow reactor (PFR) is limited only by factors such as the geometry of the reactor and the minimum flow velocity for sufficient heat transfer and sharp residence time distribution, the transfer of semi batch recipes is much more complicated. In a stirred tank there are no restrictions for the addition of the raw materials, but in a PFR the addition can only take place at a finite number of injection points. In this work the influence of the kinetics on the transferability from a semi-batch stirred tank reactor to a continuous tubular reactor is investigated. The transferability of a semi-batch recipe means the possibility of the production of the same or nearly the same product in both modes of operation. Typically, semi-batch recipes are used to cope with heat exchange limitations in batch vessels and/or to produce copolymers which are composed of monomers with largely different reaction rates. Large differences of the reactivities of the monomers require the continuous addition of one monomer into the tank reactor with low flow rates which cannot be realized in tubular reactors with side injections.

Method

The products of the processes considered here are water soluble copolymers of two different monomers that react via radical polymerization. The polymerization mechanism is modelled by the terminal model approach which is characterised by the assumption that the reactivity of an active polymer chain is only determined by the last monomer molecule in the chain. The temperature dependency of the rate constants is described by the law of Arrhenius, leading to 26 parameters that appear in the kinetic model. To describe different reaction systems and to screen the reaction systems for their transferability, the total number of parameters must be reduced. Assuming that the initiator decay is given and known, the kinetics of a reaction system of two monomers can be described by four parameters. These are the reactivity ratios (the ratios of homo- and cross propagation) of both monomers, the ratio of the homo-propagation rates of the two monomers and the ratio of the termination rates.

During a semi-batch process the faster reacting monomer and possibly the initiator are fed continuously over the batch time into the vessel, while the slower reacting monomer is usually filled into the reactor before. This procedure prevents the production of homo-polymer built from the faster reacting monomer and results in the desired copolymer composition. The composition of the copolymer chains in the product depends on the monomer concentrations during the whole batch run. The continuous addition realized in the semi-batch process can not directly be transferred to the continuous process because of the limited possibility of adding ingredients to the reactor. In this work, the number of addition points along the reactor is limited to three besides the feed point at the entry of the PFR. In total the reactor consists of 4 plug-flow reactor modules of the same length. The product quality that results from different operating parameters of the tubular reactor is computed from a model which is also used to optimize the operating conditions to reduce the difference between the characteristics of the final products obtained from the semi-batch operation and from the tubular reactor. The model describes the evolution of the concentrations of initiator, both monomers and of three kinds of radicals. The method of moments approach is used to determine different product quality parameters: the number average and weight average chain length (NACL, WACL), the total conversion in the reactor and the sequence of monomers in an average chain in the reactor.

The recipe of the semi-batch process defines the ratio of both monomers in the final product for full conversion.  This information however is not available for all reaction systems investigated here. In order to exclude unrealistic recipes and to consider only feasible ratios of the amounts of both monomers, an optimization under constraints has been performed. The variables of the optimization are the semi-batch feed rate of the faster reacting monomer, while a fixed amount of the slower reacting monomer is assumed. After full conversion of the monomers the obtained monomer ratios and product properties are used in the continuous process in order to compare both products in terms of copolymer production, conversion, and chain and sequence length distribution in the product.

The base case for the transfer from a semi-batch process to a continuous process is to simply map the time axis on the residence time (length) of the tubular reactor and to feed the corresponding amounts of initiator and monomer fed in the semi-batch process to each module.

The optimization of the feed rates to the modules of the tubular reactor with side injections considers the total amount of initiator, the distribution of initiator and monomer over the feed points and the volume flow at the outlet of the plant as degrees of freedom. While the amount of initiator mainly influences the average chain length, the chain length distribution can be influenced by the distribution of the monomer and initiator feeds. By changing the outlet flow of the plant, the residence  can be adjusted to minimize the differences in the product qualities. The cost function used in the optimization considers the mean conversions of both monomers, the NACL and WACL, the total amount of initiator and the point in time of maximum conversion.

Results

The investigation of the product differences for both modes of operation gives insight into the limitations for the semi-batch-to-continuous transfer. By the simple mapping of the semi-batch recipe to the tubular reactor and lumping of the feed amounts, the semi-batch production cannot be transferred to a continuous tubular reactor for a most reaction systems. This approach is only suitable for reaction systems where both monomers have nearly the same kinetic behaviour. The larger the kinetic differences are, the larger are the differences in the product qualities between the two reactors which is caused by the occurrence of homo-polymerization of the faster reacting monomer. The optimization of the operating conditions can reduce the differences of the products in terms of the chain length distribution and of the total composition significantly. Space-time yield improvement factors between 4 and 100 can be obtained, depending on the reaction system and under the constraints for a minimized difference in the chain length distribution.

Even when the NACL and the WACL are the same for the products from the semi-batch and from the tubular reactor,  the polymer chains have a different inner chain sequence  due to the varying concentration ratios during the reaction time for the tubular reactor. This difference in the product quality depends on the number of side injections in the tubular reactor and can only be overcome by increasing the number of side injections, which is restricted by the associated costs. The optimization based approach provides the attainable range of products for a fixed plant configuration and the limitations of the transfer from semi-batch to continuous operation  can be visualized. .

Acknowledgement

The research leading to these results was funded from the European Community's Seventh Framework Program (FP7/2007-2013) under grant agreement n° 228867

References

BUC09  F3-Factory project web page, www.f3-factory.com, S. Buchholz (project coordinator), 2009