Liliane Maria Ferrareso Lona

Neural network applications in polymerization processes

Neural networks currently play a major role in the modeling, control and optimization of polymerization processes and in polymer resin development. This paper is a brief tutorial on simple and practical procedures that can help in selecting and training neural networks and addresses complex cases where the application of neural networks has been successful in the field of polymerization.

Heterogeneous modeling for fluidized-bed polymerization reactor

The development of a heterogeneous model describing the behavior of the fluidized bed reactor in polymer production is presented. The model assumes three phases in the reactor: the gaseous bubble phase, the gaseous emulsion phase and the solids polymer particle phase. The model incorporates the interactions between phases and also predicts the physicochemical properties of the polymer.

A Comparison of Reaction Mechanisms for Reversible Addition‐Fragmentation Chain Transfer Polymerization Using Modeling Tools

A kinetic model based on a detailed reaction mechanism of the reversible addition‐fragmentation transfer (RAFT) polymerization was developed. By neglecting some reactions, or using simplifying assumptions, three different reaction mechanisms proposed in the literature can be described with this model. The resulting equations were solved using a self‐developed Fortran code. The cases were also modeled using the Predici® commercial software. Parameter sensitivity analyses and a comparison of performance of the models for the different reaction mechanisms are presented. It was found that the simplest reaction mechanisms considered in this paper cannot adequately describe the behavior of the RAFT polymerization process. It is demonstrated that the controversy in the literature regarding the “six orders of magnitude difference” in the fragmentation rate coefficient, and the “concern” about the “validity” of the commercial software package Predici® to represent the RAFT mechanism have nothing to do with model inadequacy or the alluded “empirical nature” of Predici®. Rather, these issues are merely a matter of not having precise parameter estimates in a very complex, multi‐parameter model, where similar model profiles can be generated with different combinations of model parameters.

Controlled Free‐Radical Copolymerization Kinetics of Styrene and Divinylbenzene by Bimolecular NMRP using TEMPO and Dibenzoyl Peroxide

An experimental study on the kinetics of nitroxide‐mediated free radical copolymerization (NMRP) of styrene (STY) and divinylbenzene (DVB) is presented. The experiments were carried out in bulk from a mixture of monomers, stable free radical controller (2,2,6,6‐Tetramethyl‐1‐piperidinyloxy, TEMPO), and initiator (dibenzoyl peroxide, BPO), at 120°C, without using a TEMPO‐capped prepolymer in the initial mixture. The system studied is a case of bimolecular NMRP, as opposed to the monomolecular NMRP of styrene and other crosslinker previously addressed in the literature by others. The results on total monomer conversion (polymerization rate), molecular weight development, gel fraction, and swelling index are compared against a conventional reference system (a STY/DVB copolymer, also synthesized for this study). No significant auto‐acceleration effect was observed in the early and intermediate conversion ranges of the TEMPO‐controlled copolymerization of STY/DVB, and the gelation point was significantly delayed. †On research leave from UNAM.

Another Perspective on the Nitroxide Mediated Radical Polymerization (NMRP) of Styrene Using 2,2,6,6‐Tetramethyl‐1‐piperidinyloxy (TEMPO) and Dibenzoyl Peroxide (BPO)

Polymerization conditions for the bimolecular NMRP of styrene using TEMPO and BPO were revisited and expanded with the objective of creating a more complete and reliable source of experimental data for parameter estimation and model validation purposes. Three different experimental techniques were assessed for the NMRP of styrene. The reliability of results produced in vials with inert nitrogen atmosphere was evaluated, taking as reference the more reliable technique using sealed ampoules with inert atmosphere. Polymerization rate data obtained in vials could be considered reliable if monomer loss was taken into account, but the reliability of molecular weight data at high conversions may be questionable. Polymerizations at 120 and 130°C and with TEMPO to BPO, molar ratios of 0.9 to 1.5 were carried out. Comparison of the experimental data collected against predictions obtained with a detailed kinetic model previously reported in the literature suggest that either the present understanding of the reaction system is incomplete, or some of the kinetic rate constants reported in the literature are not accurate, or both. Guidelines on how to address and design future experimental and modeling studies are offered.

Assessing the importance of diffusion-controlled effects on polymerization rate and molecular weight development in nitroxide-mediated radical polymerization of styrene

A previously derived kinetic model for the nitroxide‐mediated radical polymerization (NMRP) of styrene has been modified by considering diffusion‐controlled (DC) effects on the bimolecular radical termination, monomer propagation, dormant polymer activation, and polymer radical deactivation reactions. Free‐volume theory was used to incorporate the DC‐effects into the model. It was found that DC‐termination enhances the living behavior of the system, whereas DC‐propagation, DC‐activation and DC‐deactivation worsen it. Although the inclusion of overall DC‐effects into the kinetic model improved the performance of the model by slightly reducing the deviations obtained from experimental data of polymerization rate and molecular weight in the bimolecular NMRP of styrene with 2,2,6,6‐tetramethyl‐1‐piperidinyloxy (TEMPO) and dibenzoyl peroxide (BPO), it does not seem to justify adding the extra four free‐volume parameters. In the case of the semi‐batch addition of azo‐bis‐iso‐butyronitrile (AIBN) (several single shots at definite time intervals) in the NMRP of styrene, recently reported in the literature, it was found that DC effects are more significant, but it was observed that there was a strong dependence of polymerization rate on the frequency of addition of the shots of initiator (a maximum on polymerization rate being observed at a given frequency of addition of the shots), which could not be adequately explained in terms of DC‐effects.

APPLICATION OF NEURAL NETWORKS FOR THE DEFINITION OF THE OPERATING CONDITIONS OF FLUIDIZED BED POLYMERIZATION REACTORS

The development of polymer resins can benefit from the application of neural networks. The aim of this paper is to present how neural networks can help deal with the development of new resins, starting from the end user properties to set up the reactors operating conditions. The procedure presented in this paper consists of a network that predicts the operating conditions of the reactor with maximum error of 2%. New resins can be developed and the reactors operating conditions can be set in a much faster way, reducing the number of experiments and pilot plant tests, and hence, time and money spent on development.

Developing an educational software for heat exchangers and heat exchanger networks projects

Abstract The shell and tube heat exchanger is an equipment largely used at chemical industries. The most known methods used for its design are the methods of Kern, Bell-Delaware and Tinker, for what a lot of equations are necessary. In this way, the calculation without using any computer aid takes a lot of time. This paper present a didactical software developed for the design of shell and tube heat exchanger and heat exchangers networks, this last one using Pinch analysis. The software called heat exchanger simulator (HES), was developed to be used in undergraduate classes of Chemical Engineering. Since the use of a software can decrease the necessary time for evaluations, a fairly more time can be saved, which can be used in the analysis of the results, study of parametric sensitivity, research of backing subjects, amongst other essential tasks in the modern context of engineering. The use of the software facilitates the development of individual or group projects, in which the situation can develop its creativity through the scanning of a real problem. The benefits of such practice are several, being the starting point for the process of stand alone and continued education, unique capable to allow the future professional to remain itself brought up to date with the increasing development of new technologies.

Heterogeneous modeling of fluidized bed polymerization reactors. Influence of mass diffusion into the polymer particle

Intraparticle mass transfer can play an important role in catalyzed polymerizations. This work discusses the importance of taking this influence into account or not when modeling a fluidized bed reactor. An analysis was made of this influence on the reactor phenomena, on polymer productivity and on the polymers physicochemical properties (polymer quality). Two basic cases were simulated: low and high residence time of the polymer particle inside the reactor (low: 10 min; high: 1 h). The results indicate the need to model the reactor taking into account the intraparticle mass transfer and the concentration profile inside the particle when the fluidized bed operates under conditions of low residence time. For high residence times (above 20 min), the overall simulation results change only slightly when intraparticle mass transfer is not considered.

Development of Polymer Resins using Neural Networks

The development of polymer resins can benefit from the application of neural networks, using its great ability to correlate inputs and outputs. In this work we have developed a procedure that uses neural networks to correlate the end-user properties of a polymer with the polymerization reactors operational condition that will produce that desired polymer. This procedure is aimed at speeding up the development of new resins and help finding the appropriate operational conditions to produce a given polymer resin; reducing experimentation, pilot plant tests and therefore time and money spent on development. The procedure shown in this paper can predict the reactors operational condition with an error lower than 5%.