Model-based optimal control of the production of polyvinyl acetate
Systematic methods and tools for managing the complexity
Tools Integration - CAPE Methods & Tools (T4-10)
Keywords: Polyvinyl acetate, Model-based, control
In the environment of increasing demand and competitiveness the polymer manufacturers need to augment the production at the lowest cost. This work considers the polymerization plant in Mitol factory in Sežana, Slovenia, which is nowadays working over ninety per cent of its capacity. The preliminary analysis of the production process of polyvinyl acetate in Mitol revealed that there are not bottlenecks in the process. To increase the production without further investments a reduction of the duration of the reaction should be achieved. The main variables affecting the duration of the reaction and thus the productivity are: the temperature in the reactor, the initial amount of monomer, the subsequent feed rate of monomer and the addition of initiator.
Currently the temperature is controlled by manually adding the initiator. This can lead to the offset and oscillations of the reactor temperature, affecting both the final quality and the duration of the reaction. The ratio between the amount of monomer charged in the beginning and the amount charged during the reaction is feasible, but it has never been optimised. The feed rate of the monomer is kept constant. The initiator is introduced manually at the operators’ discretion. During the polymerization the operators have the responsibility to decide when to add more initiator, and when to start the addition of monomer. Therefore, the possibilities to increase the production of polyvinyl acetate in Mitol are the enhancement of the temperature control through the optimal addition of initiator and the optimization of the monomer addition profile. The constraints imposed on the objective function consider the desired final conversion and particle size of the polymer product. Since the real plant cannot be used to perform experiments a model has been developed on which the optimization algorithms can be applied. The details of this model can be found in a previous paper.
An optimal control strategy for the polymerization process has been designed. This strategy aims to reduce the time of the batch and so increase the productivity. The two variables controlled are the temperature and the concentration of monomer in the aqueous phase. The temperature data are collected on-line by a sensor introduced in the reactor. The concentration of monomer is not directly measured; hence it has to be estimated. A dynamic model of the reactor estimates the rate of polymerization from the calorimetric data. The resulting control strategy has been validated off-line, based on the previously developed model, and it is being implemented in the factory. The four main variables identified by the Mitol Company as characteristic for the final product quality: the final conversion, particle size, solids content and viscosity, act as the end point constraints. The maximum temperature in the reactor constitutes the path constraint.
In this paper an optimal control strategy for the addition of the initiator and the monomer has been designed. The objective of the optimization is the reduction of the time of the reaction while keeping the quality variables within acceptable values.
See the full pdf manuscript of the abstract.
Presented Thursday 20, 10:12 to 10:30, in session Tools Integration - CAPE Methods & Tools (T4-10).