To improve the quality of industrial nitrile rubbers, the copolymer chemical composition, pA(t), should ideally be kept constant along the reaction. This work proposes a closed‐loop control strategy for the semibatch operation of the reactor with the aim of regulating pA(t) within a reduced range of variability. The proposed strategy is evaluated by simulating a mathematical model of the process. To this effect, a simplified mathematical model of the reaction is first derived and then utilized to obtain a suboptimal control law and a soft‐sensor that estimates the polymerization rates. The suboptimal control law is compensated by adding a term proportional to errors in pA(t). The simulated example considers the production of the low‐composition AJLT grade, with the copolymerization reaction represented by a detailed mathematical model adjusted to an industrial plant. Due to the high performance of the soft‐sensor, the simulation results suggest that the proposed closed‐loop strategy is efficient to adequately regulate pA(t) in spite of structural and parametric uncertainties, while other quality variables remained practically unaffected.
The chemical composition of the nitrile‐butadiene rubber process can be regulated through a closed‐loop nonlinear control strategy. The nonlinear control law online calculates the monomer addition along the reaction. The required polymerization rates are estimated through an artificial neural network from online measurements of the refrigerant flow.