Industrial SBR Process. Computer Simulation Study for On-line Estimation of Steady-State Variables Using Neural Networks

ROQUE J. MINARI; GORGINA STEGMAYER; LUIS M. GUGLIOTTA; OMAR A CHIOTTI; JORGE R. VEGA

Macromolecular Reaction Engineering

Wiley

Año: 2007 vol. 1 p. 405 - 412

1862-832X

This work investigates the industrial production of Styrene-Butadiene rubber (SBR) in a continuous reactor train operated in steady-state, and proposes a soft-sensor for on-line monitoring several process and polymer quality variables in each reactor. The soft sensor includes two independent artificial neural networks (ANN). The first ANN estimates monomer conversion, solids content, polymer production, average particle diameter, and average copolymer composition; the second ANN estimates average molecular weights and average branching degrees. The required ANN inputs are: (i) the reagent feed rates into the first reactor; and (ii) the reaction heat rate in each reactor. The ANNs are trained and validated with simulation data generated with an available first-principle mathematical model, that was previously adjusted to experiments carried out in the industrial plant. The soft sensor proved robust to zero-mean noisy measurements, and to systematic measurement errors in the emulsifier and in the initiator feeds; but produces biased estimates when the monomers and/or the chain transfer agent feeds are erroneously measured. Due to the simplicity, the robustness, and the reduced response time, the proposed ANN-based soft-sensor is suitable for on-line estimation and could be used for closed-loop control strategies.