Intelligent Control on Industrial Vinyl Chloride Monomer Column: A System Identification and Artificial Intelligence Based Control Approach

• Published in: Periodica polytechnica. Chemical engineering. Vol. 69; no. 2; p. 284
• Main Authors: Ansari, Tuba Siraj, Taqvi, Syed Ali Ammar
• Format: Journal Article
• Language: English
• Published: Budapest Periodica Polytechnica, Budapest University of Technology and Economics 17.03.2025
• Subjects: Adaptive systems; Artificial intelligence; Artificial neural networks; Autoregressive models; Chemical engineering; Composition; Control algorithms; Control systems; Controllers; Data collection; Dichlorides; Efficiency; Identification; Monomers; Neural networks; Nonlinearity; Optimization; Polyvinyl chloride; Process controls; Product quality; Proportional integral derivative; R&D; Real time; Rejection; Research & development; System identification; Time lag; Tracking; Vinyl chloride
• ISSN: 0324-5853; 1587-3765
• DOI: 10.3311/PPch.39435

In the production of vinyl chloride monomer (VCM), the separation of VCM vapors from ethylene dichloride (EDC) in the distillation column is complicated due to uncertain dynamic behavior and nonlinearity of the process and results in poor controlling of the column which may overlook product quality. In this regard, the column is simulated with integrated tuned-controllers using Aspen Plus dynamics. For system identification of the VCM column, the nonlinear autoregressive model with exogenous inputs (NLARX) gives a higher Fit% for the real-time data in comparison with the first order plus time delay (FOPTD) model. The study shows the application of artificial neural network (ANN) and adaptive neuro-fuzzy inference system (ANFIS) based control strategies, alongside a traditional proportional-integral-derivative (PID) controller for the control of the top composition and bottom composition of the VCM column. The results indicate that for top composition, the ANFIS-based controller having an integral time absolute error (ITAE) value of 0.132 outperforms ANN-based controller with an ITAE value of 0.78 in terms of set point tracking, and a similar behavior is found for bottom composition. In terms of disturbance rejection, the ANFIS having an ITAE value of 0.036 outperforms ANN having an ITAE value of 1.03 for top composition and shows the same behavior for bottom composition while the PID control exhibits significantly lower performance in both set point tracking and disturbance rejection.