Dynamic process modeling and hybrid intelligent control of ethylene copolymerization in gas phase catalytic fluidized bed reactors. Issue 8 (15th May 2019)

• Record Type: Journal Article
• Authors: Abbasi, Mohammad Reza; Shamiri, Ahmad; Hussain, Mohamed Azlan; Aghay Kaboli, Seyed Hamidreza
• Is Part Of: Journal of chemical technology & biotechnology. Volume 94:Issue 8(2019)
• Page Start: 2433
• Page End: 2451
• Publication Date: 2019-05-15
• Subjects: polyethylene -- process control -- fluidized bed reactor -- dynamic modeling -- solid elutriation -- artificial intelligence; Biotechnology -- Periodicals; Chemistry, Technical -- Periodicals; Chemical engineering -- Periodicals; Industries -- Environmental aspects -- Periodicals; 660
• DOI: 10.1002/jctb.6022
• Languages: English
• ISSNs: 0268-2575

Abstract

Abstract: BACKGROUND: Polyethylene (PE) is the most extensively consumed plastic in the world, and gas phase‐based processes are widely used for its production owing to their flexibility. The sole type of reactor that can produce PE in the gas phase is the fluidized bed reactor (FBR), and effective modeling and control of FBRs are of great importance for design, scale‐up and simulation studies. This paper discusses these issues and suggests a novel advanced control structure for these systems. RESULTS: A unified process modeling and control approach is introduced for ethylene copolymerization in FBRs. The results show that our previously developed two‐phase model is well confirmed using real industrial data and is exact enough to further develop different control strategies. It is also shown that, owing to high system nonlinearities, conventional controllers are not suitable for this system, so advanced controllers are needed. Melt flow index (MFI) and reactor temperature are chosen as vital variables, and intelligent controllers were able to sufficiently control them. Performance indicators show that advanced controllers have a superior performance in comparison with conventional controllers. CONCLUSION: Based on control performance indicators, the adaptive neuro‐fuzzy inference system (ANFIS) controller for MFI control and the hybrid ANFIS–proportional‐integral‐differential (PID) controller for temperature control perform better regarding disturbance rejection and setpoint tracking in comparison with conventional controllers. © 2019 Society of Chemical Industry