Low density polyethylene tubular reactor control using state space model predictive control. Issue 4 (30th March 2021)

• Record Type: Journal Article
• Authors: Muhammad, D.; Ahmad, Z.; Aziz, N.
• Is Part Of: Chemical engineering communications. Volume 208:Issue 4(2021)
• Page Start: 500
• Page End: 516
• Publication Date: 2021-03-30
• Subjects: Model predictive control (MPC) -- low density polyethylene (LDPE) -- tubular reactor control -- state space MPC; Chemical engineering -- Periodicals; 660.205
• DOI: 10.1080/00986445.2019.1674816
• Languages: English
• ISSNs: 0098-6445

Abstract

Abstract: One of the conventional commodity polymers in the world is Low density polyethylene (LDPE). LDPE is mostly produced from free radical polymerization of ethylene gas under high pressure and temperature environment. However, the control of LDPE polymerization reactor is challenging due to the complexity of the polymerization process and the nature of the reactor itself. Thus, this study explores the application statespace model predictive control (SS-MPC) in controlling LDPE polymerization tubular reactor process. SS-MPC is a type of model predictive control (MPC) scheme which utilizes state space model as its process model. In general, MPC control scheme has a wide application in the industry due to its ability to handle multivariable control and optimize process performance. In order to implement the control scheme, a dynamic model of the reactor was developed using Aspen Dynamic. The dynamic model is then linked with MATLAB Simulink for online control implementation. The LDPE polymerization process was controlled by regulating the initiator and chain transfer agent (CTA) flowrate to control reactor temperature and polymer Melt Flow Index (MFI) grade. The state space model was identified using identification technique through two approaches; MIMO model and augmented SISO model. In order to evaluate the controller performance, MFI grade transition and disturbance rejection tests are conducted along with PID controller, as a comparison. Based on the tests, MPC had presented a notable performance in handling polymer grade transition (Integral squared error, ISE = 4530) and disturbance rejection (ISE = 99.03) in MFI profile compared to PID controller respond (ISE = 7003 for set point tracking and ISE = 12.66 for disturbance rejection). Nonetheless, PID controller had also presented a comparable performance in controlling maximum reactor temperature.