Model predictive control of a twin-screw extruder for thermoplastic vulcanizate (TPV) applications

This paper deals with the design and experimental study of a model predictive controller (MPC) for the regulation of the melt temperature (MT) and the motor load (ML) in a pilot plant co-rotating twin-screw extruder for thermoplastic vulcanizate (TPV) applications. It is shown that these two process variables have significant influence on the crosslinking of the rubber chains by initiation of curatives such as peroxide and phenolic agents. Since the measurement of the degree of crosslinking is performed only in off-line manner, the above-mentioned control methodology can offer inferential control of the process. The manipulated variables are the screw speed (SS) and the wall temperature (WT). The previously identified multi variable model ( Trifkovic et al., 2010) was used for the MPC design. In order to perform the real time control studies, the process communication was established from the local PLC (programmable logic controller) to the Advanced Control and model predictive control toolbox in MATLAB software, via OPC (OLE (Object Linking and Embedding) for Process Control) server. Both, set-point changes and disturbance effects were applied on the process for comprehensive evaluation of the controller performance.

► Inferential model predictive control of a pilot plant co-rotating twin-screw extruder for thermoplastic vulcanizate applications is studied.
► Two process variables have significant influence on the crosslinking of the rubber chains by initiation of curatives such as peroxide and phenolic agents.
► The manipulated variables are the screw speed and the wall temperature.
► Both set-point tracking and disturbance rejection were applied on the process for comprehensive evaluation of the controller performance.