Year 1996
Tomi Koskelainen

Commissioning of Model Predictive Multivariable Controllers in a Polymerization Process

The thesis discusses the commissioning of model predictive multivariable controllers at Borealis’ new polyethylene plant PE2 in Porvoo. The controllers are used for the control of two serial polymerization reactors. The commissioning consisted of two parts: identification of the process models and tuning of the controllers. The controllers have been implemented with model predictive control software NAPCON, developed by Neste Engineering. The software resides in the main process computer of the plant.

The collecting of normal operating data for the identification of the process models was begun right after the first start-up of the reactors. Separate step tests were carried out for getting models that were not successfully identified using normal operating data. The model predictive controllers use these identified SISO models for the calculation of predictions and control actions.

The commissioning and tuninng of the controllers were carried out in phases. Two so-called stabilizing controllers, which stabilize the conditions of both reactors separately, were commissioned and tuned first. The stabilizing controllers are the lowest level model predictive controllers manipulating the setpoints of the DCS-resident flow rate controllers. After the stabilizing controllers, four upper level controllers were commissioned and tuned. These in turn manipulate the setpoints of the stabilizing controllers.

The commissioning of the model predictive controllers was successful. The degree of utilization of the controllers is high except two upper level quality controllers of which commissioning is still continuing. With the model predictive controllers better quality of the produced polyethylene has been achieved because of the reduced deviation f reactor compositions and production rate ratio of the reactors. The controllers have also improved the operating efficiency of the reactors.

This info last modified 27 Apr 2024 - contact webmaster