Date of Award


Degree Type


Degree Name

Doctor of Philosophy (PhD)


Chemical Engineering


John F. MacGregor




The focus of this thesis is the modelling and control of product propelties in gas phase polyethylene reactors. The main product properties of concern are melt index (MI) and density (p), which are related to the molecular weight and composition of the ethylene/a(alpha)-olefm copolymer. A kinetic model is developed which accounts for the effects of gas composition, reactor temperature, and active site distribution of the Ziegler-Natta catalyst on the MI and p of the polymer product. The model predicts the behaviour of MI and p in an industrial reactor , as well as broadened molecular weight distribution and bimodal composition distributions, which are typical for commercial linear polyethylenes.

Because measurements of MI and p are not available on-line, a methodology is developed to infer product properties from available measurements. Simple, theoretically-based models are derived which relate MI and p to reactor operating conditions. Parameters in the models are adjusted using off-line measurements, providing an effective means for inferring both MI and p.

In a series of three product grade changeovers, dynamic optimization is used to determine optimal profiles for: hydrogen and butene feed rates, the reactor temperature setpoint, the gas bleed flow, the catalyst feed rate, and the bed level setpoint. It is shown that large transitions in MI are hampered by slow hydrogen dynamics, and that the time required for such a transition can be reduced by manipulation of the temperature setpoint and the bleed stream flow. Reduction of the bed level and catalyst feed rates during transitions can significantly decrease the quantity of off-specification polymer produced. In the absence of feedback control, disturbances and model mismatch can result in product property trajectories which differ significantly from the nominal optimal trajectory.

A novel nonlinear model-based strategy is developed for on-line product property control. This feedforward/feedback control scheme is capable of both regulating product quality about a given target and of implementing optimal transition policies with feedback. The simplified mass balance model used in the controller design contains four adjustable parameters which are updated using an extended Kalman filter (EKF). The controller and EKF provide excellent regulatory and grade transition control for the range of poleythylene products simulated. The nonlinear controller is superior to an analogous linear time-invariant internal model control (IMC) design. The control system developed in the thesis is both simple and effective, and it has great potential for improving product quality in the polymer industry.

McMaster University Library

Files over 3MB may be slow to open. For best results, right-click and select "save as..."