Next:Logic Control Up:Industrial Process Control Previous:Industrial Process Control

Design, Simulation and Implementation of Decentralized Controllers for a Multivariable Feed Blending System

This industrial control research project was concerned with the applicability of decentralized PI control to a nonlinear multivariable feed blending process at the Falconbridge nickel smelter near Sudbury, Ontario.

Falconbridge's new Raglan mine located in Northern Québec started to ship its nickel concentrate to the Sudbury smelter in the spring of 1998. A new continuous feed blending process design allowed the existing concentrate slurry from the Strathcona mine to be blended with the Raglan concentrate before being roasted and fed into the electric smelting furnace.

The basic idea behind the process was to mix dry Raglan concentrate with Strathcona slurry and process water in a repulper tank to maintain both a concentrate dry mass ratio setpoint, and a density setpoint at the output of the repulper tank. The most important controlled variables to achieve efficient roaster and furnace operation were the pulp density, followed by the Raglan to Strathcona dry mass ratio. The repulper tank level also had to be regulated.

Following the development of a nonlinear multivariable model of the feed blending process, we obtained a linearized model to design linear decentralized PI controllers. The design was carried out in the frequency domain. Robustness of the controllers was checked through extensive simulation studies with the full nonlinear process model. The feed blending control system was successfully commissioned in the spring of 1998 and has been reported to perform well since then.

B. Boulet, L. Ryan, C. Graves, B. MacKay

Annual Report

Mon Jun 26 21:22:20 GMT 2000