Distributed control at the Kidd Creek grinding plant. Part I: Control strategy design

"This two-part paper summarizes the design and implementation of a distributed PIbased control strategy installed at the B grinding circuit, at the Kidd Creek Division of Falconbridge Ltd. Part I details the design of the control strategy using a grinding circuit dynamic simulator (DYNAFRAG). The first step consisted of sampling campaigns and data processing to calibrate DYNAFRAG. Then, several control alternatives were designed and tested with the simulator. The main feature of these strategies is their ability to regulate or keep within an upper and lower limit five output variables (circulating load, rod mill density, cyclone feed and overflow densities and pump box level) with only three manipulated variables (rod mill water, pump box water and rod mill tonnage). By simulation on DYNAFRAG, it was possible to discriminate among several control structures. The preferred control strategy uses a saturation algorithm, which allows the adequate regulation of two measured variables with only one actuator. Controllers located in interacting loops are appropriately tuned, taking into account the existing process coupling. Part II will present the implementation of the selected control strategy.IntroductionSteady state simulators are used to predict the stationary state of a given process under specific operating conditions. Tradi - tionally, steady state simulation has been applied for process flowsheet design, for evaluating process modification benefits or for determining key operating parameters in process operation optimization. Dynamic simulators add a new dimension to process simulation since they allow the prediction of the process transient behaviour, resulting from the application of a different set of operating conditions and/or from natural disturbances or equipment failure. Dynamic simulators can also be used for the training of new operators or for the design of control strategies."