ABSTRACT
Successful implementation of a control scheme usually involves a number of steps: collection of process information, identification and characterization of a process model (and disturbance if required), design of a controller based on selected criteria, and implementation of the controller. A process model could change due to a number of reasons: decaying catalyst, fouling of equipment, change in process throughput, and so forth. Self-tuning control will perform all controller design and implementation steps online. Minimum variance control sometimes results in excessively large variations in the manipulated variable. A minimum variance controller tries to eliminate the deviation of the process variable from set point in one time step and can therefore cause excessively large process input changes. As a general rule the process model should be as simple as possible. It is better to start with a model that is too simple and increase the number of terms later on, than to start with a model that has too many terms.
