In this section we'll look at some of the techniques which can be used to control process plant and develop methods for analysing and predicting the performance of the control system.
Before starting this section you might want to revise the basic definitions which were introduced at the start of the course. In particular, you should clearly understand what is meant by the following: a process; a controlled variable; a manipulation; and a disturbance.
Control systems can be thought of belonging to two broad categories ( although many modern systems mix these categories to create hybrids): feedback control and feedforward control.
Feedback control is a form of control where one, or more, output of the process, the controlled variable, is measured and this measurement is supplied to the controller. The controller takes this measurement and compares the current state of the process with a desired state, or setpoint, and then calculates a change to the process which will move the output closer to its setpoint. This change is implemented through a process input, the manipulated variable. When a feedforward controller is operating to track setpoint changes (e.g. when the operator changes the setpoint on a temperature controller from 110 to 120 degC) the controller is said to be operating in servo mode. The other operating mode for a feedback controller is when is is operating in regulator or disturbance rejection mode. In this case a disturbance variable (an input which alters outside the influence of the local control system) changes and moves the controlled variable away from its setpoint - the controller then adjusts the manipulation to attempt to correct for the effect of this disturbance. When dealing with disturbances feedback control suffers from the fact that it can do nothing at all until the disturbance makes itself felt by changing the process output. If the process has a significant capacity this can be serious since the entire contents will have been affected before corrective action can be taken. With large capacity processes this can often lead to sluggish rejection of disturbances and significant deviations from setpoints.
Feedforward control measures identified and measurable disturbances rather than the controlled variable. The controller uses a model to predict the effect these disturbances will have on the controlled output and then uses another model to calculate the change in the manipulated variables which is required to precisely counteract the disturbance effects. The quality of the control achieved is entirely dependant on the models which are used. Often very simple (in many cases just steady-state) models are used, but feedforward control is combined with feedback control. The feedforward part takes care of most of the disturbance before it significantly affects the process and the feedback control mops up any residual affects due to model error. Although feedforward control isn't particularly complicated, we don't have time to cover it in this course. Feedforward, and other model-based control methods, are covered in the MEng course.