10.9 Summary of Feed-Forward Control

We have developed feed-forward control based on models of the disturbance and process transfer functions. If the disturbance and process models are not available, feed-forward control can still be implemented as a tunable gain plus lead-lag controller. You should consider the following important issues when implementing feed-forward control:

• The feed-forward controller must be physically realizable and stable, as noted by the next three items.

  • The disturbance time delay must be greater than the process time delay for perfect feed-forward compensation. If the process time delay is greater than the disturbance time delay, the feed-forward controller will have no time-delay compensation, and perfect control cannot be achieved.

  • If the process has an RHP zero, it must be factored out before designing the feed-forward controller. Otherwise, the feed-forward controller will be unstable.

  • If the process is higher order than the disturbance, fast time constants probably must be neglected in designing the feed-forward controller.

• Often a static feed-forward controller will have good performance, particularly if the disturbance dynamics are the same timescale as the process dynamics.

• A feed-forward controller does not change the closed-loop stability of the feedback system, assuming the feed-forward controller is stable. Also, a feed-forward controller does not change the setpoint response of a closed-loop system.

• Feed-forward control can be implemented with either the classical feedback (PID-type) or IMC structure.