Feed-Forward Control is an Advanced Strategy for Proactively Dealing with Upstream Disturbances
In a previous post cascade control was introduced as an effective means of limiting the lag between an upset and the associated PID control loop’s correction. As practitioners know: The longer the delay in responding, the larger the negative impact on a process. Like cascade, Feed-Forward enables the process to preemptively adjust for and counteract the effects of upstream disturbances.
A limitation of the traditional feedback controller is that it takes action only after the associated Process Variable is forced away from Set Point. Some degree of impact on the process can be expected as the PID controller’s response is always late due to the effects of Dead-Time. If disturbances consistently stem from the same upstream process, however, then Feed-Forward is an optional mechanism for proactively correcting for those disturbances and cancelling them out.
In essence, the Feed-Forward architecture utilizes a model of the disturbance to determine the timing and size of an appropriate PID controller response. By initiating the response in synch with the disturbance’s arrival, then Feed-Forward effectively neutralizes the upset and avoids any negative impact to control loop performance.
The concept of Feed-Forward is easy to grasp. Even so, there are several aspects that should be taken into consideration, including:
- Instrumentation Needs
In order to maintain awareness of the disturbances, an additional sensor is needed. Without it engineering will remain blind and the process will continue to be battered. It should be installed at the source of the upsets and capable of accurately measuring the size of each disturbance.
- Dueling Models
Modeling the process is a known necessity of PID, but modeling the upstream disturbance is an added requirement of Feed-Forward. Both models are incorporated into the Feed-Forward Element – a control architecture contained within the associated PLC or DCS. With an accurate model of the disturbance the Feed-Forward element can prepare an appropriate response.
- Dead-Time Details
It may seem obvious but the Dead-Time of the process must be shorter than the Dead-Time of the disturbance. If it’s not, then the disturbance will impact the process before the control loop’s response can be initiated. In such a situation Feed-Forward control offers no value.
- Options Galore
Implementing Feed-Forward can range from easy to difficult. The simplest way is to apply a Static or Biasing approach which limits the disturbance model to a value for Gain. No time components are involved. On the other end of the spectrum is a Fully Dynamic approach. This approach requires values for all FOPDT model components. Significant time is typically required with only marginal improvement to performance. In the end the choice should be driven by the needs of the process and the plant.
Feed-Forward control is an advanced approach that should be in every process engineer’s bag of tricks. Like cascade, Feed-Forward provides enhanced disturbance rejection capability which is essential to maintaining control of some highly dynamic, interacting processes. If you’re interested in learning more about Feed-Forward, consider enrolling in one of the many commercial workshops that cover process control.