Choosing between open- and closed-loop control

Closing the control loop provides insight into the behavior of the process.
By Vance VanDoren, PhD, PE July 31, 2017

Brew kettle example illustrating the need for closed-loop control. Courtesy: Control EngineeringClosed-loop control is an automation technique in which a computing device measures some condition of interest, decides if the measurement falls within an acceptable range, applies a corrective effort if it doesn’t, then repeats the measure-decide-actuate loop until it does. Closed-loop controllers often are used with heaters, pumps, valves, and the like to maintain the temperatures, pressures, levels, and flowrates of industrial processes at the values required to produce the highest quality product at the lowest possible cost.

An open-loop controller also attempts to influence a process but without verifying the results of its efforts. Any mechanism capable of turning a device on or off can be considered an open-loop controller, as can a closed-loop controller that has been disconnected from its sensor.

The obvious advantage of closed-loop control is the certainty of knowing how the process has reacted to the controller’s efforts. On the other hand, if there’s no doubt about what happens when a particular switch is flipped, then there’s no need to "close the loop" with a measurement.

Brew kettle example

For example, an automated brewing process requires stirring, so at some point, the controller will have to start the agitators. Barring a malfunction, the controller reasonably can assume that the agitators will, in fact, start when commanded, so it need not measure anything to determine the success of its efforts. Open-loop control is sufficient.

But during the cooking stage, the controller must measure and re-measure the temperature of the brew to ensure it remains within the optimal range. Variations in the ambient temperature and the temperature of the incoming ingredients often will cause variations in the kettle temperature that are too large to ignore. The controller cannot simply turn on the burners and assume that the cooking will proceed at the desired temperature. Closed-loop control is required.

Not as easy as it looks

Unfortunately, closed-loop control comes at a cost. In addition to the added expense of whatever sensor the controller needs to measure the condition of interest, the controller itself must be equipped with an algorithm capable of executing the measure-decide-actuate loop over and over again so as to drive the condition of interest toward the desired value and no farther. The classic proportional-integral-derivative (PID) algorithm is by far the most popular choice for industrial controllers, but it is by no means foolproof.

A PID controller must be "tuned" to match its corrective efforts to the behavior of the process. If the controller is too aggressive or the controlled process is too sensitive, the controller may end up driving the condition of interest past the desired value, perhaps to the point of creating an even greater disparity in the opposite direction. The controller must then reverse course and may end up making matters even worse if it overcorrects its initial overcorrection.

So, what can an operator do to rein in a runaway closed-loop controller? Open the loop.

Vance VanDoren, PhD, PE is a Control Engineering contributing content specialist. Reach him at controleng@msn.com. Edited by Carly Marchal, content specialist, CFE Media, cmarchal@cfemedia.com.

More advice

Key concepts

  • Closed-loop control is an automation technique in which a computing device measures some condition of interest, decides if the measurement falls within an acceptable range, applies a corrective effort if it doesn’t, then repeats the measure-decide-actuate loop until it does.
  • An open-loop controller also attempts to influence a process but without verifying the results of its efforts.
  • The classic proportional-integral-derivative (PID) algorithm is by far the most popular choice for industrial controllers, but it is by no means foolproof. 

Consider this How many closed-loop controllers in your plant are running open-loop that shouldn’t be?