Open-Loop or Closed-Loop Control
In which cases should we use open-loop control? And when closed-loop control? There are a few criteria on the basis of which this decision can be made:
Availability of the controlled variable: if the controlled variable can be measured, we generally use closed-loop control.
Acceptance of deviations: in systems where it is accepted that the controlled variable does not exactly match the setpoint, we can use open-loop control. It is then not so bad if a disturbance causes a control error. If a watch is 10 seconds off, that is socially acceptable. If it is too far off, the user has to turn a crown briefly, and then it runs correctly again for a long time. The effort for this is small.
If you set a cruise control to 80 km/h and the car then drives at 95 km/h, you probably will not find that very funny.
Complexity: if the complexity of the system is high, controllers (closed-loop) are often used. Especially when the calculation of the actuator drive signal from measured disturbances is complex.
Intensity of the disturbances: only systems with few weak disturbances that change only slightly can be managed well with purely open-loop control.
Example: open-loop or closed-loop
I show the difference between closed-loop and open-loop control of a system using the example of an archer. The archer wants to shoot an arrow into the centre of a target. The controlled variable in this case is the point where the arrow hits. The setpoint is: hit the centre. Disturbances are wind and distance. The system is the bow together with the person who shoots the arrow. The actuator is also the bow. Actuator and system are hard to separate, but that is not a problem for the system description.
In an open-loop system, the archer calculates the two launch angles (left–right and up–down) from distance and wind. The archer measures both disturbances. If the wind changes, the calculation has to be updated constantly. After shooting, the archer does not look at the result, because only the disturbances are measured, not the controlled variable. The archer shoots blind. That is typical of open-loop control. The archer shoots only once and is sure of hitting.
In a closed-loop system, the archer first shoots at the target somehow. The archer measures the controlled variable, i.e. looks at where the arrow hit, determines the control error between “centre” and point of impact and readjusts the launch angles on the basis of this information. The archer keeps shooting until, after a few attempts, the centre is hit. It does not matter whether wind or distance influenced the trajectory. If the wind changes, the archer does not notice it directly, because no disturbances are measured. The archer only notices that, for some reason, the shot went to the left, and aims a little further to the right with the next arrow.
Experience shows that the archer does not hit with the open-loop approach. If a different bow is used or the archer is replaced, everything has to be recalculated. With the closed-loop approach, it is certain that the archer will hit after a few arrows. Whatever happens. Once again, closed-loop control is superior to open-loop control. In practice, this is the case very often.