In this topic the Manual Tuning method and the Ziegler-Nichols tuning method will be explained, since they are the most used and robust solutions to the tuning problem.
Starting from the standard PI parameters, the responsiveness of the system can be maximized by increasing the P gain. The P gain can be doubled until the system starts to oscillate, and only at this point cut by half.
To reduce the duration of the transient, the I gain can be increased as well. Again, it can be doubled as many times as needed until the system becomes unstable, and then it can be halved.
As a last step, minor adjustments can be done on the two gains to reduce the overshoot (in case it is critical for the process outcome).
The Ziegler-Nichols method is based on the experiments and statistics performed by its inventors.
Based on their experience, a set of stable control loop parameters can be calculated knowing the system response to a marginally stable system.
This method is composed of different steps:
This tuning method usually provides a very aggressive response, with an overshoot. If the overshoot is not acceptable, the Kp should be reduced.