RLSTuner Example

After a week of trial-and-error tuning on a pressure controller, including usage of another tuning package, the controller still was not very stable. The pressure was manually step tested over a 1-hour period. The 5-second data was copied from a trend and pasted into excel for evaluation by the RLSTuner package. The results are shown below:

The control Aggression factor (AF) of 0.21 was chosen (mid-range – slightly aggressive) and PID control was chosen. These tuning constants were implemented. The operator did not like the noisy output, so tuning was changed to the 0.21 P+I tuning from the table. The tuning of AF = 0.21 has been used for the many months.
For most PID controllers, full analysis is not needed.

---

Curious about why this loop was difficult to tune, Full analysis and simulation were run on the loop.

The Laplace model was identified as a first order plus deadtime model, which means the loop should not be difficult to tune. When questioned the site engineers assumed a shorter settling time and had been using much faster (lower) integral times, and higher gains.
The simulation results for a simulation of a 5% SP change followed by a 0.25% load change, using tuning for AF=0.62 and no process model mismatch follows:

Simulations can be run with process model mismatch, output and input constraints, and other SP and load changes.

RLSTuner excels in tuning of processes with high order dynamics, inverse responses, and complex dynamics. Examples of results from these process dynamics are available upon request.