Industry surveys have shown that up to 30% of loops are not optimally tuned. Tuning these loops used to be time consuming and a burden for operators who were asked to participate. It is now different since software can optimize loops using normal operation data. Analyses are carried on many loops simultaneously, and operators are no longer required to disturb the process. As a result, the tuning effort is reduced by 80 to 90%.

History

In the 70's, most practitioners used trial and error methods.

In the 80's, software using bump tests appeared on the market. At the same time, controllers with autotune functions became available. However, the performances were rarely acceptable and the methodology consisted of making the loop cycle using steps on setpoint or output. These methods can only tune loops that have no defaults, which is rare.

In the 90's, software allowed to tune loops while analyzing defaults. This really improved the process optimization task. With these tools, one can optimize and tune many loops per day.

All these methods imply that:

• The process is bumped
• Operators collaborate
• Each loop is analyzed individually

Comments from users:

• Long and time consuming
• Control specialists and operators spend long periods of time
• Tools only apply standard formulas from textbooks
• Results vary depending on who does the tests
• When there are problems (stiction, dead band, looseness, non linearity, etc.) tuning does not work or becomes more artwork than science
• Not easy to stabilize the process

New

Top Control's specialists have been using all these techniques for many years, but they now use TaiJi-PID, which overcomes all the difficulties mentioned above. TaiJi-PID uses modern modeling techniques developed in the MPC (multivariable process control) world.

Matrikon has been using this approach for 10 years for MPC, and has now adapted these techniques for PID loops.

What is different?

• TaiJi-PID does not tune loops individually; it tunes entire areas.
• No need to stabilize the process first.
• TaiJi-PID analyzes the process, evaluates interactions, cascade and feedforward strategies, and identifies defaults (stiction, non linearities, etc.).
• TaiJi-PID uses normal operation data or archived data. Small excitation is added to setpoints (narrow band) or controller output, but without disturbing the process and the operator.
• All tests are automated; results do not depend of the user.
• All other methods can also be used: bump tests, setpoint changes, etc.

Contact us: our experts have applied this approach successfully and are ready to apply the same techniques at your plant and to train your personnel!