Let’s face it. We can’t always perform open loop step tests
on control loops. Even if we can, frequent load changes can make step tests impractical. Sometimes we must tune control loops by trial and error. In such cases, knowledge and experience are keys to success. The more experience you have tuning control loops using a recognized method, the better you will be at recognizing poorly tuned loops and tuning them by trial and error. With experience, you can inspect a control loop and instinctively determine if its tuning parameters are reasonable. Furthermore, if you must tune a control loop by trial and error, you will know in which direction to begin moving the tuning parameters. Gaining experience can be difficult, however, and you probably should not practice on production control loops. Consider using my PID Loop Simulator and PID Loop Analyzer to hone your loop tuning skills. Start by studying First Order Plus Dead Time (FOPDT) non-integrating processes, also known as self regulating processes (flow, for example), and then move on to integrating processes (level, for example). Simulate and tune several non-integrating processes, ranging from relatively small process gains to large ones and short process time constants to long ones. Stick with a consistent Lambda value (closed loop time constant), perhaps three times the open loop time constant. Document each one’s PID parameters and compare your results. Consider performing the same exercise utilizing the Ultimate method. You can find an example in this document. Then do the same thing with integrating processes. Finally, compare the non-integrating PID parameters to the integrating ones.
View this brief example to help you get started.