PID Tuning Fundamentals: Collecting Good Data

PostedThursday, October 4,2018 at 1:19 PM

Daniel Krupa
Daniel Krupa
PID Tuning Fundamentals: Collecting Good Data

At one time or another we’ve all heard the old adage: Garbage in, Garbage out. While it’s an old saying, it’s a good one and it’s highly relevant to PID controller tuning and process optimization. Whether you’re tuning a control loop manually or with the help of software, the quality of the data that’s used in analysis and modeling directly affects the calculation of tuning parameters. Those parameters dictate a PID’s ability to perform its job and to provide safe, efficient and profitable control.

The data collected from a bump test is what practitioners use to model a process’ dynamic behavior and subsequently to calculate PID tuning parameters. While relatively straightforward there are several considerations that are important to keep in mind when collecting that data. Here are a few that are worth a quick mention:

  • Disturbances: It’s important to limit your data to the effects of the step test. That’s what will assure a clear understanding of the cause-effect relationship existing between the Controller Output and Process Variable. Any disturbances that occur during testing will simply garble the data and hamper subsequent model and tuning parameter calculations.
  • Noise: Noise is a factor that cannot be escaped in virtually any industrial production environment and, as a result, noise can be found in the data of every process. Since that is an unavoidable reality it’s essential that the bump test take noise into consideration.
  • Testing: There are multiple bump test options ranging from the simple step and full bump to the comprehensive doublet and even the PRBS. Whichever test has applied the change to the Controller Output must result in a distinct Process Variable response. 
  • Range: Control loops are generally tuned for a specific range of operation otherwise known as the Design Level of Operation (DLO). The DLO corresponds with the range of normal operation within which the control loop spends most of its time.

With an understanding of what constitutes “good data,” the next task is to determine what type of bump test to perform. I’d previously shared a view that good, effective control should be SIMPLE. The next post in this series on PID Tuning Fundamentals will review the bump test options that are widely – and effectively – used in industry.

Dennis Nash is President and CEO of Control Station, Inc., an award-winning supplier of process diagnostic and optimization solutions. For more information about Control Station, visit the company’s profile here on the Industrial Automation Exchange.

Filed Under Process Manufacturing Process/Batch DCS Training and education Process engineering