PID Tuning Fundamentals: Bump Test Options

PostedTuesday, October 16,2018 at 9:22 AM

Daniel Krupa
Daniel Krupa
PID Tuning Fundamentals: Bump Test Options

The bump test is a critical step in any PID controller tuning method. That’s because the bump test establishes the cause-and-effect relationship between a control loop’s Process Variable (PV) and Controller Output (CO). Without a clear understanding of the PV-CO relationship it’s virtually impossible to generate tuning parameters of any meaningful value. Without an effective bump test the PID tuning process is little more than guesswork.

Bump tests essentially involve a change in CO in order to determine how the PV responds. In fact a bump test should reveal how far, how fast, and with how much delay the PV responds to a given change in CO. That’s true whether your test is performed on a self-regulating (aka non-integrating) process or an integrating process. With an established PV-CO relationship it’s possible to calculate model parameters and to convert those values into tuning parameters. Here are bump test options that are both widely and successfully used in industry:

  • Step Test: A step test is when the CO is “stepped” from one constant value to another. It results in the PV moving from one steady state to a new steady state. That’s it. The step test is limited to moving the CO in one direction. As a result there are limits to the insight that a step test can provide. There may be important information about the process’ dynamic behavior that won’t be factored into either the model parameters or tuning parameters.
  • Pulse Test: A pulse test can be thought of as two (2) step tests performed in rapid succession and in opposite direction. The CO is stepped and, as soon as the PV shows a clear response, the CO is then returned to its original value. This method is more commonly used in industry as it brings the CO back to its original value. While more complete than the step test the pulse test is still somewhat limited as it doesn’t reveal the full range a process’ dynamic behavior.
  • Doublet Test: A doublet test is essentially two (2) pulse tests. Each pulse test is performed in rapid succession and in opposite directions. The second pulse is implemented as soon as the process has shown a clear response to the first pulse. The doublet is highly effective as it fully reveals the dynamics associated with a control loop. While no surprise the test option also takes twice the time to complete.
  • PRBS Test: While a pseudo-random binary sequence (PRBS) test can provide the most insight it is the least used in industry due largely to its complexity. A PRBS can be characterized as a sequence of CO pulses that are uniform in amplitude, alternating in direction, and of random duration. It’s worth noting that the CO changes are typically smaller in size than other test options which can ultimately reveal dynamic behavior without significant upsets.

I’d previously shared a perspective that good control should be SIMPLE and offered ideas on how to collect good data. With results from an effective test using one of the options listed above it’s then possible to calculate model parameters. If calculating model parameters is of interest, then visit my PID Tuning Fundamentals blog again soon.

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.

Related Tags:

  • Process Manufacturing
  • Process Engineering
  • Training and Education
  • Process/Batch
  • DCS