The Step Test Can Be Useful When Tuning PID Loops But Be Aware – It Has Its Drawbacks Too!
Face it, more often than not tuning a PID controller is usually easier when software is applied. While not all products are created equal, the modeling capabilities within most commercial controller tuning software products can account for dynamics that are often overlooked by the human eye. Noisy and oscillatory data can thwart even the most experienced among us. So what should a practitioner do when the use of tuning software isn’t an option? The answer: Perform a step test.
The step test is an excellent option for both exposing a process’ dynamics and calculating the associated model parameters. In terms of a procedure the step test involves an adjustment to the Controller Output (CO) in a single direction while keeping disturbances calm. The Process Variable (PV) must be given sufficient time to transition from one steady state to another. A clear PV response to the adjusted CO value is needed, making graphical analysis of the CO-PV relationship possible. Common industry best-practices can then be applied to calculate values for the process Gain, Time Constant, and Dead-Time.
Although it has been a viable method used extensively in industry, the following aspects of the step test are worth considering:
- Tick, Tock
The step test can require patience. The process must start from a steady state and have time to settle at its new PV value. That implies two things. First, holding a control loop steady both before and during the test may have negative effects on other loops. Second, the process will be away from its normal operating value for an extended period of time which could result in off-spec production, safety concerns, and other potential issues. Exhibiting patience may not be a major concern when a fast responding flow loop is involved, but a slow responding temperature loop could induce a level of anxiety.
- Partial Picture
Since the CO is moved in only one direction (either up or down) the step test can only reveal half of the process’ dynamics. That can have consequences with processes where the dynamics vary. One such example is temperature control. The dynamics of heating versus cooling in certain processes can vary significantly. Exposing the full range of dynamics within the design level of operation is key to effective, efficient control. Seeing only part of the picture can be counterproductive if the objective is optimization.
- The Unforgiven
If a process’ dynamics are calculated using graphical analysis techniques, then the size of the step matters. Stepping a process adequately in order to expose the CO-PV relationship can be challenging if tolerances are tight. Even so, it’s essential to step the process sufficiently so the associated dynamics are fully exposed. Additionally, the process Gain calculation must be spot on. An inaccurate value for Gain can be unforgiving. It can cause a process to perform sluggishly on one extreme and to become unstable on the other.
If the use of software is not an option, then the step test has proven time and again to be a reliable means for modeling process dynamics and calculating new tuning parameters. For sure the step test has some potential drawbacks as noted above. But if there are no constraints on the time needed to perform the test and the resulting model adequately reflects the process’ full design level of operation, then the step test may be just fine.
