Using the Oscillation Metric to Uncover Tuning Issues

The Oscillation Metric Can Identify a PID Controller that Requires Tuning, But More Often Than Not it Provides an Indication of Loop Interaction or Mechanical Issues.

There’s a wealth of information available in most every data historian. The data can be used to evaluate the performance of a plant’s regulatory control systems in general and to uncover PID controllers that require tuning in particular. Capitalizing on that resource can help manufacturers keep their processes within designated constraints and avoid out-of-spec production.

Various key performance indicators (KPIs) help manufacturers to pinpoint under-performing PID control loops. Those KPIs identify characteristics in the data that correlate with poor tuning among other negative performance issues. Oscillatory behavior is one such characteristic that is common in industrial applications and regularly attributed to poor tuning.

The Oscillation Metric is a standard KPI used in control loop performance monitoring (CLPM) solutions and suitable as a secondary metric for identifying tuning issues. Indeed, the Oscillation Metric is a less obvious indicator of poor tuning and it generally requires corroboration from other KPIs. Consider the following:

• Rule Out Instability
  

More often than not oscillatory behavior is indicative of instability. Behavior that results in a high Oscillatory Metric value is often the result of mechanical issues. A final control element (FCE) that’s oversized or that has stiction can drive persistent and variable process behavior.

Corroboration of poor controller tuning can be found in other metrics such as Output Reversals, Output Travel, and Stiction. These other metrics focus on the FCE’s performance. If values are low for these KPIs, then a high Oscillation Metric value does suggest that tuning of the controller is an appropriate corrective action.

• Investigate Interaction
  

Behavior originating with upstream processes can filter downstream and cause other loops to oscillate in response. Most production facilities operate a tangled web of control loops that interact with each other.

Advanced forensic tools such as Power Spectrum can be used for corroboration purposes and to isolate the associated root-cause. While not a traditional KPI Power Spectrum can be used to find other loops that share the oscillation’s frequency.

• Assess Settling Time
  

Long-lasting oscillations that taper off are symptomatic of poor tuning and will trigger a high Oscillation Metric value. Such variability can be the result of a load change, an adjustment to Set Point, or other process disturbance. If not a mechanical issue, then the oscillations are most likely due to a controller that’s tuned too aggressively and that stretches out the loop’s Settling Time.

A high Oscillation Metric value can be corroborated with a similarly high value for Absolute Average Error (AAE). AAE measures a given loop’s ability to track Set Point. A loop can be deemed as being in need of tuning if these two metric values are high. Viewing the process data is an even more basic method of confirming that the controller requires tuning. If the Process Variable can be seen to taper off over time, then tuning is the correct course of action.

It’s common for practitioners to associate oscillatory loop behavior with poor controller tuning. While there is a link between a high Oscillation Metric value and poor tuning, more often than not this form of variability is tied to loop interaction and mechanical issues. The next several posts will focus on KPIs that identify the need for tuning more clearly.