Sample Rates May Vary from Loop to Loop …The Importance of Good Data Doesn’t!
Practitioners often apply simple guidelines when it comes to data collection requirements and controller tuning. These “rules of thumb” assure that sufficient data resolution exists when a given PID control loop’s dynamics are being analyzed. Without good data a process engineer’s ability to model the dynamics and tune for improved control can be undermined. If the sample rate is too slow, then most any tuning procedure will be hit-or-miss at best. So what exactly are those guidelines?
Approaches range from simplistic “rules of thumb” to scientific methods (e.g. Nyquist-Shannon sampling theorem). On the easier end of the spectrum its commonplace to stipulate sample rates of 1 second or faster for Flow loops and 1-2 seconds for Pressure. Such loops react very quickly to changes in Controller Output. Not far behind are rates of 2-5 seconds for many Temperature and Level loops which characteristically possess slower dynamics.
When considering sample rate from a scientific or mathematical perspective, the guidelines are a bit more involved and beyond the scope of this blog post. Still, it’s generally accepted best-practice for the sample rate to be based on frequency or time-based components of the process data. Basic formulas propose the use of data that is either 5-10 times faster than the Process Time Constant or 3-5 times faster than the Process Dead-Time.
Beyond these approaches it is worthwhile to keep other considerations in mind:
Aliasing is like an illusion. It occurs when the sample rate is insufficient to provide an accurate depiction of a process’ dynamics. When this type of data is collected and plotted it can depict dynamics that are altogether inaccurate and misleading. Controller tuning parameters generated using this type of data can be counterproductive and even undermine safety.
Compression is an approach to data storage that helps manufacturers keep their IT overhead under control. One approach is to eliminate data according to a schedule. Another focused on change involves the elimination of data that remains constant. Regardless of the technique applied, compression can result in aliasing by limiting the resolution of data needed for process modeling and PID controller tuning.
In order to provide good results, controller tuning software requires “good data”. So it’s no surprise that software’s value decreases steadily as the sample rate slows. Faced with poor data software can model for the Process Gain, but it loses the ability to provide meaningful values for either the Process Time Constant or the Process Dead-Time. In such instances old-fashioned trial-and-error testing is generally required to establish functional values for the associated controller tuning coefficients.
Guidelines for applying an effective sample rate are equally important to PID control loop tuning as they are to monitoring plant-wide control loop performance. If the data lacks sufficient resolution, then practitioners are in essence flying blind and limited in their ability to maintain safe and profitable operations. Best-practices for data collection and analysis are covered in the curriculum of most courses on process control.
Share this post:
Our most popular posts:
Expand your knowledge with this featured publication:
Avoiding the Traps: A Plan for Successfully Evaluating New Manufacturing Technologies
Innovation can be a manufacturer’s worst nightmare especially when it comes to conducting formal evaluations. From conflicting priorities and disagreements...Learn more