(72c) 7 Ways to Monitor Controller Variability – Which One Is Right for You?

Control systems exist, primarily, to respond to upsets and maintain the process at its operating point. In other words, to stabilize the plant and minimize variability. Millions of dollars are invested in instruments, controls, pumps, motors, and valves to accomplish this goal. The control system typically represents as much as 10% of the full cost of a new process plant.

With such a big investment, you might like to know if the control system is doing its job. It would seem a simple thing to monitor variability and answer that question. One could simply calculate the standard deviation of each variable every hour and determine if the control system is doing its job. But it turns out that such a simple solution is of limited value.

This paper presents seven methods to analyze controller variability. Each method provides insights into one or more aspects of control performance. The seven methods presented are:

• Basic statistics
• Scaled statistics
• Normalized measures
• Instrument performance
• Controller performance metrics
• Steady-State Analytics
• Signal Pattern Analysis

Attendees will learn how each different method works, how to apply these methods to controllers in their plants, and how to interpret the results from each method. Practical examples from industry are used to demonstrate the application of these methods using sampled date from process plant operations.