Over Control and Natural Variability

A common method for determining process stability, performance, and control is Statistical Process Control. This is a statistical method for determining the upper and lower control limits, and how much of the time the process is within these limits. From this it can be determined whether the process is within control or out-of-control.

Control charts are a visual representation of the process and the upper and lower limits. They show the difference between natural variation and one-off ‘special cause’ events.

Frequently processes fall into the out-of-control areas, and can often be caused by over control. This can be due a number of reasons, such as:

  • Trying to achieve a tighter target range than is possible
  • Control system reaction speed

“Over-control is defined as reacting to random ‘noise’ in the data, causing adjustments to be made when a better plan was to leave the process alone.”

John McConnell

If a business is trying to target a tighter target range than is possible within the natural variation, then this can cause operations to make process changes when they are not required. This creates a ‘saw tooth’ effect as the data swings from one extreme to another. Changing the control strategies so that only special cause events are reacted to can immedietly reduce these extreme swings and improve a chemical processing plant’s stability.

Every process has a natural variation associated with it, such as:

  • Temperature variation
  • Liquid Surge
  • Shift Changes
  • Equipment offline for maintenance
  • Equipment breakdown
  • Raw material grade changes
  • Control system logic

In many cases reducing this natural variation can result in significant business improvements.

There are uncountable methods for reducing process variation and overcontrol, but some of the main ideas can be merged into the following examples:

  • Reduce control system reaction rate through tuning
  • Avoid reacting to a single data point
  • Implement more advanced control systems
  • Align specifications with natural variability

If you have a process which you suspect is a victim of overcontrol, then first step is to prove this through the use of control charts. The second point is to remove the variation by stabilizing the system. I have seen this process work extremely well, and result in significant improvements in performance and cost reduction.

Some great examples of process variability and ways to reduce it can be found in MacKay and Steiner’s Strategies for Variability Reduction.

If you have any experience in reducing the impact of process variability and overcontrol, then please let us know.

No comments