Stop Light SPC

The Lights are On But Nobody is Home
There are many different process control methods and procedures available to the practitioner. A popular visual technique employs the stop light analogy:

Green = The process performance is stable and acceptable
Yellow = Warning, the process may be having performance stability problems
Red = The process has become unstable and the performance is unacceptable

The traffic light process control system works as follows [Prevette, 2004]:

Specifications for acceptability are established by the responsible authority
A dashboard is established with the key process input and/or output variables
Every variable has a traffic light
If the variable is in the acceptable range, then the light will be green. If the process moves into the warning zone a yellow light appears, and if the process moves into the unacceptable range, then a red light appears.

This system is simple, easy to understand, and is significantly better than just having black and white numeric data displayed in a huge table. However, this control procedure has some serious problems that will be detailed below:

The light changing specifications are generally determined by what management and/or engineering would like to see rather than what in fact is happening in the process. Hence, the light colors will usually bare little resemblance to statistically determined process control limits. As a result in many cases the lights will be stuck on red when the process is actually in control. This creates a serious problem because action taken on a stable system will tend to destabilize it. Management may think this is what they want but frequently they get a nasty surprise.
Further, if we have a large number of variables and if many lights are red, then we will have to allocate our investigative resources. However, because there is no relationship between the light color and process stability our allocation of resources may be extremely misguided.
On the other hand, if we have a large number of variables and if many lights are green, then there may well be a sense of complacence and disregard of serious process control issues.
This is a very weak control procedure because it relies on only the most recent observation to determine the state of the process. The process could be drifting slowly downward but still be green. A statistical control procedure would detect this drift and give the practitioner and early warning but the traffic light systems allows the process to drift until it finally hits a warning specification.

The traffic light system needs to be modified to reflect the actual state of statistical control so that appropriate action can be taken when and only when it is justified. Fortunately, there are powerful control algorithms such as EWMA or Adaptive Control Charts [Flaig, 1991] that can resolve all these issues and provide an excellent control system. In addition to a good control tool the practitioner needs a sound procedure for switching colors and techniques to achieve improvement [Flaig,2002]. A reasonable methodology is given in Table 1 below.

Table 1: Overview of the Possible X-Chart Outcomes

Reference:

Flaig, J. J. (1991). Adaptive Control Charts, this 1988 paper is contained in the book, “Statistical Process Control in Manufacturing”, Keats, J. B. and Montgomery, D. C., Marcel Dekker.
Flaig, J. J. (2002). Using Factor Effects Analysis to Improve SPC, Quality Engineering, Marcel Dekker, Vol. 14, No. 4, 2002
Prevette, S. (2004). Stoplight Charts with SPC inside, Quality Progress, ASQ Milwaukee, WI. Vol 37, No. 10.

Humorous Commentary
Upon reading this paper a friend sent me the following:

I have a little humor to add to your excellent paper. We must differentiate between a Traffic Light Chart and a Christmas Tree
Chart. A traffic light chart is intended to bring action and should be based on some type of statistical methodology to be effective. Your
feedback is valid for a Traffic Light Chart.

What you have in our report is what I refer to a Christmas Tree chart which is based on the concept of "what you hope Santa will bring you for
Christmas". Statistics have little relevance to the goal setting process that drives this chart because we believe in Santa.