Berg, Douglas L. (1989, ASQC) Hydra-matic Div., General Motors Corp., Ypsilanti, MI
Experienced practitioners usually have a long list of failed SPC applications. Time and again they see poorly conducted studies, inappropriate control methods, statistically unsound approaches, and end user resistance. A basic cause of these problems, which are actually symptoms, is a poor understanding of the process studied or controlled. This paper proposes a simple, common sense approach based on defining the process as it exists, a step often overlooked or bypassed in many applications. While not an exhaustive treatment of all application problems, the "back to basics" emphasis is clear through specific examples.
A different look is taken of flow or process diagrams to identify sources of variation. General application problems are presented with several illustrations.
Practitioners also encounter some poorly conceived "remedial" measures applied to overcome problems in application. Often these are unsound "statistimagic" based on assumptions rarely check, and which merely address the symptoms and not the root problems.
The solution is the straight forward, common sense application of the basics to confront the current process. Effective control reflects process dominance, encompasses the sources of variation, addresses variation sources explicitly, eliminates them where possible, and deals with the system as it is. Illustrative examples are shown.
Automobile industry,Statistical process control (SPC)