January 2002

Vol. 4, No. 1


Pre-control analysis:
Stirring the pot of controversy

Statisticians disagree about the integrity of what is known as pre-control analysis. At best, it should be accompanied by a clear understanding of its limitations. At worst, they may lead to errors in judgment and analysis of charts. Much of the disagreement about this analysis revolves around understanding the objectives and assumptions for its use, contrasted with the objectives and assumptions that underlie the use of control charts.

What is known as 'pre-control ' involves creating artificial 'control limits ''which are not really control limits at all since they are based on specification limits'to facilitate mass application to any process subject to statistical process capability and requires preventive maintenance to control variability, according to statistician Gordon K. Constable, Ph.D.

Constable emphasizes that pre-control is not a tool for quality improvement, but rather a way to keep a machine operator from producing scrap and rework. The approach begins with the assumption that the process is centered within the middle half of the specification limits; then it tries to detect shifts that might result in parts that fall outside the spec limits. Pre-control tells the operator four things:

1. When to adjust the process and when the process can be run
(after 5 consecutive pieces pass);

2. When spot checking is safe;

3. When the process needs attention;

4. When to leave the process alone.

Thus, what is known as 'pre-control ' has a very different objective from that of control charts. Control charts are used to keep a process operating to its capability with minimum variation; pre-control charts are used to keep the process producing units within specification with no concern about variation or process control as long as the parts continue to be within specifications.

To set up a pre-control chart, there are no plotting or computations necessary that are based on process data. Instead, pre-control uses a division of specification limits to determine whether action should be taken to reduce the potential for producing defective product.

Pre-control makes no assumptions about the capability of the process, but it requires that five consecutive units fall in the middle half of the specification before sampling can begin. This essentially qualifies the process as 'runnable ' without collecting a large number of data points. This is particularly helpful in situations where small lots (25-100 pieces) are produced. In cases where the process is not a capable one, getting five pieces in a row in the middle half of the specification may never happen. Here it essentially becomes 100 percent inspection. The operating rules are simple and easy to administrate.

However, particularly in the case where go/no go gauging is used, a great deal of information about the process is lost. There is no real way to tell where the process is in control, which means you cannot tell where it is centered and cannot make an accurate measurement of the spread of the process. This lack of information about the process complicates understanding of how the process characteristics interrelate to affect the output of the process. The implementation of a never-ending improvement process would be impeded by the lack of this information as well.

It would appear that the primary areas of the application of pre-control analysis are in the initial application of a quality effort in situations where the processes are more than capable and for small lot production. In the long run, control charts are preferable in understanding the process and implementing continuous improvement activities in that process.

 


 

Copyright 2002 PQ Systems.

Please direct questions or problems regarding this web site to the Webmaster.