102
PL12EN-1300 2008-03
en04d141.fm, 11.03.2008
Statistics
4
Special cause variation may be indicated by one of the following situations:
• A point outside a control limit.
• A shift, with several consecutive points inside a control limit but on the same side of the center
line.
• A trend of consecutive points that moves steadily toward a control limit.
Outside of Limit Shift Trend
Fig. 4-6: Indications of Special Cause Variation
Since we are controlling by evaluating only small sample sizes from the process, the criteria for
continuing production must be the control limits. For this reason, we do not place the specifica-
tion limits (which are recorded on the data sheet) on the control chart.
4.1.4
System Improvement
There are always two risks associated with statistical control. The first is the risk of accepting bad
product, and the second is the risk of rejecting good product. Both are undesirable, but our pri-
mary concern is usually the first since it has a direct impact on customer satisfaction.
The six sigma spread of the process described in this procedure is in general use throughout
American industry. When properly applied, it assures us that at least 99.7% of all units produced
will meet specification. This relates to a risk of delivering three substandard units for each thou-
sand produced. While this provides much better quality than can be obtained with an uncon-
trolled process and 100% inspection, it may not be good enough in the future as competition
becomes stronger in the world market.
So it's important that we continue to improve the manufacturing system - to reduce the influence
of conditions that affect variations in the performance of people, machines, and materials (see
graph 4-7). There will be practical and economic limitations to this effort, but to the extent that we
can further reduce process variation we will increase the probability that all delivered product
meets the specification.
Variation
Variation
Spec. Toler.
Spec. Toler.
Fig. 4-7: Improving The System By Reducing Variation
4.2
Statistic Symbols
•
σ
=standard deviation (or sigma)