Databases Reference
In-Depth Information
10.2
A QUICK WORD ON SIX SIGMA
Six Sigma is a quality management methodology. It allows only 3.4 defects per million opportunities
per process. In Six Sigma, a defect is defined as anything that results in customer dissatisfaction.
Six Sigma relies heavily on statistical process control (SPC) methods to measure and reduce
defects and increase quality. The central theme of Six Sigma is that if the number of defects present
in a process can be measured, the defects can systematically be eliminated and the process can (in
theory) approach a defect rate of zero.
Obviously, in any process there will be a value for production that is desired. In this example,
the resistance of each element in the Wheatstone bridge should be 1,000
. All processes, of course,
have some degree of variation. Figure 10.5 shows the Normal distribution, which illustrates the
likelihood of how close a produced part will be to its desired production value. Notice that, in a
normally distributed process, 68% of the values lie within one Standard Deviation of the mean. A
standard deviation is a computed measure of variability indicating the spread of the data set around
the mean, whereas deviation is the difference or distance of an individual observation or data value
from the center point (often the mean) in the set distribution.
Notice that two standard deviations from the mean captures 95% of all given occurrences for
a process, whereas three Standard Deviations from the mean are nearly totally inclusive at 99.7%.
This gives rise to the data in Table 10.2, which are defects per million opportunities (DPMO).
A Six Sigma process by definition encompasses all occurrences within
Ω
3 Standard Deviations of
the mean. To meet Six Sigma specifications, a process must have no more than 3.4 defects per
million products.
The
±
3 Standard Deviations from the mean has special significance in Six Sigma. These values are
called the control limits of the given process. The +3 Standard Deviations from the mean are
referred to as the UCL or Upper Control Limit, whereas -3 Standard Deviations from the mean
are referred to as the LCL or Lower Control Limit. Very loosely, when a product occurrence falls
between UCL and LCL, it is said to be in control. (In some instances this may not be true, if a
trend is present, etc.)
Therefore, if a process produced 1 million units, and only 3 of the units were defective (above
the UCL or below the LCL), such a process would be referred to as “in control.” When a process
is “in control,” it should not be tampered with. Tampering is any action taken to compensate for
variation within the control limits of a
±
. Tampering will increase rather than decrease
process variation. A stable system is one in which special-cause variation has been removed and
stable system
2.5%
-
-3
S
-2
S
-
S
X
+
S
+3
S
+3
S
68%
95%
99.7%
FIGURE 10.5
Normal distribution — occurrences from mean.
