|
This p chart shows the number of warped moldings in samples of
1000 items chosen during each shift, from one machine (#3). The
output of different machines is not mixed because the differences
between machines would mask the variation from any one of them.
The first sample on this chart was taken at 6PM on March 7. Note
that the control limits were calculated on data taken at least
a month old. Presumably there were no major changes in the process
during that time. Indications of lack of control for p charts
are the same as for x-bar R charts. In this case, the process
is in control until March 14, when there is a run of over 6 observations
below the process average. The operator wrote "new lot 3/11 1400"
to indicate a change in feedstock. The new lot of polyethelene
may have caused a reduction in the number of non-conforming items.
The p chart is strong evidence of the degree of reduction and
its timing, especially if the trend continues. New control limits
should be calculated using the next 30 samples. The exact cause
of the reduction can be investigated. |
|
|
